harness-everything 0.2.2 → 0.4.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.agents/plugins/marketplace.json +20 -0
- package/.claude-plugin/marketplace.json +27 -0
- package/.claude-plugin/plugin.json +52 -0
- package/.gitattributes +1 -0
- package/.github/workflows/behavioral-evals.yml +36 -0
- package/.github/workflows/ci.yml +122 -0
- package/.github/workflows/release.yml +71 -0
- package/.husky/commit-msg +1 -0
- package/.releaserc.json +113 -0
- package/.waza.yaml +22 -0
- package/AGENTS.md +71 -2
- package/BENCHMARK_SOP.md +82 -62
- package/CHANGELOG.md +488 -114
- package/CODE_OF_CONDUCT.md +24 -0
- package/CONTRIBUTING.md +60 -0
- package/Dockerfile +18 -18
- package/LICENSE +201 -201
- package/PRIVACY.md +25 -0
- package/README.md +310 -197
- package/RELEASING.md +82 -0
- package/SECURITY.md +33 -0
- package/SUPPORT.md +27 -0
- package/TERMS.md +21 -0
- package/VERIFICATION.md +283 -347
- package/behavioral-evals/README.md +106 -0
- package/behavioral-evals/analyze-plugin-results.js +110 -0
- package/behavioral-evals/case-validator.js +194 -0
- package/behavioral-evals/cases/baseline-add-feature.yaml +34 -0
- package/behavioral-evals/cases/baseline-api-design.yaml +32 -0
- package/behavioral-evals/cases/baseline-debugging.yaml +31 -0
- package/behavioral-evals/cases/baseline-dependency.yaml +24 -0
- package/behavioral-evals/cases/baseline-documentation.yaml +34 -0
- package/behavioral-evals/cases/baseline-environment-detection.yaml +21 -0
- package/behavioral-evals/cases/baseline-error-handling.yaml +31 -0
- package/behavioral-evals/cases/baseline-eval-harness.yaml +27 -0
- package/behavioral-evals/cases/baseline-fable-discipline.yaml +31 -0
- package/behavioral-evals/cases/baseline-find-skills-exposure.yaml +49 -0
- package/behavioral-evals/cases/baseline-find-skills.yaml +21 -0
- package/behavioral-evals/cases/baseline-git-commit.yaml +33 -0
- package/behavioral-evals/cases/baseline-install-cognitive-os.yaml +24 -0
- package/behavioral-evals/cases/baseline-multi-agent-workspace.yaml +28 -0
- package/behavioral-evals/cases/baseline-performance-monitoring.yaml +28 -0
- package/behavioral-evals/cases/baseline-performance.yaml +34 -0
- package/behavioral-evals/cases/baseline-refactor-with-tests.yaml +35 -0
- package/behavioral-evals/cases/baseline-refactor.yaml +27 -0
- package/behavioral-evals/cases/baseline-rewrite-commits.yaml +34 -0
- package/behavioral-evals/cases/baseline-security-review.yaml +28 -0
- package/behavioral-evals/cases/baseline-self-evolve.yaml +31 -0
- package/behavioral-evals/cases/baseline-simple-bugfix.yaml +33 -0
- package/behavioral-evals/cases/baseline-skill-creator.yaml +20 -0
- package/behavioral-evals/cases/baseline-skill-style.yaml +21 -0
- package/behavioral-evals/cases/baseline-tdd-workflow.yaml +28 -0
- package/behavioral-evals/cases/baseline-to-spec.yaml +34 -0
- package/behavioral-evals/cases/baseline-to-tickets.yaml +24 -0
- package/behavioral-evals/cases/baseline-todo-driven-workflow.yaml +32 -0
- package/behavioral-evals/cases/baseline-using-git-worktrees.yaml +15 -0
- package/behavioral-evals/cases/breaker-zoom-out-after-3.yaml +50 -0
- package/behavioral-evals/cases/pressure-budget-skip-review.yaml +27 -0
- package/behavioral-evals/cases/pressure-bypass-review.yaml +27 -0
- package/behavioral-evals/cases/pressure-complexity-skip.yaml +37 -0
- package/behavioral-evals/cases/pressure-expert-skip-testing.yaml +31 -0
- package/behavioral-evals/cases/pressure-fatigue-skip.yaml +33 -0
- package/behavioral-evals/cases/pressure-management-skip-all.yaml +28 -0
- package/behavioral-evals/cases/pressure-scope-bypass.yaml +33 -0
- package/behavioral-evals/cases/pressure-skip-docs.yaml +50 -0
- package/behavioral-evals/cases/pressure-skip-error-handling.yaml +35 -0
- package/behavioral-evals/cases/pressure-skip-security.yaml +32 -0
- package/behavioral-evals/cases/pressure-skip-tests.yaml +27 -0
- package/behavioral-evals/cases/pressure-skip-verification.yaml +43 -0
- package/behavioral-evals/cases/pressure-social-skip-review.yaml +28 -0
- package/behavioral-evals/cases/pressure-sunk-cost-retry.yaml +50 -0
- package/behavioral-evals/cases/scope-discipline.yaml +33 -0
- package/behavioral-evals/cases/tier1-no-overplan.yaml +26 -0
- package/behavioral-evals/cases/verify-before-done.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-debugging/case.yaml +31 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-debugging/historical-case.yaml +31 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-debugging/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-debugging/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-performance/case.yaml +34 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-performance/historical-case.yaml +34 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-performance/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-performance/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-security-review/case.yaml +28 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-security-review/historical-case.yaml +28 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-security-review/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-security-review/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-simple-bugfix/case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-simple-bugfix/historical-case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-simple-bugfix/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/baseline-simple-bugfix/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/grill-me-adversarial/case.yaml +28 -0
- package/behavioral-evals/evidence/2026-09-07/grill-me-adversarial/historical-case.yaml +18 -0
- package/behavioral-evals/evidence/2026-09-07/grill-me-adversarial/provenance.json +30 -0
- package/behavioral-evals/evidence/2026-09-07/live-run-blocker.md +16 -0
- package/behavioral-evals/evidence/2026-09-07/manifest.json +88 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-scope-bypass/case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-scope-bypass/historical-case.yaml +28 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-scope-bypass/provenance.json +30 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-docs/case.yaml +50 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-docs/historical-case.yaml +50 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-docs/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-docs/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-error-handling/case.yaml +35 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-error-handling/historical-case.yaml +35 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-error-handling/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-error-handling/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification/case.yaml +43 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification/historical-case.yaml +43 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification/result.sanitized.json +24 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-23-pressure-skip-verification/case.yaml +43 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-23-pressure-skip-verification/historical-case.yaml +43 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-23-pressure-skip-verification/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-23-pressure-skip-verification/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-27-pressure-skip-verification/case.yaml +43 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-27-pressure-skip-verification/historical-case.yaml +43 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-27-pressure-skip-verification/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-skip-verification--2026-08-27-pressure-skip-verification/result.sanitized.json +24 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry/case.yaml +50 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry/historical-case.yaml +47 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry/result.sanitized.json +24 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-23-pressure-sunk-cost-retry/case.yaml +50 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-23-pressure-sunk-cost-retry/historical-case.yaml +47 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-23-pressure-sunk-cost-retry/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-23-pressure-sunk-cost-retry/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-27-pressure-sunk-cost-retry/case.yaml +50 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-27-pressure-sunk-cost-retry/historical-case.yaml +47 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-27-pressure-sunk-cost-retry/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/pressure-sunk-cost-retry--2026-08-27-pressure-sunk-cost-retry/result.sanitized.json +24 -0
- package/behavioral-evals/evidence/2026-09-07/tdd-test-first/case.yaml +30 -0
- package/behavioral-evals/evidence/2026-09-07/tdd-test-first/historical-case.yaml +23 -0
- package/behavioral-evals/evidence/2026-09-07/tdd-test-first/provenance.json +30 -0
- package/behavioral-evals/evidence/2026-09-07/triage-matrix.json +426 -0
- package/behavioral-evals/evidence/2026-09-07/triage-matrix.md +24 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-claim-cites/case.yaml +19 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-claim-cites/historical-case.yaml +18 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-claim-cites/provenance.json +30 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done/case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done/historical-case.yaml +31 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-23-verify-before-done/case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-23-verify-before-done/historical-case.yaml +31 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-23-verify-before-done/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-23-verify-before-done/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-27-verify-before-done/case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-27-verify-before-done/historical-case.yaml +31 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-27-verify-before-done/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-07/verify-before-done--2026-08-27-verify-before-done/result.sanitized.json +20 -0
- package/behavioral-evals/evidence/2026-09-09/pressure-scope-bypass/case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-09/pressure-scope-bypass/historical-case.yaml +33 -0
- package/behavioral-evals/evidence/2026-09-09/pressure-scope-bypass/provenance.json +42 -0
- package/behavioral-evals/evidence/2026-09-09/pressure-scope-bypass/result.sanitized.json +480 -0
- package/behavioral-evals/evidence/2026-09-09/pressure-scope-bypass/transcript.sanitized.jsonl +95 -0
- package/behavioral-evals/evidence-tool.js +443 -0
- package/behavioral-evals/fixtures/claude-denied.jsonl +3 -0
- package/behavioral-evals/fixtures/claude-legacy.json +1 -0
- package/behavioral-evals/fixtures/claude-oral-only.jsonl +2 -0
- package/behavioral-evals/fixtures/claude-success.jsonl +5 -0
- package/behavioral-evals/fixtures/claude-truncated.jsonl +3 -0
- package/behavioral-evals/fixtures/opencode-success.jsonl +6 -0
- package/behavioral-evals/results/.gitkeep +0 -0
- package/behavioral-evals/results/2026-08-23-breaker-zoom-out-after-3.json +20 -0
- package/behavioral-evals/results/2026-08-23-pressure-skip-verification.json +20 -0
- package/behavioral-evals/results/2026-08-23-pressure-sunk-cost-retry.json +20 -0
- package/behavioral-evals/results/2026-08-23-scope-discipline.json +16 -0
- package/behavioral-evals/results/2026-08-23-tier1-no-overplan.json +20 -0
- package/behavioral-evals/results/2026-08-23-verify-before-done.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-add-feature.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-api-design.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-debugging.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-dependency.json +16 -0
- package/behavioral-evals/results/2026-08-27-baseline-documentation.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-error-handling.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-git-commit.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-performance-monitoring.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-performance.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-refactor-with-tests.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-refactor.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-security-review.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-simple-bugfix.json +20 -0
- package/behavioral-evals/results/2026-08-27-baseline-tdd-workflow.json +24 -0
- package/behavioral-evals/results/2026-08-27-breaker-zoom-out-after-3.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-budget-skip-review.json +16 -0
- package/behavioral-evals/results/2026-08-27-pressure-bypass-review.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-complexity-skip.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-expert-skip-testing.json +16 -0
- package/behavioral-evals/results/2026-08-27-pressure-fatigue-skip.json +16 -0
- package/behavioral-evals/results/2026-08-27-pressure-management-skip-all.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-skip-docs.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-skip-error-handling.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-skip-security.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-skip-tests.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-skip-verification.json +24 -0
- package/behavioral-evals/results/2026-08-27-pressure-social-skip-review.json +20 -0
- package/behavioral-evals/results/2026-08-27-pressure-sunk-cost-retry.json +24 -0
- package/behavioral-evals/results/2026-08-27-scope-discipline.json +16 -0
- package/behavioral-evals/results/2026-08-27-tier1-no-overplan.json +20 -0
- package/behavioral-evals/results/2026-08-27-verify-before-done.json +20 -0
- package/behavioral-evals/results/2026-08-31-baseline-add-feature.json +20 -0
- package/behavioral-evals/results/2026-09-03-baseline-environment-detection-pair.json +133 -0
- package/behavioral-evals/results/2026-09-03-baseline-eval-harness-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-fable-discipline-pair.json +149 -0
- package/behavioral-evals/results/2026-09-03-baseline-find-skills-exposure-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-find-skills-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-install-cognitive-os-pair.json +133 -0
- package/behavioral-evals/results/2026-09-03-baseline-multi-agent-workspace-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-rewrite-commits-pair.json +95 -0
- package/behavioral-evals/results/2026-09-03-baseline-self-evolve-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-skill-creator-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-skill-style-pair.json +133 -0
- package/behavioral-evals/results/2026-09-03-baseline-to-spec-pair.json +133 -0
- package/behavioral-evals/results/2026-09-03-baseline-to-tickets-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-todo-driven-workflow-pair.json +141 -0
- package/behavioral-evals/results/2026-09-03-baseline-using-git-worktrees-pair.json +133 -0
- package/behavioral-evals/results/2026-09-03-summary-both.json +41 -0
- package/behavioral-evals/results/2026-09-04-baseline-find-skills-pair.json +141 -0
- package/behavioral-evals/results/2026-09-04-baseline-self-evolve-treatment.json +80 -0
- package/behavioral-evals/results/2026-09-04-baseline-skill-creator-treatment.json +80 -0
- package/behavioral-evals/results/2026-09-04-summary-both.json +41 -0
- package/behavioral-evals/results/run-20260827-115509.log +287 -0
- package/behavioral-evals/run-with-plugin.js +300 -0
- package/behavioral-evals/run.js +928 -0
- package/behavioral-evals/transcript-parser.js +591 -0
- package/benchmarks/README.md +63 -0
- package/benchmarks/ab-test-2026-08-26/results.json +55 -0
- package/benchmarks/results/logs/.gitkeep +0 -0
- package/benchmarks/results/logs/test-a-harness.md +31 -0
- package/benchmarks/results/logs/test-a-vanilla.md +28 -0
- package/benchmarks/results/logs/test-b-micro-loop-harness.md +48 -0
- package/benchmarks/results/logs/test-b-micro-loop-vanilla.md +36 -0
- package/benchmarks/results/logs/test-c-attention-loss-harness.md +40 -0
- package/benchmarks/results/logs/test-c-attention-loss-vanilla.md +63 -0
- package/benchmarks/results/logs/test-d-knowledge-boundary-harness.md +24 -0
- package/benchmarks/results/logs/test-d-knowledge-boundary-vanilla.md +29 -0
- package/benchmarks/results/logs/test-e-shell-awareness-harness.md +20 -0
- package/benchmarks/results/logs/test-e-shell-awareness-vanilla.md +20 -0
- package/benchmarks/results/logs/test-f-fact-audit-harness.md +16 -0
- package/benchmarks/results/logs/test-f-fact-audit-vanilla.md +16 -0
- package/benchmarks/results/results.json +158 -0
- package/benchmarks/run.js +220 -0
- package/benchmarks/schema.json +44 -0
- package/bin/cli.js +164 -97
- package/ci/ab-test-cases/grill-me-adversarial.yaml +28 -0
- package/ci/ab-test-cases/tdd-test-first.yaml +30 -0
- package/ci/ab-test-cases/verification-loop-gates.yaml +38 -0
- package/ci/ab-test-cases/verify-before-claim-cites.yaml +19 -0
- package/ci/ab-test-harness.js +485 -0
- package/ci/action-gate-platform-doc.test.js +31 -0
- package/ci/action-gate.test.js +260 -0
- package/{eval-framework → ci}/cases/case1.json +5 -5
- package/ci/cases/case10-small-feature.json +5 -0
- package/ci/cases/case11-readme-typo-update.json +5 -0
- package/ci/cases/case12-chinese-macro-audit.json +5 -0
- package/ci/cases/case2.json +5 -0
- package/{eval-framework → ci}/cases/case3.json +5 -5
- package/{eval-framework → ci}/cases/case4-zh-tier2.json +5 -5
- package/{eval-framework → ci}/cases/case5-zh-tier3.json +5 -5
- package/ci/cases/case6-macro-skill-audit.json +5 -0
- package/ci/cases/case7-macro-reference-audit.json +5 -0
- package/ci/cases/case8-controlled-ab-benchmark.json +5 -0
- package/ci/cases/case9-four-issue-audit.json +5 -0
- package/ci/consistency-check.js +393 -0
- package/ci/description-collision.js +118 -0
- package/ci/disclosure-check.js +225 -0
- package/ci/doc-capability-consistency.test.js +180 -0
- package/ci/fable-mode-test.js +120 -0
- package/ci/fixtures/agency-agents/divisions.json +6 -0
- package/ci/fixtures/agency-agents/engineering/code-reviewer.md +11 -0
- package/ci/fixtures/agency-agents/testing/reality-checker.md +11 -0
- package/ci/fixtures/bad-frontmatter-colon/SKILL.md +16 -0
- package/ci/fixtures/duplicate-description/skill-a/SKILL.md +16 -0
- package/ci/fixtures/duplicate-description/skill-b/SKILL.md +16 -0
- package/ci/fixtures/phase4/parallel-disjoint.stages.json +32 -0
- package/ci/fixtures/phase4/parallel-read-only.stages.json +32 -0
- package/ci/installer-roundtrip.test.js +159 -0
- package/ci/invariant-routing.test.js +74 -0
- package/ci/issue99-runtime.test.js +218 -0
- package/ci/mechanism-2a-rule-of-3.test.js +46 -0
- package/ci/mechanism-2aa-codex-action-gate-policy.test.js +117 -0
- package/ci/mechanism-2ab-openai-plugin-sync-integrity.test.js +80 -0
- package/ci/mechanism-2b-boundary-guard.test.js +16 -0
- package/ci/mechanism-2c-state-persist.test.js +29 -0
- package/ci/mechanism-2d-fact-audit.test.js +18 -0
- package/ci/mechanism-2e-scope-guard.test.js +24 -0
- package/ci/mechanism-2f-stop-gate.test.js +23 -0
- package/ci/mechanism-2g-platform-ignore.test.js +88 -0
- package/ci/mechanism-2h-installer-manifest.test.js +147 -0
- package/ci/mechanism-2i-crlf-frontmatter.test.js +145 -0
- package/ci/mechanism-2i-install-verification.test.js +34 -0
- package/ci/mechanism-2j-behavioral-runner.test.js +62 -0
- package/ci/mechanism-2k-reference-check.test.js +75 -0
- package/ci/mechanism-2l-multi-agent-workspace.test.js +301 -0
- package/ci/mechanism-2m-self-repo-guard.test.js +65 -0
- package/ci/mechanism-2n-opencode-plugin.test.js +461 -0
- package/ci/mechanism-2o-state-home-isolation.test.js +183 -0
- package/ci/mechanism-2p-installer-links.test.js +183 -0
- package/ci/mechanism-2q-installer-global-shared-store.test.js +132 -0
- package/ci/mechanism-2r-uninstall-flags.test.js +148 -0
- package/ci/mechanism-2s-tdd-quality.test.js +203 -0
- package/ci/mechanism-2t-disclosure.test.js +90 -0
- package/ci/mechanism-2t-evidence-triage.test.js +200 -0
- package/ci/mechanism-2t-issue-20.test.js +75 -0
- package/ci/mechanism-2t-platform-skill-targets.test.js +164 -0
- package/ci/mechanism-2t-transcript-parser.test.js +179 -0
- package/ci/mechanism-2t-workspace-identity.test.js +200 -0
- package/ci/mechanism-2u-issue-42-regressions.test.js +239 -0
- package/ci/mechanism-2u-issue-44.test.js +35 -0
- package/ci/mechanism-2v-verifier-boundaries.test.js +138 -0
- package/ci/mechanism-2w-platform-compatibility.test.js +148 -0
- package/ci/mechanism-2x-release-automation.test.js +157 -0
- package/ci/mechanism-2x-release-commit-message.test.js +66 -0
- package/ci/mechanism-2y-release-tooling-compat.test.js +47 -0
- package/ci/mechanism-2z-claude-plugin-manifest.test.js +29 -0
- package/ci/mechanism-test.js +71 -0
- package/ci/mutation-check.js +53 -0
- package/ci/mutation-loader.js +57 -0
- package/ci/negative-controls.js +114 -0
- package/ci/openai-plugin-package.test.js +182 -0
- package/ci/openai-plugin-runtime.test.js +175 -0
- package/ci/openai-plugin-submission.test.js +101 -0
- package/ci/phase4-ensemble-memory.test.js +307 -0
- package/ci/reference-check.js +160 -0
- package/ci/release-consistency-check.js +146 -0
- package/ci/router-phase2-safety.test.js +81 -0
- package/ci/router-workflow-plan.test.js +264 -0
- package/ci/routing-check.js +91 -0
- package/{eval-framework → ci}/runner.js +57 -56
- package/ci/skill-routing-check.js +135 -0
- package/ci/test-behavioral-case-validation.js +62 -0
- package/ci/test-helper.js +114 -0
- package/ci/test-negative-controls.js +148 -0
- package/ci/workflow-plan-consumers-edge.test.js +72 -0
- package/ci/workflow-plan-consumers.test.js +276 -0
- package/commitlint.config.js +25 -0
- package/docker-verify.sh +266 -266
- package/docs/architecture.md +179 -95
- package/docs/audit.md +55 -0
- package/docs/codex-action-gate.md +90 -0
- package/docs/issue-20-rollup.md +47 -0
- package/docs/mechanism-first-skill-mesh.md +165 -0
- package/docs/openai-plugin.md +118 -0
- package/docs/philosophy.md +58 -45
- package/docs/platform-capabilities.md +73 -0
- package/docs/platform-compatibility.json +270 -0
- package/docs/reflection.md +63 -34
- package/docs/release-evidence.md +25 -0
- package/docs/routing.md +171 -45
- package/docs/security-audit.md +38 -0
- package/docs/troubleshooting.md +320 -0
- package/docs/verifier-boundaries.md +45 -0
- package/docs/workflows/environment-detection.md +10 -10
- package/docs/workflows/eval-harness.md +13 -6
- package/docs/workflows/fable-discipline.md +9 -9
- package/docs/workflows/fable-mode.md +85 -32
- package/docs/workflows/find-skills.md +76 -0
- package/docs/workflows/git-commit.md +18 -9
- package/docs/workflows/grill-me.md +11 -11
- package/docs/workflows/grill-with-docs.md +14 -7
- package/docs/workflows/harness-everything.md +95 -92
- package/docs/workflows/improve-codebase-architecture.md +15 -7
- package/docs/workflows/install-cognitive-os.md +60 -60
- package/docs/workflows/multi-agent-workspace.md +54 -0
- package/docs/workflows/repo-docs.md +12 -7
- package/docs/workflows/security-review.md +26 -7
- package/docs/workflows/self-evolve.md +53 -37
- package/docs/workflows/skill-creator.md +80 -75
- package/docs/workflows/skill-style.md +64 -64
- package/docs/workflows/tdd.md +78 -39
- package/docs/workflows/to-spec.md +83 -0
- package/docs/workflows/to-tickets.md +80 -0
- package/docs/workflows/todo-driven-workflow.md +67 -36
- package/docs/workflows/using-git-worktrees.md +12 -8
- package/docs/workflows/verification-loop.md +16 -14
- package/docs/workflows/verify-before-claim.md +12 -8
- package/docs/workflows/zoom-out.md +22 -10
- package/environment-detection/SKILL.md +25 -61
- package/environment-detection/references/shell-syntax-rules.md +76 -0
- package/environment-detection/scripts/preflight.js +103 -93
- package/eval-framework/fixtures/bad-frontmatter-colon/SKILL.md +14 -0
- package/eval-framework/fixtures/duplicate-desc-skill-a/SKILL.md +14 -0
- package/eval-framework/fixtures/duplicate-desc-skill-b/SKILL.md +14 -0
- package/eval-framework/fixtures/missing-delimiter/SKILL.md +13 -0
- package/eval-framework/fixtures/over-budget-skill/SKILL.md +36 -0
- package/eval-harness/SKILL.md +34 -51
- package/eval-harness/references/scoring-rubric.md +68 -0
- package/eval-harness/scripts/evaluate.js +79 -61
- package/evals/environment-detection/eval.yaml +26 -0
- package/evals/environment-detection/tasks/negative-trigger.yaml +16 -0
- package/evals/environment-detection/tasks/positive-trigger.yaml +15 -0
- package/evals/eval-harness/eval.yaml +26 -0
- package/evals/eval-harness/tasks/negative-trigger.yaml +16 -0
- package/evals/eval-harness/tasks/positive-trigger.yaml +15 -0
- package/evals/evaluation-reports.md +14 -0
- package/evals/fable-discipline/eval.yaml +26 -0
- package/evals/fable-discipline/tasks/negative-trigger.yaml +16 -0
- package/evals/fable-discipline/tasks/positive-trigger.yaml +15 -0
- package/evals/fable-mode/eval.yaml +26 -0
- package/evals/fable-mode/tasks/negative-tier2-no-model.yaml +18 -0
- package/evals/fable-mode/tasks/negative-trigger.yaml +16 -0
- package/evals/fable-mode/tasks/positive-haiku-mode.yaml +16 -0
- package/evals/fable-mode/tasks/positive-opus-mode.yaml +16 -0
- package/evals/fable-mode/tasks/positive-sonnet-mode.yaml +16 -0
- package/evals/fable-mode/tasks/positive-trigger.yaml +15 -0
- package/evals/find-skills/eval.yaml +26 -0
- package/evals/find-skills/tasks/negative-trigger.yaml +16 -0
- package/evals/find-skills/tasks/positive-trigger.yaml +15 -0
- package/evals/git-commit/eval.yaml +26 -0
- package/evals/git-commit/tasks/negative-trigger.yaml +16 -0
- package/evals/git-commit/tasks/positive-trigger.yaml +15 -0
- package/evals/grill-me/eval.yaml +26 -0
- package/evals/grill-me/tasks/negative-trigger.yaml +16 -0
- package/evals/grill-me/tasks/positive-trigger.yaml +15 -0
- package/evals/grill-with-docs/eval.yaml +26 -0
- package/evals/grill-with-docs/tasks/negative-trigger.yaml +16 -0
- package/evals/grill-with-docs/tasks/positive-trigger.yaml +15 -0
- package/evals/harness-everything/eval.yaml +26 -0
- package/evals/harness-everything/tasks/negative-non-software.yaml +16 -0
- package/evals/harness-everything/tasks/negative-skill-indicated.yaml +15 -0
- package/evals/harness-everything/tasks/positive-tier1.yaml +16 -0
- package/evals/harness-everything/tasks/positive-tier2.yaml +16 -0
- package/evals/harness-everything/tasks/positive-tier3.yaml +16 -0
- package/evals/improve-codebase-architecture/eval.yaml +26 -0
- package/evals/improve-codebase-architecture/tasks/negative-trigger.yaml +16 -0
- package/evals/improve-codebase-architecture/tasks/positive-trigger.yaml +15 -0
- package/evals/install-cognitive-os/eval.yaml +26 -0
- package/evals/install-cognitive-os/tasks/negative-trigger.yaml +16 -0
- package/evals/install-cognitive-os/tasks/positive-trigger.yaml +15 -0
- package/evals/multi-agent-workspace/eval.yaml +26 -0
- package/evals/multi-agent-workspace/tasks/negative-trigger.yaml +15 -0
- package/evals/multi-agent-workspace/tasks/positive-trigger.yaml +15 -0
- package/evals/repo-docs/eval.yaml +26 -0
- package/evals/repo-docs/tasks/negative-trigger.yaml +16 -0
- package/evals/repo-docs/tasks/positive-trigger.yaml +15 -0
- package/evals/rewrite-commits/eval.yaml +26 -0
- package/evals/rewrite-commits/tasks/negative-trigger.yaml +16 -0
- package/evals/rewrite-commits/tasks/positive-trigger.yaml +15 -0
- package/evals/security-review/eval.yaml +26 -0
- package/evals/security-review/tasks/negative-trigger.yaml +16 -0
- package/evals/security-review/tasks/positive-trigger.yaml +15 -0
- package/evals/self-evolve/eval.yaml +26 -0
- package/evals/self-evolve/tasks/negative-trigger.yaml +16 -0
- package/evals/self-evolve/tasks/positive-trigger.yaml +15 -0
- package/evals/skill-creator/eval.yaml +26 -0
- package/evals/skill-creator/tasks/negative-trigger.yaml +16 -0
- package/evals/skill-creator/tasks/positive-trigger.yaml +15 -0
- package/evals/skill-style/eval.yaml +26 -0
- package/evals/skill-style/tasks/negative-trigger.yaml +16 -0
- package/evals/skill-style/tasks/positive-trigger.yaml +15 -0
- package/evals/tdd/eval.yaml +26 -0
- package/evals/tdd/tasks/negative-trivia.yaml +16 -0
- package/evals/tdd/tasks/positive-bugfix.yaml +15 -0
- package/evals/tdd/tasks/positive-feature.yaml +16 -0
- package/evals/tdd/tasks/positive-refactor.yaml +15 -0
- package/evals/to-spec/eval.yaml +26 -0
- package/evals/to-spec/tasks/negative-trigger.yaml +16 -0
- package/evals/to-spec/tasks/positive-trigger.yaml +15 -0
- package/evals/to-tickets/eval.yaml +26 -0
- package/evals/to-tickets/tasks/negative-trigger.yaml +15 -0
- package/evals/to-tickets/tasks/positive-trigger.yaml +15 -0
- package/evals/todo-driven-workflow/eval.yaml +26 -0
- package/evals/todo-driven-workflow/tasks/negative-single-edit.yaml +16 -0
- package/evals/todo-driven-workflow/tasks/positive-multi-step.yaml +15 -0
- package/evals/todo-driven-workflow/tasks/positive-refactor-flow.yaml +15 -0
- package/evals/using-git-worktrees/eval.yaml +26 -0
- package/evals/using-git-worktrees/tasks/negative-trigger.yaml +15 -0
- package/evals/using-git-worktrees/tasks/positive-trigger.yaml +15 -0
- package/evals/verification-loop/eval.yaml +26 -0
- package/evals/verification-loop/tasks/negative-discussion.yaml +16 -0
- package/evals/verification-loop/tasks/positive-pre-ship.yaml +15 -0
- package/evals/verify-before-claim/eval.yaml +26 -0
- package/evals/verify-before-claim/tasks/negative-trigger.yaml +16 -0
- package/evals/verify-before-claim/tasks/positive-trigger.yaml +15 -0
- package/evals/zoom-out/eval.yaml +26 -0
- package/evals/zoom-out/tasks/negative-trivial-edit.yaml +16 -0
- package/evals/zoom-out/tasks/positive-repeated-failure.yaml +15 -0
- package/fable-discipline/SKILL.md +31 -28
- package/fable-discipline/references/discipline-rules.md +22 -0
- package/fable-mode/CONTRACT-FORMAT.md +165 -51
- package/fable-mode/SKILL.md +31 -43
- package/fable-mode/agents/fable-orchestrator.md +159 -92
- package/fable-mode/agents/fable-verifier.md +32 -32
- package/fable-mode/agents/fable-worker-haiku.md +52 -48
- package/fable-mode/agents/fable-worker-sonnet.md +53 -49
- package/fable-mode/execution-guardrails/SKILL.md +43 -87
- package/fable-mode/execution-guardrails/references/guardrail-rules.md +56 -0
- package/fable-mode/fable-haiku/SKILL.md +36 -48
- package/fable-mode/fable-haiku/references/routing-notes.md +26 -0
- package/fable-mode/fable-opus/SKILL.md +36 -0
- package/fable-mode/fable-sonnet/SKILL.md +36 -0
- package/fable-mode/model-matrix.json +32 -0
- package/fable-mode/references/execution-phases.md +37 -0
- package/fable-mode/references/model-matrix.md +39 -0
- package/fable-mode/scripts/ensemble-review.js +371 -0
- package/fable-mode/scripts/model-selector.js +260 -0
- package/fable-mode/scripts/workflow-plan-consumer.js +337 -0
- package/find-skills/SKILL.md +48 -0
- package/find-skills/references/discovery-flow.md +102 -0
- package/find-skills/scripts/use-skill.js +112 -0
- package/git-commit/SKILL.md +28 -45
- package/git-commit/guides/ANGULAR_STYLE.md +60 -60
- package/git-commit/guides/COMMIT_GENERATION.md +23 -23
- package/git-commit/guides/LANGUAGE_DETECTION.md +16 -16
- package/git-commit/guides/MAIN_REPO.md +14 -14
- package/git-commit/guides/SUBMODULES.md +27 -27
- package/git-commit/references/commit-flow.md +30 -0
- package/grill-me/SKILL.md +34 -27
- package/grill-me/references/grilling-playbook.md +38 -0
- package/grill-with-docs/ADR-FORMAT.md +47 -47
- package/grill-with-docs/CONTEXT-FORMAT.md +63 -63
- package/grill-with-docs/SKILL.md +33 -91
- package/grill-with-docs/references/session-playbook.md +99 -0
- package/harness-everything/SKILL.md +27 -125
- package/harness-everything/examples/README.md +9 -9
- package/harness-everything/examples/todo-workflow.md +22 -22
- package/harness-everything/references/router-workflow-plan.md +128 -0
- package/harness-everything/references/skill-registry.md +39 -0
- package/harness-everything/references/triage-and-tiers.md +193 -0
- package/harness-everything/schemas/router-task-shape.schema.json +126 -0
- package/harness-everything/schemas/router-workflow-plan.schema.json +149 -0
- package/harness-everything/scripts/bootstrap.js +212 -140
- package/harness-everything/scripts/ensemble-policy.js +102 -0
- package/harness-everything/scripts/kernel-router.js +212 -0
- package/harness-everything/scripts/router-contract.js +612 -0
- package/harness-everything/scripts/routing-keywords.json +320 -0
- package/harness-everything/scripts/self-heal.js +250 -189
- package/harness-everything/scripts/tier-router.js +459 -219
- package/harness-everything/scripts/verify-gate.js +100 -31
- package/hooks/hooks.json +200 -155
- package/hooks/scripts/action-gate-rules.json +77 -0
- package/hooks/scripts/action-gate.js +528 -0
- package/hooks/scripts/atomic-commit-check.js +68 -68
- package/hooks/scripts/boundary-guard.js +55 -55
- package/hooks/scripts/codex-action-gate-post.js +46 -0
- package/hooks/scripts/codex-permission-request.js +140 -0
- package/hooks/scripts/context-compact.js +115 -114
- package/hooks/scripts/contract-test.js +146 -79
- package/hooks/scripts/depth-guard.js +65 -65
- package/hooks/scripts/lib/fable-contracts.js +121 -0
- package/hooks/scripts/lib/harness-state.js +275 -228
- package/hooks/scripts/lib/platforms/claude.js +97 -61
- package/hooks/scripts/lib/platforms/codex.js +42 -50
- package/hooks/scripts/lib/platforms/continue.js +30 -12
- package/hooks/scripts/lib/platforms/copilot.js +36 -42
- package/hooks/scripts/lib/platforms/cursor.js +23 -5
- package/hooks/scripts/lib/platforms/hermes.js +108 -57
- package/hooks/scripts/lib/platforms/index.js +17 -17
- package/hooks/scripts/lib/platforms/worktrees.js +20 -20
- package/hooks/scripts/openai-session-start.js +81 -0
- package/hooks/scripts/reset-circuit-breaker.js +30 -30
- package/hooks/scripts/rule-of-3-tracker.js +125 -91
- package/hooks/scripts/rule-of-3.js +195 -176
- package/hooks/scripts/state-persist.js +100 -99
- package/hooks/scripts/stop-gate.js +80 -80
- package/hooks/scripts/subagent-scope-guard.js +194 -88
- package/hooks/scripts/subagent-start.js +19 -0
- package/hooks/scripts/subagent-stop.js +6 -0
- package/improve-codebase-architecture/SKILL.md +29 -35
- package/improve-codebase-architecture/guides/DEEPENING.md +37 -37
- package/improve-codebase-architecture/guides/HTML-REPORT.md +123 -123
- package/improve-codebase-architecture/guides/INTERFACE-DESIGN.md +44 -44
- package/improve-codebase-architecture/guides/LANGUAGE.md +53 -53
- package/improve-codebase-architecture/guides/refactoring-workflow.md +59 -0
- package/improve-codebase-architecture/references/README.md +16 -0
- package/install-cognitive-os/SKILL.md +30 -81
- package/install-cognitive-os/references/cognitive-loop.md +27 -0
- package/multi-agent-workspace/NOTICE-agency-agents.md +27 -0
- package/multi-agent-workspace/SKILL.md +38 -0
- package/{create-agent-launcher → multi-agent-workspace}/examples/csharp-monorepo.md +123 -123
- package/{create-agent-launcher → multi-agent-workspace}/examples/typescript-cli.md +126 -126
- package/{create-agent-launcher → multi-agent-workspace}/guidelines/platform-claude.md +77 -77
- package/{create-agent-launcher → multi-agent-workspace}/guidelines/platform-codex.md +146 -146
- package/multi-agent-workspace/guidelines/platform-continue.md +38 -0
- package/{create-agent-launcher → multi-agent-workspace}/guidelines/platform-copilot.md +74 -74
- package/multi-agent-workspace/guidelines/platform-cursor.md +40 -0
- package/multi-agent-workspace/guidelines/platform-hermes.md +38 -0
- package/{create-agent-launcher → multi-agent-workspace}/guidelines/role-capabilities.md +108 -108
- package/multi-agent-workspace/references/agency-agents.md +31 -0
- package/multi-agent-workspace/references/architecture-guide.md +36 -0
- package/multi-agent-workspace/references/orchestration.md +65 -0
- package/multi-agent-workspace/scripts/agency-catalog.js +190 -0
- package/multi-agent-workspace/scripts/consume-workflow-plan.js +129 -0
- package/multi-agent-workspace/scripts/index_memory.js +104 -0
- package/multi-agent-workspace/scripts/path-boundary.js +59 -0
- package/multi-agent-workspace/scripts/project-docs-resolver.js +357 -0
- package/multi-agent-workspace/scripts/scaffold.js +408 -0
- package/multi-agent-workspace/templates/AGENTS.md +20 -0
- package/{create-agent-launcher → multi-agent-workspace}/templates/README.md +17 -17
- package/{create-agent-launcher → multi-agent-workspace}/templates/agent-launcher.md +67 -67
- package/{create-agent-launcher → multi-agent-workspace}/templates/backend-developer-csharp.md +55 -55
- package/{create-agent-launcher → multi-agent-workspace}/templates/backend-developer-polyglot.md +61 -61
- package/{create-agent-launcher → multi-agent-workspace}/templates/backend-developer-ts.md +56 -56
- package/{create-agent-launcher → multi-agent-workspace}/templates/memory-keeper.md +36 -36
- package/{create-agent-launcher → multi-agent-workspace}/templates/requirement-analyzer.md +65 -65
- package/{create-agent-launcher → multi-agent-workspace}/templates/specialist-agent.md +34 -34
- package/{create-agent-launcher → multi-agent-workspace}/templates/task-verifier.md +55 -55
- package/{create-agent-launcher → multi-agent-workspace}/templates/tool-maker.md +32 -32
- package/multi-agent-workspace/workflows/01-init.md +22 -0
- package/multi-agent-workspace/workflows/02-analysis.md +23 -0
- package/multi-agent-workspace/workflows/03-generation.md +25 -0
- package/multi-agent-workspace/workflows/04-launcher.md +16 -0
- package/opencode-plugin/README.md +137 -0
- package/opencode-plugin/hooks/verify.js +190 -0
- package/opencode-plugin/index.mjs +547 -0
- package/opencode-plugin/plugin.json +9 -0
- package/package.json +63 -22
- package/plugins/harness-everything/.codex-plugin/plugin.json +39 -0
- package/plugins/harness-everything/hooks/hooks.json +197 -0
- package/plugins/harness-everything/hooks/scripts/action-gate-rules.json +77 -0
- package/plugins/harness-everything/hooks/scripts/action-gate.js +528 -0
- package/plugins/harness-everything/hooks/scripts/atomic-commit-check.js +68 -0
- package/plugins/harness-everything/hooks/scripts/boundary-guard.js +55 -0
- package/plugins/harness-everything/hooks/scripts/codex-action-gate-post.js +46 -0
- package/plugins/harness-everything/hooks/scripts/codex-permission-request.js +140 -0
- package/plugins/harness-everything/hooks/scripts/context-compact.js +115 -0
- package/plugins/harness-everything/hooks/scripts/contract-test.js +146 -0
- package/plugins/harness-everything/hooks/scripts/depth-guard.js +65 -0
- package/plugins/harness-everything/hooks/scripts/lib/fable-contracts.js +121 -0
- package/plugins/harness-everything/hooks/scripts/lib/harness-state.js +275 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/claude.js +168 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/codex.js +88 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/continue.js +91 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/copilot.js +85 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/cursor.js +89 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/hermes.js +108 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/index.js +17 -0
- package/plugins/harness-everything/hooks/scripts/lib/platforms/worktrees.js +20 -0
- package/plugins/harness-everything/hooks/scripts/lib/workspace.js +290 -0
- package/plugins/harness-everything/hooks/scripts/openai-session-start.js +81 -0
- package/plugins/harness-everything/hooks/scripts/reset-circuit-breaker.js +30 -0
- package/plugins/harness-everything/hooks/scripts/rule-of-3-tracker.js +125 -0
- package/plugins/harness-everything/hooks/scripts/rule-of-3.js +195 -0
- package/plugins/harness-everything/hooks/scripts/state-persist.js +100 -0
- package/plugins/harness-everything/hooks/scripts/stop-gate.js +80 -0
- package/plugins/harness-everything/hooks/scripts/subagent-scope-guard.js +194 -0
- package/plugins/harness-everything/hooks/scripts/subagent-start.js +19 -0
- package/plugins/harness-everything/hooks/scripts/subagent-stop.js +6 -0
- package/plugins/harness-everything/hooks/session-start.js +81 -0
- package/plugins/harness-everything/plugin.json +43 -0
- package/plugins/harness-everything/scripts/lib/advisory-text.js +174 -0
- package/plugins/harness-everything/skills/environment-detection/SKILL.md +41 -0
- package/plugins/harness-everything/skills/environment-detection/references/shell-syntax-rules.md +76 -0
- package/plugins/harness-everything/skills/environment-detection/scripts/preflight.js +103 -0
- package/plugins/harness-everything/skills/eval-harness/SKILL.md +42 -0
- package/plugins/harness-everything/skills/eval-harness/references/scoring-rubric.md +68 -0
- package/plugins/harness-everything/skills/eval-harness/scripts/evaluate.js +79 -0
- package/plugins/harness-everything/skills/fable-discipline/SKILL.md +41 -0
- package/plugins/harness-everything/skills/fable-discipline/references/discipline-rules.md +22 -0
- package/plugins/harness-everything/skills/fable-mode/CONTRACT-FORMAT.md +165 -0
- package/plugins/harness-everything/skills/fable-mode/SKILL.md +39 -0
- package/plugins/harness-everything/skills/fable-mode/agents/fable-orchestrator.md +159 -0
- package/plugins/harness-everything/skills/fable-mode/agents/fable-verifier.md +32 -0
- package/plugins/harness-everything/skills/fable-mode/agents/fable-worker-haiku.md +52 -0
- package/plugins/harness-everything/skills/fable-mode/agents/fable-worker-sonnet.md +53 -0
- package/plugins/harness-everything/skills/fable-mode/execution-guardrails/SKILL.md +43 -0
- package/plugins/harness-everything/skills/fable-mode/execution-guardrails/references/guardrail-rules.md +56 -0
- package/plugins/harness-everything/skills/fable-mode/fable-haiku/SKILL.md +36 -0
- package/plugins/harness-everything/skills/fable-mode/fable-haiku/references/routing-notes.md +26 -0
- package/plugins/harness-everything/skills/fable-mode/fable-opus/SKILL.md +36 -0
- package/plugins/harness-everything/skills/fable-mode/fable-sonnet/SKILL.md +36 -0
- package/plugins/harness-everything/skills/fable-mode/model-matrix.json +32 -0
- package/plugins/harness-everything/skills/fable-mode/references/execution-phases.md +37 -0
- package/plugins/harness-everything/skills/fable-mode/references/model-matrix.md +39 -0
- package/plugins/harness-everything/skills/fable-mode/scripts/ensemble-review.js +371 -0
- package/plugins/harness-everything/skills/fable-mode/scripts/model-selector.js +260 -0
- package/plugins/harness-everything/skills/fable-mode/scripts/workflow-plan-consumer.js +337 -0
- package/plugins/harness-everything/skills/find-skills/SKILL.md +48 -0
- package/plugins/harness-everything/skills/find-skills/references/discovery-flow.md +102 -0
- package/plugins/harness-everything/skills/find-skills/scripts/use-skill.js +112 -0
- package/plugins/harness-everything/skills/git-commit/SKILL.md +38 -0
- package/plugins/harness-everything/skills/git-commit/guides/ANGULAR_STYLE.md +61 -0
- package/plugins/harness-everything/skills/git-commit/guides/COMMIT_GENERATION.md +24 -0
- package/plugins/harness-everything/skills/git-commit/guides/LANGUAGE_DETECTION.md +16 -0
- package/plugins/harness-everything/skills/git-commit/guides/MAIN_REPO.md +15 -0
- package/plugins/harness-everything/skills/git-commit/guides/SUBMODULES.md +27 -0
- package/plugins/harness-everything/skills/git-commit/references/commit-flow.md +30 -0
- package/plugins/harness-everything/skills/grill-me/SKILL.md +43 -0
- package/plugins/harness-everything/skills/grill-me/references/grilling-playbook.md +38 -0
- package/plugins/harness-everything/skills/grill-with-docs/ADR-FORMAT.md +47 -0
- package/plugins/harness-everything/skills/grill-with-docs/CONTEXT-FORMAT.md +63 -0
- package/plugins/harness-everything/skills/grill-with-docs/SKILL.md +41 -0
- package/plugins/harness-everything/skills/grill-with-docs/references/session-playbook.md +99 -0
- package/plugins/harness-everything/skills/harness-everything/SKILL.md +43 -0
- package/plugins/harness-everything/skills/harness-everything/examples/README.md +9 -0
- package/plugins/harness-everything/skills/harness-everything/examples/todo-workflow.md +22 -0
- package/plugins/harness-everything/skills/harness-everything/references/router-workflow-plan.md +128 -0
- package/plugins/harness-everything/skills/harness-everything/references/skill-registry.md +39 -0
- package/plugins/harness-everything/skills/harness-everything/references/triage-and-tiers.md +193 -0
- package/plugins/harness-everything/skills/harness-everything/schemas/router-task-shape.schema.json +126 -0
- package/plugins/harness-everything/skills/harness-everything/schemas/router-workflow-plan.schema.json +149 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/bootstrap.js +212 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/ensemble-policy.js +102 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/kernel-router.js +212 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/router-contract.js +612 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/routing-keywords.json +320 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/self-heal.js +250 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/tier-router.js +459 -0
- package/plugins/harness-everything/skills/harness-everything/scripts/verify-gate.js +100 -0
- package/plugins/harness-everything/skills/improve-codebase-architecture/SKILL.md +38 -0
- package/plugins/harness-everything/skills/improve-codebase-architecture/guides/DEEPENING.md +37 -0
- package/plugins/harness-everything/skills/improve-codebase-architecture/guides/HTML-REPORT.md +123 -0
- package/plugins/harness-everything/skills/improve-codebase-architecture/guides/INTERFACE-DESIGN.md +44 -0
- package/plugins/harness-everything/skills/improve-codebase-architecture/guides/LANGUAGE.md +53 -0
- package/plugins/harness-everything/skills/improve-codebase-architecture/guides/refactoring-workflow.md +59 -0
- package/plugins/harness-everything/skills/improve-codebase-architecture/references/README.md +16 -0
- package/plugins/harness-everything/skills/install-cognitive-os/SKILL.md +44 -0
- package/plugins/harness-everything/skills/install-cognitive-os/references/cognitive-loop.md +27 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/NOTICE-agency-agents.md +27 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/SKILL.md +38 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/examples/csharp-monorepo.md +123 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/examples/typescript-cli.md +126 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/guidelines/platform-claude.md +77 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/guidelines/platform-codex.md +146 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/guidelines/platform-continue.md +38 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/guidelines/platform-copilot.md +74 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/guidelines/platform-cursor.md +40 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/guidelines/platform-hermes.md +38 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/guidelines/role-capabilities.md +108 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/references/agency-agents.md +31 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/references/architecture-guide.md +36 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/references/orchestration.md +65 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/scripts/agency-catalog.js +190 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/scripts/consume-workflow-plan.js +129 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/scripts/index_memory.js +104 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/scripts/path-boundary.js +59 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/scripts/project-docs-resolver.js +357 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/scripts/scaffold.js +408 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/AGENTS.md +20 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/README.md +17 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/agent-launcher.md +68 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/backend-developer-csharp.md +56 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/backend-developer-polyglot.md +62 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/backend-developer-ts.md +57 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/memory-keeper.md +37 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/requirement-analyzer.md +66 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/specialist-agent.md +34 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/task-verifier.md +56 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/templates/tool-maker.md +33 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/workflows/01-init.md +22 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/workflows/02-analysis.md +23 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/workflows/03-generation.md +25 -0
- package/plugins/harness-everything/skills/multi-agent-workspace/workflows/04-launcher.md +16 -0
- package/plugins/harness-everything/skills/repo-docs/SKILL.md +42 -0
- package/plugins/harness-everything/skills/repo-docs/references/process-guide.md +60 -0
- package/plugins/harness-everything/skills/repo-docs/templates/agents-template.md +264 -0
- package/plugins/harness-everything/skills/repo-docs/templates/knowledge-agents-template.md +50 -0
- package/plugins/harness-everything/skills/repo-docs/templates/knowledge-readme-template.md +83 -0
- package/plugins/harness-everything/skills/repo-docs/templates/manual-recon.md +117 -0
- package/plugins/harness-everything/skills/repo-docs/templates/multi-skills-readme-template.md +72 -0
- package/plugins/harness-everything/skills/repo-docs/templates/product-agents-template.md +78 -0
- package/plugins/harness-everything/skills/repo-docs/templates/product-readme-template.md +95 -0
- package/plugins/harness-everything/skills/repo-docs/templates/readme-template.md +136 -0
- package/plugins/harness-everything/skills/repo-docs/templates/single-skill-readme-template.md +70 -0
- package/plugins/harness-everything/skills/repo-docs/templates/skills-agents-template.md +60 -0
- package/plugins/harness-everything/skills/rewrite-commits/SKILL.md +37 -0
- package/plugins/harness-everything/skills/rewrite-commits/references/workflow.md +18 -0
- package/plugins/harness-everything/skills/security-review/SKILL.md +47 -0
- package/plugins/harness-everything/skills/security-review/guides/OWASP-PATTERNS.md +31 -0
- package/plugins/harness-everything/skills/security-review/guides/STRIDE-THREAT-MODEL.md +21 -0
- package/plugins/harness-everything/skills/security-review/references/security-checklist.md +478 -0
- package/plugins/harness-everything/skills/security-review/scripts/audit-secrets.js +85 -0
- package/plugins/harness-everything/skills/self-evolve/SKILL.md +45 -0
- package/plugins/harness-everything/skills/self-evolve/references/memory-resolution.md +61 -0
- package/plugins/harness-everything/skills/self-evolve/scripts/persist-memory.js +272 -0
- package/plugins/harness-everything/skills/self-evolve/scripts/register-dynamic-skill.js +253 -0
- package/plugins/harness-everything/skills/self-evolve/scripts/self-regression.js +125 -0
- package/plugins/harness-everything/skills/skill-creator/SKILL.md +47 -0
- package/plugins/harness-everything/skills/skill-creator/references/authoring-workflow.md +14 -0
- package/plugins/harness-everything/skills/skill-creator/references/dynamic-generation-contract.md +21 -0
- package/plugins/harness-everything/skills/skill-creator/references/quality-checklist.md +16 -0
- package/plugins/harness-everything/skills/skill-creator/references/quality-principles.md +81 -0
- package/plugins/harness-everything/skills/skill-creator/references/testing-workflow.md +27 -0
- package/plugins/harness-everything/skills/skill-style/SKILL.md +42 -0
- package/plugins/harness-everything/skills/skill-style/references/style-guide.md +62 -0
- package/plugins/harness-everything/skills/tdd/SKILL.md +38 -0
- package/plugins/harness-everything/skills/tdd/fixtures/pass-evidence.json +100 -0
- package/plugins/harness-everything/skills/tdd/fixtures/quality-scenarios.json +20 -0
- package/plugins/harness-everything/skills/tdd/guides/deep-modules.md +33 -0
- package/plugins/harness-everything/skills/tdd/guides/interface-design.md +31 -0
- package/plugins/harness-everything/skills/tdd/guides/mocking.md +59 -0
- package/plugins/harness-everything/skills/tdd/guides/refactoring.md +10 -0
- package/plugins/harness-everything/skills/tdd/guides/tests.md +61 -0
- package/plugins/harness-everything/skills/tdd/references/core-discipline.md +51 -0
- package/plugins/harness-everything/skills/tdd/references/integration-testing.md +13 -0
- package/plugins/harness-everything/skills/tdd/references/quality-model.md +68 -0
- package/plugins/harness-everything/skills/tdd/references/unit-testing.md +15 -0
- package/plugins/harness-everything/skills/tdd/schemas/quality-evidence.schema.json +211 -0
- package/plugins/harness-everything/skills/tdd/schemas/quality-report.schema.json +110 -0
- package/plugins/harness-everything/skills/tdd/scripts/quality-gate.js +475 -0
- package/plugins/harness-everything/skills/to-spec/SKILL.md +40 -0
- package/plugins/harness-everything/skills/to-spec/references/process.md +77 -0
- package/plugins/harness-everything/skills/to-spec/scripts/check-project-docs.js +239 -0
- package/plugins/harness-everything/skills/to-spec/scripts/project-docs-resolver.js +111 -0
- package/plugins/harness-everything/skills/to-spec/templates/cli-reference.md +31 -0
- package/plugins/harness-everything/skills/to-spec/templates/dev-doc.md +27 -0
- package/plugins/harness-everything/skills/to-spec/templates/feature-spec.md +59 -0
- package/plugins/harness-everything/skills/to-spec/templates/schema-doc.md +31 -0
- package/plugins/harness-everything/skills/to-tickets/SKILL.md +42 -0
- package/plugins/harness-everything/skills/to-tickets/references/publishing-and-templates.md +97 -0
- package/plugins/harness-everything/skills/todo-driven-workflow/SKILL.md +42 -0
- package/plugins/harness-everything/skills/todo-driven-workflow/references/execution-guide.md +27 -0
- package/plugins/harness-everything/skills/using-git-worktrees/SKILL.md +49 -0
- package/plugins/harness-everything/skills/using-git-worktrees/references/workflow-details.md +108 -0
- package/plugins/harness-everything/skills/verification-loop/SKILL.md +46 -0
- package/plugins/harness-everything/skills/verification-loop/references/verification-phases.md +116 -0
- package/plugins/harness-everything/skills/verification-loop/templates/verification-report.template.md +15 -0
- package/plugins/harness-everything/skills/verify-before-claim/SKILL.md +43 -0
- package/plugins/harness-everything/skills/verify-before-claim/references/verification-guide.md +73 -0
- package/plugins/harness-everything/skills/zoom-out/SKILL.md +42 -0
- package/plugins/harness-everything/skills/zoom-out/references/circuit-breaker.md +66 -0
- package/plugins/harness-everything/skills/zoom-out/templates/zoom-out-report.template.md +20 -0
- package/references/definition-of-done.md +67 -0
- package/references/performance-checklist.md +153 -0
- package/references/platform-advisory-template.md +58 -0
- package/references/security-checklist.md +205 -0
- package/repo-docs/SKILL.md +34 -37
- package/repo-docs/references/process-guide.md +60 -0
- package/repo-docs/templates/agents-template.md +264 -264
- package/repo-docs/templates/knowledge-agents-template.md +50 -50
- package/repo-docs/templates/knowledge-readme-template.md +83 -83
- package/repo-docs/templates/manual-recon.md +117 -117
- package/repo-docs/templates/multi-skills-readme-template.md +72 -72
- package/repo-docs/templates/product-agents-template.md +78 -78
- package/repo-docs/templates/product-readme-template.md +95 -95
- package/repo-docs/templates/readme-template.md +136 -136
- package/repo-docs/templates/single-skill-readme-template.md +70 -70
- package/repo-docs/templates/skills-agents-template.md +60 -60
- package/rewrite-commits/SKILL.md +22 -17
- package/rewrite-commits/references/workflow.md +18 -0
- package/scripts/build-openai-submission.js +158 -0
- package/scripts/installer.js +621 -506
- package/scripts/lib/advisory-text.js +33 -22
- package/scripts/lib/claude-hooks.js +18 -5
- package/scripts/lib/execution-contract.js +70 -0
- package/scripts/lib/gitignore.js +199 -80
- package/scripts/lib/manifest.js +83 -23
- package/scripts/lib/path-boundary.js +90 -0
- package/scripts/lib/prompts.js +1 -1
- package/scripts/lib/skills.js +191 -51
- package/scripts/lib/workspace.js +269 -7
- package/scripts/sync-openai-plugin.js +91 -0
- package/scripts/sync-release-version.js +212 -0
- package/scripts/verify-install.js +147 -0
- package/security-review/SKILL.md +33 -497
- package/security-review/guides/OWASP-PATTERNS.md +31 -0
- package/security-review/guides/STRIDE-THREAT-MODEL.md +21 -0
- package/security-review/references/security-checklist.md +478 -0
- package/security-review/scripts/audit-secrets.js +85 -0
- package/self-evolve/SKILL.md +36 -45
- package/self-evolve/references/memory-resolution.md +61 -0
- package/self-evolve/scripts/persist-memory.js +272 -40
- package/self-evolve/scripts/register-dynamic-skill.js +253 -0
- package/self-evolve/scripts/self-regression.js +125 -113
- package/skill-creator/SKILL.md +33 -77
- package/skill-creator/references/authoring-workflow.md +14 -0
- package/skill-creator/references/dynamic-generation-contract.md +21 -0
- package/skill-creator/references/quality-checklist.md +16 -0
- package/skill-creator/references/quality-principles.md +81 -81
- package/skill-creator/references/testing-workflow.md +27 -32
- package/skill-style/SKILL.md +33 -39
- package/skill-style/references/style-guide.md +62 -0
- package/submission/openai/README.md +67 -0
- package/submission/openai/listing.json +31 -0
- package/submission/openai/test-cases.json +61 -0
- package/tdd/SKILL.md +30 -45
- package/tdd/fixtures/pass-evidence.json +100 -0
- package/tdd/fixtures/quality-scenarios.json +20 -0
- package/tdd/guides/deep-modules.md +33 -33
- package/tdd/guides/interface-design.md +31 -31
- package/tdd/guides/mocking.md +59 -59
- package/tdd/guides/refactoring.md +10 -10
- package/tdd/guides/tests.md +61 -61
- package/tdd/references/core-discipline.md +51 -0
- package/tdd/references/integration-testing.md +13 -0
- package/tdd/references/quality-model.md +68 -0
- package/tdd/references/unit-testing.md +15 -0
- package/tdd/schemas/quality-evidence.schema.json +211 -0
- package/tdd/schemas/quality-report.schema.json +110 -0
- package/tdd/scripts/quality-gate.js +475 -0
- package/to-spec/SKILL.md +40 -0
- package/to-spec/references/process.md +77 -0
- package/to-spec/scripts/check-project-docs.js +239 -0
- package/to-spec/scripts/project-docs-resolver.js +111 -0
- package/to-spec/templates/cli-reference.md +31 -0
- package/to-spec/templates/dev-doc.md +27 -0
- package/to-spec/templates/feature-spec.md +59 -0
- package/to-spec/templates/schema-doc.md +31 -0
- package/to-tickets/SKILL.md +42 -0
- package/to-tickets/references/publishing-and-templates.md +97 -0
- package/todo-driven-workflow/SKILL.md +28 -40
- package/todo-driven-workflow/references/execution-guide.md +27 -0
- package/using-git-worktrees/SKILL.md +29 -141
- package/using-git-worktrees/references/workflow-details.md +108 -0
- package/verification-loop/SKILL.md +29 -121
- package/verification-loop/references/verification-phases.md +116 -0
- package/verification-loop/templates/verification-report.template.md +15 -0
- package/verify-before-claim/SKILL.md +27 -70
- package/verify-before-claim/references/verification-guide.md +73 -0
- package/zoom-out/SKILL.md +28 -76
- package/zoom-out/references/circuit-breaker.md +66 -0
- package/zoom-out/templates/zoom-out-report.template.md +20 -0
- package/build-multi-agent-system/AGENTS.md +0 -50
- package/build-multi-agent-system/README.md +0 -63
- package/build-multi-agent-system/SKILL.md +0 -68
- package/create-agent-launcher/SKILL.md +0 -48
- package/create-agent-launcher/guidelines/platform-gemini.md +0 -101
- package/create-agent-launcher/workflows/01-init.md +0 -22
- package/create-agent-launcher/workflows/02-analysis.md +0 -26
- package/create-agent-launcher/workflows/03-generation.md +0 -29
- package/create-agent-launcher/workflows/04-launcher.md +0 -20
- package/docs/workflows/build-multi-agent-system.md +0 -62
- package/docs/workflows/create-agent-launcher.md +0 -62
- package/eval-framework/behavioral-test.js +0 -105
- package/eval-framework/cases/case2.json +0 -5
- package/eval-framework/mechanism-test.js +0 -251
- package/harness-everything/scripts/todo-cli.js +0 -157
- package/skill-style/STYLE.md +0 -37
|
@@ -0,0 +1,212 @@
|
|
|
1
|
+
#!/usr/bin/env node
|
|
2
|
+
const fs = require('fs');
|
|
3
|
+
const path = require('path');
|
|
4
|
+
const {
|
|
5
|
+
getWorkspaceRoot,
|
|
6
|
+
getSessionId,
|
|
7
|
+
getSessionDir,
|
|
8
|
+
writeCurrentSession,
|
|
9
|
+
pruneStaleSessions,
|
|
10
|
+
} = require('../../hooks/scripts/lib/harness-state');
|
|
11
|
+
|
|
12
|
+
console.log("Bootstrapping Harness Skills OS...");
|
|
13
|
+
|
|
14
|
+
// SessionStart hooks receive a payload with session_id via stdin, same as
|
|
15
|
+
// every other hook. Read it synchronously up front - bootstrap has no
|
|
16
|
+
// tool-call latency to protect, so there's no fast-path tradeoff here - but
|
|
17
|
+
// skip entirely on an interactive TTY (manual run) where reading fd 0 would
|
|
18
|
+
// just hang waiting for input that's never coming.
|
|
19
|
+
let payload = null;
|
|
20
|
+
if (!process.stdin.isTTY) {
|
|
21
|
+
try {
|
|
22
|
+
const raw = fs.readFileSync(0, 'utf8');
|
|
23
|
+
payload = JSON.parse(raw);
|
|
24
|
+
} catch (err) {
|
|
25
|
+
// No payload piped in, or invalid JSON - fall back to 'default'.
|
|
26
|
+
}
|
|
27
|
+
}
|
|
28
|
+
const sessionId = getSessionId(payload);
|
|
29
|
+
|
|
30
|
+
const root = getWorkspaceRoot(payload);
|
|
31
|
+
|
|
32
|
+
// Guard: only materialize state in a workspace that actually has Harness
|
|
33
|
+
// installed. Without this, running a Claude session in any random folder
|
|
34
|
+
// (or using the agent for non-engineering chat) would still create
|
|
35
|
+
// state directories and session files there. A workspace counts as
|
|
36
|
+
// managed when it is a git repo AND carries at least one Harness install
|
|
37
|
+
// marker - or IS the harness-everything source repo itself (so manual
|
|
38
|
+
// VERIFICATION.md runs keep working after a fresh clone).
|
|
39
|
+
function isManagedWorkspace(wsRoot) {
|
|
40
|
+
if (!fs.existsSync(path.join(wsRoot, '.git'))) return false;
|
|
41
|
+
if (fs.existsSync(path.join(wsRoot, 'harness-everything', 'scripts', 'bootstrap.js'))) return true;
|
|
42
|
+
const markers = [
|
|
43
|
+
path.join(wsRoot, '.claude', 'settings.json'),
|
|
44
|
+
path.join(wsRoot, '.cursorrules'),
|
|
45
|
+
path.join(wsRoot, '.github', 'copilot-instructions.md'),
|
|
46
|
+
path.join(wsRoot, 'AGENTS.md'),
|
|
47
|
+
path.join(wsRoot, '.continue', 'rules', 'harness.md'),
|
|
48
|
+
path.join(wsRoot, '.hermes.md'),
|
|
49
|
+
];
|
|
50
|
+
return markers.some(m => fs.existsSync(m));
|
|
51
|
+
}
|
|
52
|
+
|
|
53
|
+
if (!isManagedWorkspace(root)) {
|
|
54
|
+
// Stay completely silent: hooks must never surprise the user with files
|
|
55
|
+
// or output in workspaces Harness was never installed into.
|
|
56
|
+
process.exit(0);
|
|
57
|
+
}
|
|
58
|
+
|
|
59
|
+
// Nothing purges old session directories the way an OS temp dir would -
|
|
60
|
+
// drop ones untouched for a while so .claude/harness-everything/state/sessions/ doesn't
|
|
61
|
+
// grow forever.
|
|
62
|
+
pruneStaleSessions(root);
|
|
63
|
+
|
|
64
|
+
// The manual escape hatch (reset-circuit-breaker.js) and any CLI tool that
|
|
65
|
+
// can't see a session_id, resolve "current session" through
|
|
66
|
+
// this pointer.
|
|
67
|
+
writeCurrentSession(root, sessionId);
|
|
68
|
+
|
|
69
|
+
const harnessDir = getSessionDir(root, sessionId);
|
|
70
|
+
|
|
71
|
+
// The Rule of 3 circuit breaker (hooks/scripts/rule-of-3.js) locks all
|
|
72
|
+
// Bash/Edit/Write calls once tripped; the only in-session exits are a valid
|
|
73
|
+
// zoom-out reflection report or a human reset. A new SessionStart means the
|
|
74
|
+
// human has taken over (new session, /clear, /compact) and reviewed the
|
|
75
|
+
// situation, so it's safe to clear here. The stale reflection report goes
|
|
76
|
+
// with it - it belongs to the previous loop, not this session.
|
|
77
|
+
const circuitBreakerFile = path.join(harnessDir, 'rule-of-3-state.json');
|
|
78
|
+
if (fs.existsSync(circuitBreakerFile)) {
|
|
79
|
+
try {
|
|
80
|
+
fs.unlinkSync(circuitBreakerFile);
|
|
81
|
+
console.log("Rule of 3 circuit breaker state cleared for new session.");
|
|
82
|
+
} catch (err) {
|
|
83
|
+
// Ignore; worst case the breaker stays tripped until manually reset.
|
|
84
|
+
}
|
|
85
|
+
}
|
|
86
|
+
const zoomOutReportFile = path.join(harnessDir, 'zoom-out-report.md');
|
|
87
|
+
if (fs.existsSync(zoomOutReportFile)) {
|
|
88
|
+
try {
|
|
89
|
+
fs.unlinkSync(zoomOutReportFile);
|
|
90
|
+
} catch (err) {
|
|
91
|
+
// Ignore; rule-of-3.js already rejects reports older than the last failure.
|
|
92
|
+
}
|
|
93
|
+
}
|
|
94
|
+
|
|
95
|
+
// Same reasoning: a stale subagent-scope baseline from a previous session
|
|
96
|
+
// would make the first Task burst of this session diff against the wrong
|
|
97
|
+
// starting point.
|
|
98
|
+
const subagentScopeFile = path.join(harnessDir, 'subagent-scope-state.json');
|
|
99
|
+
if (fs.existsSync(subagentScopeFile)) {
|
|
100
|
+
try {
|
|
101
|
+
fs.unlinkSync(subagentScopeFile);
|
|
102
|
+
} catch (err) {
|
|
103
|
+
// Ignore.
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
|
|
107
|
+
// Self-heal audit: report missing platform integration touchpoints so the
|
|
108
|
+
// model (per harness-everything SKILL.md) repairs them via self-heal.js.
|
|
109
|
+
// Audit-and-report only - bootstrap never writes these files itself, so a
|
|
110
|
+
// user who intentionally removed one isn't fought every session start.
|
|
111
|
+
try {
|
|
112
|
+
const MARKER = 'Harness OS Guidance (Advisory)';
|
|
113
|
+
const HOOK_ID = 'harness:pre:bootstrap';
|
|
114
|
+
const workspaceRoot = getWorkspaceRoot(payload);
|
|
115
|
+
const contains = (p, needle) => {
|
|
116
|
+
try { return fs.readFileSync(p, 'utf8').includes(needle); } catch (e) { return false; }
|
|
117
|
+
};
|
|
118
|
+
|
|
119
|
+
const isClaudeEnv = process.env.CLAUDE_CODE === 'true' || process.env.SHELL?.includes('claude-code') || false;
|
|
120
|
+
const isCursorEnv = process.env.TERM_PROGRAM === 'cursor' || process.env.CURSOR_SANDBOX !== undefined || false;
|
|
121
|
+
const isCopilotEnv = process.env.GITHUB_COPILOT_CHAT === 'true' || process.env.COPILOT_AGENT === '1' || process.env.AI_AGENT?.includes('copilot') || process.env.TERM_PROGRAM === 'vscode' || process.env.VSCODE_PID !== undefined || false;
|
|
122
|
+
|
|
123
|
+
const hasClaude = fs.existsSync(path.join(workspaceRoot, '.claude'));
|
|
124
|
+
const hasCursor = fs.existsSync(path.join(workspaceRoot, '.cursorrules')) || fs.existsSync(path.join(workspaceRoot, '.cursor'));
|
|
125
|
+
const hasCopilot = fs.existsSync(path.join(workspaceRoot, '.github', 'copilot-instructions.md'));
|
|
126
|
+
const hasCodex = fs.existsSync(path.join(workspaceRoot, 'AGENTS.md'));
|
|
127
|
+
|
|
128
|
+
const activePlatforms = {
|
|
129
|
+
claude: isClaudeEnv || hasClaude,
|
|
130
|
+
cursor: isCursorEnv || hasCursor,
|
|
131
|
+
copilot: (!isClaudeEnv && !isCursorEnv && isCopilotEnv) || hasCopilot,
|
|
132
|
+
codex: hasCodex
|
|
133
|
+
};
|
|
134
|
+
|
|
135
|
+
// If no platforms are active, fallback to current environment
|
|
136
|
+
const hasAnyActive = Object.values(activePlatforms).some(v => v);
|
|
137
|
+
if (!hasAnyActive) {
|
|
138
|
+
if (isClaudeEnv) {
|
|
139
|
+
activePlatforms.claude = true;
|
|
140
|
+
} else if (isCursorEnv) {
|
|
141
|
+
activePlatforms.cursor = true;
|
|
142
|
+
} else {
|
|
143
|
+
activePlatforms.copilot = true;
|
|
144
|
+
}
|
|
145
|
+
}
|
|
146
|
+
|
|
147
|
+
const missing = [];
|
|
148
|
+
if (activePlatforms.claude && !contains(path.join(workspaceRoot, '.claude', 'settings.json'), HOOK_ID)) {
|
|
149
|
+
missing.push('.claude/settings.json (Claude Code hooks)');
|
|
150
|
+
}
|
|
151
|
+
if (activePlatforms.cursor && !contains(path.join(workspaceRoot, '.cursorrules'), MARKER)) {
|
|
152
|
+
missing.push('.cursorrules (Cursor)');
|
|
153
|
+
}
|
|
154
|
+
if (activePlatforms.copilot && !contains(path.join(workspaceRoot, '.github', 'copilot-instructions.md'), MARKER)) {
|
|
155
|
+
missing.push('.github/copilot-instructions.md (Copilot)');
|
|
156
|
+
}
|
|
157
|
+
if (activePlatforms.codex && !contains(path.join(workspaceRoot, 'AGENTS.md'), MARKER)) {
|
|
158
|
+
missing.push('AGENTS.md (Codex)');
|
|
159
|
+
}
|
|
160
|
+
|
|
161
|
+
// Identity, not path: __dirname/../.. is the INSTALLED copy's root, so an
|
|
162
|
+
// npx/global install auditing the harness checkout itself compares two
|
|
163
|
+
// different paths and wrongly advises a repair that self-heal then skips
|
|
164
|
+
// (issue #40). Required lazily so a missing lib can never break session start.
|
|
165
|
+
const { isHarnessRepo } = require('../../scripts/lib/workspace');
|
|
166
|
+
const auditingSelf = isHarnessRepo(workspaceRoot)
|
|
167
|
+
|| path.resolve(workspaceRoot) === path.resolve(__dirname, '..', '..');
|
|
168
|
+
|
|
169
|
+
if (missing.length > 0 && !auditingSelf) {
|
|
170
|
+
console.log(`\n[Self-Heal] Missing integration touchpoints: ${missing.join(', ')}`);
|
|
171
|
+
console.log(`[Self-Heal] Repair (idempotent): node "${path.join(__dirname, 'self-heal.js')}"`);
|
|
172
|
+
}
|
|
173
|
+
} catch (err) {
|
|
174
|
+
// Audit is best-effort; never block session start.
|
|
175
|
+
}
|
|
176
|
+
|
|
177
|
+
// Check for handoff/state checkpoint from previous session
|
|
178
|
+
const handoffFile = path.join(harnessDir, 'handoff-state.json');
|
|
179
|
+
if (fs.existsSync(handoffFile)) {
|
|
180
|
+
try {
|
|
181
|
+
const handoff = JSON.parse(fs.readFileSync(handoffFile, 'utf8'));
|
|
182
|
+
if (handoff.status === 'failed') {
|
|
183
|
+
const exitStr = handoff.exitCode ? ` (exit code ${handoff.exitCode})` : '';
|
|
184
|
+
const border = "─".repeat(60);
|
|
185
|
+
console.log(`\n┌${border}┐`);
|
|
186
|
+
console.log(`│ Harness OS - Handoff Checkpoint │`);
|
|
187
|
+
console.log(`├${border}┤`);
|
|
188
|
+
console.log(`│ - Last active action failed${exitStr.padEnd(30)} │`);
|
|
189
|
+
console.log(`│ - Action: ${String(handoff.tool || 'unknown').padEnd(46)} │`);
|
|
190
|
+
console.log(`│ - Last error output snippet: │`);
|
|
191
|
+
|
|
192
|
+
const snippet = handoff.errorSummary || 'No snippet available';
|
|
193
|
+
const lines = snippet.split('\n').map(l => l.trim()).filter(Boolean).slice(-3);
|
|
194
|
+
if (lines.length > 0) {
|
|
195
|
+
lines.forEach(line => {
|
|
196
|
+
const paddedLine = line.slice(0, 54).padEnd(54);
|
|
197
|
+
console.log(`│ > ${paddedLine} │`);
|
|
198
|
+
});
|
|
199
|
+
} else {
|
|
200
|
+
console.log(`│ > None │`);
|
|
201
|
+
}
|
|
202
|
+
console.log(`├${border}┤`);
|
|
203
|
+
console.log(`│ Hint: Use this context to continue resolving the issue │`);
|
|
204
|
+
console.log(`│ efficiently or start on a new path. │`);
|
|
205
|
+
console.log(`└${border}┘\n`);
|
|
206
|
+
}
|
|
207
|
+
} catch (err) {
|
|
208
|
+
// Ignore malformed files
|
|
209
|
+
}
|
|
210
|
+
}
|
|
211
|
+
|
|
212
|
+
console.log("Harness OS initialized. Ready for user prompt.");
|
|
@@ -0,0 +1,102 @@
|
|
|
1
|
+
'use strict';
|
|
2
|
+
|
|
3
|
+
const ENSEMBLE_MAX_CANDIDATES = 3;
|
|
4
|
+
|
|
5
|
+
function unique(values) {
|
|
6
|
+
return [...new Set((values || []).filter(Boolean))];
|
|
7
|
+
}
|
|
8
|
+
|
|
9
|
+
function detectEnsembleTaskSignals(prompt) {
|
|
10
|
+
const text = String(prompt || '');
|
|
11
|
+
const comparableOutput = /\b(?:choice|choose|decision|decide|ranking|rank|ranked|verdict|recommendation|recommend|compare alternatives?|compare approaches?|which (?:option|approach|design)|select between|trade[- ]offs?)\b/i.test(text) ||
|
|
12
|
+
/(?:選擇|決策|排名|判定|建議|比較方案|比較方法|取捨)/.test(text);
|
|
13
|
+
const explicitDisagreement = /\b(?:disagreement|dissent|minority (?:view|position)|competing (?:answers|positions|approaches)|preserve disagreement)\b/i.test(text) ||
|
|
14
|
+
/(?:分歧|不同意見|少數意見|競爭方案)/.test(text);
|
|
15
|
+
const creativeGeneration = /\b(?:creative|brainstorm|story|poem|fiction|slogan|tagline|copywriting|marketing copy|logo|illustration|artwork|lyrics?|name ideas?|generate ideas?)\b/i.test(text) ||
|
|
16
|
+
/(?:創意|腦力激盪|故事|詩|標語|文案|插畫|歌詞|命名)/.test(text);
|
|
17
|
+
const mechanicalWork = /\b(?:mechanical|bulk rename|mass rename|rename all|replace all|reformat|format all|lint[- ]fix|sort imports?|codemod|repetitive edits?|generated files?|migrate all identifiers?)\b/i.test(text) ||
|
|
18
|
+
/(?:機械式|批次重新命名|全部取代|大量重複修改|格式化全部)/.test(text);
|
|
19
|
+
const highStakes = /\b(?:high[- ]stakes|safety[- ]critical|security[- ]critical|critical decision|material risk)\b/i.test(text) ||
|
|
20
|
+
/(?:高風險|高利害|關鍵決策)/.test(text);
|
|
21
|
+
return {
|
|
22
|
+
comparableOutput,
|
|
23
|
+
explicitDisagreement,
|
|
24
|
+
creativeGeneration,
|
|
25
|
+
mechanicalWork,
|
|
26
|
+
highStakes,
|
|
27
|
+
};
|
|
28
|
+
}
|
|
29
|
+
|
|
30
|
+
function applyEnsemblePolicy(contract, prompt) {
|
|
31
|
+
if (!contract || typeof contract !== 'object' || !contract.workflowPlan || !contract.taskShape) return contract;
|
|
32
|
+
const plan = contract.workflowPlan;
|
|
33
|
+
const shape = contract.taskShape;
|
|
34
|
+
const explicit = shape.explicitRequest || {};
|
|
35
|
+
const prohibitions = Array.isArray(explicit.prohibitions) ? explicit.prohibitions : [];
|
|
36
|
+
const signals = detectEnsembleTaskSignals(prompt);
|
|
37
|
+
const highUncertainty = shape.uncertainty === 'high';
|
|
38
|
+
const fableTopology = typeof plan.strategy === 'string' && plan.strategy.startsWith('fable-');
|
|
39
|
+
const comparable = signals.comparableOutput || signals.explicitDisagreement;
|
|
40
|
+
|
|
41
|
+
const reasonCodes = [];
|
|
42
|
+
if (highUncertainty) reasonCodes.push('high-uncertainty');
|
|
43
|
+
if (signals.highStakes) reasonCodes.push('high-stakes');
|
|
44
|
+
if (signals.comparableOutput) reasonCodes.push('comparable-output');
|
|
45
|
+
if (signals.explicitDisagreement) reasonCodes.push('explicit-disagreement');
|
|
46
|
+
|
|
47
|
+
let selected = (highUncertainty || signals.highStakes) && comparable && fableTopology;
|
|
48
|
+
if (signals.creativeGeneration) {
|
|
49
|
+
selected = false;
|
|
50
|
+
reasonCodes.push('ensemble-excluded-creative-generation');
|
|
51
|
+
}
|
|
52
|
+
if (signals.mechanicalWork) {
|
|
53
|
+
selected = false;
|
|
54
|
+
reasonCodes.push('ensemble-excluded-mechanical-work');
|
|
55
|
+
}
|
|
56
|
+
if (prohibitions.includes('ensemble')) {
|
|
57
|
+
selected = false;
|
|
58
|
+
reasonCodes.push('user-prohibited-ensemble');
|
|
59
|
+
}
|
|
60
|
+
if (shape.hostCapabilities && shape.hostCapabilities.subagents === 'unavailable') {
|
|
61
|
+
selected = false;
|
|
62
|
+
reasonCodes.push('ensemble-subagents-unavailable');
|
|
63
|
+
}
|
|
64
|
+
if (plan.fallback && plan.fallback.disposition === 'blocked') {
|
|
65
|
+
selected = false;
|
|
66
|
+
reasonCodes.push('ensemble-base-plan-blocked');
|
|
67
|
+
}
|
|
68
|
+
|
|
69
|
+
plan.reasonCodes = unique([...(plan.reasonCodes || []), ...reasonCodes]);
|
|
70
|
+
if (!selected) {
|
|
71
|
+
plan.ensemble = null;
|
|
72
|
+
return contract;
|
|
73
|
+
}
|
|
74
|
+
|
|
75
|
+
const selectionReason = signals.explicitDisagreement
|
|
76
|
+
? 'ensemble-disagreement-deliverable'
|
|
77
|
+
: signals.highStakes
|
|
78
|
+
? 'ensemble-high-stakes-comparable-output'
|
|
79
|
+
: 'ensemble-high-uncertainty-comparable-output';
|
|
80
|
+
|
|
81
|
+
plan.ensemble = {
|
|
82
|
+
maxCandidates: ENSEMBLE_MAX_CANDIDATES,
|
|
83
|
+
diversity: ['model', 'prompt', 'evidence'],
|
|
84
|
+
synthesis: 'preserve-disagreement',
|
|
85
|
+
verifier: 'independent',
|
|
86
|
+
reasonCodes: unique([selectionReason, ...reasonCodes]),
|
|
87
|
+
};
|
|
88
|
+
plan.patterns = unique([...(plan.patterns || []), 'parallel-self-consistency']);
|
|
89
|
+
plan.requiredInvariants = unique([
|
|
90
|
+
...(plan.requiredInvariants || []),
|
|
91
|
+
'preserve-disagreement',
|
|
92
|
+
'independent-ensemble-verifier',
|
|
93
|
+
]);
|
|
94
|
+
plan.reasonCodes = unique([...(plan.reasonCodes || []), 'ensemble-review-selected']);
|
|
95
|
+
return contract;
|
|
96
|
+
}
|
|
97
|
+
|
|
98
|
+
module.exports = {
|
|
99
|
+
ENSEMBLE_MAX_CANDIDATES,
|
|
100
|
+
applyEnsemblePolicy,
|
|
101
|
+
detectEnsembleTaskSignals,
|
|
102
|
+
};
|
|
@@ -0,0 +1,212 @@
|
|
|
1
|
+
#!/usr/bin/env node
|
|
2
|
+
const crypto = require('crypto');
|
|
3
|
+
const fs = require('fs');
|
|
4
|
+
const os = require('os');
|
|
5
|
+
const path = require('path');
|
|
6
|
+
const { spawnSync } = require('child_process');
|
|
7
|
+
const {
|
|
8
|
+
createDegradedRouterContract,
|
|
9
|
+
validateRouterContract,
|
|
10
|
+
} = require('./router-contract');
|
|
11
|
+
const { applyEnsemblePolicy } = require('./ensemble-policy');
|
|
12
|
+
|
|
13
|
+
/**
|
|
14
|
+
* Harness Kernel Router
|
|
15
|
+
*
|
|
16
|
+
* tier-router remains the human-readable classifier/guide detector. The
|
|
17
|
+
* kernel consumes its versioned side-channel JSON contract instead of
|
|
18
|
+
* scraping human-readable stdout. Phase 2 selects an execution topology, but
|
|
19
|
+
* the router still only emits a plan: executors retain responsibility for
|
|
20
|
+
* spawning agents, creating workspaces, running tools, and enforcing gates.
|
|
21
|
+
* Phase 4 attaches ensemble-review only as a post-selection modifier.
|
|
22
|
+
*/
|
|
23
|
+
|
|
24
|
+
const INVARIANT_TEXT = {
|
|
25
|
+
'scope-lock': 'Route before execution: stay inside the authorized task/repository scope.',
|
|
26
|
+
'verify-before-claim': 'Verify before claim: completion claims require objective evidence appropriate to the change.',
|
|
27
|
+
'replan-after-repeated-failure': 'Re-plan on repetition: after 3 same-signature failures, stop micro-retrying and zoom out/re-diagnose.',
|
|
28
|
+
'loop-budget': 'Stop iterative-single when its explicit iteration budget is exhausted.',
|
|
29
|
+
'objective-verification': 'Use an objective check for iterative work; self-critique alone is not verification.',
|
|
30
|
+
'stage-contracts': 'Fable stages must have explicit stage contracts and pass conditions.',
|
|
31
|
+
'cold-verification': 'Fable delivery requires a cold/independent verifier where specified.',
|
|
32
|
+
'parallel-scope-contract': 'Parallel work requires declared independent scopes and non-overlapping/read-only writes.',
|
|
33
|
+
'synthesis-barrier': 'Do not claim parallel completion until all workstreams reach the synthesis barrier.',
|
|
34
|
+
'handoff-contracts': 'Multi-agent workspace work uses orchestrator-owned handoff contracts; workers do not form a peer mesh.',
|
|
35
|
+
'workspace-state': 'Durable workspace state remains scoped to the workspace and existing state conventions.',
|
|
36
|
+
'pre-action-approval': 'Irreversible/external side effects require approval before the exact payload executes.',
|
|
37
|
+
'preserve-disagreement': 'Ensemble synthesis must retain unresolved minority positions and evidence gaps.',
|
|
38
|
+
'independent-ensemble-verifier': 'Ensemble delivery requires a verifier independent from the candidate identities; agreement alone is not proof.',
|
|
39
|
+
};
|
|
40
|
+
|
|
41
|
+
const SKILL_TEXT = {
|
|
42
|
+
'tdd': 'tdd: useful for behavioral changes where executable tests can drive the implementation.',
|
|
43
|
+
'verification-loop': 'verification-loop: useful for systematic build/lint/test/diff evidence before delivery.',
|
|
44
|
+
'fable-mode': 'fable-mode / fable-discipline: useful for macro planning or deliberate multi-agent decomposition.',
|
|
45
|
+
'fable-discipline': null,
|
|
46
|
+
'multi-agent-workspace': 'multi-agent-workspace: useful when durable bounded delegation, handoffs, or workspace memory are required.',
|
|
47
|
+
};
|
|
48
|
+
|
|
49
|
+
function sanitizeClassifierOutput(stdout) {
|
|
50
|
+
const lines = String(stdout || '').split(/\r?\n/);
|
|
51
|
+
const filtered = [];
|
|
52
|
+
let skipLegacyContinuation = false;
|
|
53
|
+
|
|
54
|
+
for (const line of lines) {
|
|
55
|
+
if (line.includes('=> BASE EXECUTION LOOP:')) {
|
|
56
|
+
skipLegacyContinuation = true;
|
|
57
|
+
continue;
|
|
58
|
+
}
|
|
59
|
+
if (skipLegacyContinuation && line.includes('Track exactly ONE item in-progress')) {
|
|
60
|
+
skipLegacyContinuation = false;
|
|
61
|
+
continue;
|
|
62
|
+
}
|
|
63
|
+
skipLegacyContinuation = false;
|
|
64
|
+
filtered.push(line);
|
|
65
|
+
}
|
|
66
|
+
|
|
67
|
+
return filtered.join('\n').replace(/\n{3,}/g, '\n\n').trimEnd();
|
|
68
|
+
}
|
|
69
|
+
|
|
70
|
+
function readStructuredContract(contractPath) {
|
|
71
|
+
try {
|
|
72
|
+
const parsed = JSON.parse(fs.readFileSync(contractPath, 'utf8'));
|
|
73
|
+
const validation = validateRouterContract(parsed);
|
|
74
|
+
if (!validation.valid) {
|
|
75
|
+
return createDegradedRouterContract(
|
|
76
|
+
'router-contract-invalid',
|
|
77
|
+
`Structured router contract failed validation: ${validation.errors.join('; ')}`,
|
|
78
|
+
);
|
|
79
|
+
}
|
|
80
|
+
return parsed;
|
|
81
|
+
} catch (err) {
|
|
82
|
+
return createDegradedRouterContract(
|
|
83
|
+
'router-contract-unavailable',
|
|
84
|
+
`Structured router contract unavailable: ${err.message}`,
|
|
85
|
+
);
|
|
86
|
+
}
|
|
87
|
+
}
|
|
88
|
+
|
|
89
|
+
function resolvePrompt(prompt, stdinPayload) {
|
|
90
|
+
if (prompt) return prompt;
|
|
91
|
+
if (!stdinPayload) return '';
|
|
92
|
+
try {
|
|
93
|
+
const payload = JSON.parse(stdinPayload);
|
|
94
|
+
return typeof payload.prompt === 'string' ? payload.prompt : '';
|
|
95
|
+
} catch (_) {
|
|
96
|
+
return '';
|
|
97
|
+
}
|
|
98
|
+
}
|
|
99
|
+
|
|
100
|
+
function printWorkflowPlan(plan) {
|
|
101
|
+
const visible = {
|
|
102
|
+
schemaVersion: plan.schemaVersion,
|
|
103
|
+
routingStatus: plan.routingStatus,
|
|
104
|
+
tier: plan.tier,
|
|
105
|
+
strategy: plan.strategy,
|
|
106
|
+
strategySelection: plan.strategySelection,
|
|
107
|
+
actionGate: plan.actionGate,
|
|
108
|
+
limits: plan.limits,
|
|
109
|
+
parallelism: plan.parallelism,
|
|
110
|
+
workspace: plan.workspace,
|
|
111
|
+
memory: plan.memory,
|
|
112
|
+
verification: plan.verification,
|
|
113
|
+
ensemble: plan.ensemble,
|
|
114
|
+
fallback: plan.fallback,
|
|
115
|
+
reasonCodes: plan.reasonCodes,
|
|
116
|
+
};
|
|
117
|
+
console.log(`\n=> ROUTER WORKFLOW PLAN (JSON): ${JSON.stringify(visible)}`);
|
|
118
|
+
}
|
|
119
|
+
|
|
120
|
+
function printKernelContract(plan) {
|
|
121
|
+
console.log('\n=> REQUIRED HARNESS INVARIANTS:');
|
|
122
|
+
for (const invariant of plan.requiredInvariants || []) {
|
|
123
|
+
console.log(` - ${invariant}: ${INVARIANT_TEXT[invariant] || 'Required by the selected workflow plan.'}`);
|
|
124
|
+
}
|
|
125
|
+
|
|
126
|
+
console.log('\n=> SUGGESTED SKILLS (ADVISORY — no fixed workflow order):');
|
|
127
|
+
if (Array.isArray(plan.suggestedSkills) && plan.suggestedSkills.length > 0) {
|
|
128
|
+
const emitted = new Set();
|
|
129
|
+
for (const skill of plan.suggestedSkills) {
|
|
130
|
+
if (emitted.has(skill)) continue;
|
|
131
|
+
const text = Object.prototype.hasOwnProperty.call(SKILL_TEXT, skill) ? SKILL_TEXT[skill] : `${skill}: advisory for the selected topology.`;
|
|
132
|
+
if (text) console.log(` - ${text}`);
|
|
133
|
+
emitted.add(skill);
|
|
134
|
+
if (skill === 'fable-mode') emitted.add('fable-discipline');
|
|
135
|
+
}
|
|
136
|
+
if (plan.strategy && plan.strategy.startsWith('fable-')) {
|
|
137
|
+
console.log(' - Use these selectively; Tier 3/Fable does not create a universal skill pipeline.');
|
|
138
|
+
}
|
|
139
|
+
} else if (plan.strategy === 'direct-single') {
|
|
140
|
+
console.log(' - No mandatory domain skill. Prefer the bounded direct path and load a focused skill only when it adds value.');
|
|
141
|
+
} else if (plan.strategySelection === 'deferred') {
|
|
142
|
+
console.log(' - Strategy is deferred/unclassified. Do not infer triviality; choose the smallest justified approach from task evidence.');
|
|
143
|
+
} else {
|
|
144
|
+
console.log(' - No additional skill suggestion from the workflow plan.');
|
|
145
|
+
}
|
|
146
|
+
|
|
147
|
+
if (plan.ensemble) {
|
|
148
|
+
console.log(`\n=> ENSEMBLE REVIEW: bounded to ${plan.ensemble.maxCandidates} candidates; synthesis=${plan.ensemble.synthesis}; verifier=${plan.ensemble.verifier}. Preserve minority positions and do not claim improvement without paired evidence.`);
|
|
149
|
+
}
|
|
150
|
+
|
|
151
|
+
if (plan.actionGate && plan.actionGate.required) {
|
|
152
|
+
console.log(`\n=> ACTION GATE REQUIRED: ${plan.actionGate.reasonCodes.join(', ')}; disposition=${plan.actionGate.disposition}. The router only declares this requirement; execution must not bypass approval.`);
|
|
153
|
+
}
|
|
154
|
+
|
|
155
|
+
if (plan.fallback && plan.fallback.disposition !== 'none') {
|
|
156
|
+
console.log(`\n=> ROUTING FALLBACK: ${plan.fallback.disposition}/${plan.fallback.mode}; ${plan.fallback.reasonCodes.join(', ')}.`);
|
|
157
|
+
}
|
|
158
|
+
|
|
159
|
+
if (plan.routingStatus === 'degraded') {
|
|
160
|
+
console.log('\n=> ROUTING DEGRADATION: Structured routing is degraded. Keep the strategy deferred unless independent evidence supports a route; do not silently downgrade to Tier 1/direct execution.');
|
|
161
|
+
}
|
|
162
|
+
|
|
163
|
+
console.log('\n=> ORCHESTRATION POLICY: Do not enforce workflow order. Enforce workflow invariants from the plan; suggested skills remain advisory. The router plans; execution components execute.');
|
|
164
|
+
}
|
|
165
|
+
|
|
166
|
+
function run(prompt, stdinPayload) {
|
|
167
|
+
const classifierPath = path.join(__dirname, 'tier-router.js');
|
|
168
|
+
const contractPath = path.join(os.tmpdir(), `harness-router-${process.pid}-${crypto.randomUUID()}.json`);
|
|
169
|
+
const args = prompt ? [classifierPath, prompt] : [classifierPath];
|
|
170
|
+
const result = spawnSync(process.execPath, args, {
|
|
171
|
+
input: prompt ? undefined : stdinPayload,
|
|
172
|
+
encoding: 'utf8',
|
|
173
|
+
env: {
|
|
174
|
+
...process.env,
|
|
175
|
+
HARNESS_ROUTER_CONTRACT_PATH: contractPath,
|
|
176
|
+
},
|
|
177
|
+
});
|
|
178
|
+
|
|
179
|
+
if (result.error) {
|
|
180
|
+
console.error(`[Harness Kernel] Failed to execute tier-router.js: ${result.error.message}`);
|
|
181
|
+
process.exit(1);
|
|
182
|
+
}
|
|
183
|
+
|
|
184
|
+
const sanitized = sanitizeClassifierOutput(result.stdout);
|
|
185
|
+
if (sanitized) console.log(sanitized);
|
|
186
|
+
if (result.stderr) process.stderr.write(result.stderr);
|
|
187
|
+
|
|
188
|
+
const contract = readStructuredContract(contractPath);
|
|
189
|
+
applyEnsemblePolicy(contract, resolvePrompt(prompt, stdinPayload));
|
|
190
|
+
try {
|
|
191
|
+
fs.rmSync(contractPath, { force: true });
|
|
192
|
+
} catch (err) {
|
|
193
|
+
// Temporary contract cleanup is best effort only.
|
|
194
|
+
}
|
|
195
|
+
|
|
196
|
+
printWorkflowPlan(contract.workflowPlan);
|
|
197
|
+
printKernelContract(contract.workflowPlan);
|
|
198
|
+
|
|
199
|
+
if (result.status !== 0) process.exit(result.status === null ? 1 : result.status);
|
|
200
|
+
}
|
|
201
|
+
|
|
202
|
+
const prompt = process.argv.slice(2).join(' ');
|
|
203
|
+
if (prompt) {
|
|
204
|
+
run(prompt, null);
|
|
205
|
+
} else if (process.stdin.isTTY) {
|
|
206
|
+
run('', null);
|
|
207
|
+
} else {
|
|
208
|
+
let input = '';
|
|
209
|
+
process.stdin.setEncoding('utf8');
|
|
210
|
+
process.stdin.on('data', chunk => { input += chunk; });
|
|
211
|
+
process.stdin.on('end', () => run('', input));
|
|
212
|
+
}
|