@zhivex-ai/harness 1.0.0-rc.7 → 1.0.0

package/docs/reports/evidence/ga-catalog-observed-plan-2026-09-19.json

@@ -0,0 +1,28 @@
+{
+  "kind": "development-catalog-observation",
+  "classification": "known task; not holdout or GA acceptance",
+  "baseline": "ga-recovery-graded-v8-qwen",
+  "model": "qwen3.8-flash",
+  "repetitions": 1,
+  "control": "unchanged mini-SWE-agent",
+  "catalog": "full 13-tool product benchmark catalog",
+  "changes": "adapter-boundary observation plus registry dependency updates; no Harness runtime or policy change",
+  "limits": {
+    "steps": 24,
+    "outputPerTurn": 2048,
+    "inputTokens": 100000,
+    "outputTokens": 16000,
+    "timeoutSeconds": 300,
+    "memoryMb": 2048,
+    "cpus": 2
+  },
+  "driverSha256": "24830996fb91c8ad34928a4b7d5da62ea1ab8f8792a77ac675cc9638d750ff2b",
+  "observerSha256": "eb79e7333f4f5804f96da87d1952382f620e7165a95b60489fcc2588debe64d8",
+  "versions": {
+    "core": "1.22.0",
+    "openai": "0.13.3",
+    "qwen": "0.14.3"
+  },
+  "lockSha256": "c213d497a98bfbea1379bd60850d0ec8ede02560df9a63bf3b737d35648c4909",
+  "authorization": "User explicitly authorized Qwen trial after registry integration"
+}

package/docs/reports/evidence/ga-catalog-observed-status-2026-09-19.json

@@ -0,0 +1,15 @@
+{
+  "kind": "development-catalog-observation-status",
+  "status": "completed",
+  "reason": null,
+  "plan": "ga-catalog-observed-plan-2026-09-19.json",
+  "liveEvidence": true,
+  "userAuthorization": "explicit user authorization received",
+  "localValidation": {
+    "command": "bun test tests/swebench-model-catalog.test.ts tests/sdk-compaction.test.ts tests/swebench.test.ts",
+    "exitCode": 0,
+    "passed": 25,
+    "failed": 0
+  },
+  "scope": "Authorized trial completed with independent grading; both samples retained"
+}

package/docs/reports/evidence/ga-complete-local-check-2026-09-19.json

@@ -0,0 +1,47 @@
+{
+  "schemaVersion": 1,
+  "kind": "complete-local-ga-preparation-check",
+  "implementationCommit": "a6d581a04eea01f7c430fe9ff1eea61caa86ecdb",
+  "command": "bun run check",
+  "exitCode": 0,
+  "logSha256": "602f0fc9c810234ecef53a5f7a8e0f889b482c26750d326b4c40fd9ba99c1a6e",
+  "tests": {
+    "passed": 584,
+    "failed": 0
+  },
+  "checks": [
+    "documentation",
+    "preparation-contract",
+    "runtime-types",
+    "tooling-types",
+    "unit-and-integration-tests",
+    "historical-migrations",
+    "deterministic-evaluation",
+    "workspace-benchmark",
+    "safe-fix-image-contract",
+    "safe-fix-benchmark",
+    "mcp-interoperability",
+    "official-mcp-sdk",
+    "oci-execution",
+    "installed-package"
+  ],
+  "dependencyAudit": {
+    "checkedPackages": 35,
+    "vulnerabilities": 0,
+    "untrustedDependenciesWithScripts": 0
+  },
+  "releaseReadiness": {
+    "passed": false,
+    "observedFailures": [
+      "RC14 changelog lacks an ISO-dated release heading",
+      "RC14 is marked Unreleased",
+      "Evidence additions make worktree dirty at check time",
+      "Checks are running on a feature branch rather than main"
+    ]
+  },
+  "limits": [
+    "Local preparation is not protected release certification.",
+    "Final representative, publication/provenance and artifact-bound security evidence remain pending.",
+    "SWE-bench fresh holdout remains 0/5 and known development rerun remains failed."
+  ]
+}

package/docs/reports/evidence/ga-completion-reminder-2026-09-19.json

@@ -0,0 +1,22 @@
+{
+  "schemaVersion": 1,
+  "kind": "bounded-completion-reminder-validation",
+  "baseCommit": "41bd1cb4ab8481a7413f096a0817f612fe9806fe",
+  "candidate": "working tree; not a release attestation",
+  "sources": {
+    "src/repair-controller.ts": "cc242f03700a9f0fe5d5dd1db3e670aba47ac2d8940910ebc93a6dbbe23265c1",
+    "src/model-budget.ts": "97c00019d7e737fc7acdd04cdac0811dca8d85c3e858de519ebba91888cf1ee7",
+    "src/runtime-checkpoints.ts": "e3b75a0b86fb5d21992b1a126b35446110ba5fcaf4317ed5005b1a44555118dc",
+    "src/harness.ts": "866115b64d90c0b2ffa3e07da6a015210afae0babe46052912d0cc288e23bc4b",
+    "tests/repair-completion-reminder.test.ts": "a3b762a0c2dd4ad06d73e0420590fbf095d0c164c6ad0f1a282b3dcb1ce6d503",
+    "tests/full-remediation.test.ts": "50d34584bab0df97381b2fdd609bda5613fb8ee034d037fafe356ce417bd6de9",
+    "bun.lock": "e370459197ce52c3961770f3c5ccaab7ff35ddbe641ef72e8bd61f87a8b8e3fe"
+  },
+  "suite": {
+    "passed": 551,
+    "failed": 0
+  },
+  "types": "passed",
+  "preparationContract": "passed",
+  "liveBoundary": "Qwen v6 failed work budget before exercising reminder; no live efficacy claim"
+}

package/docs/reports/evidence/ga-fresh-cohort-plan-2026-09-19.json

@@ -0,0 +1,54 @@
+{
+  "schemaVersion": 1,
+  "kind": "fresh-post-fix-external-cohort-plan",
+  "classification": "Five-task exploratory fresh holdout; not GA certification alone",
+  "implementationCommit": "1886644a12135348f645de634e4f10a45e247f6a",
+  "seed": 2026091902,
+  "taskCount": 5,
+  "repetitions": 1,
+  "candidates": [
+    "zhivex",
+    "mini-swe-agent"
+  ],
+  "provider": "qwen",
+  "model": "qwen3.8-flash",
+  "datasetRevision": "c104f840cc67f8b6eec6f759ebc8b2693d585d4a",
+  "excludedInstances": [
+    "astropy__astropy-14995",
+    "django__django-11265",
+    "django__django-12262",
+    "django__django-13410",
+    "django__django-14915",
+    "django__django-15128",
+    "django__django-15554",
+    "django__django-15569",
+    "django__django-15851",
+    "pydata__xarray-4695",
+    "pydata__xarray-6721",
+    "pytest-dev__pytest-8399",
+    "scikit-learn__scikit-learn-10908",
+    "scikit-learn__scikit-learn-14141",
+    "sympy__sympy-15809",
+    "sympy__sympy-20438",
+    "sympy__sympy-21930"
+  ],
+  "exclusionsSha256": "a57d36b1253ac1df5a2037b6d10cbac4a3f7453bea171b057a5ea7b96130cb2f",
+  "limits": {
+    "steps": 24,
+    "outputPerTurn": 2048,
+    "inputTokens": 100000,
+    "outputTokens": 16000,
+    "timeoutSeconds": 300,
+    "memoryMb": 2048,
+    "cpus": 2
+  },
+  "rules": [
+    "Selection fixed by sorted task IDs and seeded shuffle before any results.",
+    "Every selected task must prepare and pass preflight before paid calls.",
+    "Freeze implementation, model, images and budgets for all ten runs.",
+    "Keep failed, missing and ungraded samples in the denominator.",
+    "No per-task retries or replacement tasks.",
+    "Only public task material reaches models; independent official grader judges imported patches.",
+    "Report the entire cohort; do not infer GA from one passing task or adjust acceptance after results."
+  ]
+}

package/docs/reports/evidence/ga-fresh-cohort-preparation-2026-09-19.json

@@ -0,0 +1,242 @@
+{
+  "schemaVersion": 1,
+  "kind": "fresh-cohort-preparation",
+  "selectionSeed": 2026091902,
+  "selectionMatchesPlan": true,
+  "noPreviouslyUsedTasks": true,
+  "manifestSha256": "a6e06e6141f7151e37a0d64504447915e68209bfac9f908d29f4547406086ae9",
+  "tasks": [
+    {
+      "instance_id": "django__django-10914",
+      "repo": "django/django",
+      "base_commit": "e7fd69d051eaa67cb17f172a39b57253e9cb831a",
+      "image": "sha256:ab68f98ae10eca86803f8257a42bc6d2f9d3526afff773eabc8ad5aa3187f49d",
+      "evalImageId": "sha256:f26f0db931c6c1e785ab8287e1ea4f5f948846267db159deaf90b3fe7adea4e9",
+      "checkoutHead": "74cb72320235dc8efe51a21e64e113942f8500fc",
+      "baseTree": "51fdefd96ba482a04fc5492e30cd123252d1e658"
+    },
+    {
+      "instance_id": "pylint-dev__pylint-4604",
+      "repo": "pylint-dev/pylint",
+      "base_commit": "1e55ae64624d28c5fe8b63ad7979880ee2e6ef3f",
+      "image": "sha256:0f12a096eaf925ecd8360b4ac49de23df913dd8a82e91903bf452280012c38ed",
+      "evalImageId": "sha256:df6eb48f445388d0e228108c635625a0cc9f0cf5ed87303e89f75ab6fc3c2b39",
+      "checkoutHead": "0f266873c24f8f7270d257c8c8a8676930be0c72",
+      "baseTree": "05a6b93fcb973fb72fe368b7d43eee00ed6635da"
+    },
+    {
+      "instance_id": "psf__requests-1921",
+      "repo": "psf/requests",
+      "base_commit": "3c88e520da24ae6f736929a750876e7654accc3d",
+      "image": "sha256:1da8a3fbb935d6eed558af5c3e33a9073fe400806ee179ac7d080cf4f57d3e8a",
+      "evalImageId": "sha256:2ee0a4c706f04d2926a1622ec51b53d4ce6010b80d3ee10d414ea2a276c26cc9",
+      "checkoutHead": "040ee8c95a4bd91dde598105d7b12bc83306488a",
+      "baseTree": "cf441cdcaa76806a078893bffa79922351a92b0a"
+    },
+    {
+      "instance_id": "django__django-13964",
+      "repo": "django/django",
+      "base_commit": "f39634ff229887bf7790c069d0c411b38494ca38",
+      "image": "sha256:3b9aee6d887a7e1e5a2428e4992986fcb1baca15b9a0aabbb7da8f727106f354",
+      "evalImageId": "sha256:3bd826784556cee19b7d056d48ca3c7d793a77eac9fb311cdc460455f5d005ef",
+      "checkoutHead": "b74cd5f880b0a2a9732017426f68428b3daef92a",
+      "baseTree": "9893580bff8b44c40312575fc898ae72fb68cfd7"
+    },
+    {
+      "instance_id": "django__django-11138",
+      "repo": "django/django",
+      "base_commit": "c84b91b7603e488f7171fdff8f08368ef3d6b856",
+      "image": "sha256:74a78d2c188adb58884251594e6bd6fd5a4860d26c8da1e79c6a0cc382c7e94f",
+      "evalImageId": "sha256:ddc0080158473b72206a30e5333bcb04f19c4631c49bc039e396b721e47a11cc",
+      "checkoutHead": "44e1f73a4c7e0868cdd7bc21278780404511c3f8",
+      "baseTree": "7bb2d0d7f047e267e0146c34f89fab4cdaffaf53"
+    }
+  ],
+  "model": "qwen3.8-flash",
+  "limits": {
+    "steps": 24,
+    "outputPerTurn": 2048,
+    "inputTokens": 100000,
+    "outputTokens": 16000,
+    "timeoutSeconds": 300,
+    "memoryMb": 2048,
+    "cpus": 2
+  },
+  "modelRequestsSoFar": 0,
+  "preflight": {
+    "manifestSha256": "01110d5d894c155fa51c37e0f3153964bb09b701c2d5703377bc2473268bc89d",
+    "driverSha256": "67cee044060e78afea9fa663af8a686220aeafdb0db9ec318ef846f31aa9985d",
+    "results": [
+      {
+        "candidateSnapshot": null,
+        "candidateCaptureFailure": null,
+        "modelTimings": [],
+        "modelCatalog": {
+          "records": [],
+          "omittedCalls": 0
+        },
+        "stageTimings": [],
+        "approvalTimings": [],
+        "toolTimings": [],
+        "schemaVersion": 1,
+        "candidate": "zhivex",
+        "instanceId": "django__django-10914",
+        "status": "preflight-passed",
+        "failure": null,
+        "phase": "preflight",
+        "durationMs": 13538.983875,
+        "approvals": 0,
+        "turns": [],
+        "inputTokens": 0,
+        "outputTokens": 0,
+        "cachedInputTokens": 0,
+        "modelCalls": 0,
+        "usageComplete": true,
+        "compactions": [],
+        "terminalDiagnostic": null,
+        "thrownDiagnostic": null,
+        "guardrails": [],
+        "repairProgress": null,
+        "contextMetrics": [],
+        "omittedContextMeasurements": 0,
+        "toolCalls": 0
+      },
+      {
+        "candidateSnapshot": null,
+        "candidateCaptureFailure": null,
+        "modelTimings": [],
+        "modelCatalog": {
+          "records": [],
+          "omittedCalls": 0
+        },
+        "stageTimings": [],
+        "approvalTimings": [],
+        "toolTimings": [],
+        "schemaVersion": 1,
+        "candidate": "zhivex",
+        "instanceId": "pylint-dev__pylint-4604",
+        "status": "preflight-passed",
+        "failure": null,
+        "phase": "preflight",
+        "durationMs": 3167.9480000000003,
+        "approvals": 0,
+        "turns": [],
+        "inputTokens": 0,
+        "outputTokens": 0,
+        "cachedInputTokens": 0,
+        "modelCalls": 0,
+        "usageComplete": true,
+        "compactions": [],
+        "terminalDiagnostic": null,
+        "thrownDiagnostic": null,
+        "guardrails": [],
+        "repairProgress": null,
+        "contextMetrics": [],
+        "omittedContextMeasurements": 0,
+        "toolCalls": 0
+      },
+      {
+        "candidateSnapshot": null,
+        "candidateCaptureFailure": null,
+        "modelTimings": [],
+        "modelCatalog": {
+          "records": [],
+          "omittedCalls": 0
+        },
+        "stageTimings": [],
+        "approvalTimings": [],
+        "toolTimings": [],
+        "schemaVersion": 1,
+        "candidate": "zhivex",
+        "instanceId": "psf__requests-1921",
+        "status": "preflight-passed",
+        "failure": null,
+        "phase": "preflight",
+        "durationMs": 1251.44775,
+        "approvals": 0,
+        "turns": [],
+        "inputTokens": 0,
+        "outputTokens": 0,
+        "cachedInputTokens": 0,
+        "modelCalls": 0,
+        "usageComplete": true,
+        "compactions": [],
+        "terminalDiagnostic": null,
+        "thrownDiagnostic": null,
+        "guardrails": [],
+        "repairProgress": null,
+        "contextMetrics": [],
+        "omittedContextMeasurements": 0,
+        "toolCalls": 0
+      },
+      {
+        "candidateSnapshot": null,
+        "candidateCaptureFailure": null,
+        "modelTimings": [],
+        "modelCatalog": {
+          "records": [],
+          "omittedCalls": 0
+        },
+        "stageTimings": [],
+        "approvalTimings": [],
+        "toolTimings": [],
+        "schemaVersion": 1,
+        "candidate": "zhivex",
+        "instanceId": "django__django-13964",
+        "status": "preflight-passed",
+        "failure": null,
+        "phase": "preflight",
+        "durationMs": 14724.226542,
+        "approvals": 0,
+        "turns": [],
+        "inputTokens": 0,
+        "outputTokens": 0,
+        "cachedInputTokens": 0,
+        "modelCalls": 0,
+        "usageComplete": true,
+        "compactions": [],
+        "terminalDiagnostic": null,
+        "thrownDiagnostic": null,
+        "guardrails": [],
+        "repairProgress": null,
+        "contextMetrics": [],
+        "omittedContextMeasurements": 0,
+        "toolCalls": 0
+      },
+      {
+        "candidateSnapshot": null,
+        "candidateCaptureFailure": null,
+        "modelTimings": [],
+        "modelCatalog": {
+          "records": [],
+          "omittedCalls": 0
+        },
+        "stageTimings": [],
+        "approvalTimings": [],
+        "toolTimings": [],
+        "schemaVersion": 1,
+        "candidate": "zhivex",
+        "instanceId": "django__django-11138",
+        "status": "preflight-passed",
+        "failure": null,
+        "phase": "preflight",
+        "durationMs": 14313.235958000001,
+        "approvals": 0,
+        "turns": [],
+        "inputTokens": 0,
+        "outputTokens": 0,
+        "cachedInputTokens": 0,
+        "modelCalls": 0,
+        "usageComplete": true,
+        "compactions": [],
+        "terminalDiagnostic": null,
+        "thrownDiagnostic": null,
+        "guardrails": [],
+        "repairProgress": null,
+        "contextMetrics": [],
+        "omittedContextMeasurements": 0,
+        "toolCalls": 0
+      }
+    ]
+  }
+}