agentera 3.0.0-dev.26 → 3.0.0-dev.27

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

@@ -0,0 +1,313 @@
+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.

package/bundle/references/analysis/verification-final-proof-2026-07-21.yaml

@@ -0,0 +1,340 @@
+schemaVersion: agentera.verificationFinalProof.v1
+status: accepted
+captured: 2026-07-21
+task: ymkdnqmroi
+
+scope:
+  authoritative_policy: references/analysis/verification-policy.yaml
+  historical_baseline: references/analysis/verification-baseline-2026-07-20.yaml
+  measured_selector: AGENTERA_VITEST_RUNNER_POLICY=local-16-logical-cpu-node22
+  worker_policy: references/analysis/worker-policy-2026-07-21.yaml
+  measured_parent: 2e84ba028a3c8ec76649e06b00e21e9452b4ef60
+  parent_worktree_diff_sha256: 58d4b87c4caa29990af3a32238a96bd89503016db4d74aaaf4c088df8d18bba6
+  runner:
+    name: local-16-logical-cpu-node22
+    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
+    resolved_workers: 8
+  final_inventory_sha256: 37b16a0224f75b2160fd85094d3a9ee3d77a6ac31eaf36f2f2df600a3cb54e61
+  final_input_sha256:
+    packages/cli/test/state/planEntities.test.ts: 5b08500e57f31b13d89201e490c35aaa9fd73ee25ac70ecfbb431dc6bfef868e
+    packages/cli/test/cli/primeCompatibilityBoundary.test.ts: a1b0000b0534908ce380c4c604973689e149654faf0ef6b22b107fe53dc35ffe
+    packages/cli/test/cli/activeProtocolSurface.test.ts: 7f4f44267adaad08209c3d8346ba4fbe45cdefa223560cbe494eef5e92ece4e0
+    packages/cli/test/cli/lifecycleProtocol.test.ts: 0eb13ce63e8b4078e3aa566633bca850a0840ffafeac1ade96ab9b39344b0435
+    packages/cli/test/cli/retiredVocabulary.ts: 60d6282180579ec5edc0f8bed7e285e21629c3191d1e0995b76a0bb35462aba6
+    packages/cli/test/verification/laneOwnership.test.ts: 67388a1cb38c430351ddef22a59afdf981c616b0f40fe5cf0710718a817cb51f
+    docs/packaging/v3-packaging.md: f9fab7467c3740d2c5a5c185846bb2ab8efcdfdd48bcc69cd4958c6c1fb0f5fb
+
+controlled_parent_block:
+  purpose: establish the exact audit-1 retry tree before proof, classification, or optimization edits
+  process_conditions:
+    captured_at: 2026-07-21T15:42:00+02:00
+    one_user: true
+    competing_cpu_bound_process_observed: false
+    VITEST_MAX_WORKERS: unset
+    selector_resolution: named selector resolved to eight workers
+  command: >-
+    AGENTERA_VITEST_RUNNER_POLICY=local-16-logical-cpu-node22
+    AGENTERA_VERIFICATION_RESULT=/tmp/opencode/ymkdnqmroi-current-<phase>.json
+    pnpm -C packages/cli run test:source
+  warmup:
+    retained: false
+    wall_seconds: 45.765
+    status: pass
+  retained_runs:
+    - sample: 1
+      wall_seconds: 46.401
+      files: 194
+      tests_passed: 2369
+      tests_skipped: 1
+    - sample: 2
+      wall_seconds: 45.973
+      files: 194
+      tests_passed: 2369
+      tests_skipped: 1
+    - sample: 3
+      wall_seconds: 45.697
+      files: 194
+      tests_passed: 2369
+      tests_skipped: 1
+  median_wall_seconds: 45.973
+  result: fail
+  interpretation: >-
+    The exact audit-1 retry tree still exceeded 45 seconds without one outlier,
+    so fixture/setup optimization was required. The JSON-only owner channel did
+    not emit Vitest transform/import dimensions; those are not reconstructed.
+
+root_cause:
+  finding: >-
+    Source tests repeatedly rebuilt equivalent fixture state. The prime
+    compatibility selector ran one complete prime projection per declared field;
+    the archived-plan read test ran 21 already-covered public create/archive
+    cycles; and activeProtocolSurface regenerated every capability context even
+    though it already scans every capability instruction while lifecycleProtocol
+    separately generates every context for canonical plan projection.
+  correction: >-
+    Batch all declared fields through the documented comma-separated selector,
+    seed canonical archived read fixtures while retaining the separate public
+    lifecycle/archive acceptance, capture one representative runtime context in
+    activeProtocolSurface while retaining all-instruction scanning, and reconcile
+    retired vocabulary in lifecycleProtocol's already-generated all-capability
+    contexts alongside its canonical plan assertion.
+  assertions_removed: 0
+  tests_skipped_or_deleted: 0
+  thresholds_changed: false
+
+ownership:
+  inventory:
+    total_files: 200
+    source: 194
+    stress: 1
+    performance: 3
+    package: 2
+    mixed_files: 0
+    result: pass
+  independent_owners:
+    - owner: source
+      command: pnpm -C packages/cli run test:source
+      wall_seconds: 43.194
+      files: 194
+      passed_tests: 2369
+      skipped_tests: 1
+      result: pass
+    - owner: stress
+      command: pnpm -C packages/cli run test:stress
+      wall_seconds: 19.668
+      files: 1
+      passed_tests: 2
+      result: pass
+    - owner: performance
+      command: pnpm -C packages/cli run test:performance
+      wall_seconds: 18.495
+      files: 3
+      passed_tests: 3
+      structured_evidence: pass
+      result: pass
+    - owner: package
+      command: pnpm -C packages/cli run verify:package
+      wall_seconds: 16.533
+      files: 2
+      passed_tests: 10
+      result: pass
+
+policy_compositions:
+  result: pass
+  runs:
+    - { policy: targeted, owners: [source], wall_seconds: 42.701, result: pass }
+    - { policy: precommit, owners: [source], wall_seconds: 42.754, result: pass }
+    - { policy: fast, owners: [source], wall_seconds: 43.486, result: pass }
+    - { policy: local, owners: [source], wall_seconds: 42.343, result: pass }
+    - { policy: merge, owners: [source, package], wall_seconds: 50.191, result: pass }
+    - { policy: scheduled, owners: [source, stress, performance], wall_seconds: 81.380, result: pass }
+    - { policy: release, owners: [source, stress, performance, package], wall_seconds: 98.976, result: pass }
+  interpretation: >-
+    Every canonical policy executed the exact ordered owner list declared by the
+    policy authority. Inventory validation found no gap, overlap, or mixed file.
+
+final_measurement:
+  classification_boundary: >-
+    All code, test, packaging documentation, and evidence-file location choices
+    were final before this block. Retained evidence lives under evidence/ rather
+    than package-copied references/, so retaining reports does not alter the
+    source owner inventory or package fixture input. Only typed state records are
+    appended after this proof.
+  process_conditions:
+    captured_at: 2026-07-21T16:32:32+02:00
+    load_average: [1.11, 2.78, 5.52]
+    cpu_temperature_celsius: 44.2
+    one_user: true
+    competing_cpu_bound_process_observed: false
+    VITEST_MAX_WORKERS: unset
+    selector_resolution: named selector resolved to eight workers
+  command: >-
+    Resolve files.source from `node packages/cli/scripts/verify-lane.mjs
+    inventory --json`, then from packages/cli run `AGENTERA_VITEST_RUNNER_POLICY=
+    local-16-logical-cpu-node22 pnpm exec vp test run --config vite.config.ts
+    --reporter=default --reporter=json --outputFile=<sample>.json <exact files>`.
+  wall_definition: Bash process wall around the complete pnpm observer
+  dimensions_definition: Vitest default reporter from the same process
+  cumulative_definition: sum of JSON assertion durations; concurrent assertions overlap
+  warmup:
+    retained: false
+    wall_seconds: 41.675
+    vitest_duration_seconds: 41.09
+    transform_seconds: 4.00
+    setup_seconds: 0
+    import_seconds: 21.00
+    cumulative_test_seconds: 276.35
+    status: pass
+  retained_runs:
+    - sample: 1
+      wall_seconds: 41.803
+      vitest_duration_seconds: 41.23
+      transform_seconds: 4.14
+      setup_seconds: 0
+      import_seconds: 21.07
+      cumulative_test_seconds: 277.24
+      cumulative_assertion_seconds: 268.853
+      files: 194
+      tests_passed: 2369
+      tests_skipped: 1
+    - sample: 2
+      wall_seconds: 43.020
+      vitest_duration_seconds: 42.45
+      transform_seconds: 4.30
+      setup_seconds: 0
+      import_seconds: 22.35
+      cumulative_test_seconds: 284.94
+      cumulative_assertion_seconds: 276.478
+      files: 194
+      tests_passed: 2369
+      tests_skipped: 1
+    - sample: 3
+      wall_seconds: 42.908
+      vitest_duration_seconds: 42.33
+      transform_seconds: 4.53
+      setup_seconds: 0
+      import_seconds: 22.00
+      cumulative_test_seconds: 284.11
+      cumulative_assertion_seconds: 275.542
+      files: 194
+      tests_passed: 2369
+      tests_skipped: 1
+  median:
+    wall_seconds: 42.908
+    vitest_duration_seconds: 42.33
+    transform_seconds: 4.30
+    setup_seconds: 0
+    import_seconds: 22.00
+    cumulative_test_seconds: 284.11
+    cumulative_assertion_seconds: 275.542
+    files: 194
+    tests_total: 2370
+
+comparison_to_four_worker_baseline:
+  baseline:
+    wall_seconds: 81.466
+    vitest_duration_seconds: 80.90
+    transform_seconds: 1.76
+    setup_seconds: 0
+    import_seconds: 14.10
+    cumulative_assertion_seconds: 280.978
+    files: 191
+    tests: 2230
+    workers: 4
+  final:
+    wall_seconds: 42.908
+    vitest_duration_seconds: 42.33
+    transform_seconds: 4.30
+    setup_seconds: 0
+    import_seconds: 22.00
+    cumulative_assertion_seconds: 275.542
+    files: 194
+    tests: 2370
+    workers: 8
+  arithmetic:
+    wall_seconds_saved: 38.558
+    wall_improvement_percent: 47.330
+    file_count_change: 3
+    file_count_change_percent: 1.571
+    test_count_change: 140
+    test_count_change_percent: 6.278
+    cumulative_assertion_time_change_percent: -1.935
+  inventory_caveat: >-
+    The historical baseline has three fewer files and 140 fewer tests. Wall
+    improvement is process-level acceptance evidence; component and cumulative
+    assertion comparisons are not like-for-like speed claims.
+
+slowest_tests:
+  audit_1_before_observer:
+    - { test: planEntities active beside archived plans, milliseconds: 7717 }
+    - { test: primeCompatibility declared selector fields, milliseconds: 5533 }
+    - { test: activeProtocol retired vocabulary surface, milliseconds: 4256 }
+    - { test: lifecycleProtocol canonical startup and compact, milliseconds: 3868 }
+    - { test: listRetrieval bounded yaml cursor, milliseconds: 3522 }
+  final_target_medians:
+    - { test: planEntities active beside archived plans, milliseconds: 587.414 }
+    - { test: primeCompatibility declared selector fields, milliseconds: 359.191 }
+    - { test: activeProtocol retired vocabulary surface, milliseconds: 1290.204 }
+    - { test: lifecycleProtocol canonical startup and compact, milliseconds: 3896.026 }
+  final_top_five_medians:
+    - { test: lifecycleProtocol canonical startup and compact, milliseconds: 3896.026 }
+    - { test: listRetrieval bounded yaml cursor, milliseconds: 3593.758 }
+    - { test: listRetrieval bounded json cursor, milliseconds: 3493.647 }
+    - { test: listRetrieval bounded text cursor, milliseconds: 3396.428 }
+    - { test: prime projection emoji boundary, milliseconds: 3247.478 }
+
+raw_evidence:
+  root: evidence/ymkdnqmroi
+  compression: deterministic gzip with mtime zero
+  files:
+    current-warmup.log.gz: 9dbc7e00caece9d35f509817c8e7c6ccbefe670e69c8811a988b2cba66ffb24d
+    current-run1.log.gz: 7948e21a0106f62cedaf8f1b4183f8d44a37ab6e03992f42b8b21ef03902b989
+    current-run2.log.gz: 99f212158e94d028eff9c7e74e13686f2a505866604658fe0f0a9136ba0d90c1
+    current-run3.log.gz: ee9f52621e5a842cce784ab1b7dd549f516c563160d58b1a3da35f87f5a2ed41
+    final-warmup.log.gz: 465a27912a194b452020ac0aa74cd765bec0eed6e4108f9817d46af648177779
+    final-run1.log.gz: 824d383262fe2f80a882ba271e07ee7a8630cd4bce6a41106b9c9b026b54c6d0
+    final-run1.json.gz: eec0244f5f18ff2050d0b40d8ea594653017f5c472a7cc6dbea8753b83ade1c7
+    final-run2.log.gz: 69526e343d003d69f470c7906935d3fb183b2c2a77864b5e7eda0c5bbef8a989
+    final-run2.json.gz: f7240ced75283ef764855856eea4bc37a91dd0aac11f13416572e740e555fac3
+    final-run3.log.gz: c7ec7f4c0fbe4d544a1ab13ec996c2fceae2c49aaede16ed85a98dfe6ee8e11b
+    final-run3.json.gz: eba2c5fe1f58234ba9b20b6ffe0d417f3db6088c421d9386ff5a62f444d75385
+  original_current_json_sha256:
+    warmup: 07c1bdf72bbbee6d150cd2d3e39f4e58d43e71c092c40f798ee9f7d62b1b6865
+    run1: d3ec0f4d95351c0c889eb41e6488a351c0b5a75437bfaf8c3561958c567b35bb
+    run2: ebcec8e6e61fe75d8bc695ec7f564e24063f6a62ae841eff363340a115482cfe
+    run3: dafa94e46ccc8df16130cf3c17f62bca863046ceae445d10b1365fc2265490f5
+  recompute: >-
+    Run `sha256sum evidence/ymkdnqmroi/*`; use `gzip -dc` on each log to
+    recover timestamps, command, Vitest dimensions, and Bash wall; parse each
+    final JSON with Node, sum finite assertionResults[].duration values, and
+    median the three per-assertion durations for slow-test ranking.
+
+gates:
+  - { command: pnpm -C packages/cli run verify:generated-overlap, result: pass, detail: source 194/2370; package 2/10; build pass; zero mismatches }
+  - { command: pnpm -C packages/cli run typecheck, result: pass }
+  - { command: pnpm -C packages/cli build, result: pass }
+  - { command: node packages/cli/dist/bin/agentera.js check validate capability-contract --format json, result: pass }
+  - { command: node packages/cli/dist/bin/agentera.js check compact, result: pass }
+  - { command: pnpm -C packages/cli run pack:dry-run, result: pass, detail: 686 package entries }
+  - { command: bash packages/cli/scripts/py_ts_parity.sh --check --json, result: pass, detail: drift none }
+  - { command: bunx markdownlint docs/packaging/v3-packaging.md TODO.md, result: pass }
+
+narrow_repair_verification:
+  focused_runtime_and_ownership:
+    command: >-
+      pnpm -C packages/cli run test:source --
+      test/cli/activeProtocolSurface.test.ts test/cli/lifecycleProtocol.test.ts
+      test/verification/laneOwnership.test.ts
+    result: pass
+    detail: 3 files, 84 tests; lifecycleProtocol reconciled all 12 already-generated capability payloads, and lane ownership remained green.
+  ordinary_source:
+    command: >-
+      AGENTERA_VITEST_RUNNER_POLICY=local-16-logical-cpu-node22
+      pnpm -C packages/cli run test:source
+    result: pass
+    detail: 194 files, 2369 passed, 1 skipped; Vitest duration 42.83 seconds.
+  typecheck:
+    command: pnpm -C packages/cli run typecheck
+    result: pass
+
+acceptance:
+  owners_independent: pass
+  policy_compositions: pass
+  worker_count_at_most_eight: pass
+  median_wall_at_most_45_seconds: pass
+  wall_improvement_at_least_35_percent: pass
+  documentation_synchronized: pass
+  changelog_unchanged: pass
+  conclusion: qgkxggjber is complete; ymkdnqmroi remains blocked as audit-2 historical provenance.
+
+caveats:
+  - Approximately two seconds of ordinary wall-time variance on this non-isolated developer machine is nonblocking and not a refactor trigger; the retained meaningful improvement, eight-worker bound, and functional correctness evidence remain accepted.
+  - Decisions retrieval remained degraded to 5 of 50; omitted records were not reconstructed.
+  - Satisfaction remains unavailable for llavvlpcmb and djmjbmruyk.
+  - Vision and profile were absent; no approval was inferred.
+  - The missing NPM_TOKEN warning is nonblocking and no publication was attempted.