Format after rebase onto main

Author: TheGreatAxios

evals/compaction/fixtures.ts

@@ -20,19 +20,28 @@ export const REQUIRED_EVIDENCE: readonly Evidence[] = [
   { id: "constraint", source: "operator:initial", value: "no-schema-change" },
   { id: "decision", source: "operator:correction", value: "west-not-east" },
   { id: "failure", source: "command:diagnose", value: "unsupported-format-7" },
-  { id: "decisive", source: "file:diagnostic.log:middle", value: "route-cobalt" },
+  {
+    id: "decisive",
+    source: "file:diagnostic.log:middle",
+    value: "route-cobalt",
+  },
 ];
 
 export function evidenceText(facts: readonly Evidence[]): string {
-  return facts.map((fact) => `[[evidence:${fact.id}|${fact.source}|${fact.value}]]`).join("\n");
+  return facts
+    .map((fact) => `[[evidence:${fact.id}|${fact.source}|${fact.value}]]`)
+    .join("\n");
 }
 
 export const INITIAL =
-  "Audit the deployment. Preserve this constraint: " + evidenceText(REQUIRED_EVIDENCE.slice(0, 1));
+  "Audit the deployment. Preserve this constraint: " +
+  evidenceText(REQUIRED_EVIDENCE.slice(0, 1));
 export const CORRECTION =
-  "Correction: target west instead of east. " + evidenceText(REQUIRED_EVIDENCE.slice(1, 2));
+  "Correction: target west instead of east. " +
+  evidenceText(REQUIRED_EVIDENCE.slice(1, 2));
 export const FAILED_OUTPUT =
-  "Diagnostic preamble.\n".repeat(250) + evidenceText(REQUIRED_EVIDENCE.slice(2, 3));
+  "Diagnostic preamble.\n".repeat(250) +
+  evidenceText(REQUIRED_EVIDENCE.slice(2, 3));
 export const OVERSIZED_OUTPUT =
   "Unrelated diagnostic row.\n".repeat(1500) +
   evidenceText(REQUIRED_EVIDENCE.slice(3)) +

evals/compaction/metrics.test.ts

@@ -30,17 +30,25 @@ const baseline = {
 
 describe("compaction baseline grading", () => {
   test("requires evidence actually visible to the scripted responder", () => {
-    expect(recoverEvidence("[[evidence:region|operator:correction|west]]")).toEqual([fact]);
+    expect(
+      recoverEvidence("[[evidence:region|operator:correction|west]]"),
+    ).toEqual([fact]);
     expect(recoverEvidence("The region was mentioned earlier.")).toEqual([]);
     expect(
-      grade({ ...baseline, recovered: recoverEvidence("evidence removed") }).factualRecovery,
+      grade({ ...baseline, recovered: recoverEvidence("evidence removed") })
+        .factualRecovery,
     ).toBe(false);
   });
 
   test("rejects wrong values, sources, and altered artifacts independently", () => {
     expect(grade(baseline).passed).toBe(true);
-    expect(grade({ ...baseline, recovered: [{ ...fact, value: "east" }] }).passed).toBe(false);
-    expect(grade({ ...baseline, recovered: [{ ...fact, source: "invented" }] }).passed).toBe(false);
+    expect(
+      grade({ ...baseline, recovered: [{ ...fact, value: "east" }] }).passed,
+    ).toBe(false);
+    expect(
+      grade({ ...baseline, recovered: [{ ...fact, source: "invented" }] })
+        .passed,
+    ).toBe(false);
     const altered = grade({ ...baseline, artifact: "east\n" });
     expect(altered.completion).toBe(false);
     expect(altered.factualRecovery).toBe(true);
@@ -50,9 +58,13 @@ describe("compaction baseline grading", () => {
     expect(grade({ ...baseline, folds: [] }).qualifying).toBe(false);
     for (const fold of folds) {
       expect(qualifyingFold({ ...fold, persisted: false })).toBe(false);
-      expect(qualifyingFold({ ...fold, afterHash: fold.beforeHash })).toBe(false);
+      expect(qualifyingFold({ ...fold, afterHash: fold.beforeHash })).toBe(
+        false,
+      );
       expect(qualifyingFold({ ...fold, continuedAtCall: null })).toBe(false);
-      expect(qualifyingFold({ ...fold, afterTurns: fold.beforeTurns })).toBe(false);
+      expect(qualifyingFold({ ...fold, afterTurns: fold.beforeTurns })).toBe(
+        false,
+      );
     }
     expect(grade({ ...baseline, recovered: [] }).requiredFacts).toBe(1);
   });
@@ -89,16 +101,23 @@ describe("compaction baseline grading", () => {
   });
 
   test("missing usage is unavailable, not zero or an unlabelled estimate", () => {
-    expect(Measurement({ status: "unavailable", reason: "offline" }) instanceof type.errors).toBe(
-      false,
-    );
     expect(
-      Measurement({ status: "reported", value: -1, unit: "tokens" }) instanceof type.errors,
-    ).toBe(true);
-    expect(Measurement({ value: 0, unit: "tokens" }) instanceof type.errors).toBe(true);
+      Measurement({ status: "unavailable", reason: "offline" }) instanceof
+        type.errors,
+    ).toBe(false);
     expect(
-      Measurement({ status: "synthetic", value: 200000, unit: "trigger tokens" }) instanceof
+      Measurement({ status: "reported", value: -1, unit: "tokens" }) instanceof
         type.errors,
+    ).toBe(true);
+    expect(
+      Measurement({ value: 0, unit: "tokens" }) instanceof type.errors,
+    ).toBe(true);
+    expect(
+      Measurement({
+        status: "synthetic",
+        value: 200000,
+        unit: "trigger tokens",
+      }) instanceof type.errors,
     ).toBe(false);
   });
 });

evals/compaction/metrics.ts

@@ -1,11 +1,19 @@
 import { type } from "arktype";
 
-export const Evidence = type({ id: "string", source: "string", value: "string" });
+export const Evidence = type({
+  id: "string",
+  source: "string",
+  value: "string",
+});
 export type Evidence = typeof Evidence.infer;
 
 export const Measurement = type.or(
   { status: "'unavailable'", reason: "string" },
-  { status: "'reported' | 'estimated' | 'synthetic'", value: "number >= 0", unit: "string" },
+  {
+    status: "'reported' | 'estimated' | 'synthetic'",
+    value: "number >= 0",
+    unit: "string",
+  },
 );
 export type Measurement = typeof Measurement.infer;
 
@@ -54,8 +62,10 @@ export function repeatedWork(trace: readonly Work[]) {
     }
     if (seen.has(key)) {
       if (work.name === "read_file") repeatedReads++;
-      if (["grep", "search_files", "web_search"].includes(work.name)) repeatedSearches++;
-      if (["write_file", "edit_file", "apply_patch"].includes(work.name)) duplicatedEdits++;
+      if (["grep", "search_files", "web_search"].includes(work.name))
+        repeatedSearches++;
+      if (["write_file", "edit_file", "apply_patch"].includes(work.name))
+        duplicatedEdits++;
     }
     if (failures.has(key)) repeatedFailedAttempts++;
     seen.add(key);
@@ -81,15 +91,20 @@ export function grade(args: {
   const recoveredFacts = args.expected.filter((fact) =>
     args.recovered.some(
       (answer) =>
-        answer.id === fact.id && answer.source === fact.source && answer.value === fact.value,
+        answer.id === fact.id &&
+        answer.source === fact.source &&
+        answer.value === fact.value,
     ),
   ).length;
   const work = repeatedWork(args.trace);
   const persistedFolds = args.folds.filter(qualifyingFold).length;
   const completion = args.artifact === args.expectedArtifact;
   const factualRecovery = recoveredFacts === args.expected.length;
   const repeatedActions =
-    work.repeatedReads + work.repeatedSearches + work.repeatedFailedAttempts + work.duplicatedEdits;
+    work.repeatedReads +
+    work.repeatedSearches +
+    work.repeatedFailedAttempts +
+    work.duplicatedEdits;
   return {
     completion,
     factualRecovery,
@@ -98,14 +113,20 @@ export function grade(args: {
     persistedFolds,
     qualifying: persistedFolds >= 3,
     ...work,
-    passed: completion && factualRecovery && persistedFolds >= 3 && repeatedActions === 0,
+    passed:
+      completion &&
+      factualRecovery &&
+      persistedFolds >= 3 &&
+      repeatedActions === 0,
   };
 }
 
 /** The responder receives only inference-visible text, never grader expectations. */
 export function recoverEvidence(context: string): Evidence[] {
   const facts = new Map<string, Evidence>();
-  for (const match of context.matchAll(/\[\[evidence:([^|\]\n]+)\|([^|\]\n]+)\|([^\]\n]+)\]\]/g)) {
+  for (const match of context.matchAll(
+    /\[\[evidence:([^|\]\n]+)\|([^|\]\n]+)\|([^\]\n]+)\]\]/g,
+  )) {
     const [, id, source, value] = match;
     if (id !== undefined && source !== undefined && value !== undefined) {
       facts.set(id, { id, source, value });

src/agent/posix-tool-plugins.test.ts

@@ -393,7 +393,10 @@ describe("buildCorePosixToolPlugins", () => {
         permissionGate: gate,
         getEvidenceArchive: () => undefined,
       });
-      const shellGuardIndex = findMiddlewareIndex(plugins, "[command timed out after");
+      const shellGuardIndex = findMiddlewareIndex(
+        plugins,
+        "[command timed out after",
+      );
       const archiveIndex = findMiddlewareIndex(
         plugins,
         "evidence archive is not available in this session",

src/agent/posix-tool-plugins.ts

@@ -95,7 +95,9 @@ export function buildCorePosixToolPlugins(
   // regardless.
   const allowOutside = (): boolean => permissionGate.getSkipPermissions();
   const truncationOptions =
-    getBlobWriter !== undefined || getContextDir !== undefined || getEvidenceArchive !== undefined
+    getBlobWriter !== undefined ||
+    getContextDir !== undefined ||
+    getEvidenceArchive !== undefined
       ? {
           ...(getBlobWriter !== undefined ? { getBlobWriter } : {}),
           ...(getContextDir !== undefined ? { getContextDir } : {}),

src/agent/tools.ts

@@ -562,7 +562,8 @@ export async function createAgentToolset(
       let definition = advertiseEditFileLineRange(
         advertiseShellGuardTimeout(tool.definition, shellTimeout?.defaultMs),
       );
-      if (getEvidenceArchive !== undefined) definition = advertiseArchiveSurface(definition);
+      if (getEvidenceArchive !== undefined)
+        definition = advertiseArchiveSurface(definition);
       return { ...tool, definition };
     }),
     createListDirTool(cwd, {

src/context-compactor.test.ts

@@ -846,7 +846,10 @@ describe("createPruningCompactor — consolidated handoff (CL-7521)", () => {
       }),
     );
     turns.push(
-      makeTurn({ role: "user", content: [{ type: "text", text: "recent ask" }] }),
+      makeTurn({
+        role: "user",
+        content: [{ type: "text", text: "recent ask" }],
+      }),
     );
 
     const output1 = (await compactor.apply(turns, mockStrategyCtx)).output;
@@ -970,7 +973,9 @@ describe("createPruningCompactor — consolidated handoff (CL-7521)", () => {
     const result1 = await compactor.apply(turns, mockStrategyCtx);
     expect(result1.output).toBe(turns);
     expect(result1.record.reason).toBe("summarize failed");
-    expect(firstText(defined(result1.output[0]))).not.toContain(COMPACTED_PREFIX);
+    expect(firstText(defined(result1.output[0]))).not.toContain(
+      COMPACTED_PREFIX,
+    );
 
     const result2 = await compactor.apply(
       grow(result1.output, 16, "ok"),

src/mcp/client.ts

@@ -546,9 +546,12 @@ async function finishClient(
     serverName,
     tools,
     async callBlocks(toolName, args, signal) {
-      const context = authContext === undefined ? undefined : { ...authContext, signal };
+      const context =
+        authContext === undefined ? undefined : { ...authContext, signal };
       const result = await withHTTPAuthorizationRecovery(context, () =>
-        client.callTool({ name: toolName, arguments: args }, undefined, { signal }),
+        client.callTool({ name: toolName, arguments: args }, undefined, {
+          signal,
+        }),
       );
       return validateMcpContentBlocks(result.content);
     },

src/mcp/plugin.ts

@@ -60,7 +60,12 @@ export function mcpClientTools(
   client: MCPClient,
   spillOptions: McpSpillOptions = {},
 ): AgentTool[] {
-  const { getBlobWriter, getContextDir, excludeToolNames = [], getEvidenceArchive } = spillOptions;
+  const {
+    getBlobWriter,
+    getContextDir,
+    excludeToolNames = [],
+    getEvidenceArchive,
+  } = spillOptions;
   const excluded = new Set(excludeToolNames);
 
   return client.tools

src/plugins/evidence-archive-path-guard.test.ts

@@ -17,23 +17,42 @@ const nextHandler = async (call: ToolCall): Promise<ToolResult> => ({
 
 describe("isProtectedEvidenceLocation", () => {
   test("matches evidence-archive and tool-output/archive-* forms", () => {
-    expect(isProtectedEvidenceLocation("evidence-archive/index.jsonl")).toBe(true);
-    expect(isProtectedEvidenceLocation("/tmp/context/evidence-archive")).toBe(true);
-    expect(isProtectedEvidenceLocation("C:\\tmp\\evidence-archive\\index.jsonl")).toBe(true);
-    expect(isProtectedEvidenceLocation("tool-output/archive-sess-occ-1")).toBe(true);
-    expect(isProtectedEvidenceLocation("tool-output:///archive-sess-occ-1")).toBe(true);
-    expect(isProtectedEvidenceLocation("src/session/compaction-archive.ts")).toBe(false);
-    expect(isProtectedEvidenceLocation("tool-output:///other-spill")).toBe(false);
+    expect(isProtectedEvidenceLocation("evidence-archive/index.jsonl")).toBe(
+      true,
+    );
+    expect(isProtectedEvidenceLocation("/tmp/context/evidence-archive")).toBe(
+      true,
+    );
+    expect(
+      isProtectedEvidenceLocation("C:\\tmp\\evidence-archive\\index.jsonl"),
+    ).toBe(true);
+    expect(isProtectedEvidenceLocation("tool-output/archive-sess-occ-1")).toBe(
+      true,
+    );
+    expect(
+      isProtectedEvidenceLocation("tool-output:///archive-sess-occ-1"),
+    ).toBe(true);
+    expect(
+      isProtectedEvidenceLocation("src/session/compaction-archive.ts"),
+    ).toBe(false);
+    expect(isProtectedEvidenceLocation("tool-output:///other-spill")).toBe(
+      false,
+    );
   });
 });
 
 describe("evidenceArchivePathGuardPlugin", () => {
   test("denies path tools targeting evidence-archive or tool-output/archive-*", async () => {
     const plugin = evidenceArchivePathGuardPlugin();
-    const handler = plugin.middleware ? plugin.middleware(nextHandler) : nextHandler;
+    const handler = plugin.middleware
+      ? plugin.middleware(nextHandler)
+      : nextHandler;
     const denied = [
       makeCall("read_file", { path: "evidence-archive/index.jsonl" }),
-      makeCall("grep", { path: "/tmp/context/evidence-archive", pattern: "foo" }),
+      makeCall("grep", {
+        path: "/tmp/context/evidence-archive",
+        pattern: "foo",
+      }),
       makeCall("search_files", { path: "evidence-archive" }),
       makeCall("list_dir", { path: "evidence-archive" }),
       makeCall("write_file", { path: "evidence-archive/x", content: "nope" }),
@@ -50,7 +69,9 @@ describe("evidenceArchivePathGuardPlugin", () => {
 
   test("does not deny a grep pattern that mentions evidence-archive", async () => {
     const plugin = evidenceArchivePathGuardPlugin();
-    const handler = plugin.middleware ? plugin.middleware(nextHandler) : nextHandler;
+    const handler = plugin.middleware
+      ? plugin.middleware(nextHandler)
+      : nextHandler;
     const result = await handler(
       makeCall("grep", { path: "src", pattern: "evidence-archive" }),
       new AbortController().signal,
@@ -61,7 +82,9 @@ describe("evidenceArchivePathGuardPlugin", () => {
 
   test("passes ordinary workspace paths", async () => {
     const plugin = evidenceArchivePathGuardPlugin();
-    const handler = plugin.middleware ? plugin.middleware(nextHandler) : nextHandler;
+    const handler = plugin.middleware
+      ? plugin.middleware(nextHandler)
+      : nextHandler;
     const result = await handler(
       makeCall("read_file", { path: "src/session/compaction-archive.ts" }),
       new AbortController().signal,