fix: Dynamic Context Compaction by oyi77 · Pull Request #1368 · diegosouzapw/OmniRoute

oyi77 · 2026-04-17T13:53:37Z

This PR is a clean replacement for PR #1363, correctly rebased onto the latest main branch.

Work Included:

Code Review Fixes Applied: Added missing crypto imports, corrected memory FTS5 logic, hardened database encryption against fallback failures, fixed dangerous JSON string truncation in the Perplexity executor, and eliminated timeout resource leaks in the context handoff service.
Dynamic Context Limits: Context compaction tokens are now calculated dynamically (40% of context window) using the models.dev API integration instead of a hardcoded 8,000 limit. This ensures safe truncation for large context models like Opus or Gemini 1M.
Handle Mixed Context Sizes in Combos: When routing through a combo, the minimum context limit across all models in the combo is used for summarization, guaranteeing compatibility during fallback scenarios.

…ent literal \n\n artifacts - Changed regex quantifier from ? to * in combo.ts, comboAgentMiddleware.ts, and contextHandoff.ts to greedily strip all JSON-escaped newline sequences surrounding <omniModel> tags in SSE streaming chunks - Added \r to the character class for cross-platform robustness - Fixed Playwright strict-mode violation in combo-unification.spec.ts - Bumped OpenAPI version and CHANGELOG to 3.6.6

…w#1187/diegosouzapw#1218, diegosouzapw#1202) - fix(gemini): strip VS Code JSON Schema extensions from tool schemas (diegosouzapw#1175) Add enumDescriptions, markdownDescription, markdownEnumDescriptions, enumItemLabels and tags to UNSUPPORTED_SCHEMA_CONSTRAINTS so the Gemini sanitizer removes them before forwarding. GitHub Copilot injects these non-standard fields into tool definitions, causing Gemini to reject with 'Unknown name enumDescriptions at functionDeclarations[n].parameters'. - fix(health-check): unwrap proxy config object before passing to getAccessToken (diegosouzapw#1187 diegosouzapw#1218) resolveProxyForConnection() returns { proxy, level, levelId } but the health check loop was passing the full wrapper to getAccessToken(), which expects the inner config object (.host, .port etc). The proxy dispatcher validated .host on the wrapper (undefined) and threw 'Context proxy host is required', silently marking every connection as unhealthy every sweep. Fix mirrors the pattern already used in chatHelpers.ts: proxyResult?.proxy || null. - fix(ui): debounce models.dev sync interval slider to save only on release (diegosouzapw#1202) The slider's onChange fired updateInterval() on every drag tick, sending a PATCH per pixel of movement. Rapid API responses overwrote UI state mid-drag. Introduce draftIntervalHours for smooth visual feedback; the PATCH fires on onMouseUp / onBlur once the user releases the control.

…#1238) Integrated into release/v3.6.6

…iegosouzapw#1230) Integrated into release/v3.6.6

…diegosouzapw#1228) Integrated into release/v3.6.6

…egosouzapw#1227) Integrated into release/v3.6.6

…ouzapw#1235)

diegosouzapw#1220, diegosouzapw#1231) - fix(core): diegosouzapw#1206 inject startup guard against app/ and src/app/ conflict - fix(health): diegosouzapw#1220 add HEALTHCHECK_STAGGER_MS to prevent token refresh bursting - fix(proxy): diegosouzapw#1231 prioritize HTTP 429 over quota body heuristics - fix(sse): diegosouzapw#1211 strip leading double-newlines in responses API stream

… from PR overlaps

…egosouzapw#1165, diegosouzapw#1177)

Update package versions for the electron app and open-sse package. Sync llm.txt metadata and feature headings with the 3.6.6 release.

Add guarded outbound fetch helpers with private/local URL blocking, controlled retries, timeout normalization, and route-level status propagation for provider validation and model discovery. Introduce cooldown-aware chat retries with configurable requestRetry and maxRetryIntervalSec settings, model-scoped cooldown responses, and improved rate-limit learning from headers and error bodies so short upstream lockouts can recover automatically. Also align Antigravity and Codex header handling, require API keys for Pollinations, validate web runtime env at startup, restore sanitized Gemini tool names in translated responses, and inject a synthetic Claude text block when upstream SSE completes empty.

Introduce GLM Thinking as a first-class provider preset with shared GLM model metadata, pricing, usage sync, dashboard support, and provider request defaults for higher token budgets and longer timeouts. Use provider-side /messages/count_tokens when a Claude-compatible upstream supports it, while preserving estimated fallback behavior for missing models, missing credentials, and upstream failures. Also add startup seeding for default model aliases and normalize common cross-proxy model dialects so canonical slashful model ids do not get misrouted during resolution.

Add dedicated sync token storage, issuance, revocation, and bundle download routes backed by stable config bundle versioning and ETag support. Expose the v1 websocket handshake route and custom Next server bridge so OpenAI-compatible websocket traffic can be upgraded and proxied through the dashboard and API bridge. Expand compliance auditing with structured metadata, pagination, request context, auth and provider credential events, and SSRF-blocked validation logging.

- CHANGELOG: Add WebSocket bridge, GLM Thinking preset, safe outbound fetch/SSRF guard, cooldown-aware retries, compliance audit v2, model alias seeding, and all Internal Improvements for the 3 new commits - README: Expand v3.6.x highlights table with 10 new features; add SafeOutboundFetch, CooldownAwareRetry, SSRF guard, TPS metric, sync tokens, WebSocket bridge to Resilience/Observability/Deployment tables - ARCHITECTURE: Bump date; add new modules to executive summary, API routes, SSE core services, Auth/Security section; add SSRF/Outbound guard failure mode (section 6); expand module mapping - ENVIRONMENT: Add OMNIROUTE_CRYPT_KEY/OMNIROUTE_API_KEY_BASE64 legacy aliases, OUTBOUND_SSRF_GUARD_ENABLED, CODEX_CLIENT_VERSION, and REQUEST_RETRY/MAX_RETRY_INTERVAL_SEC cooldown retry settings - FEATURES: Add 6 new feature sections — V1 WebSocket Bridge, Sync Tokens & Config Bundle, GLM Thinking Preset, Safe Outbound Fetch & SSRF Guard, Cooldown-Aware Retries, Compliance Audit v2

Integrated into release/v3.6.6 — IPv6 proxy test fix

…osouzapw#1200 (diegosouzapw#1256) Integrated into release/v3.6.6 — Gemini custom model picker fix

…ests (diegosouzapw#1246) Integrated into release/v3.6.6 — OAuth client_id default fallbacks

…n Chat→Responses translator (diegosouzapw#1245) Integrated into release/v3.6.6 — max_tokens → max_output_tokens Responses API translation + unit tests

…egosouzapw#1258) Integrated into release/v3.6.6 — cursor-agent CLI credential source support

…eout behavior (diegosouzapw#1257) Integrated into release/v3.6.6 — CC-compatible upstream SSE restore + stream timeout fix + README table repair

…CLI tools (diegosouzapw#1263) Integrated into release/v3.6.6

Integrated into release/v3.6.6

…gosouzapw#1269) Integrated into release/v3.6.6

…zapw#1274) Integrated into release/v3.6.6

This integrates rdself's unify-provider-profile-locks PR manually to handle structural conflicts.

…kup flag

Integrated into release/v3.6.6

…del button positioning, and clarify oauth documentation

Document the latest fixes for Codex routing configuration parsing and Lobehub provider icon fallback behavior. Add the note that the remaining JavaScript test files were migrated to TypeScript ES modules to reflect the completed test stack transition.

…o avoid testing conflict

…S codebase

…ia URL parsing in test mocks

Proactively compress oversized contexts before sending to upstream providers, preventing context_length_exceeded errors. Compression triggers at 85% of model's context limit using the existing 3-layer compressContext() function. - Import compressContext, estimateTokens, getTokenLimit from contextManager - Add compression check after translation, before executor dispatch - Estimate tokens and compare against 85% threshold of model's context limit - Apply 3-layer compression (trim tools, compress thinking, purify history) - Log compression events with before/after token counts and layers applied - Audit compression events for observability - Add unit tests verifying integration behavior Closes diegosouzapw#1290

When purifyHistory() drops oldest messages to fit context window, it can split tool_use/tool_result pairs — keeping the tool_result but dropping the tool_use that initiated it. This causes upstream providers to reject the request with format errors. Add fixToolPairs() that runs after each purification pass to remove: - OpenAI format: orphaned role='tool' messages without matching tool_calls ID - Claude format: orphaned tool_result content blocks without matching tool_use ID Closes diegosouzapw#1291

… after tsx migration

- Add missing crypto imports in cursor.ts and perplexity-web.ts - Fix memory FTS5 to use rowid instead of memory_id for correct JOIN - Fix dangerous JSON string truncation in perplexity-web.ts by truncating history array before stringify - Fix timeout resource leak in contextHandoff.ts by clearing timeout - Fix encryption security issue by throwing error instead of falling back to plaintext Addresses all high and medium priority issues from gemini-code-assist review.

- Replace hardcoded 8000 token limit with dynamic calculation from models.dev - Use 40% of model's context window for history (safety margin) - Parse model string to get provider and model name - Query getTokenLimit() which checks models.dev DB for actual context limits - Fallback to 8000 tokens if model not found - Add warning log when history too large even with single message This ensures compaction works with any model size: - Small models (8K context): ~3.2K tokens for history - Medium models (128K context): ~51K tokens for history - Large models (1M+ context): ~400K+ tokens for history Fixes 'Conversation history too large to compact' errors by adapting to model capabilities.

Problem: When a combo has mixed context sizes (e.g., gpt-4o-mini 128K, gemini-3-pro 1M), using only the first model's limit could cause compaction to fail on fallback models. Solution: - Pass all combo target models to selectMessagesForSummary() - Calculate minimum context limit across all targets - Use the smallest limit to ensure compatibility with any fallback Example combo: [gpt-4o-mini (128K), claude-opus (200K), gemini-3-pro (1M)] - Before: Used 128K limit (first model only) - After: Uses 128K limit (minimum across all targets) - Result: Compacted history works with any model in the combo This ensures handoff context is compatible with all potential fallback models.

gemini-code-assist

Code Review

This pull request upgrades OmniRoute to version 3.6.6, introducing a V1 WebSocket bridge, a configuration sync system with scoped tokens, the GLM Thinking model preset, and the Perplexity Web provider. Security is significantly hardened through a new outbound SSRF guard, runtime environment validation, and a fix for a moderate vulnerability in the follow-redirects dependency. The update also adds hybrid token counting, cooldown-aware retries, and UI enhancements like pagination for memory and skills. Feedback suggests several improvements: increasing the severity of critical security audit events, aborting migrations upon detecting name mismatches, enhancing error reporting in semantic search, optimizing sequential database queries in the memory store, and implementing signal-aware sleeps in the retry logic to ensure immediate cancellation.

gemini-code-assist · 2026-04-17T14:02:28Z

+      logAuditEvent({
+        action: "auth.login.misconfigured",
+        actor: "system",
+        target: "dashboard-auth",
+        resourceType: "auth_session",
+        status: "failed",
+        ipAddress: auditContext.ipAddress || undefined,
+        requestId: auditContext.requestId,
+        metadata: { reason: "missing_jwt_secret" },
+      });


The audit event for a missing JWT secret is logged as 'system' actor. Since this is a critical configuration error preventing authentication, consider logging this with a higher severity or ensuring it triggers an alert, as it indicates a broken security setup.

gemini-code-assist · 2026-04-17T14:02:28Z

+function detectNameMismatches(
+  appliedRecords: Array<{ version: string; name: string }>,
+  files: Array<{ version: string; name: string; path: string }>
+): Array<{ version: string; appliedName: string; diskName: string }> {
+  const appliedByName = new Map(appliedRecords.map((r) => [r.version, r.name]));
+  const mismatches: Array<{ version: string; appliedName: string; diskName: string }> = [];
+
+  for (const file of files) {
+    const appliedName = appliedByName.get(file.version);
+    if (appliedName && appliedName !== file.name) {
+      mismatches.push({
+        version: file.version,
+        appliedName,
+        diskName: file.name,
+      });
+    }
+  }
+
+  return mismatches;
+}


The detectNameMismatches function only logs errors but does not prevent the migration from proceeding. If a renumbering mismatch is detected, it should likely abort the migration process to prevent inconsistent database states.

gemini-code-assist · 2026-04-17T14:02:28Z

+    case "semantic": {
+      if (config.query && ftsAvailable) {
+        const ftsQuery =
+          `SELECT m.* FROM ${tableName} m ` +
+          `JOIN memory_fts f ON m.rowid = f.rowid ` +
+          `WHERE f.memory_fts MATCH ? AND m.${columns.apiKeyId} = ? ` +
+          `AND (m.${columns.expiresAt} IS NULL OR datetime(m.${columns.expiresAt}) > datetime('now'))` +
+          (normalizedConfig.scope === "session" && config.sessionId
+            ? ` AND m.${columns.sessionId} = ?`
+            : "") +
+          (normalizedConfig.retentionDays > 0
+            ? ` AND datetime(m.${columns.createdAt}) >= datetime(?)`
+            : "") +
+          ` ORDER BY f.rank LIMIT 100`;
+        const ftsParams: any[] = [config.query, apiKeyId];
+        if (normalizedConfig.scope === "session" && config.sessionId) {
+          ftsParams.push(config.sessionId);
+        }
+        if (normalizedConfig.retentionDays > 0) {
+          const cutoff = new Date(
+            Date.now() - normalizedConfig.retentionDays * 24 * 60 * 60 * 1000
+          ).toISOString();
+          ftsParams.push(cutoff);
+        }
+        try {
+          rows = db.prepare(ftsQuery).all(...ftsParams) as MemoryRow[];
+        } catch {
+          rows = [];
+        }
+        if (rows.length === 0) {
+          query += ` ORDER BY ${columns.createdAt} DESC LIMIT 100`;
+          rows = db.prepare(query).all(...params) as MemoryRow[];
+        }
+      } else {
+        query += ` ORDER BY ${columns.createdAt} DESC LIMIT 100`;
+        rows = db.prepare(query).all(...params) as MemoryRow[];
+      }
      break;


The semantic search implementation uses a try-catch block that silently swallows errors and returns an empty array. This hides potential database issues (e.g., syntax errors in generated SQL) during FTS5 queries. It should log the error or rethrow if the database is in an invalid state.

gemini-code-assist · 2026-04-17T14:02:28Z

    whereClauses.push("api_key_id = ?");
-    params.push(filters.apiKeyId);
+    whereParams.push(filters.apiKeyId);
  }

  if (filters.type) {
    whereClauses.push("type = ?");
-    params.push(filters.type);
+    whereParams.push(filters.type);
  }

  if (filters.sessionId) {
    whereClauses.push("session_id = ?");
-    params.push(filters.sessionId);
+    whereParams.push(filters.sessionId);
+  }
+
+  if (typeof filters.query === "string" && filters.query.trim().length > 0) {
+    const likeQuery = `%${filters.query.trim().toLowerCase()}%`;
+    whereClauses.push("(LOWER(content) LIKE ? OR LOWER(key) LIKE ?)");
+    whereParams.push(likeQuery, likeQuery);
+  }
+
+  // Run COUNT query + byType aggregation in a single query
+  let countQuery = "SELECT COUNT(*) as total FROM memories";
+  if (whereClauses.length > 0) {
+    countQuery += " WHERE " + whereClauses.join(" AND ");
  }
+  const countStmt = db.prepare(countQuery);
+  const countRow = countStmt.get(...whereParams) as { total: number };
+  const total = countRow.total;

+  // Build byType aggregation (counts ALL matching rows, not just the page)
+  let byTypeQuery = "SELECT type, COUNT(*) as count FROM memories";
+  const byTypeParams: unknown[] = [...whereParams];
+  if (whereClauses.length > 0) {
+    byTypeQuery += " WHERE " + whereClauses.join(" AND ");
+  }
+  byTypeQuery += " GROUP BY type";
+  const byTypeStmt = db.prepare(byTypeQuery);
+  const byTypeRows = byTypeStmt.all(...byTypeParams) as { type: string; count: number }[];
+  const byType = Object.fromEntries(byTypeRows.map((r) => [r.type, r.count])) as Record<
+    string,
+    number
+  >;
+
+  // Calculate effective limit and offset
+  const effectiveLimit = filters.limit ?? 50;
+  const effectivePage = filters.page ?? 1;
+  const effectiveOffset = filters.offset ?? (effectivePage - 1) * effectiveLimit;
+
+  // Build SELECT query with pagination
+  let query = "SELECT * FROM memories";
  if (whereClauses.length > 0) {
    query += " WHERE " + whereClauses.join(" AND ");
  }

  // Add ordering and pagination
-  query += " ORDER BY created_at DESC";
+  query += " ORDER BY created_at DESC LIMIT ? OFFSET ?";

-  if (filters.limit !== undefined) {
-    query += " LIMIT ?";
-    params.push(filters.limit);
-  }
-
-  if (filters.offset !== undefined) {
-    if (filters.limit === undefined) {
-      query += " LIMIT -1";
-    }
-    query += " OFFSET ?";
-    params.push(filters.offset);
-  }
+  // Build params for SELECT query (WHERE params + pagination params)
+  const params = [...whereParams, effectiveLimit, effectiveOffset];

  const stmt = db.prepare(query);
  const rows = stmt.all(...params);

-  return (rows as MemoryRow[]).map(rowToMemory);
+  return {
+    data: (rows as MemoryRow[]).map(rowToMemory),
+    total,
+    byType,
+  };
 }


The listMemories function performs multiple database queries (COUNT, byType, and SELECT) sequentially. For large datasets, this could be a performance bottleneck. Consider optimizing these into a single query or using a more efficient approach if the memory table grows significantly.

gemini-code-assist · 2026-04-17T14:02:28Z

+  for (let attempt = 1; attempt <= retryConfig.attempts; attempt++) {
+    try {
+      const executeFetch = () =>
+        fetchWithTimeout(targetUrl.toString(), {
+          ...fetchOptions,
+          method,
+          redirect,
+          signal,
+          timeoutMs,
+        });
+
+      const response = proxyConfig
+        ? await runWithProxyContext(proxyConfig, executeFetch)
+        : await executeFetch();
+
+      if (!allowRedirect && response.status >= 300 && response.status < 400) {
+        const location = response.headers.get("location");
+        await cancelResponseBody(response);
+        throw new SafeOutboundFetchError(
+          `Redirect blocked for ${method} ${targetUrl.toString()} (${response.status})`,
+          {
+            code: "REDIRECT_BLOCKED",
+            url: targetUrl.toString(),
+            method,
+            attempts: attempt,
+            status: response.status,
+            location,
+            isRetryable: false,
+          }
+        );
+      }
+
+      if (
+        retryConfig.shouldRetryMethod &&
+        attempt < retryConfig.attempts &&
+        retryConfig.statusCodes.has(response.status)
+      ) {
+        await cancelResponseBody(response);
+        await sleep(getBackoffDelay(retryConfig.backoffMs, attempt));
+        continue;
+      }
+
+      return response;
+    } catch (error) {
+      const normalizedError = normalizeFetchFailure(error, targetUrl.toString(), method, attempt);
+      const shouldRetry =
+        retryConfig.shouldRetryMethod &&
+        attempt < retryConfig.attempts &&
+        normalizedError.isRetryable;
+
+      if (!shouldRetry) {
+        throw normalizedError;
+      }
+
+      await sleep(getBackoffDelay(retryConfig.backoffMs, attempt));
+    }


The retry loop uses sleep(getBackoffDelay(...)) which is not interruptible by the signal passed to safeOutboundFetch. If the request is aborted during a backoff sleep, the function will continue to wait until the sleep finishes before checking the signal. Use a signal-aware sleep or setTimeout with signal.addEventListener to ensure immediate cancellation.

diegosouzapw and others added 30 commits April 14, 2026 07:46

fix(providers): update Xiaomi MiMo token-plan endpoints (diegosouzapw…

b6e8597

…#1238) Integrated into release/v3.6.6

fix(cc-compatible): trim beta flags and preserve cache passthrough (d…

ff183f8

…iegosouzapw#1230) Integrated into release/v3.6.6

feat(memory+skills): full-featured memory & skills systems with tests (…

711812a

…diegosouzapw#1228) Integrated into release/v3.6.6

fix: forward client x-initiator header to GitHub Copilot upstream (di…

5acb40c

…egosouzapw#1227) Integrated into release/v3.6.6

feat(bailian-quota): add Alibaba Coding Plan quota monitoring (diegos…

2a84d79

…ouzapw#1235)

fix(tests): resolve memory migration and skills route pagination bugs…

9949fad

… from PR overlaps

docs: Update CHANGELOG.md with v3.6.6 features (diegosouzapw#1182, di…

d675c8f

…egosouzapw#1165, diegosouzapw#1177)

chore(release): bump version to 3.6.6

de050be

Update package versions for the electron app and open-sse package. Sync llm.txt metadata and feature headings with the 3.6.6 release.

fix: use api64 for proxy test (diegosouzapw#1255)

7fc758c

Integrated into release/v3.6.6 — IPv6 proxy test fix

fix(page): update custom models section to include all providers dieg…

08a2a15

…osouzapw#1200 (diegosouzapw#1256) Integrated into release/v3.6.6 — Gemini custom model picker fix

fix: provide default client_id fallbacks to prevent broken OAuth requ…

00a4b45

…ests (diegosouzapw#1246) Integrated into release/v3.6.6 — OAuth client_id default fallbacks

fix: translate max_tokens/max_completion_tokens → max_output_tokens i…

678298f

…n Chat→Responses translator (diegosouzapw#1245) Integrated into release/v3.6.6 — max_tokens → max_output_tokens Responses API translation + unit tests

feat(oauth): support cursor-agent CLI as Cursor credential source (di…

da9af4a

…egosouzapw#1258) Integrated into release/v3.6.6 — cursor-agent CLI credential source support

fix(cc-compatible): restore upstream SSE and correct stream/combo tim…

a17b723

…eout behavior (diegosouzapw#1257) Integrated into release/v3.6.6 — CC-compatible upstream SSE restore + stream timeout fix + README table repair

fix(cli-tools): resolve API key resolution and model mapping bugs in …

d89e38e

…CLI tools (diegosouzapw#1263) Integrated into release/v3.6.6

feat(cli-tools): add Qwen Code CLI integration (diegosouzapw#1266)

bc0ad6a

Integrated into release/v3.6.6

fix(i18n): add missing zh-CN translations and fix logger imports (die…

ca21f51

…gosouzapw#1269) Integrated into release/v3.6.6

fix(i18n): add Chinese i18n support to dashboard components (diegosou…

69e46ca

…zapw#1274) Integrated into release/v3.6.6

feat: update Pollinations to require API key, remove free tier flag (d…

2a175ec

…iegosouzapw#1177)

feat: friendly error messages for crypto/encryption failures (diegoso…

e860df1

…uzapw#1165)

feat: add TPS (tokens per second) metric column to request logs (dieg…

915341a

…osouzapw#1182)

feat: merge custom/imported models into filter list for all providers (…

efeafce

…diegosouzapw#1191)

feat(fallback): Fix provider-profile-driven lockouts (diegosouzapw#1267)

18604ac

This integrates rdself's unify-provider-profile-locks PR manually to handle structural conflicts.

diegosouzapw and others added 20 commits April 15, 2026 15:56

fix: disable mass-migration abort during unit tests based on auto-bac…

419f4c6

…kup flag

fix: update script regex in auto-update tests to use .mjs

8f86907

feat: Add Perplexity Web (Session) provider (diegosouzapw#1289)

66ace45

Integrated into release/v3.6.6

fix(cli): resolve codex routing config parsing, standardize select mo…

497014d

…del button positioning, and clarify oauth documentation

chore(release): merge diegosouzapw#1286 minor improvements manually t…

058422c

…o avoid testing conflict

chore(test): rename perplexity-web.test.mjs to .ts to maintain 100% T…

33c6c43

…S codebase

chore(docs): update CHANGELOG.md for perplexity-web provider

77e4ce1

fix(security): resolve CodeQL incomplete URL substring sanitization v…

5768e76

…ia URL parsing in test mocks

fix(tests): align reasoning expectations with GLM thinking structure

5791904

Merge release/v3.6.6 into branch

8b21c7b

Merge branch 'release/v3.6.6' into fix/compress-context-integration-1290

16e6fd8

fix(tests): supply tool_use in mock so it is not dropped

0437592

chore: convert remaining test to TypeScript

327b06e

fix(tests): restore compatibility with compressContext threshold test…

674eada

… after tsx migration

oyi77 requested a review from diegosouzapw as a code owner April 17, 2026 13:53

gemini-code-assist Bot reviewed Apr 17, 2026

View reviewed changes

oyi77 changed the title ~~fix: Re-apply PR #1363 (Code Review Fixes & Dynamic Context Compaction)~~ fix: Dynamic Context Compaction Apr 17, 2026

oyi77 closed this Apr 17, 2026

This was referenced Apr 17, 2026

fix: replace unit test with integration test for proactive context compression #1377

Closed

fix: replace unit test with integration test for proactive context compression #1378

Merged

diegosouzapw mentioned this pull request Apr 30, 2026

Release v3.7.6 #1803

Merged

Tr0sT mentioned this pull request Apr 30, 2026

fix(codex): omit compact client metadata #1822

Merged

1 of 5 tasks

diegosouzapw mentioned this pull request May 4, 2026

Release v3.7.9 #1959

Merged

diegosouzapw mentioned this pull request Jun 22, 2026

fix(gemini): preserve pattern in antigravity tool schema sanitizer #4651

Merged

5 tasks done

Navigation Menu

Repository navigation

fix: Dynamic Context Compaction - #1368

fix: Dynamic Context Compaction#1368
oyi77 wants to merge 61 commits into
diegosouzapw:mainfrom
oyi77:fix/context-compaction-reapplied

oyi77 commented Apr 17, 2026

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

12 participants

Conversation

oyi77 commented Apr 17, 2026

Work Included:

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot Apr 17, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

12 participants