@hassangameryt/freeflow 0.3.0 → 0.5.0

package/router/dist/benchmarks/context-saving-baseline.js

@@ -3,7 +3,15 @@ import { mkdir, mkdtemp, rm, writeFile } from "node:fs/promises";
 import { tmpdir } from "node:os";
 import { join, resolve } from "node:path";
 import { fileURLToPath } from "node:url";
-import { approximateTokens, defaultJsonRunReportPath, escapeMarkdownTableCell as escapeTable, formatPercent, parseBenchmarkCliArgs, reductionPercent, writeBenchmarkReportPair, } from "./benchmark-harness.js";
+import {
+  approximateTokens,
+  defaultJsonRunReportPath,
+  escapeMarkdownTableCell as escapeTable,
+  formatPercent,
+  parseBenchmarkCliArgs,
+  reductionPercent,
+  writeBenchmarkReportPair,
+} from "./benchmark-harness.js";
 import { freeflowTransform } from "../transform/engine.js";
 import { freeflowSearch } from "../tools/search.js";
 import { freeflowRun } from "../tools/run.js";
@@ -18,422 +26,445 @@ Parser decisions stay structured in details.result while model-visible text can
 Batch intermediates should be stored without flooding the model context.
 `;
 const COMMAND_STDOUT = [
-    "FAIL packages/auth/auth.test.ts",
-    "  ● refresh token keeps AUTH_TOKEN scoped",
-    "    Expected AUTH_TOKEN to be redacted before logging",
-    "PASS packages/router/run.test.ts",
-    "PASS packages/router/search.test.ts",
-    "Tests:       1 failed, 24 passed, 25 total",
-    "Snapshots:   0 total",
-    "Time:        4.21 s",
-    "Ran all test suites.",
+  "FAIL packages/auth/auth.test.ts",
+  "  ● refresh token keeps AUTH_TOKEN scoped",
+  "    Expected AUTH_TOKEN to be redacted before logging",
+  "PASS packages/router/run.test.ts",
+  "PASS packages/router/search.test.ts",
+  "Tests:       1 failed, 24 passed, 25 total",
+  "Snapshots:   0 total",
+  "Time:        4.21 s",
+  "Ran all test suites.",
 ].join("\n");
 const COMMAND_STDERR = [
-    "Error: ROUTER_BENCH_FAILED",
-    "    at refreshToken (packages/auth/auth.ts:42:11)",
-    "    at main (packages/auth/auth.test.ts:8:3)",
+  "Error: ROUTER_BENCH_FAILED",
+  "    at refreshToken (packages/auth/auth.ts:42:11)",
+  "    at main (packages/auth/auth.test.ts:8:3)",
 ].join("\n");
 const GUARDED_FIXTURES = [
-    {
-        path: "router/tests/run.test.js",
-        behaviors: [
-            "freeflowRun parser/recovery behavior",
-            "duplicate output compaction",
-            "exact failure and verification evidence preservation",
-        ],
-    },
-    {
-        path: "router/tests/tools/search.test.js",
-        behaviors: [
-            "exact repo path/range retrieval",
-            "exact vault line retrieval",
-            "vault-wide query/locate",
-            "repo/vault expand and explain behavior",
-        ],
-    },
-    {
-        path: "router/tests/transform/engine.test.js",
-        behaviors: [
-            "deterministic transform operations",
-            "transformed-output lineage and exact recovery",
-            "structured failures for invalid operations and missing sources",
-        ],
-    },
-    {
-        path: "router/tests/tools/batch.test.js",
-        behaviors: [
-            "batch accepts public run/search step kinds",
-            "batch rejects legacy retrieve/transform step kinds",
-            "query aggregation from child evidence handles",
-        ],
-    },
-    {
-        path: "router/tests/pi-extension.test.js",
-        behaviors: [
-            "public Pi tool schemas",
-            "content/details result contract",
-            "compact and expanded TUI renderers",
-        ],
-    },
-    {
-        path: "router/tests/pi/pi-extension-search.test.js",
-        behaviors: [
-            "search action=transform disabled-by-default behavior",
-            "adapter-unavailable behavior",
-            "proof-backed adapter execution path",
-        ],
-    },
-    {
-        path: "router/tests/regression-fixtures.test.js",
-        behaviors: [
-            "generated decoy avoidance",
-            "huge-line bounded evidence",
-            "exact phrase and source-prior retrieval regressions",
-        ],
-    },
-    {
-        path: "router/tests/vault-index.test.js",
-        behaviors: [
-            "command/native/observed/transformed vault record indexing",
-            "metadata-only indexing without raw recovery claims",
-            "retention and degraded-index behavior",
-        ],
-    },
+  {
+    path: "router/tests/run.test.js",
+    behaviors: [
+      "freeflowRun parser/recovery behavior",
+      "duplicate output compaction",
+      "exact failure and verification evidence preservation",
+    ],
+  },
+  {
+    path: "router/tests/tools/search.test.js",
+    behaviors: [
+      "exact repo path/range retrieval",
+      "exact vault line retrieval",
+      "vault-wide query/locate",
+      "repo/vault expand and explain behavior",
+    ],
+  },
+  {
+    path: "router/tests/transform/engine.test.js",
+    behaviors: [
+      "deterministic transform operations",
+      "transformed-output lineage and exact recovery",
+      "structured failures for invalid operations and missing sources",
+    ],
+  },
+  {
+    path: "router/tests/tools/batch.test.js",
+    behaviors: [
+      "batch accepts public run/search step kinds",
+      "batch rejects legacy retrieve/transform step kinds",
+      "query aggregation from child evidence handles",
+    ],
+  },
+  {
+    path: "router/tests/pi-extension.test.js",
+    behaviors: ["public Pi tool schemas", "content/details result contract", "compact and expanded TUI renderers"],
+  },
+  {
+    path: "router/tests/pi/pi-extension-search.test.js",
+    behaviors: [
+      "search action=transform disabled-by-default behavior",
+      "adapter-unavailable behavior",
+      "proof-backed adapter execution path",
+    ],
+  },
+  {
+    path: "router/tests/regression-fixtures.test.js",
+    behaviors: [
+      "generated decoy avoidance",
+      "huge-line bounded evidence",
+      "exact phrase and source-prior retrieval regressions",
+    ],
+  },
+  {
+    path: "router/tests/vault-index.test.js",
+    behaviors: [
+      "command/native/observed/transformed vault record indexing",
+      "metadata-only indexing without raw recovery claims",
+      "retention and degraded-index behavior",
+    ],
+  },
 ];
 export async function runContextSavingBaseline(options = {}) {
-    const root = await createTempDir("freeflow-context-saving-baseline-");
-    const repoRoot = join(root.path, "repo");
-    const vaultRoot = join(root.path, "vault");
-    const sessionId = "context-saving-baseline";
-    try {
-        await mkdir(join(repoRoot, "plugin-docs"), { recursive: true });
-        await writeFile(join(repoRoot, "plugin-docs/output-router.md"), REPO_DOC_TEXT, "utf8");
-        const commandRawBytes = byteLength(`${COMMAND_STDOUT}\n${COMMAND_STDERR}\n`);
-        const runResult = await freeflowRun({
-            command: "npm test -- --runInBand",
-            sessionId,
-            vaultRoot,
-            preserve: "important",
-            goal: "baseline model-visible output size",
-        }, {
-            async run() {
-                return {
-                    stdout: `${COMMAND_STDOUT}\n`,
-                    stderr: `${COMMAND_STDERR}\n`,
-                    combined: `STDOUT:\n${COMMAND_STDOUT}\n\nSTDERR:\n${COMMAND_STDERR}\n`,
-                    executionStatus: "failed",
-                    exitCode: 1,
-                    durationMs: 4210,
-                };
-            },
-        });
-        const outputId = typeof runResult.outputId === "string" ? runResult.outputId : "";
-        const observations = [];
-        observations.push(makeObservation({
-            id: "run-failed-test-output",
-            title: "freeflow_run failed test output",
-            tool: "freeflow_run",
-            action: "run",
-            rawBytes: commandRawBytes,
-            result: runResult,
-            requiredFacts: ["ROUTER_BENCH_FAILED", "failed", "ffout_"],
-            notes: ["Current Pi model-visible text is the full pretty-printed routed result JSON."],
-        }));
-        const repoQueryResult = await freeflowSearch({
-            action: "query",
-            source: { kind: "repo", root: repoRoot },
-            query: "Exact recovery outputId line ranges parser decisions",
-            topK: 1,
-            preserve: "important",
-        });
-        observations.push(makeObservation({
-            id: "search-repo-query",
-            title: "freeflow_search repo query",
-            tool: "freeflow_search",
-            action: "query repo",
-            rawBytes: byteLength(REPO_DOC_TEXT),
-            result: repoQueryResult,
-            requiredFacts: ["outputId", "line ranges", "details.result"],
-        }));
-        const repoExactResult = await freeflowSearch({
-            action: "retrieve",
-            source: { kind: "repo", root: repoRoot, path: "plugin-docs/output-router.md" },
-            lineRange: { start: 3, end: 5 },
-            preserve: "full",
-        });
-        observations.push(makeObservation({
-            id: "search-repo-exact-range",
-            title: "freeflow_search exact repo range",
-            tool: "freeflow_search",
-            action: "retrieve repo range",
-            rawBytes: byteLength(REPO_DOC_TEXT),
-            result: repoExactResult,
-            requiredFacts: ["stdout", "stderr", "line ranges"],
-            notes: ["This guards the existing public advanced path/range recovery behavior."],
-        }));
-        if (outputId) {
-            const vaultQueryResult = await freeflowSearch({
-                action: "query",
-                source: { kind: "vault", root: vaultRoot, sessionId, outputId, stream: "combined" },
-                query: "ROUTER_BENCH_FAILED",
-                topK: 1,
-                preserve: "important",
-            });
-            observations.push(makeObservation({
-                id: "search-vault-query",
-                title: "freeflow_search vault query",
-                tool: "freeflow_search",
-                action: "query vault",
-                rawBytes: commandRawBytes,
-                result: vaultQueryResult,
-                requiredFacts: ["ROUTER_BENCH_FAILED", outputId],
-            }));
-            const transformResult = await freeflowTransform({
-                source: { kind: "vault", outputId, stream: "combined" },
-                operation: { kind: "regexFilter", pattern: "AUTH_TOKEN|ROUTER_BENCH_FAILED|Tests:", maxMatches: 10 },
-                sessionId,
-                vaultRoot,
-                preserve: "important",
-            });
-            observations.push(makeObservation({
-                id: "transform-regex-filter",
-                title: "freeflow_search action=transform regexFilter",
-                tool: "freeflow_search action=transform",
-                action: "regexFilter vault",
-                rawBytes: commandRawBytes,
-                result: transformResult,
-                requiredFacts: ["AUTH_TOKEN", "ROUTER_BENCH_FAILED", "regexFilter"],
-            }));
-        }
-        return {
-            generatedAt: options.generatedAt ?? new Date().toISOString(),
-            summary: summarizeObservations(observations),
-            observations,
-            guardrails: GUARDED_FIXTURES,
-            notes: [
-                "This is a Slice 0 baseline for current Freeflow-owned tools before compact-output redesign.",
-                "Model-visible bytes use the current Pi routedToolText shape: JSON.stringify(result, null, 2).",
-                "Details payload bytes use the current Pi details shape: { result } with the same structured result.",
-                "Negative reduction means the current routed JSON is larger than the raw fixture bytes; this is expected baseline evidence for compact-output work.",
-            ],
-        };
-    }
-    finally {
-        await root.cleanup();
+  const root = await createTempDir("freeflow-context-saving-baseline-");
+  const repoRoot = join(root.path, "repo");
+  const vaultRoot = join(root.path, "vault");
+  const sessionId = "context-saving-baseline";
+  try {
+    await mkdir(join(repoRoot, "plugin-docs"), { recursive: true });
+    await writeFile(join(repoRoot, "plugin-docs/output-router.md"), REPO_DOC_TEXT, "utf8");
+    const commandRawBytes = byteLength(`${COMMAND_STDOUT}\n${COMMAND_STDERR}\n`);
+    const runResult = await freeflowRun(
+      {
+        command: "npm test -- --runInBand",
+        sessionId,
+        vaultRoot,
+        preserve: "important",
+        goal: "baseline model-visible output size",
+      },
+      {
+        async run() {
+          return {
+            stdout: `${COMMAND_STDOUT}\n`,
+            stderr: `${COMMAND_STDERR}\n`,
+            combined: `STDOUT:\n${COMMAND_STDOUT}\n\nSTDERR:\n${COMMAND_STDERR}\n`,
+            executionStatus: "failed",
+            exitCode: 1,
+            durationMs: 4210,
+          };
+        },
+      },
+    );
+    const outputId = typeof runResult.outputId === "string" ? runResult.outputId : "";
+    const observations = [];
+    observations.push(
+      makeObservation({
+        id: "run-failed-test-output",
+        title: "freeflow_run failed test output",
+        tool: "freeflow_run",
+        action: "run",
+        rawBytes: commandRawBytes,
+        result: runResult,
+        requiredFacts: ["ROUTER_BENCH_FAILED", "failed", "ffout_"],
+        notes: ["Current Pi model-visible text is the full pretty-printed routed result JSON."],
+      }),
+    );
+    const repoQueryResult = await freeflowSearch({
+      action: "query",
+      source: { kind: "repo", root: repoRoot },
+      query: "Exact recovery outputId line ranges parser decisions",
+      topK: 1,
+      preserve: "important",
+    });
+    observations.push(
+      makeObservation({
+        id: "search-repo-query",
+        title: "freeflow_search repo query",
+        tool: "freeflow_search",
+        action: "query repo",
+        rawBytes: byteLength(REPO_DOC_TEXT),
+        result: repoQueryResult,
+        requiredFacts: ["outputId", "line ranges", "details.result"],
+      }),
+    );
+    const repoExactResult = await freeflowSearch({
+      action: "retrieve",
+      source: { kind: "repo", root: repoRoot, path: "plugin-docs/output-router.md" },
+      lineRange: { start: 3, end: 5 },
+      preserve: "full",
+    });
+    observations.push(
+      makeObservation({
+        id: "search-repo-exact-range",
+        title: "freeflow_search exact repo range",
+        tool: "freeflow_search",
+        action: "retrieve repo range",
+        rawBytes: byteLength(REPO_DOC_TEXT),
+        result: repoExactResult,
+        requiredFacts: ["stdout", "stderr", "line ranges"],
+        notes: ["This guards the existing public advanced path/range recovery behavior."],
+      }),
+    );
+    if (outputId) {
+      const vaultQueryResult = await freeflowSearch({
+        action: "query",
+        source: { kind: "vault", root: vaultRoot, sessionId, outputId, stream: "combined" },
+        query: "ROUTER_BENCH_FAILED",
+        topK: 1,
+        preserve: "important",
+      });
+      observations.push(
+        makeObservation({
+          id: "search-vault-query",
+          title: "freeflow_search vault query",
+          tool: "freeflow_search",
+          action: "query vault",
+          rawBytes: commandRawBytes,
+          result: vaultQueryResult,
+          requiredFacts: ["ROUTER_BENCH_FAILED", outputId],
+        }),
+      );
+      const transformResult = await freeflowTransform({
+        source: { kind: "vault", outputId, stream: "combined" },
+        operation: { kind: "regexFilter", pattern: "AUTH_TOKEN|ROUTER_BENCH_FAILED|Tests:", maxMatches: 10 },
+        sessionId,
+        vaultRoot,
+        preserve: "important",
+      });
+      observations.push(
+        makeObservation({
+          id: "transform-regex-filter",
+          title: "freeflow_search action=transform regexFilter",
+          tool: "freeflow_search action=transform",
+          action: "regexFilter vault",
+          rawBytes: commandRawBytes,
+          result: transformResult,
+          requiredFacts: ["AUTH_TOKEN", "ROUTER_BENCH_FAILED", "regexFilter"],
+        }),
+      );
     }
+    return {
+      generatedAt: options.generatedAt ?? new Date().toISOString(),
+      summary: summarizeObservations(observations),
+      observations,
+      guardrails: GUARDED_FIXTURES,
+      notes: [
+        "This is a Slice 0 baseline for current Freeflow-owned tools before compact-output redesign.",
+        "Model-visible bytes use the current Pi routedToolText shape: JSON.stringify(result, null, 2).",
+        "Details payload bytes use the current Pi details shape: { result } with the same structured result.",
+        "Negative reduction means the current routed JSON is larger than the raw fixture bytes; this is expected baseline evidence for compact-output work.",
+      ],
+    };
+  } finally {
+    await root.cleanup();
+  }
 }
 export function renderContextSavingBaselineReport(report) {
-    const date = report.generatedAt.slice(0, 10);
-    const lines = [
-        "# Freeflow Context-Saving Native Tools Baseline Report - Slice 0",
-        "",
-        `Date: ${date}`,
-        "",
-        "## Scope",
-        "",
-        "Baseline current model-visible output sizes for representative `freeflow_run`, `freeflow_search`, and `freeflow_search action=transform` results before the compact-output redesign.",
-        "",
-        "The baseline intentionally measures current Pi tool text as pretty-printed routed result JSON. Later slices should reduce model-visible bytes while keeping details and exact recovery available.",
-        "",
-        "## Summary",
-        "",
-        `- Fixtures measured: ${report.summary.fixtures}`,
-        `- Facts preserved: ${report.summary.factsPreserved}/${report.summary.fixtures}`,
-        `- Recovery available: ${report.summary.recoveryAvailable}/${report.summary.fixtures}`,
-        `- Raw bytes: ${report.summary.totalRawBytes}`,
-        `- Model-visible bytes: ${report.summary.totalModelVisibleBytes}`,
-        `- Details payload bytes: ${report.summary.totalDetailsPayloadBytes}`,
-        `- Approx raw tokens: ${report.summary.totalRawTokensApprox}`,
-        `- Approx model-visible tokens: ${report.summary.totalModelVisibleTokensApprox}`,
-        `- Weighted model-visible reduction vs raw: ${formatPercent(report.summary.weightedModelVisibleReductionPercent)}`,
-        "",
-        "## Measurements",
-        "",
-        "| Fixture | Tool | Action | Raw bytes | Model-visible bytes | Details bytes | Reduction vs raw | Facts | Recovery | Status |",
-        "| --- | --- | --- | ---: | ---: | ---: | ---: | --- | --- | --- |",
-    ];
-    for (const observation of report.observations) {
-        lines.push(`| ${escapeTable(observation.id)} | ${observation.tool} | ${escapeTable(observation.action)} | ${observation.rawBytes} | ${observation.modelVisibleBytes} | ${observation.detailsPayloadBytes} | ${formatPercent(observation.modelVisibleReductionPercent)} | ${observation.factsPreserved ? "yes" : "no"} | ${observation.recoveryAvailable ? "yes" : "no"} | ${escapeTable(observation.status)} |`);
+  const date = report.generatedAt.slice(0, 10);
+  const lines = [
+    "# Freeflow Context-Saving Native Tools Baseline Report - Slice 0",
+    "",
+    `Date: ${date}`,
+    "",
+    "## Scope",
+    "",
+    "Baseline current model-visible output sizes for representative `freeflow_run`, `freeflow_search`, and `freeflow_search action=transform` results before the compact-output redesign.",
+    "",
+    "The baseline intentionally measures current Pi tool text as pretty-printed routed result JSON. Later slices should reduce model-visible bytes while keeping details and exact recovery available.",
+    "",
+    "## Summary",
+    "",
+    `- Fixtures measured: ${report.summary.fixtures}`,
+    `- Facts preserved: ${report.summary.factsPreserved}/${report.summary.fixtures}`,
+    `- Recovery available: ${report.summary.recoveryAvailable}/${report.summary.fixtures}`,
+    `- Raw bytes: ${report.summary.totalRawBytes}`,
+    `- Model-visible bytes: ${report.summary.totalModelVisibleBytes}`,
+    `- Details payload bytes: ${report.summary.totalDetailsPayloadBytes}`,
+    `- Approx raw tokens: ${report.summary.totalRawTokensApprox}`,
+    `- Approx model-visible tokens: ${report.summary.totalModelVisibleTokensApprox}`,
+    `- Weighted model-visible reduction vs raw: ${formatPercent(report.summary.weightedModelVisibleReductionPercent)}`,
+    "",
+    "## Measurements",
+    "",
+    "| Fixture | Tool | Action | Raw bytes | Model-visible bytes | Details bytes | Reduction vs raw | Facts | Recovery | Status |",
+    "| --- | --- | --- | ---: | ---: | ---: | ---: | --- | --- | --- |",
+  ];
+  for (const observation of report.observations) {
+    lines.push(
+      `| ${escapeTable(observation.id)} | ${observation.tool} | ${escapeTable(observation.action)} | ${observation.rawBytes} | ${observation.modelVisibleBytes} | ${observation.detailsPayloadBytes} | ${formatPercent(observation.modelVisibleReductionPercent)} | ${observation.factsPreserved ? "yes" : "no"} | ${observation.recoveryAvailable ? "yes" : "no"} | ${escapeTable(observation.status)} |`,
+    );
+  }
+  lines.push("", "## Recovery And Evidence Handles", "");
+  for (const observation of report.observations) {
+    lines.push(`### ${observation.id}`);
+    if (observation.outputId) {
+      lines.push(`- outputId: \`${observation.outputId}\``);
     }
-    lines.push("", "## Recovery And Evidence Handles", "");
-    for (const observation of report.observations) {
-        lines.push(`### ${observation.id}`);
-        if (observation.outputId) {
-            lines.push(`- outputId: \`${observation.outputId}\``);
-        }
-        lines.push(`- recovery: ${observation.recoveryHint || "none"}`);
-        if (observation.evidenceLocations.length > 0) {
-            lines.push(`- evidence: ${observation.evidenceLocations.map((item) => `\`${item}\``).join(", ")}`);
-        }
-        for (const note of observation.notes) {
-            lines.push(`- note: ${note}`);
-        }
-        lines.push("");
+    lines.push(`- recovery: ${observation.recoveryHint || "none"}`);
+    if (observation.evidenceLocations.length > 0) {
+      lines.push(`- evidence: ${observation.evidenceLocations.map((item) => `\`${item}\``).join(", ")}`);
     }
-    lines.push("## Guardrail Inventory", "");
-    lines.push("Existing tests/fixtures that must remain stable during the redesign:", "");
-    for (const guardrail of report.guardrails) {
-        lines.push(`- \`${guardrail.path}\``);
-        for (const behavior of guardrail.behaviors) {
-            lines.push(`  - ${behavior}`);
-        }
+    for (const note of observation.notes) {
+      lines.push(`- note: ${note}`);
     }
-    lines.push("", "## Notes", "");
-    for (const note of report.notes) {
-        lines.push(`- ${note}`);
+    lines.push("");
+  }
+  lines.push("## Guardrail Inventory", "");
+  lines.push("Existing tests/fixtures that must remain stable during the redesign:", "");
+  for (const guardrail of report.guardrails) {
+    lines.push(`- \`${guardrail.path}\``);
+    for (const behavior of guardrail.behaviors) {
+      lines.push(`  - ${behavior}`);
     }
-    return `${lines.join("\n")}\n`;
+  }
+  lines.push("", "## Notes", "");
+  for (const note of report.notes) {
+    lines.push(`- ${note}`);
+  }
+  return `${lines.join("\n")}\n`;
 }
 export async function writeContextSavingBaselineReport(report, markdownReportPath, options = {}) {
-    return writeBenchmarkReportPair({
-        report,
-        markdownReportPath,
-        jsonReportPath: options.jsonReportPath,
-        renderMarkdown: renderContextSavingBaselineReport,
-    });
+  return writeBenchmarkReportPair({
+    report,
+    markdownReportPath,
+    jsonReportPath: options.jsonReportPath,
+    renderMarkdown: renderContextSavingBaselineReport,
+  });
 }
 function makeObservation(input) {
-    const modelVisibleText = currentPiModelVisibleText(input.result);
-    const detailsPayloadText = JSON.stringify({ result: input.result }, null, 2);
-    const modelVisibleBytes = byteLength(modelVisibleText);
-    const detailsPayloadBytes = byteLength(detailsPayloadText);
-    const evidenceLocations = resultEvidenceLocations(input.result);
-    const recoveryHint = resultRecoveryHint(input.result);
-    const outputId = resultOutputId(input.result);
-    const observation = {
-        id: input.id,
-        title: input.title,
-        tool: input.tool,
-        action: input.action,
-        rawBytes: input.rawBytes,
-        modelVisibleBytes,
-        detailsPayloadBytes,
-        rawTokensApprox: approximateTokens(input.rawBytes),
-        modelVisibleTokensApprox: approximateTokens(modelVisibleBytes),
-        detailsPayloadTokensApprox: approximateTokens(detailsPayloadBytes),
-        modelVisibleReductionPercent: reductionPercent(input.rawBytes, modelVisibleBytes),
-        factsPreserved: input.requiredFacts.every((fact) => modelVisibleText.includes(fact) || detailsPayloadText.includes(fact)),
-        recoveryAvailable: Boolean(recoveryHint || outputId || evidenceLocations.length > 0),
-        recoveryHint,
-        status: resultStatus(input.result),
-        evidenceLocations,
-        notes: input.notes ?? [],
-    };
-    if (outputId) {
-        observation.outputId = outputId;
-    }
-    return observation;
+  const modelVisibleText = currentPiModelVisibleText(input.result);
+  const detailsPayloadText = JSON.stringify({ result: input.result }, null, 2);
+  const modelVisibleBytes = byteLength(modelVisibleText);
+  const detailsPayloadBytes = byteLength(detailsPayloadText);
+  const evidenceLocations = resultEvidenceLocations(input.result);
+  const recoveryHint = resultRecoveryHint(input.result);
+  const outputId = resultOutputId(input.result);
+  const observation = {
+    id: input.id,
+    title: input.title,
+    tool: input.tool,
+    action: input.action,
+    rawBytes: input.rawBytes,
+    modelVisibleBytes,
+    detailsPayloadBytes,
+    rawTokensApprox: approximateTokens(input.rawBytes),
+    modelVisibleTokensApprox: approximateTokens(modelVisibleBytes),
+    detailsPayloadTokensApprox: approximateTokens(detailsPayloadBytes),
+    modelVisibleReductionPercent: reductionPercent(input.rawBytes, modelVisibleBytes),
+    factsPreserved: input.requiredFacts.every(
+      (fact) => modelVisibleText.includes(fact) || detailsPayloadText.includes(fact),
+    ),
+    recoveryAvailable: Boolean(recoveryHint || outputId || evidenceLocations.length > 0),
+    recoveryHint,
+    status: resultStatus(input.result),
+    evidenceLocations,
+    notes: input.notes ?? [],
+  };
+  if (outputId) {
+    observation.outputId = outputId;
+  }
+  return observation;
 }
 function summarizeObservations(observations) {
-    const totalRawBytes = observations.reduce((sum, observation) => sum + observation.rawBytes, 0);
-    const totalModelVisibleBytes = observations.reduce((sum, observation) => sum + observation.modelVisibleBytes, 0);
-    const totalDetailsPayloadBytes = observations.reduce((sum, observation) => sum + observation.detailsPayloadBytes, 0);
-    return {
-        fixtures: observations.length,
-        factsPreserved: observations.filter((observation) => observation.factsPreserved).length,
-        recoveryAvailable: observations.filter((observation) => observation.recoveryAvailable).length,
-        totalRawBytes,
-        totalModelVisibleBytes,
-        totalDetailsPayloadBytes,
-        totalRawTokensApprox: approximateTokens(totalRawBytes),
-        totalModelVisibleTokensApprox: approximateTokens(totalModelVisibleBytes),
-        totalDetailsPayloadTokensApprox: approximateTokens(totalDetailsPayloadBytes),
-        weightedModelVisibleReductionPercent: reductionPercent(totalRawBytes, totalModelVisibleBytes),
-    };
+  const totalRawBytes = observations.reduce((sum, observation) => sum + observation.rawBytes, 0);
+  const totalModelVisibleBytes = observations.reduce((sum, observation) => sum + observation.modelVisibleBytes, 0);
+  const totalDetailsPayloadBytes = observations.reduce((sum, observation) => sum + observation.detailsPayloadBytes, 0);
+  return {
+    fixtures: observations.length,
+    factsPreserved: observations.filter((observation) => observation.factsPreserved).length,
+    recoveryAvailable: observations.filter((observation) => observation.recoveryAvailable).length,
+    totalRawBytes,
+    totalModelVisibleBytes,
+    totalDetailsPayloadBytes,
+    totalRawTokensApprox: approximateTokens(totalRawBytes),
+    totalModelVisibleTokensApprox: approximateTokens(totalModelVisibleBytes),
+    totalDetailsPayloadTokensApprox: approximateTokens(totalDetailsPayloadBytes),
+    weightedModelVisibleReductionPercent: reductionPercent(totalRawBytes, totalModelVisibleBytes),
+  };
 }
 function currentPiModelVisibleText(result) {
-    return JSON.stringify(result, null, 2);
+  return JSON.stringify(result, null, 2);
 }
 function resultStatus(result) {
-    const value = asRecord(result);
-    if (!value) {
-        return "unknown";
-    }
-    const statuses = [
-        stringAt(value, "toolStatus"),
-        stringAt(asRecord(value.execution), "status"),
-        stringAt(asRecord(value.routing), "status"),
-    ].filter((status) => Boolean(status));
-    return statuses.join("/") || "unknown";
+  const value = asRecord(result);
+  if (!value) {
+    return "unknown";
+  }
+  const statuses = [
+    stringAt(value, "toolStatus"),
+    stringAt(asRecord(value.execution), "status"),
+    stringAt(asRecord(value.routing), "status"),
+  ].filter((status) => Boolean(status));
+  return statuses.join("/") || "unknown";
 }
 function resultOutputId(result) {
-    const value = asRecord(result);
-    const outputId = stringAt(value, "outputId");
-    if (outputId) {
-        return outputId;
-    }
-    return stringAt(asRecord(value?.recovery), "outputId") ?? "";
+  const value = asRecord(result);
+  const outputId = stringAt(value, "outputId");
+  if (outputId) {
+    return outputId;
+  }
+  return stringAt(asRecord(value?.recovery), "outputId") ?? "";
 }
 function resultRecoveryHint(result) {
-    const value = asRecord(result);
-    return stringAt(asRecord(value?.recovery), "how") ?? "";
+  const value = asRecord(result);
+  return stringAt(asRecord(value?.recovery), "how") ?? "";
 }
 function resultEvidenceLocations(result) {
-    const value = asRecord(result);
-    const locations = [];
-    const evidence = Array.isArray(value?.evidence) ? value.evidence : [];
-    for (const packet of evidence) {
-        const record = asRecord(packet);
-        if (!record) {
-            continue;
-        }
-        const path = stringAt(record, "path") ?? stringAt(asRecord(record.source), "path") ?? stringAt(asRecord(record.source), "outputId") ?? "evidence";
-        const lines = stringAt(record, "lines");
-        locations.push(lines ? `${path}:${lines}` : path);
+  const value = asRecord(result);
+  const locations = [];
+  const evidence = Array.isArray(value?.evidence) ? value.evidence : [];
+  for (const packet of evidence) {
+    const record = asRecord(packet);
+    if (!record) {
+      continue;
     }
-    const importantLines = Array.isArray(value?.importantLines) ? value.importantLines : [];
-    for (const line of importantLines) {
-        const record = asRecord(line);
-        if (!record) {
-            continue;
-        }
-        const stream = stringAt(record, "stream") ?? "stream";
-        const linesValue = stringAt(record, "lines") ?? "?";
-        locations.push(`${stream}:${linesValue}`);
+    const path =
+      stringAt(record, "path") ??
+      stringAt(asRecord(record.source), "path") ??
+      stringAt(asRecord(record.source), "outputId") ??
+      "evidence";
+    const lines = stringAt(record, "lines");
+    locations.push(lines ? `${path}:${lines}` : path);
+  }
+  const importantLines = Array.isArray(value?.importantLines) ? value.importantLines : [];
+  for (const line of importantLines) {
+    const record = asRecord(line);
+    if (!record) {
+      continue;
     }
-    return locations;
+    const stream = stringAt(record, "stream") ?? "stream";
+    const linesValue = stringAt(record, "lines") ?? "?";
+    locations.push(`${stream}:${linesValue}`);
+  }
+  return locations;
 }
 function stringAt(record, key) {
-    const value = record?.[key];
-    return typeof value === "string" ? value : undefined;
+  const value = record?.[key];
+  return typeof value === "string" ? value : undefined;
 }
 function asRecord(value) {
-    if (value && typeof value === "object" && !Array.isArray(value)) {
-        return value;
-    }
-    return undefined;
+  if (value && typeof value === "object" && !Array.isArray(value)) {
+    return value;
+  }
+  return undefined;
 }
 async function createTempDir(prefix) {
-    const path = await mkdtemp(resolve(tmpdir(), prefix));
-    return {
-        path,
-        cleanup: async () => {
-            await rm(path, { recursive: true, force: true });
-        },
-    };
+  const path = await mkdtemp(resolve(tmpdir(), prefix));
+  return {
+    path,
+    cleanup: async () => {
+      await rm(path, { recursive: true, force: true });
+    },
+  };
 }
 function byteLength(value) {
-    return Buffer.byteLength(value, "utf8");
+  return Buffer.byteLength(value, "utf8");
 }
 function defaultReportPath() {
-    return resolve(process.cwd(), "evals/reports/runtime/context-saving-native-tools-baseline-1-report.md");
+  return resolve(process.cwd(), "router/evals/reports/context-saving-native-tools-baseline-1-report.md");
 }
 async function runCli() {
-    const { reportPath, jsonReportPath } = parseBenchmarkCliArgs(process.argv.slice(2), { reportPath: defaultReportPath() });
-    const report = await runContextSavingBaseline();
-    const reports = await writeContextSavingBaselineReport(report, reportPath, {
-        jsonReportPath: jsonReportPath === undefined ? defaultJsonRunReportPath(reportPath) : jsonReportPath,
-    });
-    const shortId = createHash("sha256").update(JSON.stringify(report.summary)).digest("hex").slice(0, 8);
-    console.log(`Freeflow context-saving baseline ${shortId}: ${report.summary.factsPreserved}/${report.summary.fixtures} facts preserved, ${report.summary.recoveryAvailable}/${report.summary.fixtures} recoverable`);
-    console.log(`Markdown report: ${reports.markdown}`);
-    if (reports.json) {
-        console.log(`JSON run data: ${reports.json}`);
-    }
-    if (report.summary.factsPreserved !== report.summary.fixtures || report.summary.recoveryAvailable !== report.summary.fixtures) {
-        process.exitCode = 1;
-    }
+  const { reportPath, jsonReportPath } = parseBenchmarkCliArgs(process.argv.slice(2), {
+    reportPath: defaultReportPath(),
+  });
+  const report = await runContextSavingBaseline();
+  const reports = await writeContextSavingBaselineReport(report, reportPath, {
+    jsonReportPath: jsonReportPath === undefined ? defaultJsonRunReportPath(reportPath) : jsonReportPath,
+  });
+  const shortId = createHash("sha256").update(JSON.stringify(report.summary)).digest("hex").slice(0, 8);
+  console.log(
+    `Freeflow context-saving baseline ${shortId}: ${report.summary.factsPreserved}/${report.summary.fixtures} facts preserved, ${report.summary.recoveryAvailable}/${report.summary.fixtures} recoverable`,
+  );
+  console.log(`Markdown report: ${reports.markdown}`);
+  if (reports.json) {
+    console.log(`JSON run data: ${reports.json}`);
+  }
+  if (
+    report.summary.factsPreserved !== report.summary.fixtures ||
+    report.summary.recoveryAvailable !== report.summary.fixtures
+  ) {
+    process.exitCode = 1;
+  }
 }
 if (process.argv[1] && fileURLToPath(import.meta.url) === resolve(process.argv[1])) {
-    await runCli();
+  await runCli();
 }