@agentskit/harness 0.3.0 → 0.5.0

package/compatibility/manifest.json

@@ -0,0 +1,17 @@
+{
+  "type": "agentskit-harness-compatibility-manifest",
+  "schemaVersion": 1,
+  "harnessVersion": "0.4.0",
+  "sourceRevision": "dad0818e5ef36be7427bdbf8f8e1522c1b2e1ec5",
+  "components": [
+    { "id": "core", "package": "@agentskit/core", "version": "1.12.9", "revision": "d89a2d74108c29f2c6227948265449e985af3a0a", "repository": "https://github.com/AgentsKit-io/agentskit", "adapterBoundary": "real-adapter", "testCommand": "pnpm --filter @agentskit/core test", "evalCommand": "pnpm --filter @agentskit/core test:coverage", "previousVersion": "@agentskit/core@1.12.8", "noHarnessBaseline": "benchmarks/harness-0.4.0-baseline.json", "migrationEvidence": "compatibility/migration.md", "rollbackEvidence": "compatibility/rollback.md" },
+    { "id": "memory", "package": "@agentskit/memory", "version": "0.11.9", "revision": "d89a2d74108c29f2c6227948265449e985af3a0a", "repository": "https://github.com/AgentsKit-io/agentskit", "adapterBoundary": "real-adapter", "testCommand": "pnpm --filter @agentskit/memory test", "evalCommand": "pnpm --filter @agentskit/memory test:coverage", "previousVersion": "@agentskit/memory@0.11.8", "noHarnessBaseline": "benchmarks/harness-0.4.0-baseline.json", "migrationEvidence": "compatibility/migration.md", "rollbackEvidence": "compatibility/rollback.md" },
+    { "id": "eval", "package": "@agentskit/eval", "version": "0.6.5", "revision": "d89a2d74108c29f2c6227948265449e985af3a0a", "repository": "https://github.com/AgentsKit-io/agentskit", "adapterBoundary": "real-adapter", "testCommand": "pnpm --filter @agentskit/eval test", "evalCommand": "pnpm --filter @agentskit/eval test:interop", "previousVersion": "@agentskit/eval@0.6.3", "noHarnessBaseline": "benchmarks/harness-0.4.0-baseline.json", "migrationEvidence": "compatibility/migration.md", "rollbackEvidence": "compatibility/rollback.md" },
+    { "id": "doc-bridge", "package": "@agentskit/doc-bridge", "version": "1.7.45", "revision": "9f9e9ddd4f736981c2bea243fef33b68be2b5002", "repository": "https://github.com/AgentsKit-io/doc-bridge", "adapterBoundary": "real-adapter", "testCommand": "pnpm test", "evalCommand": "pnpm test:verification-harness", "previousVersion": "@agentskit/doc-bridge@1.7.44", "noHarnessBaseline": "benchmarks/harness-0.4.0-baseline.json", "migrationEvidence": "compatibility/migration.md", "rollbackEvidence": "compatibility/rollback.md" },
+    { "id": "code-review", "package": "@agentskit/code-review", "version": "0.30.8", "revision": "c993f9a9359157a767d738d60c30add2b02fd09a", "repository": "https://github.com/AgentsKit-io/code-review", "adapterBoundary": "real-adapter", "testCommand": "npm test", "evalCommand": "npm run quality:matrix", "previousVersion": "@agentskit/code-review@0.30.7", "noHarnessBaseline": "benchmarks/harness-0.4.0-baseline.json", "migrationEvidence": "compatibility/migration.md", "rollbackEvidence": "compatibility/rollback.md" },
+    { "id": "adapter-boundary", "package": "@agentskit/harness", "version": "0.4.0", "revision": "dad0818e5ef36be7427bdbf8f8e1522c1b2e1ec5", "repository": "https://github.com/AgentsKit-io/harness", "adapterBoundary": "real-adapter", "testCommand": "pnpm test:adapters", "evalCommand": "pnpm test:eval-battery", "previousVersion": "@agentskit/harness@0.3.0", "noHarnessBaseline": "benchmarks/harness-0.4.0-baseline.json", "migrationEvidence": "compatibility/migration.md", "rollbackEvidence": "compatibility/rollback.md" },
+    { "id": "runtime", "package": "@agentskit/harness", "version": "0.4.0", "revision": "dad0818e5ef36be7427bdbf8f8e1522c1b2e1ec5", "repository": "https://github.com/AgentsKit-io/harness", "adapterBoundary": "real-adapter", "testCommand": "pnpm test:runtime", "evalCommand": "pnpm test:runtime-attestation", "previousVersion": "@agentskit/harness@0.3.0", "noHarnessBaseline": "benchmarks/harness-0.4.0-baseline.json", "migrationEvidence": "compatibility/migration.md", "rollbackEvidence": "compatibility/rollback.md" }
+  ],
+  "evidenceOutputs": ["compatibility/report.json", "compatibility/report.md", "compatibility/migration.md", "compatibility/rollback.md"],
+  "digest": "97d842eebdddc5c2c9c91f42c5275f28db69b2309f5e797ced653831d0b7687a"
+}

package/compatibility/migration.md

@@ -0,0 +1,10 @@
+# Compatibility migration rehearsal
+
+The manifest pins package versions and source revisions before an integration
+run. Execute each component's `testCommand` and `evalCommand` from a clean
+checkout, then attach the output to `compatibility/report.json` and compare it
+with the declared previous version and no-Harness baseline.
+
+This file is the reproducible procedure; it is not evidence of a successful
+upstream run by itself. Missing or stale command output remains `unknown` and
+blocks the compatibility gate.

package/compatibility/report.json

@@ -0,0 +1,23 @@
+{
+  "type": "agentskit-harness-compatibility-report",
+  "schemaVersion": 1,
+  "harnessVersion": "0.4.0",
+  "harnessSourceRevision": "2df60bcbda850382d1a1112f338fa53858c5d4cf",
+  "capturedAt": "2026-09-10T18:30:00Z",
+  "status": "blocked",
+  "components": [
+    { "id": "core", "package": "@agentskit/core", "version": "1.12.9", "revision": "d89a2d74108c29f2c6227948265449e985af3a0a", "status": "passed", "command": "pnpm --filter @agentskit/core test", "evidence": "34 files, 498 tests passed" },
+    { "id": "memory", "package": "@agentskit/memory", "version": "0.11.9", "revision": "d89a2d74108c29f2c6227948265449e985af3a0a", "status": "passed", "command": "pnpm --filter @agentskit/memory test", "evidence": "31 files, 228 tests passed" },
+    { "id": "eval", "package": "@agentskit/eval", "version": "0.6.5", "revision": "d89a2d74108c29f2c6227948265449e985af3a0a", "status": "passed", "command": "pnpm --filter @agentskit/eval test", "evidence": "9 files, 105 tests passed; build and bundler preparation passed" },
+    { "id": "doc-bridge", "package": "@agentskit/doc-bridge", "version": "1.7.45", "revision": "9f9e9ddd4f736981c2bea243fef33b68be2b5002", "status": "passed", "command": "pnpm build && pnpm test && pnpm typecheck", "evidence": "50 files, 351 tests passed; Cursor/Claude/Copilot/portable plugin contracts passed; typecheck passed" },
+    { "id": "code-review", "package": "@agentskit/code-review", "version": "0.30.8", "revision": "c993f9a9359157a767d738d60c30add2b02fd09a", "status": "failed", "command": "npm test; npm run quality:matrix -- --input <real-runner-artifact>", "evidence": "241 tests passed. Real run ceaf007a-ea6e-4dc9-a601-f816702b301c reached BLOCKED: detection 1/4, precision 1/4, batch-efficiency 1/4, integration 1/4; no comparable baseline." },
+    { "id": "adapter-boundary", "package": "@agentskit/harness", "version": "0.4.0", "revision": "2df60bcbda850382d1a1112f338fa53858c5d4cf", "status": "passed", "command": "pnpm test:adapters", "evidence": "covered by ak-verify run 1789065264843-3627-dgz8xb" },
+    { "id": "runtime", "package": "@agentskit/harness", "version": "0.4.0", "revision": "2df60bcbda850382d1a1112f338fa53858c5d4cf", "status": "passed", "command": "pnpm test:runtime && pnpm test:runtime-attestation", "evidence": "covered by ak-verify run 1789065264843-3627-dgz8xb" }
+  ],
+  "blockers": [
+    "code-review real-run quality matrix has no passing detection/precision/baseline evidence",
+    "harness-vs-no-Harness cohort is not yet comparable (10-task owner-approved cohort required)",
+    "published-registry smoke is pending the main-branch release"
+  ],
+  "baseline": { "path": "benchmarks/harness-0.4.0-baseline.json", "status": "not-run", "doNotInfer": true }
+}

package/compatibility/report.md

@@ -0,0 +1,22 @@
+# Compatibility report
+
+**Release:** `@agentskit/harness@0.4.0`  
+**Harness revision:** `2df60bcbda850382d1a1112f338fa53858c5d4cf`
+**Captured:** 2026-09-10  
+**Decision:** BLOCKED (fail closed)
+
+## Passed boundaries
+
+- `@agentskit/core@1.12.9` at `d89a2d7`: 498/498 tests.
+- `@agentskit/memory@0.11.9` at `d89a2d7`: 228/228 tests.
+- `@agentskit/eval@0.6.5` at `d89a2d7`: 105/105 tests, including build/bundler preparation.
+- `@agentskit/doc-bridge@1.7.45` at `9f9e9d`: 351/351 tests plus plugin contracts and typecheck.
+- Harness adapter and runtime boundaries: covered by `ak-verify` run `1789063362018-21016-775q93` (21/21 checks).
+
+## Blocked evidence
+
+`@agentskit/code-review@0.30.8` passed its 241-test suite. A real fixed-provider run (`codex-cli`, `gpt-5.6-luna`, high, run `ceaf007a-ea6e-4dc9-a601-f816702b301c`) completed coverage, safety, reliability, memory, configuration, and actionability checks, but the quality matrix blocked on detection, precision, batch efficiency, integration, and missing comparable baselines. The result is retained as evidence, not promoted to a pass.
+
+The 0.4.0 no-Harness comparison remains unavailable because the owner-approved ten-task cohort and matching criterion-level logs have not been captured. Registry smoke can only run after the main Trusted Publishing workflow publishes the package.
+
+See [`migration.md`](migration.md), [`rollback.md`](rollback.md), and the machine-readable [`report.json`](report.json).

package/compatibility/rollback.md

@@ -0,0 +1,8 @@
+# Compatibility rollback rehearsal
+
+Rollback is a clean checkout at the pinned `previousVersion`, followed by the
+same adapter-boundary smoke and eval commands. Record the resulting revision,
+commands, exit codes, and evidence digest in `compatibility/report.json`.
+
+No external mutation is performed by the Harness. A missing rehearsal record
+is `unknown` and blocks release rather than being inferred as a pass.