agentera 3.0.0-dev.27 → 3.0.0-dev.29

package/bundle/references/analysis/evidence/verification-baseline-2026-07-20/recompute.mjs

@@ -1,98 +0,0 @@
-#!/usr/bin/env node
-import fs from "node:fs";
-import path from "node:path";
-import zlib from "node:zlib";
-
-const root = import.meta.dirname;
-const median = (values) => [...values].sort((a, b) => a - b)[Math.floor(values.length / 2)];
-const round = (value) => Math.round(value * 1000) / 1000;
-const sources = [];
-const assertionsByIdentity = new Map();
-
-for (let sample = 1; sample <= 3; sample += 1) {
-  const report = JSON.parse(
-    zlib.gunzipSync(fs.readFileSync(path.join(root, `source-${sample}.json.gz`))),
-  );
-  const assertions = report.testResults.flatMap((file) =>
-    file.assertionResults.map((assertion) => ({
-      file: file.name.split("/packages/cli/").at(-1),
-      name: assertion.fullName,
-      duration_ms: assertion.duration ?? 0,
-    })),
-  );
-  const jsonElapsedMs =
-    Math.max(...report.testResults.map((file) => file.endTime)) - report.startTime;
-  sources.push({
-    sample,
-    files: report.testResults.length,
-    tests: report.numTotalTests,
-    cumulative_assertion_ms: round(
-      assertions.reduce((total, assertion) => total + assertion.duration_ms, 0),
-    ),
-    json_elapsed_ms: round(jsonElapsedMs),
-  });
-  for (const assertion of assertions) {
-    const identity = `${assertion.file}\n${assertion.name}`;
-    const entry = assertionsByIdentity.get(identity) ?? {
-      file: assertion.file,
-      test: assertion.name,
-      samples_ms: [],
-    };
-    entry.samples_ms.push(assertion.duration_ms);
-    assertionsByIdentity.set(identity, entry);
-  }
-}
-
-const slowest = [...assertionsByIdentity.values()]
-  .map((entry) => ({
-    ...entry,
-    samples_ms: entry.samples_ms.map(round),
-    median_ms: round(median(entry.samples_ms)),
-  }))
-  .sort((a, b) => b.median_ms - a.median_ms)
-  .slice(0, 10);
-
-function hook(name) {
-  const text = zlib.gunzipSync(fs.readFileSync(path.join(root, name))).toString("utf8");
-  const wrapper = JSON.parse(text.trim().split("\n").at(-1));
-  const lefthook = Number(/done in ([0-9.]+) seconds/.exec(text)?.[1]) * 1000;
-  const policies = [...text.matchAll(/✔️ ([^\n]+?) \(([0-9.]+) seconds\)/g)].map((match) => ({
-    name: match[1],
-    duration_ms: Number(match[2]) * 1000,
-  }));
-  return {
-    wall_ms: wrapper.wall_ms,
-    lefthook_ms: lefthook,
-    exit_status: wrapper.exit_status,
-    policies,
-  };
-}
-
-const route = zlib
-  .gunzipSync(fs.readFileSync(path.join(root, "precommit-cli-route.log.gz")))
-  .toString("utf8")
-  .trim()
-  .split("\n");
-
-console.log(
-  JSON.stringify(
-    {
-      source_samples: sources,
-      source_medians: {
-        cumulative_assertion_ms: round(
-          median(sources.map((sample) => sample.cumulative_assertion_ms)),
-        ),
-        json_elapsed_ms: round(median(sources.map((sample) => sample.json_elapsed_ms))),
-        files: median(sources.map((sample) => sample.files)),
-        tests: median(sources.map((sample) => sample.tests)),
-      },
-      slowest_tests_by_median: slowest,
-      pre_commit: {
-        representative_cli: { ...hook("precommit-cli.log.gz"), route },
-        representative_documentation: hook("precommit-documentation.log.gz"),
-      },
-    },
-    null,
-    2,
-  ),
-);

package/bundle/references/analysis/migration-pagination-proportional-2026-07-20.yaml

@@ -1,64 +0,0 @@
-schemaVersion: agentera.verificationMeasurement.v1
-status: pass
-captured: 2026-07-20
-
-comparison:
-  test: uses snapshot-bound whole-entry pagination to recover every entry without gaps or duplicates
-  metric: assertion_duration_ms
-  direction: lower
-  before:
-    authority: references/analysis/verification-baseline-2026-07-20.yaml
-    fixture_entries: 400
-    samples_ms: [18325, 16575, 16680]
-    median_ms: 16680
-  after:
-    fixture_entries: 40
-    samples_ms: [179.449348, 164.137058, 178.642103]
-    median_ms: 178.642103
-    raw_report_compression: gzip_lossless_mtime_zero
-    raw_reports:
-      - references/analysis/evidence/migration-pagination-proportional-2026-07-20/source-focused-1.json.gz
-      - references/analysis/evidence/migration-pagination-proportional-2026-07-20/source-focused-2.json.gz
-      - references/analysis/evidence/migration-pagination-proportional-2026-07-20/source-focused-3.json.gz
-  reduction_ms: 16501.357897
-  reduction_percent: 98.929004
-  speedup: 93.371046
-
-protocol:
-  inherited_from: references/analysis/verification-baseline-2026-07-20.yaml#protocol.source
-  environment:
-    VITEST_MAX_WORKERS: "4"
-  warmup_runs: 1
-  retained_runs: 3
-  command: >-
-    pnpm -C packages/cli exec vp test run --config vite.config.ts
-    test/state/entityMigrationPreview.test.ts
-    -t "uses snapshot-bound whole-entry pagination" --reporter=json
-  note: >-
-    The before samples are the exact assertion durations retained by Task 1's
-    full-source protocol. The after samples isolate the same assertion so the
-    proportional fixture cost is directly reproducible without rerunning
-    unrelated source tests.
-
-coverage:
-  proportional_source:
-    owner: source
-    path: packages/cli/test/state/entityMigrationPreview.test.ts
-    fixture_entries: 40
-    guarantees: [multi_page, whole_entry, byte_bound, no_gaps, no_duplicates, continuation_command]
-  cursor_mutations:
-    owner: source
-    mutations: [source, authority, filter, order, project]
-    guarantees: [continuation_rejected, restart_without_cursor, zero_write]
-  production_cap:
-    owner: performance
-    path: packages/cli/test/performance/entityMigrationPreviewCap.test.ts
-    fixture_entries: 400
-    response_cap_bytes: 32768
-    owner_command: pnpm -C packages/cli test:performance
-    policies: [scheduled, release]
-
-caveats:
-  - The decisions query remained degraded; missing decision detail was not reconstructed.
-  - Build startup omitted decisions, vision, and profile state; no refresh or inferred approval was performed.
-  - Decision satisfaction remained unavailable for llavvlpcmb and djmjbmruyk.

package/bundle/references/analysis/verification-baseline-2026-07-20.yaml

@@ -1,313 +0,0 @@
-schemaVersion: agentera.verificationBaseline.v1
-status: baseline
-captured: 2026-07-20
-source_revision: af9cf773
-
-runner:
-  name: local-16-logical-cpu-node22
-  plan_label: named 16-core Node 22 runner
-  plan_label_mapping: >-
-    The plan's "16-core" label denotes the 16 logical scheduling CPUs reported
-    by nproc and os.cpus(), not 16 physical cores. The hardware is truthfully an
-    8-physical-core, 16-thread processor; acceptance wording is preserved.
-  os: Linux 7.1.4-1-cachyos x86_64
-  cpu: AMD Ryzen 7 9800X3D 8-Core Processor
-  physical_cores: 8
-  logical_cpus: 16
-  node: v22.23.1
-  pnpm: 10.30.3
-
-protocol:
-  source:
-    command: pnpm -C packages/cli exec vp test run --config vite.config.ts --reporter=default --reporter=json
-    environment:
-      VITEST_MAX_WORKERS: "4"
-    warmup_runs: 1
-    retained_runs: 3
-    warmup_result:
-      wall_ms: 109757
-      status: pass
-      retained: false
-    wall_definition: elapsed time around the complete pnpm process
-    runner_duration_definition: Vitest duration from startup through test completion
-    cumulative_assertion_definition: sum of JSON reporter assertion durations; concurrent work overlaps
-    setup_definition: Vitest setup component reported by the default reporter
-    slowest_definition: median assertion duration across the three retained runs
-    raw_report_format: Vitest JSON reporter output
-    raw_report_compression: gzip (RFC 1952), lossless, deterministic mtime zero
-  pre_commit:
-    command: pnpm exec lefthook run pre-commit
-    method: >-
-      Stage one reversible representative change, run the complete hook, record
-      lefthook policy timings and outer wall time, then restore the index and
-      worktree. Intentional plan entities were not staged for these samples.
-
-motivation_reconciliation:
-  plan_observation:
-    wall_ms: 79100
-    classification: exploratory_single_run
-  task_1_baseline:
-    wall_median_ms: 81466
-    classification: authoritative_three_run_process_wall_baseline
-  resolution: >-
-    The plan motivation's 79.1-second value was an exploratory one-run
-    observation used to justify measurement work. Task 1's protocol supersedes
-    it with the retained three-run 81.466-second median. The immutable plan
-    intent remains unchanged; later comparisons use 81.466 seconds.
-
-source_samples:
-  - sample: 1
-    status: pass
-    wall_ms: 86237
-    runner_duration_ms: 85620
-    json_elapsed_ms: 85628.146
-    setup_ms: 0
-    transform_ms: 2050
-    import_ms: 14550
-    cumulative_assertion_ms: 292922
-    files: 191
-    tests: 2230
-  - sample: 2
-    status: pass
-    wall_ms: 81090
-    runner_duration_ms: 80520
-    json_elapsed_ms: 80526.117
-    setup_ms: 0
-    transform_ms: 1760
-    import_ms: 14100
-    cumulative_assertion_ms: 280110
-    files: 191
-    tests: 2230
-  - sample: 3
-    status: pass
-    wall_ms: 81466
-    runner_duration_ms: 80900
-    json_elapsed_ms: 80904.892
-    setup_ms: 0
-    transform_ms: 1730
-    import_ms: 13970
-    cumulative_assertion_ms: 280978
-    files: 191
-    tests: 2230
-
-source_median:
-  wall_ms: 81466
-  runner_duration_ms: 80900
-  json_elapsed_ms: 80904.892
-  setup_ms: 0
-  transform_ms: 1760
-  import_ms: 14100
-  cumulative_assertion_ms: 280978
-  workers: 4
-  files: 191
-  tests: 2230
-
-slowest_tests_by_median:
-  - file: test/state/entityStorage.test.ts
-    test: repeatedly gives simultaneous stale-lock reclaimers one publisher and one explicit loser
-    samples_ms: [21905, 18628, 18791]
-    median_ms: 18791
-  - file: test/state/entityMigrationPreview.test.ts
-    test: uses snapshot-bound whole-entry pagination to recover every entry without gaps or duplicates
-    samples_ms: [18325, 16575, 16680]
-    median_ms: 16680
-  - file: test/cli/primeProjectionContract.test.ts
-    test: keeps cold entity-mode startup and retrieval within authority budgets at declared scales
-    samples_ms: [11256, 10960, 11049]
-    median_ms: 11049
-  - file: test/analytics/tierReader.test.ts
-    test: reports oversized with the artifact path when a shard exceeds the cap
-    samples_ms: [13089, 10497, 10245]
-    median_ms: 10497
-  - file: test/analytics/extractCorpusParity.test.ts
-    test: matches record_count, earliest, and latest across probe shapes for TS
-    samples_ms: [10715, 9891, 9864]
-    median_ms: 9891
-  - file: test/analytics/tierReader.test.ts
-    test: retains an oversized record in its own shard and reports oversized state
-    samples_ms: [9350, 8554, 8589]
-    median_ms: 8589
-  - file: test/analytics/extractCorpusParity.test.ts
-    test: matches TypeScript extractor probe on seeded opencode.db
-    samples_ms: [8242, 7416, 7161]
-    median_ms: 7416
-  - file: test/analytics/extractCorpusParity.test.ts
-    test: matches generated Python extractor probe on seeded opencode.db
-    samples_ms: [6416, 6397, 6400]
-    median_ms: 6400
-  - file: test/state/planEntities.test.ts
-    test: keeps one active plan unambiguous beside more than twenty archived plans
-    samples_ms: [5629, 5632, 5725]
-    median_ms: 5632
-  - file: test/cli/primeCompatibilityBoundary.test.ts
-    test: declares exactly the fields the prime selector accepts
-    samples_ms: [4158, 4055, 4055]
-    median_ms: 4055
-
-pre_commit_samples:
-  - sample: representative_cli
-    staged_path: packages/cli/src/cli/todoMarkdown.ts
-    status: pass
-    wall_ms: 6227
-    lefthook_ms: 5900
-    selected_policies:
-      - name: agentera
-        command: pnpm -C packages/cli build && node packages/cli/dist/bin/agentera.js check compact
-        duration_ms: 2950
-      - name: py-ts-parity
-        command: bash packages/cli/scripts/py_ts_parity.sh --check --json
-        duration_ms: 30
-      - name: tests
-        command: bash scripts/precommit-vitest.sh packages/cli/src/cli/todoMarkdown.ts
-        duration_ms: 2920
-        route: targeted
-        targets:
-          - test/cli/todoMarkdown.test.ts
-          - test/cli/state.test.ts
-          - test/cli/orientation.test.ts
-  - sample: representative_documentation
-    staged_path: README.md
-    status: pass
-    wall_ms: 3421
-    lefthook_ms: 3100
-    selected_policies:
-      - name: agentera
-        command: pnpm -C packages/cli build && node packages/cli/dist/bin/agentera.js check compact
-        duration_ms: 2960
-      - name: docs
-        command: bunx markdownlint --dot --fix
-        duration_ms: 100
-      - name: py-ts-parity
-        command: bash packages/cli/scripts/py_ts_parity.sh --check --json
-        duration_ms: 30
-    unselected_policies:
-      - name: tests
-        reason: root README.md is outside the pre-commit tests glob
-
-raw_evidence:
-  root: references/analysis/evidence/verification-baseline-2026-07-20
-  source_reports:
-    - sample: 1
-      path: references/analysis/evidence/verification-baseline-2026-07-20/source-1.json.gz
-      format: vitest_json
-      compression: gzip_lossless_mtime_zero
-      compressed_bytes: 94678
-      compressed_sha256: 8431150fe302c63493387e7f6b0893309f1fc5c56bb32e5ed7a54ce842f5c277
-      uncompressed_bytes: 769383
-      uncompressed_sha256: 65014942bfe5111d48b2d871a326bb995b5ae258d94dd2d54273dc6273705eee
-    - sample: 2
-      path: references/analysis/evidence/verification-baseline-2026-07-20/source-2.json.gz
-      format: vitest_json
-      compression: gzip_lossless_mtime_zero
-      compressed_bytes: 94578
-      compressed_sha256: ed1dab5caabf0155e3ebc54f15830028abf0d6405bda13e1b9544679d1a07af0
-      uncompressed_bytes: 769397
-      uncompressed_sha256: a2672026c9d41ae69861145fe41087ddabcff7ee7dfd8dfb78fb73bfb67ea74c
-    - sample: 3
-      path: references/analysis/evidence/verification-baseline-2026-07-20/source-3.json.gz
-      format: vitest_json
-      compression: gzip_lossless_mtime_zero
-      compressed_bytes: 94695
-      compressed_sha256: 6b9f33dff97d886a4f004474b8049b4fc81e524d56e0329f2f431694cfeadefe
-      uncompressed_bytes: 769322
-      uncompressed_sha256: e31078a8eddb48a199b699b7d84739984fc5ec4a1a2d56b20183fabf4460aae5
-  pre_commit_reports:
-    - sample: representative_cli
-      path: references/analysis/evidence/verification-baseline-2026-07-20/precommit-cli.log.gz
-      format: lefthook_text_with_json_wall_footer
-      compression: gzip_lossless_mtime_zero
-      compressed_bytes: 176
-      compressed_sha256: 5515ce69ebd4c2700f112868322db8cbe7e804c5a952b0596bee3def9bba7c41
-      uncompressed_bytes: 210
-      uncompressed_sha256: eecbcb3fac333762ada9a7a6071fea791e26e093c025c83c4298cc3d2d26fe88
-      route_path: references/analysis/evidence/verification-baseline-2026-07-20/precommit-cli-route.log.gz
-      route_format: precommit_vitest_plain_text_route
-      route_compression: gzip_lossless_mtime_zero
-      route_compressed_bytes: 83
-      route_compressed_sha256: a50db85abdfcd27743bd181d19a3d2ce183c6822d9cd90b7b6a47d688d521797
-      route_uncompressed_bytes: 116
-      route_uncompressed_sha256: d8bf45b9e554ea03d2b5fa4efd90f397c31620dfdaf0c005d597c766f6bfd4b2
-    - sample: representative_documentation
-      path: references/analysis/evidence/verification-baseline-2026-07-20/precommit-documentation.log.gz
-      format: lefthook_text_with_json_wall_footer
-      compression: gzip_lossless_mtime_zero
-      compressed_bytes: 180
-      compressed_sha256: 047cbeb4e445f7792872a10c0dd640fa191a51c34097f4b319dd653806dca9b4
-      uncompressed_bytes: 219
-      uncompressed_sha256: acc0f47b9582805b74fd29cf5a54611a8149b07bbccb550736b7dce78eba66b2
-  recomputation:
-    metrics_command: node references/analysis/evidence/verification-baseline-2026-07-20/recompute.mjs
-    compressed_hash_command: sha256sum references/analysis/evidence/verification-baseline-2026-07-20/*.gz
-    uncompressed_hash_command: for f in references/analysis/evidence/verification-baseline-2026-07-20/*.gz; do gzip -dc "$f" | sha256sum; done
-    source_metrics: >-
-      The script recomputes exact JSON elapsed times, cumulative assertion sums,
-      file/test totals, three-run medians, and the top-ten median assertion
-      ranking directly from all three reports.
-    hook_metrics: >-
-      The script parses lefthook total and policy durations plus the outer wall
-      JSON footer, and reports the retained CLI route targets.
-
-production_scale_ownership:
-  - requirement: entity startup, bounded list, and exact retrieval at 100 and 1000 entities across every declared boundary, five cold-process repetitions
-    authority: references/artifacts/state-storage-authority.yaml#entity_target.measurement_contract
-    current_owner: test/performance/entityAuthorityPerformance.test.ts
-    current_policy: scheduled_and_release
-    intended_gates: [scheduled, release]
-    merge_evidence: retain a bounded functional source smoke when the performance owner is separated
-  - requirement: archive enumeration at 100 and 1000 entries within latency, heap, and 32768-byte response bounds
-    authority: references/artifacts/state-storage-authority.yaml#budgets.list.benchmark
-    current_owner: test/state/listRetrieval.test.ts
-    current_policy: merge
-    intended_gates: [scheduled, release]
-    merge_evidence: retain deterministic response-bound and pagination correctness coverage
-  - requirement: whole-entry migration pagination recovers every identity without gaps or duplicates under the production response cap
-    authority: references/artifacts/state-storage-authority.yaml#budgets.projection
-    current_owner: test/state/entityMigrationPreview.test.ts
-    current_policy: merge
-    intended_gates: [merge, release]
-    merge_evidence: production-cap correctness remains merge-blocking until an authority amendment says otherwise
-  - requirement: 64 MiB analytics shard, signal, and reader bounds reject oversized input without loading it whole
-    authority: references/analysis/evidence-tier-authority.yaml#bounds
-    current_owner: test/analytics/oversizedRegression.test.ts and test/analytics/tierReader.test.ts
-    current_policy: merge
-    intended_gates: [merge, release]
-    merge_evidence: sparse files and constant assertions make the production bound deterministic and proportional
-  - requirement: dead-process writer-lock reclamation preserves one publisher, one explicit loser, and no residue under repetition
-    authority: references/artifacts/state-storage-authority.yaml#entity_target.storage_boundary.shared_primitives.project_wide_id_claim.interrupted_writer_recovery
-    current_owner: test/state/entityStorage.test.ts
-    current_policy: merge
-    intended_gates: [scheduled, release]
-    merge_evidence: retain deterministic lock correctness and run stress on relevant writer changes until the owner split is proven
-
-risk_decisions:
-  fast_policy:
-    decision: fast is an execution policy that composes evidence owners; it never owns evidence
-  source_package_boundary:
-    decision: source and package proof remain distinct owners and commands
-  worker_count:
-    state: unresolved
-    current_default_workers_on_runner: 4
-    upper_bound_warning: 12 workers previously failed under contention; this task did not reproduce or supersede that exploratory result
-    resolution_criteria:
-      - isolate generated outputs before testing higher concurrency
-      - benchmark each CI runner rather than infer capacity from logical CPU count
-      - consider only candidates at or below 8 workers on this runner
-      - require three consecutive warm passes with no timeout, retry, shared-output race, or hidden failure
-      - select only a candidate that meets the plan's final wall-time and improvement targets
-  lanes:
-    state: unresolved
-    resolution_criteria:
-      - trace every test file to exactly one source, stress, performance, or package owner
-      - split mixed files before assigning policy
-      - keep deterministic correctness and security failures merge-blocking
-      - move probabilistic stress and machine-sensitive performance only after proportional merge evidence exists
-      - compose named owners for targeted, local, merge, scheduled, and release policies
-      - keep fast as a policy composition, never an owner
-
-caveats:
-  - The decisions query was degraded and compacted; missing decision detail was not reconstructed.
-  - Firm runtime, merge, and lifecycle constraints were preserved, but llavvlpcmb and djmjbmruyk had no satisfaction record.
-  - Build orchestration context lacked decisions, vision, and profile families; no refresh or inferred approval was performed.
-  - Source timings include normal local-run variability; the excluded warmup was materially slower than retained runs.
-  - Outer process-wall samples are the measurement wrapper's structured records; retained JSON reports independently reproduce the separately reported Vitest duration median within 5ms and exactly reproduce assertion, ranking, and count metrics.
-  - Earlier 8-worker and 12-worker exploration is context only, not baseline evidence from this protocol.