feat(compression): add Responses tool-output engine

config/quality/file-size-baseline.json

@@ -178,7 +178,9 @@
     "open-sse/handlers/search.ts": 1546,
     "open-sse/handlers/sseParser.ts": 830,
     "open-sse/handlers/videoGeneration.ts": 1275,
-    "open-sse/mcp-server/schemas/tools.ts": 1497,
+    "_rebaseline_2026_07_22_8010_codex_responses_engine": "PR #8010 (@JxnLexn) own growth: open-sse/mcp-server/schemas/tools.ts 1497->1505 (+8 = threading the new \"codex-responses\" literal into the compressionConfigureInput strategy/autoTriggerMode Zod enums and setCompressionEngineInput engine enum, mirroring the existing rtk/omniglyph enum entries; no new tool). open-sse/services/compression/strategySelector.ts 1043->1054 (+11 = one new `if (mode === \"codex-responses\")` dispatch branch in runCompression that delegates 100% to the new codexResponsesEngine.apply, mirroring the existing rtk single-mode dispatch, plus threading config.codexResponsesConfig.preserveToolNames into the shared adaptBodyForCompression call at the 3 existing call sites). src/lib/db/compression.ts (untracked, new-file cap 800) 794->845 (+51 = normalizeCodexResponsesConfig, mirroring the existing normalizeRtkConfig normalizer, plus registering \"codex-responses\" in the COMPRESSION_MODES/STACKED_PIPELINE_ENGINE_IDS/SINGLE_MODE_ENGINE sets and the getCompressionSettings load/save switch) — added to the baseline at its current size. All three are cohesive dispatch/normalizer wiring at existing chokepoints (mirroring the prior compression-mode rebaselines #6534/#6556), not extractable without hiding the mode-dispatch boundary. Covered by tests/unit/compression/codex-responses.test.ts (6) + omniglyph-registries.test.ts/types.test.ts (22, updated for the new mode).",
+    "src/lib/db/compression.ts": 845,
+    "open-sse/mcp-server/schemas/tools.ts": 1505,
     "open-sse/mcp-server/server.ts": 1555,
     "open-sse/mcp-server/tools/advancedTools.ts": 1120,
     "_rebaseline_2026_06_27_5193_antigravity_basered": "Base-red (pre-existing release drift, fast-gate PR->release skips check:file-size): accountFallback.ts 1773->1777 and src/app/api/providers/[id]/test/route.ts 924->940 were already over their frozen caps on release/v3.8.39 independent of any antigravity change. Owner chose to rebaseline (keep the documented issue-reference comments #1846/#1449/#347 etc.) rather than accept the contributor comment-stripping in #5200/#5198. Reverted #5200 to restore the comments; bumped these two frozen caps to the actual base sizes. No logic change.",
@@ -194,7 +196,7 @@
     "_rebaseline_2026_06_26_fidelity_gate_extraction": "Milestone-B fidelity-gate wiring residual: bodyToText+gateAdvance extracted to fidelityGateStep.ts (889->854, -35), but the StackOptions.fidelityGate field, the `const fidelityGate` reads at the two stacked-loop dispatch chokepoints, and the import of FidelityGateConfig are irreducible wiring that cannot leave strategySelector without an architectural refactor of the pre-existing stacked pipeline. Net: 889->854 (+6 vs the pre-Milestone-B frozen 848). Covered by tests/unit/compression/*.test.ts (940 pass).",
     "_rebaseline_2026_06_28_5243_risk_gate_prepass": "PR #5243 (compression risk-gate pre-pass) own growth: open-sse/services/compression/strategySelector.ts 854->899 (+45). The three exported entry points (applyCompression/applyStackedCompression/applyStackedCompressionAsync) become thin wrappers over pure-extracted private bodies (runCompression/runStackedCompression/runStackedCompressionAsync) so the risk-gate mask->run->restore wrapper sits strictly OUTSIDE the per-step loop — a single universal integration point. The wrapper logic itself (resolveRiskGate/withRiskGate) lives in the new riskGate/strategyWrap.ts (<cap); the residual growth is the duplicated thin-wrapper signatures + the extracted bodies' dispatch boundary, guarded by a byte-identical parity test (riskGateIntegration). Default off (DEFAULT_COMPRESSION_CONFIG unchanged). Not extractable without hiding the dispatch boundary, mirroring prior compression rebaselines. Structural shrink tracked in #3501.",
     "_rebaseline_2026_06_29_5286_memoization": "PR #5286 own growth: strategySelector.ts 899->960 (+61 = the opt-in result-memoization branches in applyCompression/applyCompressionAsync — principal+determinism gate, makeMemoKey lookup/store with model+supportsVision folded into the key, recompute-with-memo-off). Default off (memoizeCompressionResults), so zero behavior change. The memo helpers live in the leaf resultMemo.ts (<cap); the chokepoint wiring here is not extractable. Structural shrink of this hot-path file tracked in #3501.",
-    "open-sse/services/compression/strategySelector.ts": 1043,
+    "open-sse/services/compression/strategySelector.ts": 1054,
     "_rebaseline_2026_07_10_gcf_v3_2_decode": "PR #6838 own growth: new vendored file open-sse/services/compression/engines/headroom/gcf/decode_generic.ts frozen at 880 (> 800 cap). It is the vendored GCF generic-profile decoder (spec v3.2 nested flattening plus the prototype-pollution / hasOwnProperty hardening added in this PR's Gemini review). Kept as one file faithful to upstream gcf-typescript so re-vendoring stays a clean copy rather than a re-split each cycle (sibling generic.ts/scalar.ts stay < cap; extraction would also fragment the file's frozen eslint no-explicit-any suppressions). Round-trip + prototype-pollution regression coverage in tests/unit/compression/headroom-smartcrusher.test.ts. Frozen: only shrinks from here.",
     "open-sse/services/compression/engines/headroom/gcf/decode_generic.ts": 880,
     "open-sse/services/rateLimitManager.ts": 1035,

open-sse/mcp-server/schemas/tools.ts

@@ -1095,11 +1095,31 @@ export const compressionStatusTool: McpToolDefinition<
 export const compressionConfigureInput = z.object({
   enabled: z.boolean().optional(),
   strategy: z
-    .enum(["off", "lite", "standard", "aggressive", "ultra", "rtk", "stacked", "omniglyph"])
+    .enum([
+      "off",
+      "lite",
+      "standard",
+      "aggressive",
+      "ultra",
+      "rtk",
+      "codex-responses",
+      "stacked",
+      "omniglyph",
+    ])
     .optional()
     .describe("Compression mode"),
   autoTriggerMode: z
-    .enum(["off", "lite", "standard", "aggressive", "ultra", "rtk", "stacked", "omniglyph"])
+    .enum([
+      "off",
+      "lite",
+      "standard",
+      "aggressive",
+      "ultra",
+      "rtk",
+      "codex-responses",
+      "stacked",
+      "omniglyph",
+    ])
     .optional(),
   maxTokens: z
     .number()
@@ -1132,7 +1152,7 @@ export const compressionConfigureTool: McpToolDefinition<
 > = {
   name: "omniroute_compression_configure",
   description:
-    "Configure compression settings at runtime. Supports enabling/disabling compression, changing strategy (off/lite/standard/aggressive/ultra/rtk/stacked), adjusting maxTokens threshold, targetRatio, auto-trigger mode, system prompt preservation, and MCP description compression.",
+    "Configure compression settings at runtime. Supports enabling/disabling compression, changing strategy (off/lite/standard/aggressive/ultra/rtk/codex-responses/stacked), adjusting maxTokens threshold, targetRatio, auto-trigger mode, system prompt preservation, and MCP description compression.",
   inputSchema: compressionConfigureInput,
   outputSchema: compressionConfigureOutput,
   scopes: ["write:compression"],
@@ -1142,7 +1162,7 @@ export const compressionConfigureTool: McpToolDefinition<
 };
 
 export const setCompressionEngineInput = z.object({
-  engine: z.enum(["off", "caveman", "rtk", "stacked"]).optional(),
+  engine: z.enum(["off", "caveman", "rtk", "codex-responses", "stacked"]).optional(),
   cavemanIntensity: z.enum(["lite", "full", "ultra"]).optional(),
   rtkIntensity: z.enum(["minimal", "standard", "aggressive"]).optional(),
   outputMode: z.boolean().optional(),

open-sse/mcp-server/tools/compressionTools.ts

@@ -428,26 +428,39 @@ export async function handleSetCompressionEngine(
   args: z.infer<typeof setCompressionEngineInput>
 ): Promise<{ success: boolean; settings: Record<string, unknown> }> {
   const updates: Record<string, unknown> = { enabled: true };
+  const current = await getCompressionSettings();
   if (args.engine) {
     updates.defaultMode = args.engine === "caveman" ? "standard" : args.engine;
-    if (args.engine === "off") updates.enabled = false;
+    if (args.engine === "off") {
+      updates.enabled = false;
+    } else if (args.engine !== "stacked") {
+      const selectedEngine = args.engine === "caveman" ? "caveman" : args.engine;
+      updates.engines = Object.fromEntries(
+        Object.entries(current.engines).map(([id, toggle]) => [
+          id,
+          {
+            ...toggle,
+            enabled: id === selectedEngine,
+            ...(id === "caveman" && args.cavemanIntensity ? { level: args.cavemanIntensity } : {}),
+            ...(id === "rtk" && args.rtkIntensity ? { level: args.rtkIntensity } : {}),
+          },
+        ])
+      );
+    }
   }
   if (args.cavemanIntensity) {
-    const current = await getCompressionSettings();
     updates.cavemanConfig = {
       ...(current.cavemanConfig ?? {}),
       intensity: args.cavemanIntensity,
     };
   }
   if (args.rtkIntensity) {
-    const current = await getCompressionSettings();
     updates.rtkConfig = {
       ...(current.rtkConfig ?? {}),
       intensity: args.rtkIntensity,
     };
   }
   if (args.outputMode !== undefined) {
-    const current = await getCompressionSettings();
     updates.cavemanOutputMode = {
       ...(current.cavemanOutputMode ?? {}),
       enabled: args.outputMode,

open-sse/services/compression/bodyAdapter.ts

@@ -6,6 +6,17 @@ type MessageLike = {
   [key: string]: unknown;
 };
 
+export const CODEX_RESPONSE_ITEM_META = Symbol("codexResponseItemMeta");
+
+export type CodexResponseItemMeta = {
+  type: string;
+  eligible: boolean;
+};
+
+type CodexMessageLike = MessageLike & {
+  [CODEX_RESPONSE_ITEM_META]?: CodexResponseItemMeta;
+};
+
 type ResponsesItem = {
   type?: unknown;
   role?: unknown;
@@ -18,6 +29,8 @@ const RESPONSES_MESSAGE_TYPES = new Set([
   "message",
   "function_call_output",
   "custom_tool_call_output",
+  "local_shell_call_output",
+  "apply_patch_call_output",
 ]);
 const COMPRESSION_INPUT_INDEX = Symbol("compressionInputIndex");
 
@@ -100,11 +113,16 @@ function responsesToolOutputField(item: ResponsesItem): "output" | "content" {
   return item.output !== null && item.output !== undefined ? "output" : "content";
 }
 
-function responsesItemToMessage(item: ResponsesItem): MessageLike | null {
+function responsesItemToMessage(item: ResponsesItem): CodexMessageLike | null {
   const type = typeof item.type === "string" ? item.type : "message";
   if (!RESPONSES_MESSAGE_TYPES.has(type)) return null;
 
-  if (type === "function_call_output" || type === "custom_tool_call_output") {
+  if (
+    type === "function_call_output" ||
+    type === "custom_tool_call_output" ||
+    type === "local_shell_call_output" ||
+    type === "apply_patch_call_output"
+  ) {
     const rawOutput = item.output ?? item.content;
     // OpenAI Responses shape (Codex): body.input holds Responses items. When
     // output is a JSON object (not a string or content array), serialise it so
@@ -123,6 +141,7 @@ function responsesItemToMessage(item: ResponsesItem): MessageLike | null {
           : isObjectOutput
             ? JSON.stringify(rawOutput)
             : toChatContent(rawOutput),
+      [CODEX_RESPONSE_ITEM_META]: { type, eligible: false },
     };
   }
 
@@ -132,9 +151,65 @@ function responsesItemToMessage(item: ResponsesItem): MessageLike | null {
   };
 }
 
+const DEFAULT_CODEX_PROTECTED_TOOL_NAMES = new Set([
+  "read",
+  "glob",
+  "grep",
+  "write",
+  "edit",
+  "websearch",
+  "webfetch",
+  "web_search",
+  "web_fetch",
+]);
+function markCodexResponseEligibility(
+  messages: CodexMessageLike[],
+  inputItems: unknown[],
+  preserveToolNames: string[] = []
+): void {
+  const protectedNames = new Set([
+    ...DEFAULT_CODEX_PROTECTED_TOOL_NAMES,
+    ...preserveToolNames.map((name) => name.trim().toLowerCase()),
+  ]);
+  const functionCalls = new Map<string, string>();
+  const skippedCallIds = new Set<string>();
+  for (const raw of inputItems) {
+    if (!isRecord(raw) || raw.type !== "function_call") continue;
+    if (typeof raw.call_id !== "string" || raw.call_id.length === 0) continue;
+    const name = typeof raw.name === "string" ? raw.name : "";
+    functionCalls.set(raw.call_id, name);
+    if (
+      protectedNames.has(name.trim().toLowerCase()) ||
+      name === "headless_retrieval" ||
+      name.endsWith("__headless_retrieval")
+    ) {
+      skippedCallIds.add(raw.call_id);
+    }
+  }
+
+  for (const message of messages) {
+    const meta = message[CODEX_RESPONSE_ITEM_META];
+    if (!meta) continue;
+    if (meta.type === "local_shell_call_output" || meta.type === "apply_patch_call_output") {
+      meta.eligible = true;
+      continue;
+    }
+    if (meta.type !== "function_call_output") continue;
+    const rawIndex = message[COMPRESSION_INPUT_INDEX];
+    const rawItem = typeof rawIndex === "number" ? inputItems[rawIndex] : null;
+    const callId = isRecord(rawItem) && typeof rawItem.call_id === "string" ? rawItem.call_id : "";
+    meta.eligible = callId.length > 0 && functionCalls.has(callId) && !skippedCallIds.has(callId);
+  }
+}
+
 function messageToResponsesItem(message: MessageLike, originalItem: ResponsesItem): ResponsesItem {
   const type = typeof originalItem.type === "string" ? originalItem.type : "message";
-  if (type === "function_call_output" || type === "custom_tool_call_output") {
+  if (
+    type === "function_call_output" ||
+    type === "custom_tool_call_output" ||
+    type === "local_shell_call_output" ||
+    type === "apply_patch_call_output"
+  ) {
     const outputField = responsesToolOutputField(originalItem);
     const originalOutput = originalItem[outputField];
     return {
@@ -166,7 +241,10 @@ export type CompressionBodyAdapter = {
   restore(compressedBody: Record<string, unknown>): Record<string, unknown>;
 };
 
-export function adaptBodyForCompression(body: Record<string, unknown>): CompressionBodyAdapter {
+export function adaptBodyForCompression(
+  body: Record<string, unknown>,
+  preserveToolNames: string[] = []
+): CompressionBodyAdapter {
   if (Array.isArray(body.messages)) {
     return {
       body,
@@ -205,6 +283,8 @@ export function adaptBodyForCompression(body: Record<string, unknown>): Compress
     messages.push({ ...message, [COMPRESSION_INPUT_INDEX]: index });
   });
 
+  markCodexResponseEligibility(messages, inputItems, preserveToolNames);
+
   if (messages.length === 0) {
     return {
       body,

open-sse/services/compression/deriveDefaultPlan.ts

@@ -8,16 +8,12 @@ const SINGLE_MODE_OF: Record<string, string> = {
   aggressive: "aggressive",
   ultra: "ultra",
   rtk: "rtk",
+  "codex-responses": "codex-responses",
   omniglyph: "omniglyph",
 };
 
 export type CompressionSource =
-  | "request-header"
-  | "routing-override"
-  | "active-profile"
-  | "auto-trigger"
-  | "default"
-  | "off";
+  "request-header" | "routing-override" | "active-profile" | "auto-trigger" | "default" | "off";
 
 export interface DerivedPlan {
   mode: string;
@@ -37,7 +33,7 @@ export interface DerivedPlan {
  */
 export function deriveDefaultPlan(
   engines: Record<string, EngineToggle>,
-  masterEnabled: boolean,
+  masterEnabled: boolean
 ): DerivedPlan {
   if (!masterEnabled) return { mode: "off", stackedPipeline: [] };
 

open-sse/services/compression/engineCatalog.ts

@@ -22,8 +22,8 @@ export interface EngineMeta {
   id: string;
   label: string;
   stackPriority: number;
-  levels?: string[];        // intensity options; undefined = no level selector
-  isSingleMode: boolean;    // can be the effective mode when it is the only engine on
+  levels?: string[]; // intensity options; undefined = no level selector
+  isSingleMode: boolean; // can be the effective mode when it is the only engine on
   description: string;
   guidance: EngineGuidance;
 }
@@ -82,6 +82,19 @@ export const ENGINE_CATALOG: Record<string, EngineMeta> = {
       cacheImpact: "moderate",
     },
   },
+  "codex-responses": {
+    id: "codex-responses",
+    label: "Responses Tool Output",
+    stackPriority: 12,
+    isSingleMode: true,
+    description: "Conservative compression for supported Responses tool outputs.",
+    guidance: {
+      tradeoffs:
+        "Lossless-first JSON and bounded diagnostic compression for shell, patch, search, and build outputs. Protected tools and uncertain shapes pass through unchanged.",
+      lossy: true,
+      cacheImpact: "low",
+    },
+  },
   headroom: {
     id: "headroom",
     label: "Headroom",