diff --git a/AGENTS.md b/AGENTS.md
index ae23985b4b8..86c7c0a749b 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -45,6 +45,7 @@ Package-specific guides:
 | Build plugin | `cd nemoclaw && npm run build` |
 | Watch mode | `cd nemoclaw && npm run dev` |
 | Run all tests | `npm test` |
+| Render behavior-oriented test tree | `npm run test:spec` |
 | Run fast source tests | `npm run test:fast` |
 | Run integration tests | `npm run test:integration` |
 | Run package contracts | `npm run test:package` |
@@ -87,6 +88,7 @@ When writing tests:
 - Plugin tests use TypeScript and are co-located with their source files
 - Import CLI source from ordinary tests. Put genuine compiled-artifact assertions under `test/package-contract/`.
 - Keep project globs disjoint; `npm run test:projects:check` derives membership from Vitest and rejects overlap.
+- Write behavior-oriented titles, put local issue references in a final `(#1234)` suffix, and use `npm run test:spec` for the hierarchical specification view.
 - Mock external dependencies; don't call real NVIDIA APIs in unit tests
 - E2E tests run on ephemeral Brev cloud instances
 
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 66d15f587f2..1af00b86b46 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -121,6 +121,7 @@ These are the primary `make` and `npm` targets for day-to-day development:
 | `make format` | Auto-format TypeScript and Python source |
 | `npm run typecheck:cli` | Type-check CLI TypeScript using `tsconfig.cli.json` (`bin/`, `scripts/`, `src/`, `test/`, `nemoclaw-blueprint/scripts/`) |
 | `npm test` | Build package artifacts and run every non-live Vitest project |
+| `npm run test:spec` | Run every non-live test with hierarchical behavior-oriented output |
 | `npm run test:fast` | Clean `dist/` and run source CLI, plugin, and E2E-support tests |
 | `npm run test:integration` | Clean-build the CLI and run root integration and installer tests |
 | `npm run test:package` | Clean-build CLI/plugin artifacts and run compiled-package contracts |
@@ -132,6 +133,18 @@ These are the primary `make` and `npm` targets for day-to-day development:
 | `npm run docs:deps` | Print the pinned Fern CLI version used by docs commands |
 | `npx prek run --all-files` | Run all hooks from `.pre-commit-config.yaml` — see below |
 
+### Test Titles as Behavioral Documentation
+
+Write `describe` and `it` titles so the Vitest tree reads as behavioral documentation. Start test
+titles with behavior or context rather than issue numbers, flags, or scenario labels, and put local
+issue references in a final suffix such as `(#1234)`. Prefer
+`it("reticulates splines correctly (#1234)")` over
+`it("#1234 fixes spline reticulation")`.
+
+Run `npm run test:spec` to render the suite with Vitest's hierarchical tree reporter. Run
+`npm run test:titles:check` to enforce the objective title-shape conventions without attempting to
+lint subjective English grammar.
+
 ### Git hooks (prek)
 
 All git hooks are managed by [prek](https://prek.j178.dev/), a fast, single-binary pre-commit hook runner installed as a devDependency (`@j178/prek`). The `npm install` step runs `prek install` automatically via the `prepare` script, which wires up the following hooks from [`.pre-commit-config.yaml`](.pre-commit-config.yaml):
diff --git a/nemoclaw/src/blueprint/ssrf.test.ts b/nemoclaw/src/blueprint/ssrf.test.ts
index 5d52b3dd26c..b1db007870e 100644
--- a/nemoclaw/src/blueprint/ssrf.test.ts
+++ b/nemoclaw/src/blueprint/ssrf.test.ts
@@ -258,7 +258,7 @@ describe("isPrivateIp – CIDR boundary precision", () => {
     ["128.0.0.0", false], // just above 127.0.0.0/8
     ["192.167.255.255", false], // just below 192.168.0.0/16
     ["192.169.0.0", false], // just above 192.168.0.0/16
-  ])("boundary %s → private=%s", (ip, expected) => {
+  ])("classifies boundary address %s as private=%s", (ip, expected) => {
     expect(isPrivateIp(ip)).toBe(expected);
   });
 });
@@ -277,7 +277,7 @@ describe("isPrivateIp – IPv6 edge cases", () => {
     ["fd00::0", true], // first address in fd00::/8 (within fc00::/7)
     ["fdff:ffff:ffff:ffff:ffff:ffff:ffff:ffff", true], // last address in fc00::/7 ULA range
     ["fe00::1", false], // just above fc00::/7 (link-local starts at fe80::)
-  ])("IPv6 %s → private=%s", (ip, expected) => {
+  ])("classifies IPv6 address %s as private=%s", (ip, expected) => {
     expect(isPrivateIp(ip)).toBe(expected);
   });
 
diff --git a/nemoclaw/src/security/secret-scanner.test.ts b/nemoclaw/src/security/secret-scanner.test.ts
index f87eef71333..a962049a448 100644
--- a/nemoclaw/src/security/secret-scanner.test.ts
+++ b/nemoclaw/src/security/secret-scanner.test.ts
@@ -29,132 +29,132 @@ const FAKE = {
 };
 
 describe("scanForSecrets", () => {
-  describe("detects known secret patterns", () => {
-    it("NVIDIA API key", () => {
+  describe("known secret patterns", () => {
+    it("detects an NVIDIA API key", () => {
       const matches = scanForSecrets(`my key is ${FAKE.nvidia}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("NVIDIA API key");
     });
 
-    it("OpenAI API key", () => {
+    it("detects an OpenAI API key", () => {
       const matches = scanForSecrets(`export OPENAI_API_KEY=${FAKE.openai}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("OpenAI API key");
     });
 
-    it("OpenAI project API key", () => {
+    it("detects an OpenAI project API key", () => {
       const matches = scanForSecrets(`export OPENAI_API_KEY=${FAKE.openaiProject}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("OpenAI API key");
     });
 
-    it("GitHub personal access token", () => {
+    it("detects a GitHub personal access token", () => {
       const matches = scanForSecrets(`token: ${FAKE.github}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("GitHub token");
     });
 
-    it("AWS access key", () => {
+    it("detects an AWS access key", () => {
       const matches = scanForSecrets(`aws_access_key_id = ${FAKE.aws}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("AWS access key");
     });
 
-    it("Slack bot token", () => {
+    it("detects a Slack bot token", () => {
       const matches = scanForSecrets(`SLACK_TOKEN=${FAKE.slack}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Slack token");
     });
 
-    it("Slack app token", () => {
+    it("detects a Slack app token", () => {
       const matches = scanForSecrets(`SLACK_APP_TOKEN=${FAKE.slackApp}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Slack token");
     });
 
-    it("npm token", () => {
+    it("detects an npm token", () => {
       const matches = scanForSecrets(`//registry.npmjs.org/:_authToken=${FAKE.npm}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("npm token");
     });
 
-    it("private key (PEM RSA)", () => {
+    it("detects a PEM RSA private key", () => {
       const matches = scanForSecrets(FAKE.pemRsa);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Private key");
     });
 
-    it("private key (OpenSSH)", () => {
+    it("detects an OpenSSH private key", () => {
       const matches = scanForSecrets(FAKE.pemOpenssh);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Private key");
     });
 
-    it("Telegram bot token", () => {
+    it("detects a Telegram bot token", () => {
       const matches = scanForSecrets(`bot token: ${FAKE.telegram}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Telegram bot token");
     });
 
-    it("Google API key", () => {
+    it("detects a Google API key", () => {
       const matches = scanForSecrets(`GOOGLE_API_KEY=${FAKE.google}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Google API key");
     });
 
-    it("Anthropic API key", () => {
+    it("detects an Anthropic API key", () => {
       const matches = scanForSecrets(`key: ${FAKE.anthropic}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Anthropic API key");
     });
 
-    it("HuggingFace token", () => {
+    it("detects a HuggingFace token", () => {
       const matches = scanForSecrets(`HF_TOKEN=${FAKE.huggingface}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("HuggingFace token");
     });
 
-    it("Discord bot token", () => {
+    it("detects a Discord bot token", () => {
       const matches = scanForSecrets(`DISCORD_TOKEN=${FAKE.discord}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Discord bot token");
     });
 
-    it("AWS secret key", () => {
+    it("detects an AWS secret key", () => {
       const matches = scanForSecrets(`aws_secret_access_key = ${FAKE.awsSecret}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("AWS secret key");
     });
 
-    it("Authorization header", () => {
+    it("detects an Authorization header", () => {
       const matches = scanForSecrets(`Authorization: Bearer ${FAKE.authHeader}`);
       expect(matches).toHaveLength(1);
       expect(matches[0].pattern).toBe("Authorization header");
     });
   });
 
-  describe("does not false-positive on safe content", () => {
-    it("normal markdown text", () => {
+  describe("safe content", () => {
+    it("allows normal markdown text", () => {
       expect(scanForSecrets("# My Project\n\nThis is a regular markdown file.")).toHaveLength(0);
     });
 
-    it("code blocks without secrets", () => {
+    it("allows code blocks without secrets", () => {
       expect(scanForSecrets("```python\nprint('hello world')\n```")).toHaveLength(0);
     });
 
-    it("short tokens that don't meet minimum length", () => {
+    it("ignores short tokens below the minimum length", () => {
       expect(scanForSecrets("sk-short")).toHaveLength(0);
     });
 
-    it("URLs with path segments", () => {
+    it("allows URLs with path segments", () => {
       expect(scanForSecrets("https://github.com/NVIDIA/NemoClaw/pull/1121")).toHaveLength(0);
     });
 
-    it("UUIDs", () => {
+    it("allows UUIDs", () => {
       expect(scanForSecrets("id: 550e8400-e29b-41d4-a716-446655440000")).toHaveLength(0);
     });
 
-    it("git commit hashes", () => {
+    it("allows git commit hashes", () => {
       expect(scanForSecrets("commit 24a8b5a3f1e2d3c4b5a6f7e8d9c0b1a2")).toHaveLength(0);
     });
   });
diff --git a/package.json b/package.json
index e2864b9ace0..6861a9f1bf6 100644
--- a/package.json
+++ b/package.json
@@ -21,12 +21,14 @@
   "scripts": {
     "preinstall": "node scripts/check-node-version.js",
     "test": "npm run clean:cli && npm --prefix nemoclaw run clean && npm run build:cli && npm --prefix nemoclaw run build && vitest run --project cli --project integration --project installer-integration --project package-contract --project plugin --project e2e-vitest-support",
+    "test:spec": "npm test -- --reporter=tree",
     "test:fast": "npm run clean:cli && vitest run --project cli --project plugin --project e2e-vitest-support",
     "test:integration": "npm run clean:cli && npm run build:cli && vitest run --project integration --project installer-integration",
     "test:package": "npm run clean:cli && npm --prefix nemoclaw run clean && npm run build:cli && npm --prefix nemoclaw run build && vitest run --project package-contract",
     "test:live-e2e": "NEMOCLAW_RUN_E2E_SCENARIOS=1 vitest run --project e2e-scenarios-live",
     "test:imports:check": "tsx scripts/checks/no-test-dist-imports.ts",
     "test:projects:check": "tsx scripts/checks/vitest-project-overlap.ts",
+    "test:titles:check": "tsx scripts/checks/test-title-style.ts",
     "check": "npx prek run --all-files",
     "checks": "tsx scripts/checks/run.ts",
     "lint": "npx @biomejs/biome lint . && npm run checks",
diff --git a/scripts/checks/run.ts b/scripts/checks/run.ts
index 67f0eea7873..a7272ad31c4 100644
--- a/scripts/checks/run.ts
+++ b/scripts/checks/run.ts
@@ -46,6 +46,11 @@ const CHECKS: readonly CheckCommand[] = [
     command: TSX,
     args: ["scripts/checks/vitest-project-overlap.ts"],
   },
+  {
+    name: "test-title-style",
+    command: TSX,
+    args: ["scripts/checks/test-title-style.ts"],
+  },
 ];
 
 function main(): void {
diff --git a/scripts/checks/test-title-style.ts b/scripts/checks/test-title-style.ts
new file mode 100755
index 00000000000..6bae0dbf086
--- /dev/null
+++ b/scripts/checks/test-title-style.ts
@@ -0,0 +1,187 @@
+#!/usr/bin/env -S npx tsx
+// SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// SPDX-License-Identifier: Apache-2.0
+
+import { existsSync, readdirSync, readFileSync, statSync } from "node:fs";
+import path from "node:path";
+import { fileURLToPath, pathToFileURL } from "node:url";
+
+import ts from "typescript";
+
+export type TestTitleRule =
+  | "issue-reference-suffix"
+  | "leading-metadata"
+  | "placeholder-only"
+  | "result-arrow";
+
+export type TestTitleViolation = {
+  readonly file: string;
+  readonly line: number;
+  readonly column: number;
+  readonly call: "describe" | "it" | "test";
+  readonly title: string;
+  readonly rule: TestTitleRule;
+  readonly message: string;
+};
+
+const REPO_ROOT = path.resolve(path.dirname(fileURLToPath(import.meta.url)), "../..");
+const DEFAULT_SCAN_ROOTS = Object.freeze(["src", "test", "nemoclaw/src"]);
+const TEST_FILE_PATTERN = /\.(?:test|spec)\.(?:[cm]?[jt]sx?)$/;
+const TEST_CALL_NAMES = new Set(["describe", "it", "test"]);
+const SKIP_DIRS = new Set([".git", ".venv", "coverage", "dist", "node_modules"]);
+const LEADING_METADATA_PATTERN =
+  /^(?:#\d+\b|issue\s+#?\d+\b|regression\s+#?\d+\b|--\S+|-\w\b|\[[^\]]+\]|scenario\b)/i;
+const LOCAL_ISSUE_REFERENCE_PATTERN = /(?<![\w/-])#\d+\b/;
+const ISSUE_SUFFIX_PATTERN = /\s\(#\d+(?:\s*(?:,|\/|and)\s*#\d+)*\)$/;
+const PLACEHOLDER_ONLY_PATTERN = /^(?:%[sdifjo]|\$\{…\})$/;
+
+function scriptKindFor(filePath: string): ts.ScriptKind {
+  if (/\.tsx$/i.test(filePath)) return ts.ScriptKind.TSX;
+  if (/\.jsx$/i.test(filePath)) return ts.ScriptKind.JSX;
+  if (/\.[cm]?js$/i.test(filePath)) return ts.ScriptKind.JS;
+  return ts.ScriptKind.TS;
+}
+
+function rootCallName(expression: ts.Expression): string | null {
+  if (ts.isIdentifier(expression)) return expression.text;
+  if (ts.isPropertyAccessExpression(expression)) return rootCallName(expression.expression);
+  if (ts.isCallExpression(expression)) return rootCallName(expression.expression);
+  return null;
+}
+
+function literalTitle(argument: ts.Expression | undefined): string | null {
+  if (argument === undefined) return null;
+  if (ts.isStringLiteral(argument) || ts.isNoSubstitutionTemplateLiteral(argument)) {
+    return argument.text;
+  }
+  if (!ts.isTemplateExpression(argument)) return null;
+
+  return `${argument.head.text}${argument.templateSpans
+    .map((span) => `\${…}${span.literal.text}`)
+    .join("")}`;
+}
+
+function titleRules(title: string): readonly { rule: TestTitleRule; message: string }[] {
+  const trimmed = title.trim();
+  const violations: { rule: TestTitleRule; message: string }[] = [];
+
+  if (LOCAL_ISSUE_REFERENCE_PATTERN.test(trimmed) && !ISSUE_SUFFIX_PATTERN.test(trimmed)) {
+    violations.push({
+      rule: "issue-reference-suffix",
+      message: "move local issue references to a final '(#1234)' suffix",
+    });
+  }
+  if (LEADING_METADATA_PATTERN.test(trimmed)) {
+    violations.push({
+      rule: "leading-metadata",
+      message: "start with behavior or context instead of metadata, flags, or scenario labels",
+    });
+  }
+  if (PLACEHOLDER_ONLY_PATTERN.test(trimmed)) {
+    violations.push({
+      rule: "placeholder-only",
+      message: "add behavior around the parameter placeholder",
+    });
+  }
+  if (/\s→\s/.test(trimmed)) {
+    violations.push({
+      rule: "result-arrow",
+      message: "describe the expected result as a sentence instead of an input-to-output label",
+    });
+  }
+
+  return violations;
+}
+
+export function scanTestTitleStyle(file: string, source: string): readonly TestTitleViolation[] {
+  const sourceFile = ts.createSourceFile(
+    file,
+    source,
+    ts.ScriptTarget.Latest,
+    true,
+    scriptKindFor(file),
+  );
+  const violations: TestTitleViolation[] = [];
+
+  function visit(node: ts.Node): void {
+    if (ts.isCallExpression(node)) {
+      const call = rootCallName(node.expression);
+      const title = literalTitle(node.arguments[0]);
+      if (call !== null && TEST_CALL_NAMES.has(call) && title !== null) {
+        const location = sourceFile.getLineAndCharacterOfPosition(node.getStart(sourceFile));
+        for (const violation of titleRules(title)) {
+          violations.push({
+            file,
+            line: location.line + 1,
+            column: location.character + 1,
+            call: call as TestTitleViolation["call"],
+            title,
+            ...violation,
+          });
+        }
+      }
+    }
+    ts.forEachChild(node, visit);
+  }
+
+  visit(sourceFile);
+  return violations;
+}
+
+function isSkipped(absolutePath: string): boolean {
+  const segments = path.relative(REPO_ROOT, absolutePath).split(path.sep);
+  return segments.some((segment) => SKIP_DIRS.has(segment));
+}
+
+function* walkTestFiles(directory: string): Generator<string> {
+  if (!existsSync(directory) || isSkipped(directory)) return;
+
+  for (const entry of readdirSync(directory)) {
+    const absolutePath = path.join(directory, entry);
+    if (isSkipped(absolutePath)) continue;
+    const stats = statSync(absolutePath);
+    if (stats.isDirectory()) {
+      yield* walkTestFiles(absolutePath);
+    } else if (stats.isFile() && TEST_FILE_PATTERN.test(entry)) {
+      yield absolutePath;
+    }
+  }
+}
+
+export function findTestTitleStyleViolations(
+  roots: readonly string[] = DEFAULT_SCAN_ROOTS,
+): readonly TestTitleViolation[] {
+  const violations: TestTitleViolation[] = [];
+  for (const root of roots) {
+    const absoluteRoot = path.resolve(REPO_ROOT, root);
+    for (const absolutePath of walkTestFiles(absoluteRoot)) {
+      const file = path.relative(REPO_ROOT, absolutePath).split(path.sep).join("/");
+      violations.push(...scanTestTitleStyle(file, readFileSync(absolutePath, "utf8")));
+    }
+  }
+  return violations;
+}
+
+function main(): void {
+  const violations = findTestTitleStyleViolations();
+  if (violations.length === 0) {
+    console.log("Test title style check passed.");
+    return;
+  }
+
+  for (const violation of violations) {
+    console.error(
+      `${violation.file}:${violation.line}:${violation.column} [${violation.rule}] ${violation.message}: ${JSON.stringify(violation.title)}`,
+    );
+  }
+  console.error(`Found ${violations.length} test title style violation(s).`);
+  process.exitCode = 1;
+}
+
+const invokedPath = process.argv[1];
+if (
+  invokedPath !== undefined &&
+  import.meta.url === pathToFileURL(path.resolve(invokedPath)).href
+) {
+  main();
+}
diff --git a/src/lib/actions/sandbox/agent/passthrough.test.ts b/src/lib/actions/sandbox/agent/passthrough.test.ts
index 62e171cda1b..792bf71da96 100644
--- a/src/lib/actions/sandbox/agent/passthrough.test.ts
+++ b/src/lib/actions/sandbox/agent/passthrough.test.ts
@@ -402,7 +402,7 @@ describe("runAgentPassthrough", () => {
     expect(exit).toHaveBeenCalledWith(2);
   });
 
-  it("prints recovery hints with exit 1 before selector rejection when the sandbox phase is non-Ready (covers the literal #5655 stopped-sandbox repro `agent -m ping`)", async () => {
+  it("prints recovery hints with exit 1 before selector rejection for the literal stopped-sandbox repro `agent -m ping` (#5655)", async () => {
     ensureLiveMock.mockResolvedValueOnce({ output: "Phase: Error" });
     getSandboxMock.mockReturnValueOnce({ agent: "openclaw" });
     const { writes, exit, proc } = makeProcMock();
diff --git a/src/lib/actions/sandbox/auto-pair-approval.test.ts b/src/lib/actions/sandbox/auto-pair-approval.test.ts
index 83fa1b1eb1d..21e2c17a4c8 100644
--- a/src/lib/actions/sandbox/auto-pair-approval.test.ts
+++ b/src/lib/actions/sandbox/auto-pair-approval.test.ts
@@ -268,7 +268,7 @@ process.exit(2);
     }
   });
 
-  it("does not recover approval failures without the #4462 compatibility signature", () => {
+  it("does not recover approval failures without the compatibility signature (#4462)", () => {
     if (spawnSync("sh", ["-c", "command -v python3"], { stdio: "ignore" }).status !== 0) {
       return;
     }
diff --git a/src/lib/actions/sandbox/auto-pair-warmup.test.ts b/src/lib/actions/sandbox/auto-pair-warmup.test.ts
index d3d7c8f5384..d0c8e4c2a76 100644
--- a/src/lib/actions/sandbox/auto-pair-warmup.test.ts
+++ b/src/lib/actions/sandbox/auto-pair-warmup.test.ts
@@ -22,7 +22,7 @@ import { WARMUP_SESSION_ID_PREFIX } from "./warmup-session";
 // the OpenShell-exec wrapping the leaf depends on — and the
 // finalization.test.ts ordering tests pin the provoke→approve wiring.
 
-describe("scope-upgrade warm-up timeout bound (#4504-v2)", () => {
+describe("scope-upgrade warm-up timeout bound v2 (#4504)", () => {
   it("uses a fixed 30s outer cap so a wedged warm-up can never block onboard", () => {
     // The `-m "ping"` one-shot returns fast even when it falls back to embedded
     // mode; 30s covers gateway-connect + the scope-upgrade request plus
@@ -42,7 +42,7 @@ describe("scope-upgrade warm-up timeout bound (#4504-v2)", () => {
   });
 });
 
-describe("warm-up payload survives OpenShell exec (#4504-v2)", () => {
+describe("warm-up payload survives OpenShell exec in v2 (#4504)", () => {
   // The leaf wraps its in-sandbox script with the shared `wrapSandboxShellScript`
   // (OpenShell exec rejects newline-bearing args). These cases pin that wrapper
   // contract — the exact mechanism the warm-up exec relies on — without needing
@@ -83,7 +83,7 @@ describe("warm-up tags its throwaway session for user-facing filters (#5511)", (
     expect(WARMUP_SCRIPT).toContain(`--session-id "${WARMUP_SESSION_ID_PREFIX}$$-$(date +%s)"`);
   });
 
-  it("keeps the #4504-v2 provoke run foreground and within the original budget", () => {
+  it("keeps the v2 provoke run foreground and within the original budget (#4504)", () => {
     expect(WARMUP_SCRIPT).toContain('openclaw agent --agent main -m "ping" \\');
     expect(WARMUP_SCRIPT).toContain(">/dev/null 2>&1 || true");
     expect(WARMUP_SCRIPT).not.toContain("setsid");
diff --git a/src/lib/actions/sandbox/gateway-wedge-diagnostics.test.ts b/src/lib/actions/sandbox/gateway-wedge-diagnostics.test.ts
index f322798f932..77d43cd8e68 100644
--- a/src/lib/actions/sandbox/gateway-wedge-diagnostics.test.ts
+++ b/src/lib/actions/sandbox/gateway-wedge-diagnostics.test.ts
@@ -6,7 +6,7 @@ import { describe, expect, it } from "vitest";
 // Import from compiled dist for parity with the other CLI tests in this project.
 import { collectGatewayWedgeDiagnostics, sanitizeWedgeLogLine } from "./gateway-wedge-diagnostics";
 
-describe("collectGatewayWedgeDiagnostics — #4710 wedge signature", () => {
+describe("collectGatewayWedgeDiagnostics wedge signature (#4710)", () => {
   it("returns the matching gateway.log lines, trimmed", () => {
     const lines = collectGatewayWedgeDiagnostics("my-sandbox", () => ({
       status: 0,
diff --git a/src/lib/actions/sandbox/policy-channel-conflict.test.ts b/src/lib/actions/sandbox/policy-channel-conflict.test.ts
index 48754d95177..df3c0a11960 100644
--- a/src/lib/actions/sandbox/policy-channel-conflict.test.ts
+++ b/src/lib/actions/sandbox/policy-channel-conflict.test.ts
@@ -480,7 +480,7 @@ describe("addSandboxChannel cross-sandbox conflict check (#4305)", () => {
   });
 
   // Scenario 4
-  it("--force bypasses the conflict even in non-interactive mode", async () => {
+  it("bypasses the conflict with --force even in non-interactive mode", async () => {
     arrangeRegistry({
       current: makeEmptyEntry("alpha"),
       others: [
@@ -566,7 +566,7 @@ describe("addSandboxChannel cross-sandbox conflict check (#4305)", () => {
   });
 
   // Scenario 7
-  it("--dry-run never runs the conflict check or touches credentials", async () => {
+  it("avoids the conflict check and credentials with --dry-run", async () => {
     arrangeRegistry({
       current: makeEmptyEntry("alpha"),
       others: [
@@ -706,7 +706,7 @@ describe("addSandboxChannel cross-sandbox conflict check (#4305)", () => {
     expect(upsertMock).not.toHaveBeenCalled();
   });
 
-  it("--force proceeds when the conflict check throws", async () => {
+  it("proceeds with --force when the conflict check throws", async () => {
     arrangeRegistry({ current: makeEmptyEntry("alpha"), others: [] });
     getCredentialMock.mockReturnValue(TELEGRAM_TOKEN);
     listSandboxesMock.mockImplementation(() => {
diff --git a/src/lib/actions/sandbox/process-recovery.test.ts b/src/lib/actions/sandbox/process-recovery.test.ts
index e7ba277bd15..cbfe901000a 100644
--- a/src/lib/actions/sandbox/process-recovery.test.ts
+++ b/src/lib/actions/sandbox/process-recovery.test.ts
@@ -9,7 +9,7 @@ import {
   waitForRecoveredSandboxGateway,
 } from "./process-recovery";
 
-describe("probeSandboxInferenceGatewayHealth — #3265 gateway-chain subprobe", () => {
+describe("probeSandboxInferenceGatewayHealth gateway-chain subprobe (#3265)", () => {
   const makeExec =
     (stdout: string, status = 0) =>
     async () => ({ status, stdout, stderr: "" });
@@ -58,7 +58,7 @@ describe("probeSandboxInferenceGatewayHealth — #3265 gateway-chain subprobe",
   });
 });
 
-describe("waitForRecoveredSandboxGateway — #4710 settle-window confirm", () => {
+describe("waitForRecoveredSandboxGateway settle-window confirmation (#4710)", () => {
   const ENV_KEYS = [
     "NEMOCLAW_GATEWAY_RECOVERY_WAIT_SECONDS",
     "NEMOCLAW_GATEWAY_RECOVERY_POLL_INTERVAL_SECONDS",
diff --git a/src/lib/actions/sandbox/rebuild-env-isolation.test.ts b/src/lib/actions/sandbox/rebuild-env-isolation.test.ts
index 0b27a76cb51..24956edb006 100644
--- a/src/lib/actions/sandbox/rebuild-env-isolation.test.ts
+++ b/src/lib/actions/sandbox/rebuild-env-isolation.test.ts
@@ -10,7 +10,7 @@ import {
   sanitizeEnvValueForDisplay,
 } from "./rebuild-env-isolation.js";
 
-describe("sanitizeEnvValueForDisplay (#5735 PRA-7)", () => {
+describe("sanitizeEnvValueForDisplay PRA-7 (#5735)", () => {
   it("collapses a multi-line / ANSI value into a single safe line", () => {
     // Untrusted NEMOCLAW_AGENT with a newline + CR + ANSI escape that tries to
     // paint a fake "Installation complete" status line.
@@ -34,7 +34,7 @@ describe("sanitizeEnvValueForDisplay (#5735 PRA-7)", () => {
   });
 });
 
-describe("AMBIENT_RECREATE_ENV_VARS contract (#5735 PRA-4)", () => {
+describe("AMBIENT_RECREATE_ENV_VARS contract PRA-4 (#5735)", () => {
   it("pins the exact onboard-selection env set the recreate must isolate", () => {
     // Mirrors the ambient selection env vars `onboard --resume` reads at its
     // source boundary. Adding a new onboard-selection env var must be a conscious
diff --git a/src/lib/actions/sandbox/rebuild-flow.test.ts b/src/lib/actions/sandbox/rebuild-flow.test.ts
index 695b25e6f4e..60f3ab9b3d2 100644
--- a/src/lib/actions/sandbox/rebuild-flow.test.ts
+++ b/src/lib/actions/sandbox/rebuild-flow.test.ts
@@ -554,7 +554,7 @@ describe("rebuildSandbox flow", () => {
     }
   });
 
-  it("recreates a matching-session custom-endpoint sandbox from a validated session endpoint, ignoring hostile ambient endpoint/provider/model (#5735 PRA-4)", async () => {
+  it("recreates a matching-session custom-endpoint sandbox from a validated session endpoint while ignoring hostile ambient values for PRA-4 (#5735)", async () => {
     // Matching session (sandboxName === target) with a custom endpoint recorded
     // in that session. Hostile ambient NEMOCLAW_ENDPOINT_URL/PROVIDER/MODEL must
     // be absent during recreate so onboard --resume uses the validated session
diff --git a/src/lib/actions/uninstall/run-plan.test.ts b/src/lib/actions/uninstall/run-plan.test.ts
index 153cf502ff9..50dc57e74aa 100644
--- a/src/lib/actions/uninstall/run-plan.test.ts
+++ b/src/lib/actions/uninstall/run-plan.test.ts
@@ -848,7 +848,7 @@ describe("uninstall run plan", () => {
     expect(logs).not.toContain("Swap file removed");
   });
 
-  it("#3456 sub-bug #4: gateway destroy no-op uses the 'already removed' wording, not 'Destroyed ... skipped'", () => {
+  it("uses the 'already removed' wording instead of 'Destroyed ... skipped' for gateway destroy no-ops, sub-bug 4 (#3456)", () => {
     // When `openshell gateway destroy -g nemoclaw` returns non-zero (gateway
     // already gone), the previous code printed `Destroyed gateway 'nemoclaw'
     // skipped` — self-contradictory. The fix routes this branch to an onSkip
diff --git a/src/lib/actions/update.test.ts b/src/lib/actions/update.test.ts
index 4befc314d46..28a4fa3d0c5 100644
--- a/src/lib/actions/update.test.ts
+++ b/src/lib/actions/update.test.ts
@@ -15,7 +15,7 @@ import {
 } from "./update";
 
 describe("runUpdateAction", () => {
-  it("--check reports update availability without running the installer", async () => {
+  it("reports update availability without running the installer for --check", async () => {
     const spawnSyncImpl = vi.fn();
     const log = vi.fn();
 
@@ -42,7 +42,7 @@ describe("runUpdateAction", () => {
     expect(log).toHaveBeenCalledWith(expect.stringContaining("Latest maintained version: 0.2.0"));
   });
 
-  it("--check renders NemoHermes branding and installer guidance when the Hermes alias is active", async () => {
+  it("renders NemoHermes branding and installer guidance for --check when the Hermes alias is active", async () => {
     const log = vi.fn();
 
     const result = await runUpdateAction(
@@ -66,7 +66,7 @@ describe("runUpdateAction", () => {
     );
   });
 
-  it("--check renders NemoDeepAgents branding and installer guidance when the Deep Agents alias is active", async () => {
+  it("renders NemoDeepAgents branding and installer guidance for --check when the Deep Agents alias is active", async () => {
     const log = vi.fn();
 
     const result = await runUpdateAction(
@@ -173,7 +173,7 @@ describe("runUpdateAction", () => {
     );
   });
 
-  it("--yes runs the maintained installer without prompting", async () => {
+  it("runs the maintained installer without prompting for --yes", async () => {
     const prompt = vi.fn(async () => "no");
     const spawnSyncImpl = vi.fn(
       () => ({ status: 0, stdout: "", stderr: "", signal: null }) as never,
diff --git a/src/lib/adapters/openshell/resolve.test.ts b/src/lib/adapters/openshell/resolve.test.ts
index 5a35838e9f5..03b11e21561 100644
--- a/src/lib/adapters/openshell/resolve.test.ts
+++ b/src/lib/adapters/openshell/resolve.test.ts
@@ -64,7 +64,7 @@ describe("lib/resolve-openshell", () => {
     ).toBe("/usr/local/bin/openshell");
   });
 
-  it("falls back to /opt/homebrew/bin (Apple Silicon Homebrew prefix, #5334)", () => {
+  it("falls back to the Apple Silicon Homebrew prefix at /opt/homebrew/bin (#5334)", () => {
     expect(
       resolveOpenshell({
         commandVResult: null,
diff --git a/src/lib/adapters/openshell/timeouts.test.ts b/src/lib/adapters/openshell/timeouts.test.ts
index 65ea905188e..36850657ccd 100644
--- a/src/lib/adapters/openshell/timeouts.test.ts
+++ b/src/lib/adapters/openshell/timeouts.test.ts
@@ -32,7 +32,7 @@ describe("openshell-timeouts", () => {
     expect(OPENSHELL_DOWNLOAD_TIMEOUT_MS).toBeLessThan(OPENSHELL_HEAVY_TIMEOUT_MS);
   });
 
-  it("uses the same probe constant name as PR #2454 for forward compatibility", () => {
+  it("uses the same probe constant name for forward compatibility (#2454)", () => {
     // PR #2454 introduces OPENSHELL_PROBE_TIMEOUT_MS = 15_000 locally.
     // This ensures the shared module stays aligned so #2454 can import it after rebase.
     expect(OPENSHELL_PROBE_TIMEOUT_MS).toBe(15_000);
diff --git a/src/lib/agent/onboard.test.ts b/src/lib/agent/onboard.test.ts
index 5e6feb8861c..55f3c0f1a20 100644
--- a/src/lib/agent/onboard.test.ts
+++ b/src/lib/agent/onboard.test.ts
@@ -95,7 +95,7 @@ const buildUrlsLoopback = (token: string | null, port: number): string[] => {
   return [`http://127.0.0.1:${port}/${hash}`];
 };
 
-describe("printDashboardUi — regression for #2078 (port 8642 is not a chat UI)", () => {
+describe("printDashboardUi with port 8642 outside the chat UI (#2078)", () => {
   const logSpy = vi.spyOn(console, "log").mockImplementation(() => {});
   const noteSpy = vi.fn();
 
diff --git a/src/lib/agent/runtime.test.ts b/src/lib/agent/runtime.test.ts
index 08ac29eb294..6c98edc1881 100644
--- a/src/lib/agent/runtime.test.ts
+++ b/src/lib/agent/runtime.test.ts
@@ -215,7 +215,7 @@ describe("buildRecoveryScript", () => {
   // swallowed sourcing errors via `2>/dev/null`, leaving respawned gateways
   // guard-less and crash-looping on the next library error from ciao,
   // model-pricing, or anything else hitting a sandboxed syscall.
-  describe("#2478 hardened library-guard preload chain", () => {
+  describe("hardened library-guard preload chain (#2478)", () => {
     it("sources the generated recovery env after validating the gateway env file", () => {
       const script = buildRecoveryScript(minimalAgent, 19000);
       expect(script).toContain("_nemoclaw_validate_recovery_proxy_env /tmp/nemoclaw-proxy-env.sh");
diff --git a/src/lib/build-context.test.ts b/src/lib/build-context.test.ts
index 19bf57ac5b0..41516f5c9e3 100644
--- a/src/lib/build-context.test.ts
+++ b/src/lib/build-context.test.ts
@@ -98,7 +98,7 @@ describe("printSandboxCreateRecoveryHints", () => {
   // This recovery path runs ONLY after the OpenShell upload failure is
   // classified; ordinary x86_64 happy-path onboards never reach it. The tests
   // below assert that branch deterministically by injecting platform/arch.
-  it("prints the local-registry workaround with the preserved built image tag for the #3266 upload 404", () => {
+  it("prints the local-registry workaround with the preserved built image tag for an upload 404 (#3266)", () => {
     printSandboxCreateRecoveryHints(
       [
         "  Built image openshell/sandbox-from-nemoclaw:abcd1234",
@@ -128,7 +128,7 @@ describe("printSandboxCreateRecoveryHints", () => {
     expect(out).toContain("onboard --resume");
   });
 
-  it("adds the Linux ARM64 (aarch64) note for the #3266 upload 404 only on Linux arm64", () => {
+  it("adds the Linux ARM64 note for an upload 404 only on Linux arm64 (#3266)", () => {
     printSandboxCreateRecoveryHints("failed to upload image tar into container", {
       platform: "linux",
       arch: "arm64",
@@ -136,7 +136,7 @@ describe("printSandboxCreateRecoveryHints", () => {
     expect(stderr()).toContain("known limitation on Linux ARM64 (aarch64)");
   });
 
-  it("omits the ARM64 note for the #3266 upload 404 on x86_64 hosts", () => {
+  it("omits the ARM64 note for an upload 404 on x86_64 hosts (#3266)", () => {
     printSandboxCreateRecoveryHints("failed to upload image tar into container", {
       platform: "linux",
       arch: "x64",
diff --git a/src/lib/channel-runtime-status.test.ts b/src/lib/channel-runtime-status.test.ts
index 0027d4be7b1..93fa024a56d 100644
--- a/src/lib/channel-runtime-status.test.ts
+++ b/src/lib/channel-runtime-status.test.ts
@@ -311,7 +311,7 @@ describe("probeChannelRuntimeStatus", () => {
     expect(result.configuredButNotRunning).toEqual([]);
   });
 
-  it("flags a configured channel as not-running when the gateway log never mentions it (#4156 reporter case)", () => {
+  it("flags a configured channel as not-running when the gateway log never mentions it in the reporter case (#4156)", () => {
     // Reporter symptom: openclaw.json had the telegram block but the
     // dashboard rendered "No channels found." This is the failure mode —
     // configured but the OpenClaw runtime never logged anything for it.
diff --git a/src/lib/cli/oclif-runner.test.ts b/src/lib/cli/oclif-runner.test.ts
index f14e9b06d7c..10f6b7118c2 100644
--- a/src/lib/cli/oclif-runner.test.ts
+++ b/src/lib/cli/oclif-runner.test.ts
@@ -195,7 +195,7 @@ describe("runOclifCommandById", () => {
     expect(errorLine).not.toHaveBeenCalled();
   });
 
-  it("#2666: surfaces errors that happen to carry oclif.exit === 0 instead of swallowing them", async () => {
+  it("surfaces errors that happen to carry oclif.exit === 0 instead of swallowing them (#2666)", async () => {
     // Before #2666 this branch silently set exit 0 and produced no output.
     // The bug was an arbitrary error riding the same `oclif.exit === 0`
     // channel, e.g. propagated from inside a command's run(). Surface the
@@ -216,7 +216,7 @@ describe("runOclifCommandById", () => {
     );
   });
 
-  it("#2666: falls back to a generic line when the error message is empty", async () => {
+  it("falls back to a generic line when the error message is empty (#2666)", async () => {
     // Closes the residual silent path: if a non-ExitError(0) carries an
     // empty message (or one that trims to empty), still emit *something*
     // so the user is never left looking at exit 0 + blank stdout/stderr.
diff --git a/src/lib/core/stdin.test.ts b/src/lib/core/stdin.test.ts
index 7583f8bc0b5..a434e53455a 100644
--- a/src/lib/core/stdin.test.ts
+++ b/src/lib/core/stdin.test.ts
@@ -48,7 +48,7 @@ describe("readLineFromStdin", () => {
     ["returns buffered bytes when EOF arrives before a newline", ["y", 0], "y"],
     ["returns null on a hard error with no buffered bytes", ["EBADF"], null],
     ["returns buffered bytes when a hard error interrupts mid-line", ["y", "e", "EBADF"], "ye"],
-  ] as const)("%s", (_label, events, expected) => {
+  ] as const)("handles the table case: %s", (_label, events, expected) => {
     expect(readLineFromStdin({ readSync: makeReadSync([...events]), sleep: vi.fn() })).toBe(
       expected,
     );
diff --git a/src/lib/core/version.test.ts b/src/lib/core/version.test.ts
index 60200f856af..b9e7ae65fd6 100644
--- a/src/lib/core/version.test.ts
+++ b/src/lib/core/version.test.ts
@@ -66,7 +66,7 @@ describe("lib/version", () => {
     rmSync(join(testDir, ".version"));
   });
 
-  it("regression #1239: returns .version even when package.json is stale", () => {
+  it("returns .version even when package.json is stale (#1239)", () => {
     // npm-published tarballs ship with a stale package.json version (0.1.0)
     // and a .version file stamped from the git tag at publish time. The
     // installed CLI must report the .version contents, not the package.json
diff --git a/src/lib/domain/sandbox/logs.test.ts b/src/lib/domain/sandbox/logs.test.ts
index f7f9f1aea4d..234b761a2aa 100644
--- a/src/lib/domain/sandbox/logs.test.ts
+++ b/src/lib/domain/sandbox/logs.test.ts
@@ -125,7 +125,7 @@ describe("mergeTailLogLines", () => {
     expect(mergeTailLogLines(["[1] a\n"], 0)).toBe("[1] a\n");
   });
 
-  it("caps the merged output at maxLines (closes #4100)", () => {
+  it("caps the merged output at maxLines (#4100)", () => {
     const gateway = ["[1] g1", "[3] g2", "[5] g3"].join("\n") + "\n";
     const openshell = ["[2] o1", "[4] o2", "[6] o3"].join("\n") + "\n";
     const merged = mergeTailLogLines([gateway, openshell], 3);
diff --git a/src/lib/domain/uninstall/paths.test.ts b/src/lib/domain/uninstall/paths.test.ts
index e85641554a6..77b371e026b 100644
--- a/src/lib/domain/uninstall/paths.test.ts
+++ b/src/lib/domain/uninstall/paths.test.ts
@@ -53,7 +53,7 @@ describe("uninstall paths", () => {
       path.join("/home/test", ".config", "nemoclaw"),
     ]);
   });
-  it("#3456: exposes the Linux Docker-driver gateway state dir so uninstall can clean it", () => {
+  it("exposes the Linux Docker-driver gateway state dir so uninstall can clean it (#3456)", () => {
     // ~/.local/state/nemoclaw/ holds the openshell-gateway PID file, SQLite
     // database, audit log, and vm-driver/ state. Documented as
     // NEMOCLAW_OPENSHELL_GATEWAY_STATE_DIR in docs/reference/commands.mdx.
diff --git a/src/lib/inference/config.test.ts b/src/lib/inference/config.test.ts
index 5acf9c49f7e..63d56ea1930 100644
--- a/src/lib/inference/config.test.ts
+++ b/src/lib/inference/config.test.ts
@@ -399,7 +399,7 @@ describe("planInferenceRouteReconcile", () => {
     expect(planInferenceRouteReconcile(null, recorded)).toEqual({ kind: "repair" });
   });
 
-  it("flags divergence when the gateway model differs (the #3726 case)", () => {
+  it("flags divergence when the gateway model differs (#3726)", () => {
     const live = {
       provider: "nvidia-prod",
       model: "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
diff --git a/src/lib/inference/nim.test.ts b/src/lib/inference/nim.test.ts
index afaa8809819..ee2cc913786 100644
--- a/src/lib/inference/nim.test.ts
+++ b/src/lib/inference/nim.test.ts
@@ -632,7 +632,7 @@ describe("nim", () => {
     // hosts, so mixed-GPU machines (RTX PRO 6000 + GB300 on the QA verification
     // host) dropped the model info entirely. We keep `name` undefined to avoid
     // misattribution but now surface the per-GPU breakdown via `gpus`.
-    it("drops name and populates gpus breakdown on mixed-model hosts (regression #2669)", () => {
+    it("drops name and populates the gpus breakdown on mixed-model hosts (#2669)", () => {
       const runCapture = vi.fn((cmd: string | string[]) => {
         if (!Array.isArray(cmd)) throw new Error("expected argv array");
         if (cmd[0] === "nvidia-smi" && cmd.some((a: string) => a.includes("name,memory.total"))) {
diff --git a/src/lib/inference/onboard-probes.test.ts b/src/lib/inference/onboard-probes.test.ts
index a8e4868fcc0..24ddad5fc9b 100644
--- a/src/lib/inference/onboard-probes.test.ts
+++ b/src/lib/inference/onboard-probes.test.ts
@@ -445,7 +445,7 @@ describe("OpenAI-compatible inference probes", () => {
     });
   });
 
-  describe("retriable HTTP statuses (issues #2980, #3033)", () => {
+  describe("retriable HTTP statuses (#2980, #3033)", () => {
     it("retries 429 (rate limit)", () => {
       expect(RETRIABLE_HTTP_PROBE_STATUSES.has(429)).toBe(true);
     });
@@ -465,7 +465,7 @@ describe("OpenAI-compatible inference probes", () => {
       expect(RETRIABLE_HTTP_PROBE_STATUSES.has(200)).toBe(false);
     });
 
-    it("recovers when an upstream 502 clears on retry (regression #2980)", () => {
+    it("recovers when an upstream 502 clears on retry (#2980)", () => {
       const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-502-probe-"));
       const fakeBin = path.join(tmpDir, "bin");
       const counter = path.join(tmpDir, "counter");
diff --git a/src/lib/inventory/index.test.ts b/src/lib/inventory/index.test.ts
index 8a7c4116caa..ba2d70b8339 100644
--- a/src/lib/inventory/index.test.ts
+++ b/src/lib/inventory/index.test.ts
@@ -135,7 +135,7 @@ describe("inventory commands", () => {
     expect(getLiveInference).not.toHaveBeenCalled();
   });
 
-  it("shows agent as 'unknown' for a gateway-recovered sandbox (#5714), not the OpenClaw default", async () => {
+  it("shows agent as 'unknown' instead of the OpenClaw default for a gateway-recovered sandbox (#5714)", async () => {
     const inventory = await getSandboxInventory({
       recoverRegistryEntries: async () => ({
         sandboxes: [
@@ -269,7 +269,7 @@ describe("inventory commands", () => {
     );
   });
 
-  it("#2753: suppresses last-onboarded hint when sandbox step never completed", async () => {
+  it("suppresses the last-onboarded hint when the sandbox step never completed (#2753)", async () => {
     // The session retains a sandbox name from an interrupted onboard
     // (pre-fix sessions on disk, or any in-progress write between steps).
     // Surfacing it as the "last onboarded sandbox" would resurrect the
@@ -763,7 +763,7 @@ describe("inventory commands", () => {
     expect(showServiceStatus).toHaveBeenCalledWith({ sandboxName: "alpha" });
   });
 
-  describe("#1077 — env-resolved default sandbox", () => {
+  describe("env-resolved default sandbox (#1077)", () => {
     const savedSandboxName = process.env.SANDBOX_NAME;
     const savedNemoclawSandboxName = process.env.NEMOCLAW_SANDBOX_NAME;
     const savedNemoclawSandbox = process.env.NEMOCLAW_SANDBOX;
diff --git a/src/lib/onboard/bridge-dns-preflight.test.ts b/src/lib/onboard/bridge-dns-preflight.test.ts
index a12f7489f78..ee82544190f 100644
--- a/src/lib/onboard/bridge-dns-preflight.test.ts
+++ b/src/lib/onboard/bridge-dns-preflight.test.ts
@@ -26,7 +26,7 @@ describe("printDockerBridgeContainerStartFailure", () => {
     vi.restoreAllMocks();
   });
 
-  it("uses the active CLI branding in the verify-outside hint (#3630 CodeRabbit)", () => {
+  it("uses the active CLI branding in the verify-outside hint per CodeRabbit review (#3630)", () => {
     setOnboardBrandingAgent("hermes");
     process.env.NEMOCLAW_AGENT = "hermes";
     process.env.NEMOCLAW_INVOKED_AS = "nemohermes";
@@ -49,7 +49,7 @@ describe("printDockerBridgeContainerStartFailure", () => {
     expect(verifyLine).not.toContain("Verify outside NemoClaw:");
   });
 
-  it("renders Linux daemon.json remediation without the bare-echo clobber fallback (#3630 CodeRabbit)", () => {
+  it("renders Linux daemon.json remediation without the bare-echo clobber fallback per CodeRabbit review (#3630)", () => {
     const messages: string[] = [];
     const errSpy = vi.spyOn(console, "error").mockImplementation((arg?: unknown) => {
       messages.push(String(arg ?? ""));
@@ -79,7 +79,7 @@ describe("printDockerBridgeContainerStartFailure", () => {
     expect(blob).toMatch(/\{"dns":\["[^"]+"\]\}/);
   });
 
-  it("renders WSL-without-systemd remediation without using systemctl steps (#3630 CodeRabbit)", () => {
+  it("renders WSL-without-systemd remediation without using systemctl steps per CodeRabbit review (#3630)", () => {
     const messages: string[] = [];
     const errSpy = vi.spyOn(console, "error").mockImplementation((arg?: unknown) => {
       messages.push(String(arg ?? ""));
@@ -103,7 +103,7 @@ describe("printDockerBridgeContainerStartFailure", () => {
     expect(blob).toContain("jq -n");
   });
 
-  it("renders WSL-with-systemd remediation with the Linux systemd path (#3630 CodeRabbit)", () => {
+  it("renders WSL-with-systemd remediation with the Linux systemd path per CodeRabbit review (#3630)", () => {
     const messages: string[] = [];
     const errSpy = vi.spyOn(console, "error").mockImplementation((arg?: unknown) => {
       messages.push(String(arg ?? ""));
@@ -120,7 +120,7 @@ describe("printDockerBridgeContainerStartFailure", () => {
     expect(blob).toContain("sudo systemctl restart docker");
   });
 
-  it("uses the pinned BusyBox digest in the manual verify-fix commands (#3630 CodeRabbit)", () => {
+  it("uses the pinned BusyBox digest in manual verify-fix commands per CodeRabbit review (#3630)", () => {
     const messages: string[] = [];
     const errSpy = vi.spyOn(console, "error").mockImplementation((arg?: unknown) => {
       messages.push(String(arg ?? ""));
@@ -139,7 +139,7 @@ describe("printDockerBridgeContainerStartFailure", () => {
     expect(blob).not.toMatch(/docker run --rm busybox\s+nslookup/);
   });
 
-  it("uses the pinned BusyBox digest in the verify-outside hint after a bridge failure (#3630 CodeRabbit)", () => {
+  it("uses the pinned BusyBox digest in the post-failure verify-outside hint per CodeRabbit review (#3630)", () => {
     const messages: string[] = [];
     const errSpy = vi.spyOn(console, "error").mockImplementation((arg?: unknown) => {
       messages.push(String(arg ?? ""));
@@ -158,7 +158,7 @@ describe("printDockerBridgeContainerStartFailure", () => {
     expect(blob).not.toMatch(/busybox:latest true/);
   });
 
-  it("renders macOS Docker Desktop daemon.json remediation without bare-echo clobber (#3630 CodeRabbit)", () => {
+  it("renders macOS Docker Desktop daemon.json remediation without bare-echo clobber per CodeRabbit review (#3630)", () => {
     const messages: string[] = [];
     const errSpy = vi.spyOn(console, "error").mockImplementation((arg?: unknown) => {
       messages.push(String(arg ?? ""));
diff --git a/src/lib/onboard/docker-gpu-local-inference.test.ts b/src/lib/onboard/docker-gpu-local-inference.test.ts
index 319a15a98a2..59ee505a1e8 100644
--- a/src/lib/onboard/docker-gpu-local-inference.test.ts
+++ b/src/lib/onboard/docker-gpu-local-inference.test.ts
@@ -173,7 +173,7 @@ describe("verifyDockerGpuSandboxLocalInference", () => {
     expect(script).not.toContain("docker exec");
   });
 
-  it("fails on a 4xx — route reached but not usable (auth/route misconfig), not the #4509 proof", () => {
+  it("fails on a 4xx because the route is reached but unusable instead of satisfying the proof (#4509)", () => {
     const result = verifyDockerGpuSandboxLocalInference(
       GPU_CONFIG,
       "ollama-local",
@@ -186,7 +186,7 @@ describe("verifyDockerGpuSandboxLocalInference", () => {
     }
   });
 
-  it("fails (unreachable) and retries on HTTP 000 — the #4509 regression", () => {
+  it("fails as unreachable and retries on HTTP 000 (#4509)", () => {
     const execInSandbox = execEmitting("HTTP_000");
     const sleep = vi.fn();
     const result = verifyDockerGpuSandboxLocalInference(
diff --git a/src/lib/onboard/docker-gpu-patch.test.ts b/src/lib/onboard/docker-gpu-patch.test.ts
index 2a11000f24a..2c67062fbbe 100644
--- a/src/lib/onboard/docker-gpu-patch.test.ts
+++ b/src/lib/onboard/docker-gpu-patch.test.ts
@@ -764,7 +764,7 @@ describe("docker-gpu-patch sandbox DNS fallback (#3579)", () => {
     );
   });
 
-  it("regression manifest: host.openshell.internal + google.com + gateway.discord.gg + integrate.api.nvidia.com (#3579 manager spec)", () => {
+  it("includes every hostname from the manager-provided regression manifest (#3579)", () => {
     // The four hostnames called out in #3579's manager-provided spec:
     //   host.openshell.internal      → resolved via --add-host (mount namespace)
     //   google.com                   → public DNS via embedded Docker resolver
diff --git a/src/lib/onboard/dockerfile-patch.test.ts b/src/lib/onboard/dockerfile-patch.test.ts
index c0b1f6a8503..2d860d17b97 100644
--- a/src/lib/onboard/dockerfile-patch.test.ts
+++ b/src/lib/onboard/dockerfile-patch.test.ts
@@ -689,7 +689,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("regression #1904: BASE_IMAGE must reference sandbox-base, not openshell-community", () => {
+  it("requires BASE_IMAGE to reference sandbox-base instead of openshell-community (#1904)", () => {
     // This is the exact bug that broke all e2e tests in PR #1937:
     // the code read a digest from blueprint.yaml (openshell-community registry)
     // and applied it to nemoclaw/sandbox-base (different registry).
@@ -823,7 +823,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("regression #1409: bakes NEMOCLAW_PROXY_HOST/PORT env into the staged Dockerfile", () => {
+  it("bakes NEMOCLAW_PROXY_HOST/PORT env into the staged Dockerfile (#1409)", () => {
     const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-onboard-dockerfile-proxy-"));
     const dockerfilePath = path.join(tmpDir, "Dockerfile");
     fs.writeFileSync(
@@ -873,7 +873,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("regression #1409: leaves Dockerfile defaults when proxy env is unset", () => {
+  it("leaves Dockerfile defaults when proxy env is unset (#1409)", () => {
     const tmpDir = fs.mkdtempSync(
       path.join(os.tmpdir(), "nemoclaw-onboard-dockerfile-proxy-default-"),
     );
@@ -918,7 +918,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("regression #2421: bakes NEMOCLAW_INFERENCE_INPUTS into the staged Dockerfile when env is set", () => {
+  it("bakes NEMOCLAW_INFERENCE_INPUTS into the staged Dockerfile when env is set (#2421)", () => {
     const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-onboard-dockerfile-inputs-"));
     const dockerfilePath = path.join(tmpDir, "Dockerfile");
     fs.writeFileSync(
@@ -959,7 +959,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("regression #2421: rejects malformed NEMOCLAW_INFERENCE_INPUTS and keeps default", () => {
+  it("rejects malformed NEMOCLAW_INFERENCE_INPUTS and keeps the default (#2421)", () => {
     const tmpDir = fs.mkdtempSync(
       path.join(os.tmpdir(), "nemoclaw-onboard-dockerfile-inputs-bad-"),
     );
@@ -1018,7 +1018,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("regression #1409: rejects malformed NEMOCLAW_PROXY_HOST/PORT and keeps defaults", () => {
+  it("rejects malformed NEMOCLAW_PROXY_HOST/PORT and keeps defaults (#1409)", () => {
     const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-onboard-dockerfile-proxy-bad-"));
     const dockerfilePath = path.join(tmpDir, "Dockerfile");
     fs.writeFileSync(
@@ -1070,7 +1070,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("#2281: bakes NEMOCLAW_AGENT_TIMEOUT env into the staged Dockerfile", () => {
+  it("bakes NEMOCLAW_AGENT_TIMEOUT env into the staged Dockerfile (#2281)", () => {
     const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-onboard-dockerfile-timeout-"));
     const dockerfilePath = path.join(tmpDir, "Dockerfile");
     fs.writeFileSync(
@@ -1111,7 +1111,7 @@ describe("dockerfile patch helpers", () => {
     }
   });
 
-  it("#2880: bakes NEMOCLAW_AGENT_HEARTBEAT_EVERY env into the staged Dockerfile", () => {
+  it("bakes NEMOCLAW_AGENT_HEARTBEAT_EVERY env into the staged Dockerfile (#2880)", () => {
     const tmpDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-onboard-dockerfile-heartbeat-"));
     const dockerfilePath = path.join(tmpDir, "Dockerfile");
     const baseDockerfile = [
diff --git a/src/lib/onboard/gateway-sandbox-reachability.test.ts b/src/lib/onboard/gateway-sandbox-reachability.test.ts
index bdc99b516d6..066279004d0 100644
--- a/src/lib/onboard/gateway-sandbox-reachability.test.ts
+++ b/src/lib/onboard/gateway-sandbox-reachability.test.ts
@@ -125,7 +125,7 @@ describe("isSandboxBridgeGatewayReachable", () => {
     expect(result.detail).toContain("operation not supported");
   });
 
-  it("does not misclassify unrelated 'veth' or 'operation not supported' output as veth_unsupported (#3630 CodeRabbit)", async () => {
+  it("does not misclassify unrelated 'veth' or 'operation not supported' output as veth_unsupported per CodeRabbit review (#3630)", async () => {
     // Generic veth status lines, or `operation not supported` from
     // other syscalls (mount, ioctl, etc.) must fall through to the
     // existing inconclusive path, not be reported as fatal Jetson veth.
@@ -215,7 +215,7 @@ describe("isSandboxBridgeGatewayReachable", () => {
     expect(result.reason).toBe("tcp_failed");
   });
 
-  it("downgrades a slow-registry pre-pull timeout to probe_unavailable (not fatal probe_timeout) (#3630 codex review)", async () => {
+  it("downgrades a slow-registry pre-pull timeout to probe_unavailable instead of fatal probe_timeout per Codex review (#3630)", async () => {
     const result = await isSandboxBridgeGatewayReachable({
       inspectNetworkImpl: () => ({ subnet: "172.19.0.0/16", gatewayIp: "172.19.0.1" }),
       usesHostGatewayRouteImpl: () => false,
@@ -231,7 +231,7 @@ describe("isSandboxBridgeGatewayReachable", () => {
     expect(result.detail).toContain("timed out");
   });
 
-  it("classifies docker-daemon-connect failures from the probe run as fatal docker_daemon_unreachable (#3630 CodeRabbit)", async () => {
+  it("classifies docker-daemon-connect failures from the probe run as fatal docker_daemon_unreachable per CodeRabbit review (#3630)", async () => {
     // The image-cache pre-pull succeeded (or was bypassed), but the
     // actual `docker run` probe failed with the daemon-down signature.
     // This must surface as docker_daemon_unreachable (fatal), not slip
@@ -278,7 +278,7 @@ describe("isSandboxBridgeGatewayReachable", () => {
     expect(result.reason).toBe("docker_daemon_unreachable");
   });
 
-  it("escalates inspect_unavailable to fatal docker_daemon_unreachable (#3630 codex review)", async () => {
+  it("escalates inspect_unavailable to fatal docker_daemon_unreachable per Codex review (#3630)", async () => {
     const result = await isSandboxBridgeGatewayReachable({
       inspectNetworkImpl: () => ({ subnet: "172.19.0.0/16", gatewayIp: "172.19.0.1" }),
       usesHostGatewayRouteImpl: () => false,
@@ -294,7 +294,7 @@ describe("isSandboxBridgeGatewayReachable", () => {
     expect(result.detail).toContain("Cannot connect to the Docker daemon");
   });
 
-  it("uses inspect-specific fallback detail when inspect_unavailable has no details (#3630 CodeRabbit)", async () => {
+  it("uses inspect-specific fallback detail when inspect_unavailable has no details per CodeRabbit review (#3630)", async () => {
     const result = await isSandboxBridgeGatewayReachable({
       inspectNetworkImpl: () => ({ subnet: "172.19.0.0/16", gatewayIp: "172.19.0.1" }),
       usesHostGatewayRouteImpl: () => false,
@@ -408,7 +408,7 @@ describe("formatSandboxBridgeUnreachableMessage", () => {
     expect(msg).toContain("enable integration for this distro");
   });
 
-  it("uses cliDisplayName() and cliName() in fatal messages instead of hardcoded NemoClaw branding (#3630 CodeRabbit)", () => {
+  it("uses cliDisplayName() and cliName() in fatal messages instead of hardcoded branding per CodeRabbit review (#3630)", () => {
     const savedAgent = process.env.NEMOCLAW_AGENT;
     const savedInvoked = process.env.NEMOCLAW_INVOKED_AS;
     process.env.NEMOCLAW_AGENT = "hermes";
diff --git a/src/lib/onboard/gateway-start-failure-integration.test.ts b/src/lib/onboard/gateway-start-failure-integration.test.ts
index d28e6678532..a9c3cfec56d 100644
--- a/src/lib/onboard/gateway-start-failure-integration.test.ts
+++ b/src/lib/onboard/gateway-start-failure-integration.test.ts
@@ -186,7 +186,7 @@ describe("startGatewayWithOptions docker-unreachable abort (#2347)", () => {
   // confirms the chain bottoms out at exitProcess(1) with the recovery
   // message printed.
 
-  describe("composition: classify → handleFinal → exit 1", () => {
+  describe("composition classifies, handles the final state, and exits 1", () => {
     let capturedClassifyLog: string[];
     let capturedPrintError: string[];
 
diff --git a/src/lib/onboard/hermes-dashboard.test.ts b/src/lib/onboard/hermes-dashboard.test.ts
index 400a198a8ab..1075eb73f35 100644
--- a/src/lib/onboard/hermes-dashboard.test.ts
+++ b/src/lib/onboard/hermes-dashboard.test.ts
@@ -71,7 +71,7 @@ describe("onboard Hermes dashboard helpers", () => {
     );
   });
 
-  it("routes the #4984 rejection through fail() so onboarding exits non-zero", () => {
+  it("routes the reserved-port rejection through fail() so onboarding exits non-zero (#4984)", () => {
     const fail = vi.fn((message: string): never => {
       throw new Error(message);
     });
diff --git a/src/lib/onboard/install-ollama-linux.test.ts b/src/lib/onboard/install-ollama-linux.test.ts
index 90a17e7436f..fad151e61ed 100644
--- a/src/lib/onboard/install-ollama-linux.test.ts
+++ b/src/lib/onboard/install-ollama-linux.test.ts
@@ -99,7 +99,7 @@ describe("decideInstallOllamaLinuxMode", () => {
     expect(decideInstallOllamaLinuxMode(opts)).toBe("system");
   });
 
-  it("returns user-local when non-interactive without passwordless sudo (issue #4114 repro)", () => {
+  it("returns user-local when non-interactive without passwordless sudo (#4114)", () => {
     const opts = makeOpts({
       canSudoNonInteractive: () => false,
       isNonInteractive: () => true,
diff --git a/src/lib/onboard/machine/handlers/finalization.test.ts b/src/lib/onboard/machine/handlers/finalization.test.ts
index 6dff275a6be..9fea41cd87c 100644
--- a/src/lib/onboard/machine/handlers/finalization.test.ts
+++ b/src/lib/onboard/machine/handlers/finalization.test.ts
@@ -293,7 +293,7 @@ describe("handleFinalizationState", () => {
   // (provoke / create pending) → autoPairScopeApproval (approve / clear
   // pending). Reversing warmup and approval makes the approval pass a no-op and
   // the user's first real run falls back — exactly the bug v2 fixes.
-  it("provokes the scope upgrade after recovery and before the approval pass (#4504-v2)", async () => {
+  it("provokes the scope upgrade after recovery and before the approval pass in v2 (#4504)", async () => {
     const { deps, calls } = createDeps();
 
     await handleFinalizationState(baseOptions(deps));
@@ -318,7 +318,7 @@ describe("handleFinalizationState", () => {
   // proceeds straight to the approval pass, verification, and the dashboard.
   // The dep returning nothing useful (no pending provoked, gateway slow) does
   // not change the downstream flow: behavior degrades to v1, never blocks.
-  it("does not depend on the warm-up succeeding; finalization still completes (#4504-v2)", async () => {
+  it("completes v2 finalization without depending on the warm-up succeeding (#4504)", async () => {
     // The default warm-up mock returns undefined (e.g. gateway not up → the
     // production leaf swallowed and provoked nothing). Finalization must be
     // unaffected.
@@ -338,7 +338,7 @@ describe("handleFinalizationState", () => {
   // upgrade is provoked regardless of which agent the sandbox runs (the
   // contract says run it unconditionally; idempotent once operator.write is
   // paired).
-  it("provokes the scope upgrade regardless of agent type (#4504-v2)", async () => {
+  it("provokes the v2 scope upgrade regardless of agent type (#4504)", async () => {
     const { deps: depsHermes, calls: callsHermes } = createDeps();
     await handleFinalizationState({ ...baseOptions(depsHermes), agent: { name: "hermes" } });
     expect(callsHermes.warmupScopeUpgrade).toHaveBeenCalledWith("my-assistant");
diff --git a/src/lib/onboard/machine/transitions.test.ts b/src/lib/onboard/machine/transitions.test.ts
index ac1bd22cbbd..92698f55aec 100644
--- a/src/lib/onboard/machine/transitions.test.ts
+++ b/src/lib/onboard/machine/transitions.test.ts
@@ -37,7 +37,7 @@ const canonicalDirectTransitions = [
 ] as const;
 
 describe("onboard machine vocabulary", () => {
-  it("defines the initial coarse state vocabulary from issue #3802", () => {
+  it("defines the initial coarse state vocabulary (#3802)", () => {
     expect(ONBOARD_MACHINE_STATES).toEqual([
       "init",
       "preflight",
@@ -55,7 +55,7 @@ describe("onboard machine vocabulary", () => {
     ]);
   });
 
-  it("defines the initial observe-only event vocabulary from issue #3802", () => {
+  it("defines the initial observe-only event vocabulary (#3802)", () => {
     expect(ONBOARD_MACHINE_EVENT_TYPES).toEqual([
       "onboard.started",
       "onboard.resumed",
diff --git a/src/lib/onboard/messaging-channel-setup.test.ts b/src/lib/onboard/messaging-channel-setup.test.ts
index ea749a1e2c0..e8025633ff9 100644
--- a/src/lib/onboard/messaging-channel-setup.test.ts
+++ b/src/lib/onboard/messaging-channel-setup.test.ts
@@ -75,7 +75,7 @@ describe("setupSelectedMessagingChannels", () => {
     vi.restoreAllMocks();
   });
 
-  it("#4068 prints Telegram group privacy-mode setup guidance during onboarding", async () => {
+  it("prints Telegram group privacy-mode setup guidance during onboarding (#4068)", async () => {
     process.env.TELEGRAM_BOT_TOKEN = "123456:ABC-test-token";
     process.env.TELEGRAM_REQUIRE_MENTION = "1";
     process.env.TELEGRAM_ALLOWED_IDS = "123456789";
@@ -573,7 +573,7 @@ describe("setupMessagingChannels", () => {
     expect(output).toContain("slack — already configured");
   });
 
-  it("#5696 exits with code 1 when TELEGRAM_GROUP_POLICY is set to an unrecognised value", async () => {
+  it("exits with code 1 when TELEGRAM_GROUP_POLICY is set to an unrecognised value (#5696)", async () => {
     process.env.TELEGRAM_BOT_TOKEN = "123456:ABC-test-token";
     process.env.TELEGRAM_GROUP_POLICY = "lockdown";
     const errors: string[] = [];
diff --git a/src/lib/onboard/model-router-process.test.ts b/src/lib/onboard/model-router-process.test.ts
index 054dac3fe61..2ece137e4d5 100644
--- a/src/lib/onboard/model-router-process.test.ts
+++ b/src/lib/onboard/model-router-process.test.ts
@@ -6,7 +6,7 @@ import { describe, expect, it } from "vitest";
 import { findModelRouterPidForPort } from "./model-router-process";
 
 describe("findModelRouterPidForPort", () => {
-  it("returns the PID when a model-router proxy is found via proc scan (direct, #5169)", () => {
+  it("returns the PID when a model-router proxy is found via direct proc scan (#5169)", () => {
     const pid = findModelRouterPidForPort(4000, {
       readProcCommandLine: (p) =>
         p === 12345
@@ -17,7 +17,7 @@ describe("findModelRouterPidForPort", () => {
     expect(pid).toBe(12345);
   });
 
-  it("returns the PID when model-router is Python-interpreted (args[1], #5169)", () => {
+  it("returns the PID when model-router is Python-interpreted through args[1] (#5169)", () => {
     const pid = findModelRouterPidForPort(4000, {
       readProcCommandLine: (p) =>
         p === 12345
diff --git a/src/lib/onboard/model-router-python.test.ts b/src/lib/onboard/model-router-python.test.ts
index ec0a56c0a3d..ea168ff63bc 100644
--- a/src/lib/onboard/model-router-python.test.ts
+++ b/src/lib/onboard/model-router-python.test.ts
@@ -85,7 +85,7 @@ describe("pickHostPython", () => {
     assert.equal(result.overrideRequested, false);
   });
 
-  it("returns every healthy candidate in priority order so the caller can fall back on venv failure (#3786 Codex P2)", () => {
+  it("returns every healthy candidate in priority order so the caller can fall back on venv failure per Codex P2 review (#3786)", () => {
     const which = (cmd: string) =>
       ({
         "python3.13": "/usr/bin/python3.13",
@@ -176,7 +176,7 @@ describe("pickHostPython", () => {
     assert.equal(probeCount, 1);
   });
 
-  it("treats NEMOCLAW_MODEL_ROUTER_PYTHON as a strict pin and does not fall back to PATH (#3786 Codex P3)", () => {
+  it("treats NEMOCLAW_MODEL_ROUTER_PYTHON as a strict pin without falling back to PATH per Codex P3 review (#3786)", () => {
     const which = (cmd: string) => (cmd === "python3.12" ? "/usr/bin/python3.12" : null);
     const probe = (executable: string) => {
       if (executable === "/opt/custom/python3.10") {
@@ -299,7 +299,7 @@ describe("supported version window", () => {
     assert.deepEqual([...MIN_PYTHON_VERSION], [3, 10]);
   });
 
-  it("excludes 3.14 to dodge the macOS Homebrew pyexpat regression in #3781", () => {
+  it("excludes 3.14 to avoid the macOS Homebrew pyexpat regression (#3781)", () => {
     assert.deepEqual([...MAX_PYTHON_EXCLUSIVE], [3, 14]);
   });
 });
diff --git a/src/lib/onboard/ollama-startup.test.ts b/src/lib/onboard/ollama-startup.test.ts
index 1b3ae8b0d89..a745a36e997 100644
--- a/src/lib/onboard/ollama-startup.test.ts
+++ b/src/lib/onboard/ollama-startup.test.ts
@@ -16,7 +16,7 @@ import {
 const wait = require("../core/wait");
 const runner = require("../runner");
 
-describe("runOllamaStartupOrGate (#4365 steer hint)", () => {
+describe("runOllamaStartupOrGate steer hint (#4365)", () => {
   let originalWaitForHttp: typeof wait.waitForHttp;
   let originalRunShell: typeof runner.runShell;
   let originalProviderEnv: string | undefined;
diff --git a/src/lib/onboard/preflight-cdi.test.ts b/src/lib/onboard/preflight-cdi.test.ts
index 7f66830e0b8..5930a94b24b 100644
--- a/src/lib/onboard/preflight-cdi.test.ts
+++ b/src/lib/onboard/preflight-cdi.test.ts
@@ -429,7 +429,7 @@ describe("planHostRemediation — CDI", () => {
   });
 });
 
-describe("shouldEnforceCdiNvidiaGpuSpec (#5489 enforcement gate)", () => {
+describe("shouldEnforceCdiNvidiaGpuSpec enforcement gate (#5489)", () => {
   it("enforces when the spec is missing and the operator did not explicitly opt out", () => {
     // The #5489 scenario: GPU hardware present (so cdiNvidiaGpuSpecMissing is
     // true) with sandbox GPU AUTO-disabled (nvidia-smi unavailable). Auto-disable
diff --git a/src/lib/onboard/preflight.test.ts b/src/lib/onboard/preflight.test.ts
index 925c7243c48..9cea07f9e4d 100644
--- a/src/lib/onboard/preflight.test.ts
+++ b/src/lib/onboard/preflight.test.ts
@@ -1040,7 +1040,7 @@ describe("probeContainerDns", () => {
     expect(isFatalContainerDnsProbeFailure(result)).toBe(true);
   });
 
-  it("downgrades unrelated docker output (no resolver evidence) from fatal resolution_failed to inconclusive error (#3630 CodeRabbit)", () => {
+  it("downgrades unrelated docker output without resolver evidence from fatal resolution_failed to an inconclusive error per CodeRabbit review (#3630)", () => {
     // No "Server:" header — nslookup never produced a resolver response.
     // The output is some docker-side message unrelated to DNS, so we
     // must not abort onboarding with the systemd-resolved remediation.
@@ -1183,7 +1183,7 @@ describe("probeContainerDns", () => {
     expect(isFatalContainerDnsProbeFailure(result)).toBe(true);
   });
 
-  it("classifies a wedged Docker daemon (inspect_unavailable) as fatal docker_daemon_unreachable (#3630 codex review)", () => {
+  it("classifies a wedged Docker daemon with inspect_unavailable as fatal docker_daemon_unreachable per Codex review (#3630)", () => {
     const result = probeContainerDns({
       ensureImageCachedOverride: {
         ok: false,
@@ -1196,7 +1196,7 @@ describe("probeContainerDns", () => {
     expect(isFatalContainerDnsProbeFailure(result)).toBe(true);
   });
 
-  it("does not treat a registry TCP timeout (i/o timeout on :443) as a fatal DNS failure (#3630 codex review)", () => {
+  it("does not treat a registry TCP timeout on port 443 as a fatal DNS failure per Codex review (#3630)", () => {
     // dial tcp <ip>:443 errors are TCP connectivity, NOT DNS — must not
     // be routed to UDP:53/systemd-resolved remediation.
     const result = probeContainerDns({
@@ -1303,7 +1303,7 @@ describe("probeContainerDns", () => {
     expect(seenScript).toContain("nslookup pinned-test.invalid");
   });
 
-  it("rejects shell metacharacters in probeName to prevent sh -c injection (#3630 CodeRabbit)", () => {
+  it("rejects shell metacharacters in probeName to prevent sh -c injection per CodeRabbit review (#3630)", () => {
     const injections = [
       "x; touch /tmp/pwned",
       "x && touch /tmp/pwned",
@@ -1415,7 +1415,7 @@ describe("probeDockerBridgeContainerStart", () => {
     expect(result.exitCode).toBe(125);
   });
 
-  it("does not misclassify unrelated 'veth' mentions as fatal veth_unsupported (#3630 CodeRabbit)", () => {
+  it("does not misclassify unrelated 'veth' mentions as fatal veth_unsupported per CodeRabbit review (#3630)", () => {
     // Output references "veth" in passing — without the bridge-create
     // signature, it must stay on the generic-error path, not the fatal
     // Jetson remediation path.
@@ -1432,7 +1432,7 @@ describe("probeDockerBridgeContainerStart", () => {
     expect(result.reason).not.toBe("veth_unsupported");
   });
 
-  it("does not misclassify generic 'operation not supported' errors as veth_unsupported (#3630 CodeRabbit)", () => {
+  it("does not misclassify generic 'operation not supported' errors as veth_unsupported per CodeRabbit review (#3630)", () => {
     // Generic OS-level "operation not supported" (e.g., from a cgroup
     // mount or unrelated syscall) must not be promoted to fatal veth.
     const result = probeDockerBridgeContainerStart({
@@ -1448,7 +1448,7 @@ describe("probeDockerBridgeContainerStart", () => {
     expect(result.reason).not.toBe("veth_unsupported");
   });
 
-  it("flags bridge container kill-by-signal (no timeout) as reason 'killed' (#3630 CodeRabbit)", () => {
+  it("flags a bridge container killed by signal without a timeout as reason 'killed' per CodeRabbit review (#3630)", () => {
     const result = probeDockerBridgeContainerStart({
       executionOverride: {
         stdout: "",
@@ -1507,7 +1507,7 @@ describe("probeDockerBridgeContainerStart", () => {
     expect(seenOpts?.timeout).toBe(20_000);
   });
 
-  it("reports image_pull_failed (not bridge timeout) when the busybox pre-pull times out (#3630 codex review)", () => {
+  it("reports image_pull_failed instead of bridge timeout when the busybox pre-pull times out per Codex review (#3630)", () => {
     const result = probeDockerBridgeContainerStart({
       ensureImageCachedOverride: {
         ok: false,
@@ -1533,7 +1533,7 @@ describe("probeDockerBridgeContainerStart", () => {
     expect(result.ok).toBe(true);
   });
 
-  it("reports a wedged Docker daemon (inspect_unavailable) as fatal docker_daemon_unreachable (#3630 codex review)", () => {
+  it("reports a wedged Docker daemon with inspect_unavailable as fatal docker_daemon_unreachable per Codex review (#3630)", () => {
     const result = probeDockerBridgeContainerStart({
       ensureImageCachedOverride: {
         ok: false,
@@ -1565,7 +1565,7 @@ describe("ensureProbeImageCached", () => {
     expect(result.alreadyCached).toBe(true);
   });
 
-  it("classifies an inspect spawn timeout (ETIMEDOUT) as inspect_unavailable without falling through to pull (#3630 CodeRabbit)", () => {
+  it("classifies an ETIMEDOUT inspect spawn as inspect_unavailable without falling through to pull per CodeRabbit review (#3630)", () => {
     const result = ensureProbeImageCached("busybox:latest", {
       inspectProbeImpl: () => ({
         stdout: "",
@@ -1584,7 +1584,7 @@ describe("ensureProbeImageCached", () => {
     expect(result.reason).toBe("inspect_unavailable");
   });
 
-  it("classifies 'Cannot connect to the Docker daemon' inspect stderr as inspect_unavailable (#3630 codex review)", () => {
+  it("classifies 'Cannot connect to the Docker daemon' inspect stderr as inspect_unavailable per Codex review (#3630)", () => {
     const result = ensureProbeImageCached("busybox:latest", {
       inspectProbeImpl: () => ({
         stdout: "",
@@ -1799,7 +1799,7 @@ describe("isDockerUnderProvisioned", () => {
   });
 });
 
-describe("assessHost — container runtime resource detection (regression #2514)", () => {
+describe("assessHost container runtime resource detection (#2514)", () => {
   it("flags default Colima (2 CPU / 2 GiB) as under-provisioned", () => {
     const result = assessHost({
       platform: "darwin",
diff --git a/src/lib/registry-recovery-action.test.ts b/src/lib/registry-recovery-action.test.ts
index fd82378cb98..e23e6fab92d 100644
--- a/src/lib/registry-recovery-action.test.ts
+++ b/src/lib/registry-recovery-action.test.ts
@@ -73,7 +73,7 @@ import { recoverRegistryEntries } from "./registry-recovery-action.js";
 import { parseLiveSandboxEntries } from "./runtime-recovery.js";
 import { loadSession } from "./state/onboard-session.js";
 
-describe("recoverRegistryEntries (#2753 seed-time guard)", () => {
+describe("recoverRegistryEntries seed-time guard (#2753)", () => {
   beforeEach(() => {
     mockRegistryState.sandboxes = {};
     mockRegistryState.defaultSandbox = null;
@@ -226,7 +226,7 @@ describe("recoverRegistryEntries (#2753 seed-time guard)", () => {
   });
 });
 
-describe("recoverRegistryEntries (#5714 empty-registry live gateway recovery)", () => {
+describe("recoverRegistryEntries empty-registry live gateway recovery (#5714)", () => {
   beforeEach(() => {
     vi.clearAllMocks();
     mockRegistryState.sandboxes = {};
@@ -347,7 +347,7 @@ describe("recoverRegistryEntries (#5714 empty-registry live gateway recovery)",
     expect(result.recoveredFromGateway).toBe(0);
   });
 
-  it("does NOT persist unseeded gateway recoveries to the on-disk registry (#5714 agent safety)", async () => {
+  it("does not persist unseeded gateway recoveries to the on-disk registry for agent safety (#5714)", async () => {
     // `openshell sandbox list` does not expose the agent type, so persisting a
     // recovered entry would default agent to "openclaw" everywhere downstream
     // and permanently misclassify a Deep Agents/Hermes sandbox. Recovery is
diff --git a/src/lib/sandbox-base-image.test.ts b/src/lib/sandbox-base-image.test.ts
index b033bb06b4b..82110080319 100644
--- a/src/lib/sandbox-base-image.test.ts
+++ b/src/lib/sandbox-base-image.test.ts
@@ -162,7 +162,7 @@ describe("sandbox base image helpers", () => {
     expect(output).toContain("the --mount option requires BuildKit");
   });
 
-  it("surfaces stdout-only build diagnostics — BuildKit can land errors there (Codex review on #3584)", () => {
+  it("surfaces stdout-only build diagnostics because BuildKit can put errors there per Codex review (#3584)", () => {
     const output = formatBuildFailureDiagnostics({
       stderr: "",
       stdout:
diff --git a/src/lib/shields/timer.test.ts b/src/lib/shields/timer.test.ts
index 16480768f80..ab5e97203fd 100644
--- a/src/lib/shields/timer.test.ts
+++ b/src/lib/shields/timer.test.ts
@@ -381,7 +381,7 @@ describe("shields timer authorization", () => {
   // (This narrows the revert window; it does not close the TOCTOU.)
   // -------------------------------------------------------------------------
 
-  it("#4663 re-verifies the auto-restore lock after settle so a reconciler reverting .config-hash perms is caught", async () => {
+  it("re-verifies the auto-restore lock after settle so a reconciler reverting .config-hash perms is caught (#4663)", async () => {
     const stateDir = path.join(tmpHome, ".nemoclaw", "state");
     fs.mkdirSync(stateDir, { recursive: true });
 
@@ -443,7 +443,7 @@ describe("shields timer authorization", () => {
     expect(JSON.parse(fs.readFileSync(stateFile, "utf-8")).shieldsDown).toBe(false);
   });
 
-  it("#4663 leaves shields DOWN and audits when the post-settle re-lock cannot hold .config-hash perms", async () => {
+  it("leaves shields DOWN and audits when the post-settle re-lock cannot hold .config-hash perms (#4663)", async () => {
     const stateDir = path.join(tmpHome, ".nemoclaw", "state");
     fs.mkdirSync(stateDir, { recursive: true });
 
diff --git a/src/lib/state/onboard-session.test.ts b/src/lib/state/onboard-session.test.ts
index ae7d6f28b4d..ba8c7505e1a 100644
--- a/src/lib/state/onboard-session.test.ts
+++ b/src/lib/state/onboard-session.test.ts
@@ -476,7 +476,7 @@ describe("onboard session", () => {
   // disk and the next rebuild preflight demanded a credential the current
   // sandbox did not need.
 
-  it("clears credentialEnv when provider-selection update passes null (GH #2625)", () => {
+  it("clears credentialEnv when a provider-selection update passes null (#2625)", () => {
     // Seed with a prior remote-provider onboard state.
     session.saveSession(session.createSession());
     markStepCompleteLegacy(session, stepMutation, "provider_selection", {
@@ -776,7 +776,7 @@ describe("onboard session", () => {
     expect(fresh.messagingPlan).toBeNull();
   });
 
-  it("#1737: persists telegramConfig across save/load roundtrips (requireMention=true)", () => {
+  it("persists telegramConfig across save/load roundtrips with requireMention=true (#1737)", () => {
     const created = session.createSession();
     created.telegramConfig = { requireMention: true };
     session.saveSession(created);
@@ -785,7 +785,7 @@ describe("onboard session", () => {
     expect(loaded.telegramConfig).toEqual({ requireMention: true });
   });
 
-  it("#1737: persists telegramConfig across save/load roundtrips (requireMention=false)", () => {
+  it("persists telegramConfig across save/load roundtrips with requireMention=false (#1737)", () => {
     const created = session.createSession();
     created.telegramConfig = { requireMention: false };
     session.saveSession(created);
@@ -794,7 +794,7 @@ describe("onboard session", () => {
     expect(loaded.telegramConfig).toEqual({ requireMention: false });
   });
 
-  it("#1737: rejects malformed telegramConfig on load", () => {
+  it("rejects malformed telegramConfig on load (#1737)", () => {
     // Simulate a hand-edited session file with garbage in telegramConfig.
     // Going through saveSession() would re-normalize the value before it
     // hits disk, so write raw JSON directly to exercise the load-time
@@ -809,7 +809,7 @@ describe("onboard session", () => {
     expect(loaded.telegramConfig).toBeNull();
   });
 
-  it("#1737: defaults telegramConfig to null for fresh sessions", () => {
+  it("defaults telegramConfig to null for fresh sessions (#1737)", () => {
     const fresh = session.createSession();
     expect(fresh.telegramConfig).toBeNull();
   });
@@ -986,7 +986,7 @@ describe("onboard session", () => {
     }
   });
 
-  it("regression #1281: stale-cleanup race does not unlink a fresh lock claimed by another process", () => {
+  it("does not unlink a fresh lock claimed by another process during a stale-cleanup race (#1281)", () => {
     // Reproduces the race: the lock file we read as 'stale' gets replaced
     // with a fresh claim from a faster concurrent process between our
     // read and our unlink. The slower process must NOT unlink the fresh
@@ -1211,7 +1211,7 @@ describe("onboard session", () => {
     expect(loaded.messagingPlan).toBeNull();
   });
 
-  it("#1737: filterSafeUpdates routes telegramConfig through markStepComplete", () => {
+  it("routes telegramConfig through markStepComplete in filterSafeUpdates (#1737)", () => {
     session.saveSession(session.createSession());
     markStepCompleteLegacy(session, stepMutation, "provider_selection", {
       telegramConfig: { requireMention: true },
@@ -1226,7 +1226,7 @@ describe("onboard session", () => {
     expect(cleared.telegramConfig).toBeNull();
   });
 
-  it("#1737: filterSafeUpdates drops malformed telegramConfig values", () => {
+  it("drops malformed telegramConfig values in filterSafeUpdates (#1737)", () => {
     session.saveSession(session.createSession());
     // Non-boolean requireMention — must not leak through.
     markStepCompleteLegacy(session, stepMutation, "provider_selection", {
@@ -1267,7 +1267,7 @@ describe("onboard session", () => {
     expect(loaded.wechatConfig).toBeNull();
   });
 
-  it("createSession with messagingPlan override", () => {
+  it("creates a session with a messagingPlan override", () => {
     const plan = makeMessagingPlan("my-assistant", ["telegram", "slack"]);
     const created = session.createSession({ messagingPlan: plan });
     expect(created.messagingPlan).toEqual(plan);
diff --git a/src/lib/tunnel/services.test.ts b/src/lib/tunnel/services.test.ts
index 6c5e503dbb2..e2a49fbf093 100644
--- a/src/lib/tunnel/services.test.ts
+++ b/src/lib/tunnel/services.test.ts
@@ -147,7 +147,7 @@ describe("sandbox name validation", () => {
   });
 });
 
-describe("#1077 — status host service PID dir matches start/stop env", () => {
+describe("status host service PID dir matches start/stop env (#1077)", () => {
   const savedSandboxName = process.env.SANDBOX_NAME;
   const savedNemoclawSandbox = process.env.NEMOCLAW_SANDBOX;
   const savedNemoclawSandboxName = process.env.NEMOCLAW_SANDBOX_NAME;
diff --git a/src/lib/validation.test.ts b/src/lib/validation.test.ts
index 869fcdb0e4d..19f3b9d5e62 100644
--- a/src/lib/validation.test.ts
+++ b/src/lib/validation.test.ts
@@ -59,7 +59,7 @@ describe("classifyValidationFailure", () => {
     });
   });
 
-  it("classifies 400 + expired key message as credential (#1942 — Gemini)", () => {
+  it("classifies a Gemini 400 with an expired-key message as a credential error (#1942)", () => {
     // Gemini returns HTTP 400 with this exact message when the API key has expired.
     // Must classify as "credential" so the onboard wizard prompts to re-enter the
     // key instead of looping back to provider selection.
@@ -74,7 +74,7 @@ describe("classifyValidationFailure", () => {
     });
   });
 
-  it("classifies 400 + API_KEY_INVALID message as credential (#1942 — Gemini)", () => {
+  it("classifies a Gemini 400 with API_KEY_INVALID as a credential error (#1942)", () => {
     // Gemini also uses "API_KEY_INVALID" as the status string for revoked keys.
     expect(
       classifyValidationFailure({
@@ -87,7 +87,7 @@ describe("classifyValidationFailure", () => {
     });
   });
 
-  it("classifies bare 'API key not valid' message as credential (#1942 — Gemini .message only)", () => {
+  it("classifies a bare Gemini 'API key not valid' .message as a credential error (#1942)", () => {
     // When the message field is extracted without the API_KEY_INVALID status
     // prefix, the bare wording must still classify as credential. Flagged by
     // CodeRabbit on #2132.
@@ -279,7 +279,7 @@ describe("classifySandboxCreateFailure", () => {
     expect(result.kind).toBe("image_upload_container_missing");
   });
 
-  it("does NOT classify an unrelated 404 as image_upload_container_missing (#3266 regression guard)", () => {
+  it("does not classify an unrelated 404 as image_upload_container_missing (#3266)", () => {
     // A generic 404 with no upload-tar phrase and no gateway container name
     // must not be mistaken for the ARM64 upload failure.
     expect(
diff --git a/src/lib/verify-deployment.test.ts b/src/lib/verify-deployment.test.ts
index 97b2e2e9b07..32a5894cbdc 100644
--- a/src/lib/verify-deployment.test.ts
+++ b/src/lib/verify-deployment.test.ts
@@ -34,7 +34,7 @@ describe("verifyDeployment", () => {
     expect(result.verification.dashboardReachable).toBe(true);
   });
 
-  it("treats HTTP 401 as gateway alive (device auth enabled — fixes #2342)", async () => {
+  it("treats HTTP 401 as a live gateway with device auth enabled (#2342)", async () => {
     const deps = makeDeps({
       executeSandboxCommand: () => ({ status: 0, stdout: "401", stderr: "" }),
       probeHostPort: () => 401,
@@ -261,7 +261,7 @@ describe("verifyDeployment", () => {
     expect(msgDiag?.hint).toContain("rebuild");
   });
 
-  it("surfaces an inconclusive runtime probe as a messaging warn (catches malformed openclaw.json #4156)", async () => {
+  it("surfaces an inconclusive runtime probe as a messaging warning for malformed openclaw.json (#4156)", async () => {
     const deps = makeDeps({
       getMessagingChannels: () => ["telegram"],
       providerExistsInGateway: () => true,
diff --git a/test/channels-add-preset.test.ts b/test/channels-add-preset.test.ts
index 5557cb5c6c4..6a7ed85ee85 100644
--- a/test/channels-add-preset.test.ts
+++ b/test/channels-add-preset.test.ts
@@ -281,7 +281,7 @@ module.exports = {
 `;
 }
 
-describe("channels add applies matching policy preset (issue #3437)", () => {
+describe("channels add applies a matching policy preset (#3437)", () => {
   it("plans channel enrollment through the messaging manifest workflow", () => {
     const script = `${buildPreamble()}
 const ctx = module.exports;
@@ -1576,7 +1576,7 @@ const ctx = module.exports;
 // startup breadcrumb confirmation or an actionable warning. These tests
 // drive the verifier through stubbed sandbox-exec output so the contract
 // is pinned regardless of OpenClaw/OpenShell runtime availability.
-describe("channels add verifies bridge startup after rebuild (issue #4314, #4390)", () => {
+describe("channels add verifies bridge startup after rebuild (#4314, #4390)", () => {
   function buildInteractivePreamble(): string {
     return String.raw`
 const resolver = require(${j("adapters/openshell/resolve.js")});
diff --git a/test/channels-remove-full-teardown.test.ts b/test/channels-remove-full-teardown.test.ts
index c9adb0a733f..06b5be1b265 100644
--- a/test/channels-remove-full-teardown.test.ts
+++ b/test/channels-remove-full-teardown.test.ts
@@ -218,7 +218,7 @@ module.exports = {
 `;
 }
 
-describe("channels remove full teardown (issue #3998)", () => {
+describe("channels remove full teardown (#3998)", () => {
   for (const sandboxAgent of ["openclaw", "hermes"] as const) {
     it(`strips '${sandboxAgent}' session.policyPresets and clears the in-sandbox whatsapp state dir`, () => {
       const script = `${buildPreamble({ sandboxAgent })}
diff --git a/test/cli/connect-recovery-settle.test.ts b/test/cli/connect-recovery-settle.test.ts
index f7188465ea6..bd0f4787464 100644
--- a/test/cli/connect-recovery-settle.test.ts
+++ b/test/cli/connect-recovery-settle.test.ts
@@ -15,7 +15,7 @@ import path from "node:path";
 import { runWithEnv, writeSandboxRegistry } from "./helpers";
 
 describe("CLI dispatch", () => {
-  it("fails probe-only when the gateway serves once and then drops its listener (#4710 wedge)", () => {
+  it("fails probe-only when a wedged gateway serves once and then drops its listener (#4710)", () => {
     const home = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-cli-connect-probe-wedge-"));
     const localBin = path.join(home, "bin");
     const markerFile = path.join(home, "openshell-calls");
diff --git a/test/cli/dispatch-basics.test.ts b/test/cli/dispatch-basics.test.ts
index 85021a586fd..9330df7613c 100644
--- a/test/cli/dispatch-basics.test.ts
+++ b/test/cli/dispatch-basics.test.ts
@@ -169,7 +169,7 @@ describe("CLI dispatch", () => {
     expect(r.out).toContain("langchain-deepagents-code");
   });
 
-  it("--help exits 0", () => {
+  it("exits 0 for --help", () => {
     expect(run("--help").code).toBe(0);
   });
 
@@ -179,7 +179,7 @@ describe("CLI dispatch", () => {
     expect(r.out.trim()).toMatch(/^nemoclaw v/);
   });
 
-  it("-h exits 0", () => {
+  it("exits 0 for -h", () => {
     expect(run("-h").code).toBe(0);
   });
 
diff --git a/test/cli/onboard-compatibility.test.ts b/test/cli/onboard-compatibility.test.ts
index 93b953f8a05..1f30478589c 100644
--- a/test/cli/onboard-compatibility.test.ts
+++ b/test/cli/onboard-compatibility.test.ts
@@ -156,7 +156,7 @@ describe("CLI onboard compatibility", () => {
     expect(r.out.includes("nemoclaw onboard")).toBeTruthy();
   });
 
-  it("#2753: refuses non-interactive --resume when sandbox step never completed and no name is provided", () => {
+  it("refuses non-interactive --resume when the sandbox step never completed and no name is provided (#2753)", () => {
     const home = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-cli-resume-no-name-"));
     const localBin = path.join(home, "bin");
     const nemoclawDir = path.join(home, ".nemoclaw");
@@ -180,7 +180,7 @@ describe("CLI onboard compatibility", () => {
     expect(r.out.includes("--name <sandbox>")).toBeTruthy();
   });
 
-  it("#2753: whitespace-only NEMOCLAW_SANDBOX_NAME does not satisfy the resume guard", () => {
+  it("does not let whitespace-only NEMOCLAW_SANDBOX_NAME satisfy the resume guard (#2753)", () => {
     // The env-var ingest pipeline trims and rejects whitespace-only values
     // before populating requestedSandboxName, so the guard sees no recovered
     // name and fires correctly.
diff --git a/test/credentials.test.ts b/test/credentials.test.ts
index e5d5e890bfb..f4e4e611a76 100644
--- a/test/credentials.test.ts
+++ b/test/credentials.test.ts
@@ -464,8 +464,8 @@ describe("legacy credentials.json migration (two-phase: stage then remove)", ()
   });
 });
 
-describe("removeLegacyCredentialsFileIfEmpty (post-upgrade cleanup, #3105)", () => {
-  it("removes an empty {} legacy file (regression #3105)", async () => {
+describe("removeLegacyCredentialsFileIfEmpty post-upgrade cleanup (#3105)", () => {
+  it("removes an empty legacy file containing {} (#3105)", async () => {
     const home = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-creds-"));
     const credsDir = path.join(home, ".nemoclaw");
     const legacyFile = path.join(credsDir, "credentials.json");
diff --git a/test/destroy-wipe-sandbox-state.test.ts b/test/destroy-wipe-sandbox-state.test.ts
index c6f10b256fa..f8a8b29d027 100644
--- a/test/destroy-wipe-sandbox-state.test.ts
+++ b/test/destroy-wipe-sandbox-state.test.ts
@@ -106,7 +106,7 @@ describe("wipeSandboxState (#5449)", () => {
   // shell-quoted but fed straight into `rm -rf -- ...` inside `cd ${dir}`,
   // where the relative form would traverse outside the agent config dir.
   // Validate paths against the resolved config dir and skip with a warning.
-  it("skips a state_dir whose resolved path escapes the agent config dir (#5455 PRA-6)", () => {
+  it("skips a state_dir whose resolved path escapes the agent config dir for PRA-6 (#5455)", () => {
     const { deps, runOpenshell } = buildDeps({
       loadAgent: vi.fn(() => ({
         configPaths: { dir: "/sandbox/.openclaw" },
@@ -138,7 +138,7 @@ describe("wipeSandboxState (#5449)", () => {
     }
   });
 
-  it("skips a state_file whose resolved path escapes the agent config dir (#5455 PRA-6)", () => {
+  it("skips a state_file whose resolved path escapes the agent config dir for PRA-6 (#5455)", () => {
     const { deps, runOpenshell } = buildDeps({
       loadAgent: vi.fn(() => ({
         configPaths: { dir: "/sandbox/.openclaw" },
@@ -168,7 +168,7 @@ describe("wipeSandboxState (#5449)", () => {
   // script intact, single-quoted, with no expansion or word-splitting risk.
   // shellQuote already handles this; the assertion locks the contract in so a
   // future refactor of the targets-construction can't accidentally drop it.
-  it("shell-quotes accepted manifest paths so metacharacters cannot break out of `rm -rf` (#5455 PRA-3)", () => {
+  it("shell-quotes accepted manifest paths so metacharacters cannot break out of `rm -rf` for PRA-3 (#5455)", () => {
     const { deps, runOpenshell } = buildDeps({
       loadAgent: vi.fn(() => ({
         configPaths: { dir: "/sandbox/.openclaw" },
@@ -209,7 +209,7 @@ describe("wipeSandboxState (#5449)", () => {
       dir: "/sandbox/.openclaw/../../etc",
       label: "absolute path escapes after agent subdir via `..`",
     },
-  ])("refuses to wipe when the agent config dir is unsafe ($label) (#5455 PRA-2)", ({ dir }) => {
+  ])("refuses to wipe when the $label agent config dir is unsafe for PRA-2 (#5455)", ({ dir }) => {
     const { deps, runOpenshell } = buildDeps({
       loadAgent: vi.fn(() => ({
         configPaths: { dir },
@@ -239,7 +239,7 @@ describe("wipeSandboxState (#5449)", () => {
   // re-onboard must NOT inherit USER.md from the prior sandbox. The proof
   // here is that the wipe script targets workspace/ under the agent config
   // dir AND contains no path escape that could rm -rf outside it.
-  it("targets workspace/ under the agent config dir and never contains a `..` escape (#5455 PRA-7)", () => {
+  it("targets workspace/ under the agent config dir without a `..` escape for PRA-7 (#5455)", () => {
     const { deps, runOpenshell } = buildDeps();
 
     destroy.wipeSandboxState("test-sb", deps as never);
@@ -269,7 +269,7 @@ describe("wipeSandboxState (#5449)", () => {
   // clean state. Skips on Windows because the `cd ... && rm -rf` script
   // is POSIX-shell-only.
   it.skipIf(process.platform === "win32")(
-    "actually deletes USER.md / SOUL.md when the constructed script is executed (#5455 PRA-1/PRA-2 behavioral)",
+    "deletes USER.md and SOUL.md when the constructed script executes for PRA-1 and PRA-2 (#5455)",
     () => {
       const tmpRoot = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-wipe-behavioral-"));
       try {
@@ -390,7 +390,7 @@ describe("wipeSandboxState (#5449)", () => {
       stateFiles: [{ path: "config.toml" }, { path: "hooks.json" }],
       label: "langchain-deepagents-code",
     },
-  ])("wipes the shipped $label manifest shape under its own /sandbox/<agent> dir (#5455 Ultra PRA-2)", ({
+  ])("wipes the shipped $label manifest shape under its own /sandbox/<agent> dir for Ultra PRA-2 (#5455)", ({
     agent,
     configDir,
     stateDirs,
@@ -418,7 +418,7 @@ describe("wipeSandboxState (#5449)", () => {
   // state_dirs and state_files must still issue the wipe so the multi-agent
   // `workspace-*` glob runs, but the `rm -rf --` argv must not collapse into
   // a syntactically broken command.
-  it("issues a syntactically valid wipe even when state_dirs and state_files are both empty (#5455 Ultra PRA-2)", () => {
+  it("issues a syntactically valid wipe with empty state_dirs and state_files for Ultra PRA-2 (#5455)", () => {
     const { deps, runOpenshell } = buildDeps({
       loadAgent: vi.fn(() => ({
         configPaths: { dir: "/sandbox/.openclaw" },
diff --git a/test/e2e-scenario/live/openclaw-tui-chat-correlation.test.ts b/test/e2e-scenario/live/openclaw-tui-chat-correlation.test.ts
index 26034af662a..82736136578 100644
--- a/test/e2e-scenario/live/openclaw-tui-chat-correlation.test.ts
+++ b/test/e2e-scenario/live/openclaw-tui-chat-correlation.test.ts
@@ -488,7 +488,7 @@ async function runLiveIssue2603ReproWithEventCaptureRetry(
 // ─── The live regression guard ─────────────────────────────────────
 
 test(
-  "openclaw-tui-chat-correlation: rapid TUI/webchat sends stay correlated on a real OpenClaw sandbox (#2603 + #3145)",
+  "openclaw-tui-chat-correlation keeps rapid TUI and webchat sends correlated on a real OpenClaw sandbox (#2603, #3145)",
   async ({ artifacts, environment, onboard, sandbox, secrets }) => {
     secrets.required("NVIDIA_INFERENCE_API_KEY");
 
diff --git a/test/e2e-scenario/support-tests/ci-compatible-inference.test.ts b/test/e2e-scenario/support-tests/ci-compatible-inference.test.ts
index 421c09820fb..95f8026f131 100644
--- a/test/e2e-scenario/support-tests/ci-compatible-inference.test.ts
+++ b/test/e2e-scenario/support-tests/ci-compatible-inference.test.ts
@@ -21,7 +21,7 @@ describe("gateway-managed compatible inference detection", () => {
         NVIDIA_INFERENCE_API_KEY: "hosted-compatible-test-key",
       },
     },
-  ])("skips the live issue #4434 sandbox-egress repro for $label", ({ env: hostedEnv }) => {
+  ])("skips the live sandbox-egress repro for $label (#4434)", ({ env: hostedEnv }) => {
     const {
       NEMOCLAW_E2E_USE_HOSTED_INFERENCE: _hostedSentinel,
       NEMOCLAW_PROVIDER: _provider,
diff --git a/test/e2e-scenario/support-tests/e2e-scenario-matrix.test.ts b/test/e2e-scenario/support-tests/e2e-scenario-matrix.test.ts
index 791d02f8f0d..8e9dd128f61 100644
--- a/test/e2e-scenario/support-tests/e2e-scenario-matrix.test.ts
+++ b/test/e2e-scenario/support-tests/e2e-scenario-matrix.test.ts
@@ -135,7 +135,7 @@ describe("live Vitest scenario matrix", () => {
     ]);
   });
 
-  it("--emit-live-matrix prints a single-line JSON array for supported live Vitest scenarios", () => {
+  it("prints a single-line JSON array of supported live Vitest scenarios for --emit-live-matrix", () => {
     const result = runEmitLiveMatrix();
     expect(result.status, result.stderr).toBe(0);
     const lines = result.stdout.trim().split("\n");
@@ -148,7 +148,7 @@ describe("live Vitest scenario matrix", () => {
     ]);
   });
 
-  it("--emit-live-matrix honors explicit scenario selections", () => {
+  it("honors explicit scenario selections for --emit-live-matrix", () => {
     const result = runEmitLiveMatrix(["--scenarios", "ubuntu-repo-cloud-hermes"]);
     expect(result.status, result.stderr).toBe(0);
     const parsed = JSON.parse(result.stdout.trim());
diff --git a/test/gateway-state-reconcile-2276.test.ts b/test/gateway-state-reconcile-2276.test.ts
index 11ff6948089..ea92feba053 100644
--- a/test/gateway-state-reconcile-2276.test.ts
+++ b/test/gateway-state-reconcile-2276.test.ts
@@ -293,7 +293,7 @@ afterEach(() => {
 });
 
 // ─── Scenario 1 ─── connect is now non-destructive (#4497) ─────────────────
-describe("Scenario 1: connect — healthy nemoclaw active + sandbox NotFound truly gone", () => {
+describe("connect with a healthy active nemoclaw gateway and a truly gone sandbox in scenario 1", () => {
   it("preserves the registry entry and session, points at rebuild/destroy, and exits 1", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -324,7 +324,7 @@ describe("Scenario 1: connect — healthy nemoclaw active + sandbox NotFound tru
 });
 
 // ─── Scenario 2 ─── passive `status` must preserve registry state ─────────
-describe("Scenario 2: status — healthy nemoclaw active + sandbox NotFound truly gone", () => {
+describe("status with a healthy active nemoclaw gateway and a truly gone sandbox in scenario 2", () => {
   it("reports the missing live sandbox without removing local registry state", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -352,7 +352,7 @@ describe("Scenario 2: status — healthy nemoclaw active + sandbox NotFound trul
 });
 
 // ─── Scenario 3 ─── self-heal via gateway select succeeds ──────────────────
-describe("Scenario 3: status — select succeeds, sandbox reappears, registry intact", () => {
+describe("status preserves the registry when selection succeeds and the sandbox reappears in scenario 3", () => {
   it("attempts `gateway select nemoclaw`, re-queries, proceeds; registry preserved", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -386,7 +386,7 @@ describe("Scenario 3: status — select succeeds, sandbox reappears, registry in
 });
 
 // ─── Scenario 4 ─── select fails → wrong_gateway_active, registry intact ───
-describe("Scenario 4: connect — select fails, sandbox still NotFound", () => {
+describe("connect when selection fails and the sandbox remains NotFound in scenario 4", () => {
   it("surfaces wrong_gateway_active guidance, preserves registry, exits 1", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -419,7 +419,7 @@ describe("Scenario 4: connect — select fails, sandbox still NotFound", () => {
 });
 
 // ─── Scenario 5 ─── exact #2276 repro: registry entry still present ────────
-describe("Scenario 5: #2276 repro — failed connect must leave registry entry intact", () => {
+describe("failed connect leaves the registry entry intact in scenario 5 (#2276)", () => {
   it("after a failed connect triggered by drifted gateway, entry is still present", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -446,7 +446,7 @@ describe("Scenario 5: #2276 repro — failed connect must leave registry entry i
 });
 
 // ─── Scenario 6 ─── nemoclaw gateway missing + NotFound ────────────────────
-describe("Scenario 6: connect — nemoclaw gateway missing after restart", () => {
+describe("connect with a missing nemoclaw gateway after restart in scenario 6", () => {
   it("returns gateway_missing_after_restart, preserves registry, exits 1", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -476,7 +476,7 @@ describe("Scenario 6: connect — nemoclaw gateway missing after restart", () =>
 });
 
 // ─── Scenario 7 ─── nemoclaw gateway unreachable + NotFound ────────────────
-describe("Scenario 7: connect — nemoclaw gateway unreachable after restart", () => {
+describe("connect with an unreachable nemoclaw gateway after restart in scenario 7", () => {
   it("returns gateway_unreachable_after_restart, preserves registry, exits 1", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -506,7 +506,7 @@ describe("Scenario 7: connect — nemoclaw gateway unreachable after restart", (
 });
 
 // ─── Scenario 8 ─── gateway info fails / unparseable ───────────────────────
-describe("Scenario 8: gateway info fails — safe default, registry preserved", () => {
+describe("gateway info failure preserves the registry with a safe default in scenario 8", () => {
   it("non-zero exit on `openshell gateway info -g nemoclaw` still preserves registry", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -533,8 +533,8 @@ describe("Scenario 8: gateway info fails — safe default, registry preserved",
 });
 
 // ─── Scenario 9 ─── openshell status empty / malformed ─────────────────────
-describe("Scenario 9: empty or malformed status — registry untouched", () => {
-  it("empty status + gateway info missing → registry preserved, no removal", {
+describe("empty or malformed status leaves the registry untouched in scenario 9", () => {
+  it("preserves the registry without removal when status is empty and gateway info is missing", {
     timeout: TIMEOUT_MS,
   }, () => {
     writeStubOpenshell({
@@ -557,7 +557,7 @@ describe("Scenario 9: empty or malformed status — registry untouched", () => {
     assert.doesNotMatch(r.stderr, /Removed stale local registry entry/);
   });
 
-  it("malformed status + malformed gateway info → registry preserved", {
+  it("preserves the registry when status and gateway info are malformed", {
     timeout: TIMEOUT_MS,
   }, () => {
     writeStubOpenshell({
@@ -581,7 +581,7 @@ describe("Scenario 9: empty or malformed status — registry untouched", () => {
 });
 
 // ─── Scenario 10 ─── non-interactive mode: no prompts ──────────────────────
-describe("Scenario 10: non-interactive mode — deterministic exit, no prompts", () => {
+describe("non-interactive mode exits deterministically without prompts in scenario 10", () => {
   it("NEMOCLAW_NON_INTERACTIVE=1 does not block on user input and exits 1", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -609,7 +609,7 @@ describe("Scenario 10: non-interactive mode — deterministic exit, no prompts",
 });
 
 // ─── Scenario 11 ─── cross-command parity: status drifts same way ──────────
-describe("Scenario 11: status — wrong gateway active yields guidance, not removal", () => {
+describe("status gives guidance instead of removal for the wrong active gateway in scenario 11", () => {
   it("drift case under `status` preserves registry and prints guidance", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -641,7 +641,7 @@ describe("Scenario 11: status — wrong gateway active yields guidance, not remo
 });
 
 // ─── Scenario 12 ─── cross-command parity: skill install drifts same way ───
-describe("Scenario 12: skill install — wrong gateway active yields guidance, not removal", () => {
+describe("skill install gives guidance instead of removal for the wrong active gateway in scenario 12", () => {
   it("skill install under drift preserves registry, exits 1 with guidance", {
     timeout: TIMEOUT_MS,
   }, () => {
@@ -710,7 +710,7 @@ describe("Scenario 12: skill install — wrong gateway active yields guidance, n
 // (a) locates the preserved entry (no "does not exist"), (b) does NOT dead-end
 // at "Cannot back up state", and (c) reports the stale state and proceeds to
 // recreate from the preserved registry metadata instead of aborting.
-describe("Scenario 14 (#4497): connect preserves registry so rebuild can recover", () => {
+describe("connect preserves the registry so rebuild can recover in scenario 14 (#4497)", () => {
   it("after a non-destructive connect, `rebuild --yes` recovers the stale sandbox", {
     timeout: TIMEOUT_MS,
   }, () => {
diff --git a/test/generate-hermes-config.test.ts b/test/generate-hermes-config.test.ts
index 90438f406b5..ce06ea66b7d 100644
--- a/test/generate-hermes-config.test.ts
+++ b/test/generate-hermes-config.test.ts
@@ -380,7 +380,7 @@ describe("agents/hermes/generate-config.ts", () => {
     ).toEqual(["INTERNAL_API line 3"]);
   });
 
-  it("regression #4230: configures Anthropic Messages routing for Hermes managed inference", () => {
+  it("configures Anthropic Messages routing for Hermes managed inference (#4230)", () => {
     const { config } = runConfigScript({
       NEMOCLAW_PROVIDER_KEY: "anthropic",
       NEMOCLAW_INFERENCE_BASE_URL: "https://inference.local",
diff --git a/test/generate-openclaw-config.test.ts b/test/generate-openclaw-config.test.ts
index f518a6e2794..1c3ef8ac1d9 100644
--- a/test/generate-openclaw-config.test.ts
+++ b/test/generate-openclaw-config.test.ts
@@ -356,13 +356,13 @@ describe("generate-openclaw-config.mts: config generation", () => {
     expect(config.gateway.controlUi.allowedOrigins).toEqual(["http://127.0.0.1:18789"]);
   });
 
-  it("#3256: emits gateway.port from a non-default CHAT_UI_URL port", () => {
+  it("emits gateway.port from a non-default CHAT_UI_URL port (#3256)", () => {
     const config = runConfigScript({ CHAT_UI_URL: "http://127.0.0.1:18790" });
     expect(config.gateway.port).toBe(18790);
     expect(config.gateway.controlUi.allowedOrigins).toEqual(["http://127.0.0.1:18790"]);
   });
 
-  it("#3256: lets NEMOCLAW_DASHBOARD_PORT drive gateway.port when set", () => {
+  it("lets NEMOCLAW_DASHBOARD_PORT drive gateway.port when set (#3256)", () => {
     const config = runConfigScript({
       CHAT_UI_URL: "",
       NEMOCLAW_DASHBOARD_PORT: "18790",
@@ -402,7 +402,7 @@ describe("generate-openclaw-config.mts: config generation", () => {
     expect(config.gateway.controlUi.allowedOrigins).toContain("http://remote.example");
   });
 
-  it("includes portless origin for reverse-proxy access (Fixes #3000)", () => {
+  it("includes a portless origin for reverse-proxy access (#3000)", () => {
     const config = runConfigScript({
       CHAT_UI_URL: "https://nemoclaw0-abc123.brevlab.com:18789",
     });
@@ -616,7 +616,7 @@ describe("generate-openclaw-config.mts: config generation", () => {
     expect(config.channels.discord.accounts.default.proxy).toBeUndefined();
   });
 
-  it("#3894: routes Discord gateway traffic through OpenClaw's managed proxy", () => {
+  it("routes Discord gateway traffic through OpenClaw's managed proxy (#3894)", () => {
     const channels = Buffer.from(JSON.stringify(["discord"])).toString("base64");
     const config = runConfigScript({
       NEMOCLAW_MESSAGING_CHANNELS_B64: channels,
@@ -844,7 +844,7 @@ describe("generate-openclaw-config.mts: config generation", () => {
     expect(config.agents.defaults.heartbeat).toEqual({ every: "30m" });
   });
 
-  it("disables heartbeat when set to 0m (NemoClaw#2880)", () => {
+  it("disables heartbeat when set to 0m (#2880)", () => {
     const config = runConfigScript({ NEMOCLAW_AGENT_HEARTBEAT_EVERY: "0m" });
     expect(config.agents.defaults.heartbeat).toEqual({ every: "0m" });
   });
@@ -1818,13 +1818,13 @@ describe("generate-openclaw-config.mts: config generation", () => {
     expect(config.plugins.entries.xai.enabled).toBe(false);
   });
 
-  it("#4246: enables the discord plugin entry when Discord channel is configured", () => {
+  it("enables the discord plugin entry when Discord is configured (#4246)", () => {
     const channels = Buffer.from(JSON.stringify(["discord"])).toString("base64");
     const config = runConfigScript({ NEMOCLAW_MESSAGING_CHANNELS_B64: channels });
     expect(config.plugins.entries.discord).toEqual({ enabled: true });
   });
 
-  it("#4246: omits the discord plugin entry when Discord channel is not configured", () => {
+  it("omits the discord plugin entry when Discord is not configured (#4246)", () => {
     const config = runConfigScript();
     expect(config.plugins.entries.discord).toBeUndefined();
   });
diff --git a/test/generate-platform-docs.test.ts b/test/generate-platform-docs.test.ts
index 2db0c2be6f9..30f477d8099 100644
--- a/test/generate-platform-docs.test.ts
+++ b/test/generate-platform-docs.test.ts
@@ -271,7 +271,7 @@ print(module.generate_platform_table_full(platforms))
     expect(full).toContain("WSL");
   });
 
-  it("--check exits non-zero on placeholder owner in real matrix", () => {
+  it("exits non-zero for --check on a placeholder owner in the real matrix", () => {
     const tmp = mkdtempSync(path.join(tmpdir(), "genplatform-"));
     const matrixPath = path.join(tmp, "matrix.json");
     writeFileSync(
diff --git a/test/hermes-gateway-wrapper.test.ts b/test/hermes-gateway-wrapper.test.ts
index 64626fce5a9..13d04c0b74c 100644
--- a/test/hermes-gateway-wrapper.test.ts
+++ b/test/hermes-gateway-wrapper.test.ts
@@ -108,7 +108,7 @@ describe.skipIf(!canRun)("agents/hermes/hermes-wrapper.sh", () => {
     expect(run.realInvoked).toBe(false);
   });
 
-  it("cannot be bypassed by shadowing python3 on PATH (#4981 review)", () => {
+  it("cannot be bypassed by shadowing python3 on PATH after review (#4981)", () => {
     // PATH is part of the untrusted env; a planted python3 that exits 0 must not
     // let the gateway start with a raw secret. The wrapper uses a trusted
     // absolute interpreter, so the guard still refuses.
diff --git a/test/host-artifact-cleanup.test.ts b/test/host-artifact-cleanup.test.ts
index 262a550a38d..675f3cb466e 100644
--- a/test/host-artifact-cleanup.test.ts
+++ b/test/host-artifact-cleanup.test.ts
@@ -47,7 +47,7 @@ afterEach(() => {
   vi.unstubAllEnvs();
 });
 
-describe("cleanupStaleHostFiles (post-upgrade sweep, #3105)", () => {
+describe("cleanupStaleHostFiles post-upgrade sweep (#3105)", () => {
   it("removes an empty legacy credentials.json and logs the removal", () => {
     const home = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-cleanup-"));
     const credsDir = path.join(home, ".nemoclaw");
diff --git a/test/http-proxy-fix-rewrite.test.ts b/test/http-proxy-fix-rewrite.test.ts
index 7dae7ca38de..a5f24a7e9c6 100644
--- a/test/http-proxy-fix-rewrite.test.ts
+++ b/test/http-proxy-fix-rewrite.test.ts
@@ -51,7 +51,7 @@ function loadWrapper() {
   require(FIX_PATH);
 }
 
-describe("http-proxy-fix rewrite (deepinfra-style failure, follow-up to #2344)", () => {
+describe("http-proxy-fix rewrite for a deepinfra-style failure (#2344)", () => {
   let origHttpRequest: typeof http.request;
   let httpsSpy: ReturnType<typeof vi.spyOn>;
   let captured: RewrittenOptions | null;
diff --git a/test/install-preflight.test.ts b/test/install-preflight.test.ts
index a2504274080..13a53359b39 100644
--- a/test/install-preflight.test.ts
+++ b/test/install-preflight.test.ts
@@ -414,7 +414,7 @@ exit 98
     expect(output).not.toMatch(/deprecated compatibility wrapper/);
   });
 
-  it("--help exits 0 and shows install usage", () => {
+  it("exits 0 and shows install usage for --help", () => {
     const result = spawnSync("bash", [INSTALLER, "--help"], {
       cwd: path.join(import.meta.dirname, ".."),
       encoding: "utf-8",
@@ -449,7 +449,7 @@ exit 98
     expect(output).toMatch(/aliases: cloud -> build, nim -> nim-local/);
   });
 
-  it("--version exits 0 and prints the version number", () => {
+  it("exits 0 and prints the version number for --version", () => {
     const result = spawnSync("bash", [INSTALLER, "--version"], {
       cwd: path.join(import.meta.dirname, ".."),
       encoding: "utf-8",
@@ -461,7 +461,7 @@ exit 98
     expect(output).not.toMatch(/0\.1\.0/);
   });
 
-  it("-v exits 0 and prints the version number", () => {
+  it("exits 0 and prints the version number for -v", () => {
     const result = spawnSync("bash", [INSTALLER, "-v"], {
       cwd: path.join(import.meta.dirname, ".."),
       encoding: "utf-8",
@@ -982,7 +982,7 @@ fi`,
   // #2430: --fresh is the escape hatch. Even with a session file on disk
   // (failed or otherwise), the installer should skip the auto-resume check
   // and let the onboard command create a new session.
-  it("--fresh skips auto-resume regardless of session state (#2430)", () => {
+  it("skips auto-resume with --fresh regardless of session state (#2430)", () => {
     const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-install-fresh-"));
     const fakeBin = path.join(tmp, "bin");
     const prefix = path.join(tmp, "prefix");
@@ -2900,7 +2900,7 @@ describe("installer flag parsing", () => {
     expect(output).toMatch(/NemoClaw Installer/); // usage was printed
   });
 
-  it("--help shows NEMOCLAW_INSTALL_TAG in environment section", () => {
+  it("shows NEMOCLAW_INSTALL_TAG in the --help environment section", () => {
     const result = spawnSync("bash", [INSTALLER, "--help"], {
       cwd: path.join(import.meta.dirname, ".."),
       encoding: "utf-8",
@@ -3125,7 +3125,7 @@ exit 0`,
     return { result, args };
   }
 
-  it("#2670: ACCEPT_THIRD_PARTY_SOFTWARE=1 alone clears the notice in non-TTY mode", () => {
+  it("clears the notice in non-TTY mode with ACCEPT_THIRD_PARTY_SOFTWARE=1 alone (#2670)", () => {
     const { result, args } = callShowUsageNotice({
       // Simulates curl|bash mode: stdin is not a TTY, NON_INTERACTIVE is unset,
       // and only --yes-i-accept-third-party-software was passed.
@@ -3160,7 +3160,7 @@ exit 0`,
     expect(output).not.toMatch(/\/dev\/tty/);
   });
 
-  it("#3058: error message includes a working curl|bash example users can copy-paste", () => {
+  it("includes a working curl|bash example users can copy-paste in the error message (#3058)", () => {
     // The reporter on #3058 hit this error with `curl ... | bash` on a
     // non-TTY box and was left guessing how to combine the env var with
     // the documented one-liner. The fix surfaces the exact invocations
@@ -3772,7 +3772,7 @@ sys.exit(exit_code)
     return runInstallerWithTty(answer, "tty");
   }
 
-  it("#2671: headless curl|bash with no flags exits 1 BEFORE phase 1 (atomic — no Node/CLI install)", () => {
+  it("exits 1 before phase 1 for headless curl|bash with no flags and installs nothing (#2671)", () => {
     const { result, phases } = runInstaller({});
     expect(result.status).not.toBe(0);
     const output = `${result.stdout}${result.stderr}`;
@@ -3855,7 +3855,7 @@ sys.exit(exit_code)
     expect(state).toBe("");
   });
 
-  it("--non-interactive alone with a controlling TTY still stops before phase 1", () => {
+  it("stops before phase 1 for --non-interactive alone with a controlling TTY", () => {
     const { result, phases, state } = runInstallerWithTty("yes\n", "pipe", {
       NEMOCLAW_NON_INTERACTIVE: "1",
     });
@@ -3871,7 +3871,7 @@ sys.exit(exit_code)
     expect(state).toBe("");
   });
 
-  it("--yes-i-accept-third-party-software alone is sufficient to clear the fail-fast gate", () => {
+  it("clears the fail-fast gate with --yes-i-accept-third-party-software alone", () => {
     // The flag implies non-interactive intent (set by main() before the
     // preflight check), so it must clear the gate AND let the install
     // progress past preflight into phase 1 — assert phases is non-empty
@@ -3883,7 +3883,7 @@ sys.exit(exit_code)
     expect(phases).not.toBe("");
   });
 
-  it("--non-interactive alone does not clear the fail-fast gate", () => {
+  it("does not clear the fail-fast gate with --non-interactive alone", () => {
     const { result, phases } = runInstaller({ NEMOCLAW_NON_INTERACTIVE: "1" });
     const output = `${result.stdout}${result.stderr}`;
     expect(result.status).not.toBe(0);
diff --git a/test/install-stage-from-stdin.test.ts b/test/install-stage-from-stdin.test.ts
index b6f66c1fb9e..48b7c113387 100644
--- a/test/install-stage-from-stdin.test.ts
+++ b/test/install-stage-from-stdin.test.ts
@@ -126,7 +126,7 @@ function runEntryGuard(opts: {
   };
 }
 
-describe("install.sh entry-guard staging — #4414 curl|bash stdin self-stage", () => {
+describe("install.sh entry-guard staging for curl|bash stdin self-stage (#4414)", () => {
   it("stages to /tmp and would exec bash on the staged file when invoked via curl|bash", () => {
     // Pipe-mode invocation: BASH_SOURCE[0] empty. Without staging,
     // ensure_docker's sg(1) re-exec from #4419 has no file to point at
diff --git a/test/install-upgrade-sandboxes-severity.test.ts b/test/install-upgrade-sandboxes-severity.test.ts
index 7ba627b4c58..355ea81f136 100644
--- a/test/install-upgrade-sandboxes-severity.test.ts
+++ b/test/install-upgrade-sandboxes-severity.test.ts
@@ -95,7 +95,7 @@ describe("install.sh print_done — auto-upgrade severity (#5735)", () => {
   });
 });
 
-describe("install.sh finalize_install — fatal exit on failed auto-upgrade (#5735 PRA-5)", () => {
+describe("install.sh finalize_install fatal exit on failed auto-upgrade for PRA-5 (#5735)", () => {
   it("exits zero and prints the clean banner when no upgrade failed", () => {
     const result = runFinalizeInstall(false);
     expect(result.status, result.stderr).toBe(0);
diff --git a/test/issue-4434-tui-unreachable-inference.test.ts b/test/issue-4434-tui-unreachable-inference.test.ts
index e679af4615a..84a02e8cc0f 100644
--- a/test/issue-4434-tui-unreachable-inference.test.ts
+++ b/test/issue-4434-tui-unreachable-inference.test.ts
@@ -98,7 +98,7 @@ function driveMockOpenClawGatewayChatPath(params: {
   };
 }
 
-describe("issue #4434 unreachable inference TUI behavior", () => {
+describe("unreachable inference TUI behavior (#4434)", () => {
   it("classifies the captured spinner plus connected status with no error as the broken signature", () => {
     const capture = [
       "  flibbertigibbeting... - 3m 42s | connected",
diff --git a/test/issue-5667-hosted-inference-model-namespace.test.ts b/test/issue-5667-hosted-inference-model-namespace.test.ts
index 1fd1b9b1e5e..bbe15e7f082 100644
--- a/test/issue-5667-hosted-inference-model-namespace.test.ts
+++ b/test/issue-5667-hosted-inference-model-namespace.test.ts
@@ -121,7 +121,7 @@ function writeFakeDeepAgentsCodeModule(tmpDir: string): string {
   return pythonPath;
 }
 
-describe("issue #5667: hosted inference default model namespace", () => {
+describe("hosted inference default model namespace (#5667)", () => {
   // Snapshot the whole environment and restore it wholesale so the teardown
   // stays linear (no per-key conditional): clear every key, then repopulate
   // from the snapshot. Keys added during a test are dropped; original values
diff --git a/test/messaging-build-applier.test.ts b/test/messaging-build-applier.test.ts
index 4649c6996eb..c32ecd8a5bc 100644
--- a/test/messaging-build-applier.test.ts
+++ b/test/messaging-build-applier.test.ts
@@ -721,7 +721,7 @@ describe("messaging-build-applier.mts: agent-install", () => {
     expect(result.stderr).toContain("unexpected-package==1.2.3");
   });
 
-  it("#4246: messaging post-agent-install render reaches the mocked OpenClaw doctor boundary", () => {
+  it("reaches the mocked OpenClaw doctor boundary during post-agent-install messaging render (#4246)", () => {
     const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-discord-runtime-contract-"));
     const tracePath = path.join(tmp, "openclaw.trace");
     const fakeOpenclaw = path.join(tmp, "openclaw");
diff --git a/test/nemo-deepagents-alias.test.ts b/test/nemo-deepagents-alias.test.ts
index 07c9ac0e5d7..69677ad67a7 100644
--- a/test/nemo-deepagents-alias.test.ts
+++ b/test/nemo-deepagents-alias.test.ts
@@ -83,7 +83,7 @@ describe("nemo-deepagents alias", () => {
     expect(stat.mode & 0o100).not.toBe(0);
   });
 
-  it("--version outputs nemo-deepagents branding", () => {
+  it("outputs nemo-deepagents branding for --version", () => {
     const { code, out } = runDeepAgents("--version");
     expect(code).toBe(0);
     expect(out).toMatch(/^nemo-deepagents v[\d.]+/);
diff --git a/test/nemoclaw-start-gateway-health.test.ts b/test/nemoclaw-start-gateway-health.test.ts
index 878c37aa111..552b9e40aff 100644
--- a/test/nemoclaw-start-gateway-health.test.ts
+++ b/test/nemoclaw-start-gateway-health.test.ts
@@ -358,7 +358,7 @@ describe("gateway serving watchdog (#4710)", () => {
 });
 
 describe("record_gateway_pid", () => {
-  it("replaces a planted symlink without writing through it (#4710 pidfile race)", () => {
+  it("replaces a planted symlink without writing through it during the pidfile race (#4710)", () => {
     // In root mode the pidfile lives in sticky /tmp; a sandbox process can
     // plant a symlink at that path between respawns. The update must replace
     // the symlink as a directory entry (atomic rename), never open it.
diff --git a/test/nemoclaw-start-plugin-refresh.test.ts b/test/nemoclaw-start-plugin-refresh.test.ts
index a9158caf5ff..aa27cad13c6 100644
--- a/test/nemoclaw-start-plugin-refresh.test.ts
+++ b/test/nemoclaw-start-plugin-refresh.test.ts
@@ -268,7 +268,7 @@ describe("plugin refresh log preparation", () => {
   });
 });
 
-describe("plugin registry refresh workaround (#2021, openclaw/openclaw#89606)", () => {
+describe("plugin registry refresh workaround for openclaw/openclaw#89606 (#2021)", () => {
   it("invokes `openclaw plugins registry --refresh` once the gateway reports ready", () => {
     const { result, refreshLog, callLog, tmpDir } = runRefreshBlock();
     try {
diff --git a/test/nemoclaw-start.test.ts b/test/nemoclaw-start.test.ts
index 7f05a16439e..192325648d0 100644
--- a/test/nemoclaw-start.test.ts
+++ b/test/nemoclaw-start.test.ts
@@ -484,7 +484,7 @@ describe("nemoclaw-start non-root fallback", () => {
     expect(result.stdout).not.toContain("SHOULD_NOT_CONFIGURE");
   });
 
-  it("#3256: only requires early gateway token generation for gateway and OpenClaw commands", () => {
+  it("only requires early gateway token generation for gateway and OpenClaw commands (#3256)", () => {
     const src = fs.readFileSync(START_SCRIPT, "utf-8");
     const script = [
       "set -euo pipefail",
@@ -507,7 +507,7 @@ describe("nemoclaw-start non-root fallback", () => {
     expect(result.stdout).toContain("no:bash");
   });
 
-  it("#4517: refreshes startup tokens but only ensures direct OpenClaw command tokens", () => {
+  it("refreshes startup tokens but only ensures direct OpenClaw command tokens (#4517)", () => {
     const src = fs.readFileSync(START_SCRIPT, "utf-8");
     const script = [
       "set -euo pipefail",
@@ -752,7 +752,7 @@ describe("nemoclaw-start gateway token export (#1114)", () => {
     expect(envFile).not.toContain(".profile");
   });
 
-  it("#3256: writes gateway port and URL into the runtime shell env", () => {
+  it("writes the gateway port and URL into the runtime shell env (#3256)", () => {
     const { result, envFile } = runGatewayTokenHarness(
       JSON.stringify({ gateway: { auth: { token: "token" } } }),
       "stale-token",
@@ -766,7 +766,7 @@ describe("nemoclaw-start gateway token export (#1114)", () => {
     expect(envFile).toContain("export OPENCLAW_GATEWAY_TOKEN='token'");
   });
 
-  it("#3730: writes OpenClaw state env for connect-shell pairing approval", () => {
+  it("writes OpenClaw state env for connect-shell pairing approval (#3730)", () => {
     const { result, envFile } = runGatewayTokenHarness(
       JSON.stringify({ gateway: { auth: { token: "token" } } }),
     );
@@ -781,7 +781,7 @@ describe("nemoclaw-start gateway token export (#1114)", () => {
     );
   });
 
-  it("#3256: generates a gateway token before writing the runtime shell env", () => {
+  it("generates a gateway token before writing the runtime shell env (#3256)", () => {
     const { result, envFile, configAfter, hashAfter } = runGatewayTokenHarness(
       JSON.stringify({ gateway: { auth: {} } }),
       "stale-token",
@@ -800,7 +800,7 @@ describe("nemoclaw-start gateway token export (#1114)", () => {
     expect(hashAfter).toMatch(/ openclaw\.json\n$/);
   });
 
-  it("#4517: rotates an existing gateway token before writing the runtime shell env", () => {
+  it("rotates an existing gateway token before writing the runtime shell env (#4517)", () => {
     const oldToken = "old-token-before-rebuild";
     const { result, envFile, configAfter, hashAfter } = runGatewayTokenHarness(
       JSON.stringify({ gateway: { auth: { token: oldToken } } }),
@@ -820,7 +820,7 @@ describe("nemoclaw-start gateway token export (#1114)", () => {
     expect(hashAfter).toMatch(/ openclaw\.json\n$/);
   });
 
-  it("#4517: rotates an existing gateway token from JSON5 config", () => {
+  it("rotates an existing gateway token from JSON5 config (#4517)", () => {
     const oldToken = "old-json5-token-before-rebuild";
     const { result, envFile, configAfter, hashAfter } = runGatewayTokenHarness(
       [
@@ -1028,7 +1028,7 @@ describe("nemoclaw-start configure guard behavior", () => {
       fs.rmSync(setup.tmpDir, { recursive: true, force: true });
     }
   });
-  it("#4462: unsets gateway env and recovers constrained replacement state", () => {
+  it("unsets gateway env and recovers constrained replacement state (#4462)", () => {
     const setup = writeProxyEnvWithGuard();
     const stateDir = path.join(setup.tmpDir, "openclaw-state");
     const devicesDir = path.join(stateDir, "devices");
@@ -1103,7 +1103,7 @@ exit 1
   // existing test above only exercises `add slack`. Lock in coverage for every
   // (channel × op) combo so the guard cannot regress for any one of them
   // while passing for another.
-  it("#2592: blocks every (channel × op) mutating combo and surfaces the host-side hint", () => {
+  it("blocks every mutating channel-operation combination and surfaces the host-side hint (#2592)", () => {
     const setup = writeProxyEnvWithGuard();
     try {
       const channels = ["slack", "telegram", "discord", "wechat", "whatsapp"];
diff --git a/test/nemohermes-alias.test.ts b/test/nemohermes-alias.test.ts
index f355a1db11c..4ae5fbb4068 100644
--- a/test/nemohermes-alias.test.ts
+++ b/test/nemohermes-alias.test.ts
@@ -77,7 +77,7 @@ describe("nemohermes alias", () => {
     expect(stat.mode & 0o100).not.toBe(0);
   });
 
-  it("--version outputs nemohermes branding", () => {
+  it("outputs nemohermes branding for --version", () => {
     const { code, out } = runHermes("--version");
     expect(code).toBe(0);
     expect(out).toMatch(/^nemohermes v[\d.]+/);
diff --git a/test/ollama-tools-capability.test.ts b/test/ollama-tools-capability.test.ts
index 865ce12ecfd..ada1675eb67 100644
--- a/test/ollama-tools-capability.test.ts
+++ b/test/ollama-tools-capability.test.ts
@@ -347,7 +347,7 @@ describe("checkOllamaModelToolSupport", () => {
     }
   });
 
-  it("interactive yes → {ok:true, allowToolsIncompatible:true}", async () => {
+  it("allows a tools-incompatible model after interactive confirmation", async () => {
     const h = loadProxyWithStubs();
     h.setProbeResult({
       source: "api",
@@ -365,7 +365,7 @@ describe("checkOllamaModelToolSupport", () => {
     expect(h.promptCalls.length).toBeGreaterThan(0);
   });
 
-  it("interactive no → {ok:false} with 'choose a tools-capable model' message", async () => {
+  it("returns ok=false with guidance after interactive rejection", async () => {
     const h = loadProxyWithStubs();
     h.setProbeResult({
       source: "api",
@@ -379,7 +379,7 @@ describe("checkOllamaModelToolSupport", () => {
     expect(out.message!.toLowerCase()).toContain("tools-capable");
   });
 
-  it("non-interactive default → {ok:false} with stderr containing NEMOCLAW_OLLAMA_REQUIRE_TOOLS=0", async () => {
+  it("returns ok=false with override guidance by default in non-interactive mode", async () => {
     process.env.NEMOCLAW_NON_INTERACTIVE = "1";
     const h = loadProxyWithStubs();
     h.setProbeResult({
@@ -392,7 +392,7 @@ describe("checkOllamaModelToolSupport", () => {
     expect(h.errors.some((e) => e.includes("NEMOCLAW_OLLAMA_REQUIRE_TOOLS=0"))).toBe(true);
   });
 
-  it("non-interactive + NEMOCLAW_OLLAMA_REQUIRE_TOOLS=0 → {ok:true, allowToolsIncompatible:true} after stderr warning", async () => {
+  it("allows a tools-incompatible model with NEMOCLAW_OLLAMA_REQUIRE_TOOLS=0 after warning", async () => {
     process.env.NEMOCLAW_NON_INTERACTIVE = "1";
     process.env.NEMOCLAW_OLLAMA_REQUIRE_TOOLS = "0";
     const h = loadProxyWithStubs();
@@ -411,7 +411,7 @@ describe("checkOllamaModelToolSupport", () => {
     expect(matched).toBe(true);
   });
 
-  it("NEMOCLAW_YES=1 → {ok:true, allowToolsIncompatible:true} after note", async () => {
+  it("allows a tools-incompatible model with NEMOCLAW_YES=1 after a note", async () => {
     process.env.NEMOCLAW_YES = "1";
     const h = loadProxyWithStubs();
     h.setProbeResult({
@@ -512,7 +512,7 @@ describe("checkOllamaModelToolSupport", () => {
     expect(confirm).not.toHaveBeenCalled();
   });
 
-  it("probe failed (capabilities unknown) → {ok:true} (graceful degradation)", async () => {
+  it("returns ok=true for graceful degradation after a capability probe failure", async () => {
     const h = loadProxyWithStubs();
     h.setProbeResult({
       source: "unknown",
@@ -665,8 +665,8 @@ describe("validateOllamaModel — no-tools override propagation (#4241)", () =>
 // validateOllamaModel does not know about the override and re-rejects.
 // ─────────────────────────────────────────────────────────────────
 
-describe("override propagation across checkOllamaModelToolSupport → validateOllamaModel (#4241)", () => {
-  it("user accepts override → later validateOllamaModel does NOT reject for the same tools-incompatible error", async () => {
+describe("override propagation from checkOllamaModelToolSupport to validateOllamaModel (#4241)", () => {
+  it("does not reject the same tools-incompatible error after the user accepts an override", async () => {
     const h = loadProxyWithStubs();
     h.setProbeResult({
       source: "api",
@@ -702,7 +702,7 @@ describe("override propagation across checkOllamaModelToolSupport → validateOl
     expect(result.ok).toBe(true);
   });
 
-  it("user declines override → both stages refuse and no later validation runs", async () => {
+  it("refuses both stages without later validation after the user declines an override", async () => {
     const h = loadProxyWithStubs();
     h.setProbeResult({
       source: "api",
diff --git a/test/onboard-ollama-autostart.test.ts b/test/onboard-ollama-autostart.test.ts
index 7a35c3d8332..a627af8421e 100644
--- a/test/onboard-ollama-autostart.test.ts
+++ b/test/onboard-ollama-autostart.test.ts
@@ -346,8 +346,8 @@ process.exit = (code) => {
   return JSON.parse(lastBraceLine);
 }
 
-describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
-  it("Scenario A: stopped Ollama + flag set → no spawn, warning, falls back to DEFAULT_OLLAMA_MODEL", {
+describe("nemoclaw onboard --no-ollama-autostart (#3751)", () => {
+  it("avoids spawning stopped Ollama, warns, and falls back to DEFAULT_OLLAMA_MODEL with the flag in scenario A", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     const payload = runOllamaAutostartScenario({
@@ -403,7 +403,7 @@ describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
     );
   });
 
-  it("Scenario B: stopped Ollama + flag NOT set → existing spawn path preserved", {
+  it("preserves the existing spawn path for stopped Ollama without the flag in scenario B", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     const payload = runOllamaAutostartScenario({
@@ -436,7 +436,7 @@ describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
     );
   });
 
-  it("Scenario C (flag unset): Ollama already running → behavior unchanged, no spawn, no warning", {
+  it("leaves running Ollama unchanged without spawning or warning when the flag is unset in scenario C", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     const payload = runOllamaAutostartScenario({
@@ -465,7 +465,7 @@ describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
     assert.equal(payload.sentinelTripped, true);
   });
 
-  it("Scenario C (flag set): Ollama already running + flag set → no warning, no spawn", {
+  it("avoids warning or spawning when Ollama is running and the flag is set in scenario C", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     const payload = runOllamaAutostartScenario({
@@ -486,7 +486,7 @@ describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
     assert.equal(payload.sentinelTripped, true);
   });
 
-  it("Scenario D: non-interactive + flag set → no process.exit, warning, model = DEFAULT_OLLAMA_MODEL", {
+  it("warns and selects DEFAULT_OLLAMA_MODEL without process.exit in non-interactive scenario D", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     const payload = runOllamaAutostartScenario({
@@ -526,7 +526,7 @@ describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
     assert.equal(payload.sentinelTripped, false);
   });
 
-  it("Scenario G (#4365): pinned-provider runner crash exits instead of looping on Ollama model selection", {
+  it("exits instead of looping on Ollama model selection after a pinned-provider runner crash in scenario G (#4365)", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     // Reporter's second-step: Ollama responds, user reaches model selection,
@@ -565,7 +565,7 @@ describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
     );
   });
 
-  it("Scenario H (#4365): pinned-provider runner crash also exits when NEMOCLAW_PROVIDER uses a casing variant", {
+  it("exits after a pinned-provider runner crash with a casing variant in scenario H (#4365)", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     // NEMOCLAW_PROVIDER=OLLAMA is accepted by getNonInteractiveProvider's
@@ -597,7 +597,7 @@ describe("nemoclaw onboard --no-ollama-autostart (issue #3751)", () => {
     );
   });
 
-  it("Scenario E: stopped Ollama + flag NOT set + NEMOCLAW_PROVIDER=ollama + waitForHttp timeout → process.exit, no selectionLoop re-entry", {
+  it("exits without re-entering selectionLoop after an Ollama waitForHttp timeout in scenario E", {
     timeout: OLLAMA_AUTOSTART_TEST_TIMEOUT_MS,
   }, () => {
     // Reporter scenario: provider pinned via env, gate not set, Ollama
diff --git a/test/onboard-preset-diff.test.ts b/test/onboard-preset-diff.test.ts
index cc00fba546c..60b58d3e01b 100644
--- a/test/onboard-preset-diff.test.ts
+++ b/test/onboard-preset-diff.test.ts
@@ -128,7 +128,7 @@ const { setupPoliciesWithSelection } = require(${onboardPath});
 `;
 }
 
-describe("setupPoliciesWithSelection preset-diff (issue #2177)", () => {
+describe("setupPoliciesWithSelection preset diff (#2177)", () => {
   // In non-interactive mode a user who runs onboard twice — first with Balanced
   // defaults (applies 5 presets), second with NEMOCLAW_POLICY_PRESETS=npm —
   // expects the final sandbox to have ONLY npm. Previously-applied presets
diff --git a/test/onboard.test.ts b/test/onboard.test.ts
index 51552087d86..4057cdcb861 100644
--- a/test/onboard.test.ts
+++ b/test/onboard.test.ts
@@ -437,7 +437,7 @@ startGateway(null).catch(() => {});
     }
   });
 
-  it("#2753: ignores an incomplete session sandbox name when checking resume conflicts", () => {
+  it("ignores an incomplete session sandbox name when checking resume conflicts (#2753)", () => {
     // A pre-fix on-disk session may carry sandboxName even though the
     // sandbox step never completed. Treating that as a conflict source
     // would block users from running `--resume --name <new>` to recover.
@@ -4667,7 +4667,7 @@ const { setupInference } = require(${onboardPath});
     assert.equal(commands.length, 4);
   });
 
-  it("regression #1904: pullAndResolveBaseImageDigest uses sandbox-base registry", () => {
+  it("uses the sandbox-base registry in pullAndResolveBaseImageDigest (#1904)", () => {
     // Structural check: verify the constant matches the Dockerfile default
     // and does NOT reference the openshell-community registry.
     assert.ok(
diff --git a/test/openclaw-chat-send-patch.test.ts b/test/openclaw-chat-send-patch.test.ts
index d19ed827958..d3500b0a5bc 100644
--- a/test/openclaw-chat-send-patch.test.ts
+++ b/test/openclaw-chat-send-patch.test.ts
@@ -686,7 +686,7 @@ describe("OpenClaw chat.send compatibility patch", () => {
     }
   });
 
-  it("--audit reports all recognizers as would-apply on fresh fixtures without mutating files", () => {
+  it("reports all recognizers as would-apply on fresh fixtures without mutation for --audit", () => {
     const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-openclaw-chat-send-audit-fresh-"));
     const dist = path.join(tmp, "dist");
     fs.mkdirSync(dist);
@@ -720,7 +720,7 @@ describe("OpenClaw chat.send compatibility patch", () => {
     }
   });
 
-  it("--audit reports all recognizers as already-applied on a patched dist", () => {
+  it("reports all recognizers as already-applied on a patched dist for --audit", () => {
     const tmp = fs.mkdtempSync(
       path.join(os.tmpdir(), "nemoclaw-openclaw-chat-send-audit-applied-"),
     );
@@ -748,7 +748,7 @@ describe("OpenClaw chat.send compatibility patch", () => {
     }
   });
 
-  it("--audit exits non-zero and surfaces a per-recognizer miss without mutating files", () => {
+  it("exits non-zero and surfaces a per-recognizer miss without mutation for --audit", () => {
     const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-openclaw-chat-send-audit-miss-"));
     const dist = path.join(tmp, "dist");
     fs.mkdirSync(dist);
@@ -780,7 +780,7 @@ describe("OpenClaw chat.send compatibility patch", () => {
     }
   });
 
-  it("--audit exits non-zero and reports each missing file when selectors fail", () => {
+  it("exits non-zero and reports each missing file for --audit when selectors fail", () => {
     const tmp = fs.mkdtempSync(
       path.join(os.tmpdir(), "nemoclaw-openclaw-chat-send-audit-not-found-"),
     );
@@ -800,7 +800,7 @@ describe("OpenClaw chat.send compatibility patch", () => {
     }
   });
 
-  it("--audit rejects extra positional arguments", () => {
+  it("rejects extra positional arguments for --audit", () => {
     const result = spawnSync(process.execPath, [PATCH_SCRIPT, "--audit", "/nonexistent", "extra"], {
       encoding: "utf-8",
       timeout: 10000,
diff --git a/test/openclaw-tui-chat-correlation.test.ts b/test/openclaw-tui-chat-correlation.test.ts
index aca870d36d1..c948091f245 100644
--- a/test/openclaw-tui-chat-correlation.test.ts
+++ b/test/openclaw-tui-chat-correlation.test.ts
@@ -579,7 +579,7 @@ function runLiveIssue2603ReproWithEventCaptureRetry(sandboxName: string): LiveIs
 }
 
 describe("OpenClaw TUI chat correlation regression (#2603)", () => {
-  it("classifies the observed #2603 gateway trace as broken", () => {
+  it("classifies the observed gateway trace as broken (#2603)", () => {
     const analysis = analyzeIssue2603Trace(capturedIssue2603Trace);
 
     expect(analysis.emptyFinalsForSubmittedRuns).toEqual([
diff --git a/test/package-contract/cli/policy-dispatch.test.ts b/test/package-contract/cli/policy-dispatch.test.ts
index fef7099de84..c673473157a 100644
--- a/test/package-contract/cli/policy-dispatch.test.ts
+++ b/test/package-contract/cli/policy-dispatch.test.ts
@@ -239,7 +239,7 @@ Promise.resolve(require(${CLI_PATH}).mainPromise).finally(() => {
       expect(result.stderr).toMatch(/Aborting --from-dir/);
     });
 
-    it("--from-dir skips hidden dotfile yaml presets", () => {
+    it("skips hidden dotfile YAML presets for --from-dir", () => {
       const dir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-from-dir-hidden-"));
       fs.writeFileSync(path.join(dir, ".bad.yaml"), "preset:\n  name: bad\nnetwork_policies: {}\n");
       fs.writeFileSync(
@@ -260,7 +260,7 @@ Promise.resolve(require(${CLI_PATH}).mainPromise).finally(() => {
       expect(result.stderr).toMatch(/Directory not found/);
     });
 
-    it("--from-dir skips sub-directories whose names end in .yaml/.yml", () => {
+    it("skips subdirectories ending in .yaml or .yml for --from-dir", () => {
       const dir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-from-dir-skipdir-"));
       // A real preset file and a directory that happens to match the yaml glob.
       fs.writeFileSync(
diff --git a/test/package-contract/repro-2010.test.ts b/test/package-contract/repro-2010.test.ts
index c663af8ac5e..a20c7153215 100644
--- a/test/package-contract/repro-2010.test.ts
+++ b/test/package-contract/repro-2010.test.ts
@@ -153,7 +153,7 @@ function callGetGatewayPresets(
   return JSON.parse(stdout.trim());
 }
 
-describe("issue #2010 — policy state inconsistency", () => {
+describe("policy state inconsistency (#2010)", () => {
   describe("getGatewayPresets — matching logic", () => {
     it("returns telegram when gateway has telegram policy loaded", () => {
       const result = callGetGatewayPresets(buildGatewayYaml(["telegram"]));
diff --git a/test/package-contract/ssrf-parity.test.ts b/test/package-contract/ssrf-parity.test.ts
index 6ce37378a97..518d11c9180 100644
--- a/test/package-contract/ssrf-parity.test.ts
+++ b/test/package-contract/ssrf-parity.test.ts
@@ -295,7 +295,7 @@ describe("CLI and plugin isPrivateHostname agree on every CIDR boundary", () =>
   ];
 
   for (const [addr, expected, label] of vectors) {
-    it(`${label}: ${addr} → ${String(expected)}`, () => {
+    it(`classifies ${label} at ${addr} as ${String(expected)}`, () => {
       expect(pluginHelper.isPrivateHostname(addr)).toBe(expected);
       expect(cliHelper.isPrivateHostname(addr)).toBe(expected);
     });
@@ -332,7 +332,7 @@ describe("CLI and plugin isPrivateHostname agree on wrapper-level cases", () =>
   ];
 
   for (const [addr, expected, label] of extras) {
-    it(`${label}: ${JSON.stringify(addr)} → ${String(expected)}`, () => {
+    it(`classifies ${label} at ${JSON.stringify(addr)} as ${String(expected)}`, () => {
       expect(pluginHelper.isPrivateHostname(addr)).toBe(expected);
       expect(cliHelper.isPrivateHostname(addr)).toBe(expected);
     });
diff --git a/test/platform.test.ts b/test/platform.test.ts
index ff68772a624..cf9e6771e2f 100644
--- a/test/platform.test.ts
+++ b/test/platform.test.ts
@@ -236,7 +236,7 @@ describe("platform helpers", () => {
       });
     });
 
-    it("discovers the bare ~/.colima/docker.sock layout (regression for #3503)", () => {
+    it("discovers the bare ~/.colima/docker.sock layout (#3503)", () => {
       // The reporter's Colima setup puts the socket at the top-level
       // ~/.colima/docker.sock rather than under ~/.colima/default/. Before
       // this fix, detection returned null and the gateway fell back to
diff --git a/test/policies.test.ts b/test/policies.test.ts
index 68755169487..df83d4922cc 100644
--- a/test/policies.test.ts
+++ b/test/policies.test.ts
@@ -862,7 +862,7 @@ exit 1
   // (the installer's user-local location) but PATH from a non-interactive shell does
   // not include ~/.local/bin/, buildPolicySetCommand / buildPolicyGetCommand must
   // resolve openshell to an absolute path so spawnSync does not raise ENOENT.
-  describe("issue 4224: spawnSync openshell ENOENT in non-interactive shells", () => {
+  describe("spawnSync openshell ENOENT in non-interactive shells (#4224)", () => {
     let tmpHome: string;
     let fakeOpenshell: string;
     let origHome: string | undefined;
@@ -1003,7 +1003,7 @@ exit 1
     });
   });
 
-  describe("issue 4586: preset apply must not overwrite a live policy that could not be read", () => {
+  describe("preset apply must not overwrite a live policy that could not be read (#4586)", () => {
     const registryModule = requireForTest(
       path.join(REPO_ROOT, "src", "lib", "state", "registry.ts"),
     ) as Record<string, any>;
@@ -1104,7 +1104,7 @@ exit 1
     });
   });
 
-  describe("issue 4510: policy-add --from-file false success when the sandbox is absent from the registry", () => {
+  describe("policy-add --from-file false success when the sandbox is absent from the registry (#4510)", () => {
     const registryModule = requireForTest(
       path.join(REPO_ROOT, "src", "lib", "state", "registry.ts"),
     ) as Record<string, any>;
diff --git a/test/policy-tiers.test.ts b/test/policy-tiers.test.ts
index 1bff6d702af..df79f318903 100644
--- a/test/policy-tiers.test.ts
+++ b/test/policy-tiers.test.ts
@@ -71,7 +71,7 @@ describe("tiers", () => {
       expect(listTiers()).toHaveLength(3);
     });
 
-    it("tiers are ordered restricted → balanced → open", () => {
+    it("orders tiers as restricted, balanced, then open", () => {
       const names = listTiers().map((tier: Tier) => tier.name);
       expect(names).toEqual(["restricted", "balanced", "open"]);
     });
diff --git a/test/rebuild-credential-hydration.test.ts b/test/rebuild-credential-hydration.test.ts
index f3b50b4fe9b..3434de2a4d1 100644
--- a/test/rebuild-credential-hydration.test.ts
+++ b/test/rebuild-credential-hydration.test.ts
@@ -106,7 +106,7 @@ process.stdout.write(JSON.stringify(payload));
   return { result, tmpDir };
 }
 
-describe("Issue #2273 Layer 1: credential hydration from legacy storage", () => {
+describe("credential hydration from legacy storage, layer 1 (#2273)", () => {
   // Test each provider's credential env to ensure parametric coverage
   const providers = [
     {
diff --git a/test/rebuild-credential-preflight.test.ts b/test/rebuild-credential-preflight.test.ts
index 18a0c768413..3e3dc716bec 100644
--- a/test/rebuild-credential-preflight.test.ts
+++ b/test/rebuild-credential-preflight.test.ts
@@ -359,7 +359,7 @@ function registryHasSandbox(fixture: ReturnType<typeof createFixture>): boolean
   }
 }
 
-describe("Issue #2273: atomic rebuild", () => {
+describe("atomic rebuild (#2273)", () => {
   describe("Layer 2: preflight credential check", () => {
     it("cancels interactive rebuild before credential preflight or backup on non-affirmative input", {
       timeout: 60_000,
@@ -616,7 +616,7 @@ describe("Issue #2273: atomic rebuild", () => {
     it.each([
       ["ollama-local"],
       ["vllm-local"],
-    ])("migrates legacy %s sandbox off OPENAI_API_KEY (GH #2519)", (provider) => {
+    ])("migrates a legacy %s sandbox off OPENAI_API_KEY (#2519)", (provider) => {
       // Pre-fix sandboxes recorded credentialEnv="OPENAI_API_KEY" even
       // though local inference never actually needed it. After the fix,
       // the wizard records null. Rebuild must accept the legacy value,
diff --git a/test/rebuild-shields-auto-unlock.test.ts b/test/rebuild-shields-auto-unlock.test.ts
index cb510657aae..8aebb38ee11 100644
--- a/test/rebuild-shields-auto-unlock.test.ts
+++ b/test/rebuild-shields-auto-unlock.test.ts
@@ -308,7 +308,7 @@ function runRebuild(fixture: ReturnType<typeof createFixture>) {
   );
 }
 
-describe("Issue #3113: rebuild auto-unlocks when shields are UP", () => {
+describe("rebuild auto-unlocks when shields are UP (#3113)", () => {
   it("detects locked shields and prints auto-unlock notice", { timeout: 60_000 }, () => {
     const f = createFixture({ shieldsLocked: true });
     const r = runRebuild(f);
diff --git a/test/rebuild-stale-recovery.test.ts b/test/rebuild-stale-recovery.test.ts
index 945d6528dbf..23f85e44e05 100644
--- a/test/rebuild-stale-recovery.test.ts
+++ b/test/rebuild-stale-recovery.test.ts
@@ -205,7 +205,7 @@ function registryHasSandbox(fixture: { nemoclawDir: string; sandboxName: string
   }
 }
 
-describe("Issue #4497: stale sandbox rebuild recovery", () => {
+describe("stale sandbox rebuild recovery (#4497)", () => {
   it("does NOT abort with 'Cannot back up state' when the live sandbox is gone", {
     timeout: 90_000,
   }, () => {
diff --git a/test/repro-1751-extra.test.ts b/test/repro-1751-extra.test.ts
index 110a5d1c053..74912a8b917 100644
--- a/test/repro-1751-extra.test.ts
+++ b/test/repro-1751-extra.test.ts
@@ -40,7 +40,7 @@ afterEach(() => {
   }
 });
 
-describe("Issue #1751 — GPU passthrough session persistence", () => {
+describe("GPU passthrough session persistence (#1751)", () => {
   it("filterSafeUpdates: gpuPassthrough=true is propagated to safe", () => {
     session.saveSession(session.createSession());
     markStepCompleteLegacy(session, stepMutation, "provider_selection", { gpuPassthrough: true });
diff --git a/test/repro-2201.test.ts b/test/repro-2201.test.ts
index 7efc4dac9f7..ccf100709a0 100644
--- a/test/repro-2201.test.ts
+++ b/test/repro-2201.test.ts
@@ -316,7 +316,7 @@ function readSessionMessagingPlan(
   return readSession(fixture).messagingPlan;
 }
 
-describe("Issue #2201: rebuild syncs agent from registry, not stale session", () => {
+describe("rebuild syncs agent from registry instead of a stale session (#2201)", () => {
   it("rebuild openclaw after hermes was onboarded last (reporter scenario)", {
     timeout: 60_000,
   }, () => {
@@ -361,7 +361,7 @@ describe("Issue #2201: rebuild syncs agent from registry, not stale session", ()
   });
 });
 
-describe("Issue #2301: rebuild forwards stored --from Dockerfile to onboard", () => {
+describe("rebuild forwards the stored --from Dockerfile to onboard (#2301)", () => {
   it("rebuild does not hit fromDockerfile conflict when session has a stored --from path", {
     timeout: 60_000,
   }, () => {
diff --git a/test/repro-2376.test.ts b/test/repro-2376.test.ts
index bd27a552c63..449afd1515d 100644
--- a/test/repro-2376.test.ts
+++ b/test/repro-2376.test.ts
@@ -59,7 +59,7 @@ function runRcFile(
   }
 }
 
-describe("Issue #2376: Hermes rc files source HERMES_HOME from proxy-env", () => {
+describe("Hermes rc files source HERMES_HOME from proxy-env (#2376)", () => {
   for (const rcFileName of [".bashrc", ".profile"] as const) {
     it(`${rcFileName} exports HERMES_HOME when proxy-env exists`, () => {
       const out = runRcFile(rcFileName, "export HERMES_HOME=/sandbox/.hermes\n");
diff --git a/test/repro-2666-silent-list-status.test.ts b/test/repro-2666-silent-list-status.test.ts
index 3786322d166..5a8a0251f05 100644
--- a/test/repro-2666-silent-list-status.test.ts
+++ b/test/repro-2666-silent-list-status.test.ts
@@ -67,7 +67,7 @@ function buildDepsWithThrowingRecovery(): ListSandboxesCommandDeps {
   };
 }
 
-describe("#2666 — silent empty output regression", () => {
+describe("silent empty output regression (#2666)", () => {
   it("nemoclaw list renders the registry-only listing when recovery fails", async () => {
     const deps = buildDepsWithThrowingRecovery();
     const inventory = await getSandboxInventory(deps);
@@ -90,7 +90,7 @@ describe("#2666 — silent empty output regression", () => {
   });
 });
 
-describe("#2666 — list-command-deps resilience wrapper", () => {
+describe("list-command-deps resilience wrapper (#2666)", () => {
   // Exercises the actual exported `recoverRegistryEntriesWithFallback` from
   // src/lib/list-command-deps.ts, not a parallel re-implementation. If the
   // production wrapper regresses, these tests fail.
@@ -145,7 +145,7 @@ describe("#2666 — list-command-deps resilience wrapper", () => {
   });
 });
 
-describe("#2666 — subprocess regression: simulated (container-stopped + foreign-port-holder)", () => {
+describe("simulated container-stopped and foreign-port-holder subprocess regression (#2666)", () => {
   // End-to-end test that runs the real `nemoclaw` binary against a fake
   // `openshell` shell script simulating the bug repro: the openshell sandbox
   // container is stopped AND a foreign listener holds port 8080. In that
diff --git a/test/repro-2749-extra.test.ts b/test/repro-2749-extra.test.ts
index 1791712e748..c069cad47fa 100644
--- a/test/repro-2749-extra.test.ts
+++ b/test/repro-2749-extra.test.ts
@@ -43,8 +43,8 @@ function listPolicyFiles(): string[] {
   return files;
 }
 
-describe("Issue #2749 — additional coverage on top of existing tls:terminate guard", () => {
-  it("PARSE SAFETY: every policy YAML input still parses after deletions", () => {
+describe("keeps policy YAML valid while removing `tls: terminate` (#2749)", () => {
+  it("parses every policy YAML input after deletions", () => {
     for (const file of listPolicyFiles()) {
       const content = fs.readFileSync(file, "utf-8");
       // js-yaml throws on syntactic damage (dangling list markers, broken
@@ -57,7 +57,7 @@ describe("Issue #2749 — additional coverage on top of existing tls:terminate g
     }
   });
 
-  it("OVER-DELETION GUARD: `tls: skip` entries for WS pass-through are preserved", () => {
+  it("preserves `tls: skip` entries for WebSocket pass-through", () => {
     // The PR removes `tls: terminate` (deprecated) but the body explicitly
     // calls out that `tls: skip` for WebSocket pass-through should stay.
     // Confirm at least one built-in preset still has `tls: skip` so a
diff --git a/test/repro-4538-raw-doctor-perms.test.ts b/test/repro-4538-raw-doctor-perms.test.ts
index f08c6b4529c..efa643f5133 100644
--- a/test/repro-4538-raw-doctor-perms.test.ts
+++ b/test/repro-4538-raw-doctor-perms.test.ts
@@ -121,7 +121,7 @@ function seedTightenedConfigTree(): { tmpDir: string; configDir: string; configF
   return { tmpDir, configDir, configFile };
 }
 
-describe("#4538 raw `openclaw doctor --fix` mutable-perm restore", () => {
+describe("raw `openclaw doctor --fix` mutable-perm restore (#4538)", () => {
   const src = fs.readFileSync(START_SCRIPT, "utf-8");
 
   it("restore helper re-asserts 2770/660 after the tree is tightened to 700/600", () => {
@@ -362,7 +362,7 @@ function resolveSandboxImage(): string | null {
 }
 
 describe.skipIf(!RUN_DOCKER_E2E || !dockerAvailable())(
-  "#4538 reporter workflow E2E (real sandbox image, raw openclaw doctor --fix)",
+  "reporter workflow E2E with a real sandbox image and raw openclaw doctor --fix (#4538)",
   () => {
     it("restores 2770/660 after a connect-shell `openclaw doctor --fix`", () => {
       const image = resolveSandboxImage();
diff --git a/test/runner.test.ts b/test/runner.test.ts
index aa2d5881210..bef0190f7be 100644
--- a/test/runner.test.ts
+++ b/test/runner.test.ts
@@ -229,7 +229,7 @@ describe("runner env merging", () => {
     expect(firstCall[2]?.env?.PATH).toBe("/usr/local/bin:/usr/bin");
   });
 
-  it("#2616: runCaptureEx injects NO_PROXY=localhost,127.0.0.1 when http_proxy is set", () => {
+  it("injects NO_PROXY=localhost,127.0.0.1 in runCaptureEx when http_proxy is set (#2616)", () => {
     // Regression for the macOS Privoxy scenario: validateOllamaModel calls
     // runCaptureEx with a curl probe against http://localhost:11434. Before
     // the fix, runCaptureEx merged raw process.env (including the user's
diff --git a/test/sandbox-init.test.ts b/test/sandbox-init.test.ts
index c7386ec5835..a5dd71e7e65 100644
--- a/test/sandbox-init.test.ts
+++ b/test/sandbox-init.test.ts
@@ -475,7 +475,7 @@ EOF
     // Default (no NEMOCLAW_REQUIRE_CAP_DROP): warns and CONTINUES even though
     // dangerous caps remain — preserving the zero-regression posture for
     // CAP_SETPCAP-less hosts. report_residual_capabilities still names them.
-    it("warns but does NOT refuse to start when CAP_SETPCAP is unavailable (issue #3280)", () => {
+    it("warns without refusing to start when CAP_SETPCAP is unavailable (#3280)", () => {
       const { stdout } = runWithLib(
         [
           "TMP=$(mktemp -d)",
diff --git a/test/sandbox-provisioning.test.ts b/test/sandbox-provisioning.test.ts
index be21d9d7dda..848fa5d277a 100644
--- a/test/sandbox-provisioning.test.ts
+++ b/test/sandbox-provisioning.test.ts
@@ -1338,7 +1338,7 @@ describe("Hermes sandbox provisioning", () => {
     return { result, tmp };
   }
 
-  it("regression #4230: installs Hermes' native Anthropic provider dependency", () => {
+  it("installs Hermes' native Anthropic provider dependency (#4230)", () => {
     const { result, tmp } = runHermesUvExtrasExpansion();
     try {
       expect(result.status).toBe(0);
diff --git a/test/security-sandbox-tar-traversal.test.ts b/test/security-sandbox-tar-traversal.test.ts
index a8693d1c5c5..12c8308ae5d 100644
--- a/test/security-sandbox-tar-traversal.test.ts
+++ b/test/security-sandbox-tar-traversal.test.ts
@@ -391,7 +391,7 @@ describe("Fix: safeTarExtract blocks malicious archives and extracts safe ones",
   // is extracted on the host, these absolute targets don't exist on the host
   // and were falsely rejected as escapes. The fix maps /sandbox/ paths onto
   // the extraction root before checking, matching the sandbox-internal view.
-  it("regression #2317: allows known-safe /sandbox/.openclaw-data symlinks in backup archives", async () => {
+  it("allows known-safe /sandbox/.openclaw-data symlinks in backup archives (#2317)", async () => {
     const { safeTarExtract } = await loadSandboxState();
     const workDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-2317-"));
     try {
@@ -414,7 +414,7 @@ describe("Fix: safeTarExtract blocks malicious archives and extracts safe ones",
     }
   });
 
-  it("regression #2317: still blocks absolute symlinks outside /sandbox/.openclaw-data", async () => {
+  it("still blocks absolute symlinks outside /sandbox/.openclaw-data (#2317)", async () => {
     const { safeTarExtract } = await loadSandboxState();
     const workDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-2317-block-"));
     try {
@@ -515,7 +515,7 @@ describe("Fix: safeTarExtract blocks malicious archives and extracts safe ones",
     }
   });
 
-  it("regression #2317: blocks path traversal within allowed prefix (/sandbox/.openclaw-data/../../etc/passwd)", async () => {
+  it("blocks path traversal within the allowed /sandbox/.openclaw-data prefix (#2317)", async () => {
     const { safeTarExtract } = await loadSandboxState();
     const workDir = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-2317-traversal-"));
     try {
diff --git a/test/service-env.test.ts b/test/service-env.test.ts
index 128a49d944d..ab73f6e57a7 100644
--- a/test/service-env.test.ts
+++ b/test/service-env.test.ts
@@ -176,7 +176,7 @@ describe("service environment", () => {
     });
   });
 
-  describe("GIT_SSL_CAINFO for proxy CA trust (issue #2270)", () => {
+  describe("GIT_SSL_CAINFO for proxy CA trust (#2270)", () => {
     const sandboxInitSource = `source ${JSON.stringify(join(import.meta.dirname, "../scripts/lib/sandbox-init.sh"))}`;
 
     it("entrypoint exports GIT_SSL_CAINFO when SSL_CERT_FILE points to a real file", () => {
@@ -380,7 +380,7 @@ describe("service environment", () => {
     });
   });
 
-  describe("XDG and tool cache redirects (issue #804)", () => {
+  describe("XDG and tool cache redirects (#804)", () => {
     it("entrypoint pre-creates redirected dirs and restricts GNUPGHOME permissions", () => {
       const scriptPath = join(import.meta.dirname, "../scripts/nemoclaw-start.sh");
       const src = readFileSync(scriptPath, "utf-8");
@@ -438,7 +438,7 @@ describe("service environment", () => {
     });
   });
 
-  describe("proxy environment variables (issue #626)", () => {
+  describe("proxy environment variables (#626)", () => {
     // The proxy persistence block calls emit_sandbox_sourced_file from the
     // shared library. Wrappers that execute the extracted block must source it.
     const sandboxInitSource = `source ${JSON.stringify(join(import.meta.dirname, "../scripts/lib/sandbox-init.sh"))}`;
@@ -1098,7 +1098,7 @@ describe("service environment", () => {
       }
     });
 
-    it("[simulation] sourcing proxy-env.sh overrides narrow NO_PROXY and no_proxy", () => {
+    it("overrides narrow NO_PROXY and no_proxy while sourcing proxy-env.sh in simulation", () => {
       const fakeDataDir = mkdtempSync(join(tmpdir(), "nemoclaw-bashi-test-"));
       try {
         const envContent = [
@@ -1138,7 +1138,7 @@ describe("service environment", () => {
       }
     });
 
-    it("regression #2109: proxy-env.sh includes NODE_OPTIONS --require when NODE_USE_ENV_PROXY=1", () => {
+    it("includes NODE_OPTIONS --require in proxy-env.sh when NODE_USE_ENV_PROXY=1 (#2109)", () => {
       const fakeDataDir = join(tmpdir(), `nemoclaw-http-fix-test-${process.pid}`);
       execFileSync("mkdir", ["-p", fakeDataDir]);
       const tmpFile = join(tmpdir(), `nemoclaw-http-fix-env-${process.pid}.sh`);
@@ -1182,7 +1182,7 @@ describe("service environment", () => {
       }
     });
 
-    it("regression #2109: proxy-env.sh does NOT include NODE_OPTIONS when NODE_USE_ENV_PROXY is unset", () => {
+    it("omits NODE_OPTIONS from proxy-env.sh when NODE_USE_ENV_PROXY is unset (#2109)", () => {
       const fakeDataDir = join(tmpdir(), `nemoclaw-http-noop-test-${process.pid}`);
       execFileSync("mkdir", ["-p", fakeDataDir]);
       const tmpFile = join(tmpdir(), `nemoclaw-http-noop-env-${process.pid}.sh`);
diff --git a/test/snapshot-restore-existing-dest.test.ts b/test/snapshot-restore-existing-dest.test.ts
index 938642594cd..fc30cafb93a 100644
--- a/test/snapshot-restore-existing-dest.test.ts
+++ b/test/snapshot-restore-existing-dest.test.ts
@@ -242,7 +242,7 @@ describe("snapshot restore --to existing destination (#3756)", () => {
     expect(log).not.toMatch(/sandbox delete dst/);
   });
 
-  it("refuses by default before running source-image preflight (Codex #3796 P2)", () => {
+  it("refuses by default before running source-image preflight per Codex P2 review (#3796)", () => {
     // Existing destination + unresolvable source image. The user must see the
     // precise "destination exists" error, not the "cannot resolve image"
     // misdirection that would land if the refusal came after preflight.
diff --git a/test/stale-dist-check.test.ts b/test/stale-dist-check.test.ts
index 6d6c8774c3f..9014ca70784 100644
--- a/test/stale-dist-check.test.ts
+++ b/test/stale-dist-check.test.ts
@@ -80,7 +80,7 @@ describe("stale-dist-check", () => {
     expect(checkStaleDist(root)).toBeNull();
   });
 
-  it("warnIfStale writes a build:cli hint mentioning #1958", () => {
+  it("writes a build:cli hint mentioning the tracked issue in warnIfStale (#1958)", () => {
     writeFile(path.join(root, "dist", "lib", "foo.js"), "x", 1_000_000);
     writeFile(path.join(root, "src", "lib", "foo.ts"), "x", 5_000_000);
     const chunks: string[] = [];
diff --git a/test/test-title-style.test.ts b/test/test-title-style.test.ts
new file mode 100644
index 00000000000..81b6f35acda
--- /dev/null
+++ b/test/test-title-style.test.ts
@@ -0,0 +1,68 @@
+// SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// SPDX-License-Identifier: Apache-2.0
+
+import { describe, expect, it } from "vitest";
+
+import { scanTestTitleStyle } from "../scripts/checks/test-title-style";
+
+function rulesFor(source: string): string[] {
+  return scanTestTitleStyle("test/virtual-title-style.test.ts", source).map(
+    (violation) => violation.rule,
+  );
+}
+
+describe("enforces behavior-oriented Vitest titles", () => {
+  it("detects issue-first, metadata-first, placeholder-only, and arrow-label titles", () => {
+    const rules = rulesFor(`
+      import { describe, it } from "vitest";
+      describe("issue #1234 spline behavior", () => {
+        it("#1234: fixes splines", () => {});
+        it("--force bypasses validation", () => {});
+        it("Scenario A: spline fixed", () => {});
+        it.each([["spline"]])("%s", () => {});
+        it("input → output", () => {});
+      });
+    `);
+
+    expect(rules).toEqual([
+      "issue-reference-suffix",
+      "leading-metadata",
+      "issue-reference-suffix",
+      "leading-metadata",
+      "leading-metadata",
+      "leading-metadata",
+      "placeholder-only",
+      "result-arrow",
+    ]);
+  });
+
+  it("accepts behavior-oriented titles and final issue suffixes through Vitest modifiers", () => {
+    const violations = scanTestTitleStyle(
+      "test/virtual-title-style.test.ts",
+      `
+        import { describe, it } from "vitest";
+        describe.skipIf(false)("spline behavior (#1234)", () => {
+          it("reticulates splines correctly (#1234)", () => {});
+          it.each([["cubic"]])("reticulates %s splines", () => {});
+        });
+      `,
+    );
+
+    expect(violations).toEqual([]);
+  });
+
+  it("ignores external repository issue references and nonliteral titles", () => {
+    const violations = scanTestTitleStyle(
+      "test/virtual-title-style.test.ts",
+      `
+        import { describe, it } from "vitest";
+        const generated = "generated elsewhere";
+        describe("upstream owner/repo#123 behavior", () => {
+          it(generated, () => {});
+        });
+      `,
+    );
+
+    expect(violations).toEqual([]);
+  });
+});
diff --git a/test/uninstall.test.ts b/test/uninstall.test.ts
index ec9f99058f2..9ba76ea4e5e 100644
--- a/test/uninstall.test.ts
+++ b/test/uninstall.test.ts
@@ -19,7 +19,7 @@ describe("uninstall CLI flags", () => {
     }
   }
 
-  it("--help exits 0 and shows usage", () => {
+  it("exits 0 and shows usage for --help", () => {
     const result = spawnSync("bash", [UNINSTALL_SCRIPT, "--help"], {
       cwd: path.join(import.meta.dirname, ".."),
       encoding: "utf-8",
@@ -33,7 +33,7 @@ describe("uninstall CLI flags", () => {
     expect(output).toMatch(/--delete-models/);
   });
 
-  it("--help uses NemoHermes branding when Hermes is the active agent", () => {
+  it("uses NemoHermes branding for --help when Hermes is active", () => {
     const result = spawnSync("bash", [UNINSTALL_SCRIPT, "--help"], {
       cwd: path.join(import.meta.dirname, ".."),
       encoding: "utf-8",
@@ -52,7 +52,7 @@ describe("uninstall CLI flags", () => {
     expect(output).not.toMatch(/NemoClaw Uninstaller/);
   });
 
-  it("--yes skips the confirmation prompt and completes successfully", () => {
+  it("skips the confirmation prompt and completes successfully for --yes", () => {
     const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "nemoclaw-uninstall-yes-"));
     const fakeBin = path.join(tmp, "bin");
     writeFakeTools(fakeBin);
@@ -81,7 +81,7 @@ describe("uninstall CLI flags", () => {
     }
   }, 60_000);
 
-  it("--yes uses NemoHermes branding when Hermes is the active agent", () => {
+  it("uses NemoHermes branding for --yes when Hermes is active", () => {
     const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "nemohermes-uninstall-yes-"));
     const fakeBin = path.join(tmp, "bin");
     writeFakeTools(fakeBin);
diff --git a/test/validate-blueprint.test.ts b/test/validate-blueprint.test.ts
index 76fbad0e1e2..70db9adfc94 100644
--- a/test/validate-blueprint.test.ts
+++ b/test/validate-blueprint.test.ts
@@ -133,7 +133,7 @@ describe("blueprint.yaml", () => {
     expect(Object.keys(defined ?? {}).length).toBeGreaterThan(0);
   });
 
-  it("regression #1438: sandbox image is pinned by digest, not by mutable tag", () => {
+  it("pins the sandbox image by digest instead of a mutable tag (#1438)", () => {
     // The blueprint MUST NOT pull a sandbox image by a mutable tag like
     // ":latest" — a registry compromise or accidental force-push could
     // silently swap the image. Pin via @sha256:... so the image cannot
@@ -151,7 +151,7 @@ describe("blueprint.yaml", () => {
     expect(digestMatch).not.toBeNull();
   });
 
-  it("regression #1438: top-level digest field is populated and matches the image digest", () => {
+  it("populates the top-level digest field with the image digest (#1438)", () => {
     // The top-level `digest:` field at the top of blueprint.yaml is
     // documented as "Computed at release time" and was empty on main,
     // which left blueprint-level integrity unverifiable. Mirror the
@@ -205,12 +205,12 @@ describe("blueprint.yaml", () => {
 describe("Model Router pool config", () => {
   const pool = loadYaml<RouterPoolConfig>(ROUTER_POOL_CONFIG_PATH);
 
-  it("regression #3255: routes NVIDIA API keys to the public NVIDIA inference endpoint", () => {
+  it("routes NVIDIA API keys to the public NVIDIA inference endpoint (#3255)", () => {
     const apiBases = new Set((pool.models ?? []).map((model) => model.api_base));
     expect(apiBases).toEqual(new Set(["https://integrate.api.nvidia.com/v1"]));
   });
 
-  it("regression #3255: uses valid LiteLLM NVIDIA model identifiers", () => {
+  it("uses valid LiteLLM NVIDIA model identifiers (#3255)", () => {
     const modelsByName = new Map(
       (pool.models ?? []).map((model) => [model.name, model.litellm_model]),
     );
@@ -313,12 +313,12 @@ describe("base sandbox policy", () => {
     return out;
   }
 
-  it("regression #1437: base policy does not expose sentry.io by default", () => {
+  it("does not expose sentry.io in the base policy by default (#1437)", () => {
     const sentryEndpoints = findEndpoints((h) => h === "sentry.io");
     expect(sentryEndpoints).toEqual([]);
   });
 
-  it("regression #1583: base policy does not silently grant GitHub access", () => {
+  it("does not silently grant GitHub access in the base policy (#1583)", () => {
     // Until #1583, github.com / api.github.com plus the git/gh
     // binaries lived in network_policies and were therefore included
     // in every sandbox regardless of user opt-in. The fix moves the
@@ -336,7 +336,7 @@ describe("base sandbox policy", () => {
     expect(githubHosts).toEqual([]);
   });
 
-  it("regression #2663: managed_inference policy allows inference.local:443 GET and POST", () => {
+  it("allows inference.local:443 GET and POST in the managed_inference policy (#2663)", () => {
     // inference.local is the OpenShell gateway's managed inference virtual
     // hostname — the gateway proxies it to the configured provider (OpenAI,
     // NVIDIA, etc.). Every sandbox uses this route regardless of provider.
@@ -360,7 +360,7 @@ describe("base sandbox policy", () => {
     expect(hasPost).toBe(true);
   });
 
-  it("regression #2663: managed_inference allows openclaw and tool binaries", () => {
+  it("allows openclaw and tool binaries in the managed_inference policy (#2663)", () => {
     const np = policy.network_policies ?? {};
     const binaries = (np.managed_inference?.binaries ?? []).map((b) => b.path).sort();
     expect(binaries).toEqual([
@@ -377,7 +377,7 @@ describe("base sandbox policy", () => {
     expect(serialized).not.toContain("/usr/local/bin/claude");
   });
 
-  it("regression #2180: base policy does not silently grant Telegram access", () => {
+  it("does not silently grant Telegram access in the base policy (#2180)", () => {
     // Until #1705 (later regressed by #1700 and re-surfaced in #2180),
     // `api.telegram.org` plus a /usr/local/bin/node binary lived in the
     // base network_policies, so every sandbox could call the Telegram
@@ -393,7 +393,7 @@ describe("base sandbox policy", () => {
     expect(telegramHosts).toEqual([]);
   });
 
-  it("regression #2180: base policy does not silently grant Discord access", () => {
+  it("does not silently grant Discord access in the base policy (#2180)", () => {
     // Parallel to the Telegram regression above. Discord (discord.com,
     // gateway.discord.gg, cdn.discordapp.com, media.discordapp.net) is
     // the opt-in preset path, not baseline. Re-adding these endpoints
@@ -413,7 +413,7 @@ describe("base sandbox policy", () => {
     expect(discordHosts).toEqual([]);
   });
 
-  it("regression #2180: base policy does not silently grant Slack access", () => {
+  it("does not silently grant Slack access in the base policy (#2180)", () => {
     // Slack was never in the baseline, but guard against it being added
     // in the same merge-conflict-resolution pattern that re-added
     // Telegram and Discord after #1705. Slack access is in
@@ -431,7 +431,7 @@ describe("base sandbox policy", () => {
     expect(slackHosts).toEqual([]);
   });
 
-  it("regression #1458: baseline npm_registry must not include npm or node binaries", () => {
+  it("omits npm and node binaries from the baseline npm_registry policy (#1458)", () => {
     const np = policy.network_policies ?? {};
     const npmRegistry = np.npm_registry;
     expect(npmRegistry).toBeDefined();
@@ -485,7 +485,7 @@ describe("permissive sandbox policy", () => {
     expect(policy.network_policies).toBeDefined();
   });
 
-  it("regression #2513: managed_inference block allows inference.local:443", () => {
+  it("allows inference.local:443 in the managed_inference block (#2513)", () => {
     const np = policy.network_policies ?? {};
     expect(np.managed_inference).toBeDefined();
     const endpoints = np.managed_inference?.endpoints ?? [];
@@ -499,7 +499,7 @@ describe("permissive sandbox policy", () => {
     expect(inferenceEp?.enforcement).toBe("enforce");
   });
 
-  it("regression #2513: managed_inference uses permissive '/**' binary allowlist", () => {
+  it("uses a permissive '/**' binary allowlist for managed_inference (#2513)", () => {
     const np = policy.network_policies ?? {};
     const binaries = (np.managed_inference?.binaries ?? []).map((b) => b.path);
     // Matches the permissive-file convention used by every other block
@@ -541,7 +541,7 @@ describe("Hermes sandbox policy", () => {
     ]);
   }
 
-  it("regression #4230: managed_inference keeps a narrow inference API allowlist", () => {
+  it("keeps a narrow inference API allowlist for managed_inference (#4230)", () => {
     expectManagedInferenceSecurityShape();
   });
 
@@ -567,7 +567,7 @@ describe("github preset", () => {
     import.meta.url,
   );
 
-  it("regression #1583: github preset file exists and parses", () => {
+  it("parses the existing github preset file (#1583)", () => {
     const parsed = loadYaml<PolicyPreset>(PRESET_PATH);
     expect(parsed).toEqual(expect.objectContaining({}));
     const meta = parsed.preset;
@@ -576,7 +576,7 @@ describe("github preset", () => {
     expect(np && "github" in np).toBe(true);
   });
 
-  it("regression #2179: github preset only advertises the installed git binary", () => {
+  it("only advertises the installed git binary in the github preset (#2179)", () => {
     const parsed = loadYaml<PolicyPreset>(PRESET_PATH);
     const meta = parsed.preset;
     expect(meta?.description).toBe("GitHub.com and GitHub API access (git)");
@@ -612,7 +612,7 @@ describe("huggingface preset", () => {
     return Array.isArray(hf?.endpoints) ? hf.endpoints : [];
   }
 
-  it("regression #1432: huggingface.co has no POST allow rule", () => {
+  it("omits POST allow rules for huggingface.co (#1432)", () => {
     const endpoints = presetEndpoints().filter((ep) => ep.host === "huggingface.co");
     expect(endpoints.length).toBeGreaterThan(0);
     for (const ep of endpoints) {
@@ -628,7 +628,7 @@ describe("huggingface preset", () => {
     }
   });
 
-  it("regression #1432: huggingface.co retains GET so downloads still work", () => {
+  it("retains GET for huggingface.co so downloads still work (#1432)", () => {
     const endpoints = presetEndpoints().filter((ep) => ep.host === "huggingface.co");
     for (const ep of endpoints) {
       const rules = Array.isArray(ep.rules) ? ep.rules : [];
@@ -651,7 +651,7 @@ describe("jira preset", () => {
   );
   const jiraPreset = loadYaml<PolicyPreset>(JIRA_PRESET_PATH);
 
-  it("regression #3758: Jira allows Node but not curl", () => {
+  it("allows Node but not curl for Jira (#3758)", () => {
     const binaries = (jiraPreset.network_policies?.atlassian?.binaries ?? [])
       .map((binary) => binary.path)
       .sort();
@@ -766,7 +766,7 @@ describe("npm preset", () => {
   const REGISTRY_HOSTS = ["registry.npmjs.org", "registry.yarnpkg.com"];
 
   for (const host of REGISTRY_HOSTS) {
-    it(`regression #2767: ${host} uses L4 tunnel (access: full, tls: skip) for CONNECT compatibility`, () => {
+    it(`uses an L4 tunnel for CONNECT compatibility on ${host} (access: full, tls: skip) (#2767)`, () => {
       const endpoints = npmEndpoints().filter((ep) => ep.host === host);
       expect(endpoints.length).toBeGreaterThan(0);
       for (const ep of endpoints) {
diff --git a/test/wsl2-probe-timeout.test.ts b/test/wsl2-probe-timeout.test.ts
index aabb8c35582..60aa6b76888 100644
--- a/test/wsl2-probe-timeout.test.ts
+++ b/test/wsl2-probe-timeout.test.ts
@@ -30,7 +30,7 @@ if (!isOnboardValidationInternals(onboardValidationInternals)) {
 }
 const { getValidationProbeCurlArgs } = onboardValidationInternals;
 
-describe("WSL2 inference verification timeouts (issue #987)", () => {
+describe("WSL2 inference verification timeouts (#987)", () => {
   describe("getValidationProbeCurlArgs", () => {
     it("returns standard timeouts on non-WSL platforms", () => {
       expect(getValidationProbeCurlArgs({ isWsl: false })).toEqual([