From 67830d0f612b59c8acf926e2a83dff00af38e2fa Mon Sep 17 00:00:00 2001 From: Dotta Date: Fri, 2 Oct 2026 23:36:17 -0500 Subject: [PATCH] Publish immutable native recovery admission and bounded manifest Co-Authored-By: Paperclip --- ...-native-completion-master-qualification.md | 12 + .../baseline.json | 312 ++++++++++++++++ .../candidate.json | 336 ++++++++++++++++++ .../failed-anchor-baseline.json | 312 ++++++++++++++++ .../failed-anchor-candidate.json | 336 ++++++++++++++++++ .../failed-sandbox-baseline.json | 312 ++++++++++++++++ .../failed-sandbox-candidate.json | 336 ++++++++++++++++++ .../hosted-source-reproduction.json | 75 ++++ ...e-completion-master-recovery-manifest.json | 218 ++++++++++++ ...10-02-stock-harness-paperclip-checklist.md | 5 + 10 files changed, 2254 insertions(+) create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-admission/baseline.json create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-admission/candidate.json create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-baseline.json create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-candidate.json create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-baseline.json create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-candidate.json create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-admission/hosted-source-reproduction.json create mode 100644 doc/plans/2026-10-02-native-completion-master-recovery-manifest.json diff --git a/doc/plans/2026-10-02-native-completion-master-qualification.md b/doc/plans/2026-10-02-native-completion-master-qualification.md index fdb2142d09a..16bd73b0e7e 100644 --- a/doc/plans/2026-10-02-native-completion-master-qualification.md +++ b/doc/plans/2026-10-02-native-completion-master-qualification.md @@ -1,3 +1,15 @@ +# Native completion qualification on current master context + +**TL;DR: no model outcomes have been measured in this context yet.** The original twelve hosted cells all failed before providers. Corrected candidate `0a9c5a75164cd0b02115ff12273aadb6a86c9464` and historical `00a761b967f9f73b0f45069c0ba828fae277a76e` now pass actual launcher prepare-and-verify admission. Six paired outcomes remain pending; no non-regression claim is established. + +The [recovery manifest](2026-10-02-native-completion-master-recovery-manifest.json) freezes exactly six cells per variant, twelve expected provider turns, one attempt per cell, unchanged profiles/models/auth/permissions/defaults and existing deadlines/budgets. Exactly the original eleven native implementation/unit-test paths differ. Both variants carry the same strict named finishing-call oracle and closed terminal-name projection. The original anonymous OpenCode evidence is preserved and cannot retrospectively prove this identity. + +Both exact-source admissions pass with zero providers: candidate 132 / historical 127 selected TypeScript assertions, 127 Node calibrations, one Rust normalization calibration, E2E typecheck, manifest checks, six-cell discovery and actual public receipt verification. Each has 257 explicitly skipped assertions, not coverage. Fresh Rust builds record the actually selected binary digest. Hosted cells instead verify the trusted same-run archive/source and selected restored executable; they explicitly record Rust calibration not executed in that cell. + +The sandbox-denied loopback attempts and stale anchor-expectation attempts are retained under the [admission folder](2026-10-02-native-completion-master-recovery-admission). Sources or assertions were not relaxed to pass either failure. The final public launcher identity calibration covers both prepare and verify subprocesses while excluding all credential/ambient override names. + +## Preserved original setup cohort + # Native completion guidance: master-context comparison **TL;DR:** Task performance is unmeasured: all twelve hosted cells stopped before provider execution. Zero model turns ran; zero task pairs were graded. Six pairs remain pending recovery. Local exact-source admission passed, but hosted checkout metadata did not satisfy the gate. This comparison measures native finish/block descriptions on unchanged master defaults; it does not measure removal of native fixed prompts or the reduced manuals in #14948. diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-admission/baseline.json b/doc/plans/2026-10-02-native-completion-master-recovery-admission/baseline.json new file mode 100644 index 00000000000..aeae8674b18 --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-admission/baseline.json @@ -0,0 +1,312 @@ +{ + "schema": "paperclip.native-completion-preflight.v1", + "sourceSha": "00a761b967f9f73b0f45069c0ba828fae277a76e", + "sourceFingerprint": "07d39fd1b0e9fc15455193e0c77550b5da21fe67d4cdf4abb1599ecea786799b", + "fixtureFingerprint": "6568138fe6b1abb1d44a48f9678140c2e63adb2edb09e07047492250cf5e40d6", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "variant": "historical", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "archiveSha": "9138f570c341c251a5727c32d6615ce238bc8e03", + "sourceErrors": [], + "immutable": true, + "layering": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "full_ancestry", + "shallow": true, + "sha": "00a761b967f9f73b0f45069c0ba828fae277a76e", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": null + }, + "hydration": { + "present": false, + "verified": false, + "archiveSha256": null, + "checksumSha256": null + }, + "worktreeEntries": [], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "ea106c05e935318121fbf1d0d1b9f310ab2ec67b1f99ea49ae65aa76f2957151", + "measuredAt": "2026-10-03T04:33:40.047Z", + "providerCalls": 0, + "live": "not_run", + "expectedCells": 6, + "expectedTurns": 6, + "maximumAttemptsPerCell": 1, + "setup": { + "passed": true, + "sdkExitCode": 0, + "runnerTypeScriptExitCode": 0, + "runnerdExitCode": 0, + "evidence": [ + { + "file": "setup-sdk.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + }, + { + "file": "setup-runner-typescript.txt", + "sha256": "0b439b2bd5b65d06783d0a61482269e16d6af4e6e5e3f98ec20138bb1982c5ed" + }, + { + "file": "setup-runnerd.txt", + "sha256": "a96c2934e450fa2ba5e8ba158398b3e1474389c4ecb630c0f44c53caebf592cd" + } + ], + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "8e4d54800e3b4b829621b12a9d8ce434c89445492abe7584d4465922e3ff5cea", + "binaryBytes": 84271504, + "archiveMemberBytes": null, + "sourceSha": "00a761b967f9f73b0f45069c0ba828fae277a76e", + "sourceFingerprint": "07d39fd1b0e9fc15455193e0c77550b5da21fe67d4cdf4abb1599ecea786799b", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "buildOutputFingerprint": "b131fe332da31bb4c1b3b8fbaf70f98e96f70de12c426dd087b47037f649b7d4", + "runnerdSha256": "8e4d54800e3b4b829621b12a9d8ce434c89445492abe7584d4465922e3ff5cea", + "runnerdSelectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "runnerdSourceSha": "00a761b967f9f73b0f45069c0ba828fae277a76e", + "runnerdSourceFingerprint": "07d39fd1b0e9fc15455193e0c77550b5da21fe67d4cdf4abb1599ecea786799b" + }, + "gates": [ + { + "id": "NC-schemas", + "name": "Variant-native completion schemas", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/contracts/completion-result.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "distinguishes user-facing answer content", + "passed": true + }, + { + "name": "propagates answer and wait descriptions", + "passed": true + }, + { + "name": "requires a reason code for verification that was not run", + "passed": true + } + ], + "total": 13, + "passedTests": 13, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-schemas.json", + "sha256": "8c81de6a87c7573e97f5ba45aa01a5bec23406bf962fd9c908f8d055f4330f67" + } + }, + { + "id": "NC-tools", + "name": "Native tool delivery and retained provider catalogs", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/drivers/codex/codex-app-server-driver.lifecycle.test.ts", + "passed": true + }, + { + "file": "src/drivers/runner-tool-bridge.test.ts", + "passed": true + }, + { + "file": "src/drivers/opencode/mcp-bridge.test.ts", + "passed": true + }, + { + "file": "src/live/runnerd-codex-transport.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "runner semantic MCP bridge binds to loopback", + "passed": true + }, + { + "name": "OpenCode MCP bridge binds to loopback", + "passed": true + }, + { + "name": "preserves stock Codex instructions on task recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on prepared recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on direct recovery", + "passed": true + }, + { + "name": "includes ACPX terminal tools in the authenticated bridge catalog", + "passed": true + } + ], + "total": 222, + "passedTests": 7, + "failedTests": 0, + "pendingTests": 215, + "evidence": { + "file": "NC-tools.json", + "sha256": "513a932e15034385d7748f3f48a2e2466cf34643f3dcc2c473eeef1b70db6a39" + } + }, + { + "id": "NC-resume", + "name": "Variant-native fingerprint and checkpoint refresh", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "server/src/services/native-runtime/native-session-resume.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "refreshes retained 'completion descriptions'", + "passed": true + }, + { + "name": "refreshes retained 'task-bound human-input description'", + "passed": true + } + ], + "total": 44, + "passedTests": 2, + "failedTests": 0, + "pendingTests": 42, + "evidence": { + "file": "NC-resume.json", + "sha256": "9d9ee9309475fd77057dca6031165e7d8eae176693305934da8dfe542665bad2" + } + }, + { + "id": "NC-eval", + "name": "Independent native oracle, defaults, admission and attempt policy", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "tests/runner-e2e/native-completion-scoring.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-defaults.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-admission.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/automatic-retry.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-evidence.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-flow-logs.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "allows toolchain paths and excludes every present or future credential", + "passed": true + }, + { + "name": "retains credential-free prerequisites inside the exact campaign root", + "passed": true + }, + { + "name": "retains the first transient_infrastructure failure without a second attempt", + "passed": true + }, + { + "name": "retains the first provider_variance failure without a second attempt", + "passed": true + } + ], + "total": 105, + "passedTests": 105, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-eval.json", + "sha256": "42a0d667f54a7997696ac219632d74802d98ac262c51248f41df1e2d04245ecd" + } + }, + { + "id": "NC-node", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-node.txt", + "sha256": "5bf50ce428a8a8492daa1336af60abca71e3d85a4fc772915a4f27719a26e4a9" + } + }, + { + "id": "NC-typecheck", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-typecheck.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + } + }, + { + "id": "NC-rust-carrier", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-rust-carrier.txt", + "sha256": "3effafc1fffad102ec1e2031539c389aa9e3bc912bc64c0e21439b6115ea414e" + } + }, + { + "id": "NC-discovery", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-discovery.txt", + "sha256": "3097ab7907c3b49d43920fab50d0b180b70cb6895c0bd1a58f72fd740fec6567" + } + }, + { + "id": "NC-manifest", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-manifest.txt", + "sha256": "7f5d044db6a7813857e842095c5e3aa749a98570a96ff9939483de9452322502" + } + } + ], + "passed": true +} diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-admission/candidate.json b/doc/plans/2026-10-02-native-completion-master-recovery-admission/candidate.json new file mode 100644 index 00000000000..7de44aaad92 --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-admission/candidate.json @@ -0,0 +1,336 @@ +{ + "schema": "paperclip.native-completion-preflight.v1", + "sourceSha": "0a9c5a75164cd0b02115ff12273aadb6a86c9464", + "sourceFingerprint": "977d88b7318da67938c3e1e4cff7b10481b739b34cd6a13e12b1be5f833506b2", + "fixtureFingerprint": "6568138fe6b1abb1d44a48f9678140c2e63adb2edb09e07047492250cf5e40d6", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "variant": "candidate", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "archiveSha": "9138f570c341c251a5727c32d6615ce238bc8e03", + "sourceErrors": [], + "immutable": true, + "layering": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "full_ancestry", + "shallow": true, + "sha": "0a9c5a75164cd0b02115ff12273aadb6a86c9464", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": null + }, + "hydration": { + "present": false, + "verified": false, + "archiveSha256": null, + "checksumSha256": null + }, + "worktreeEntries": [], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "981430add4b66e9cb9f28927b129946ba671537a4d59583630563897af1ce59d", + "measuredAt": "2026-10-03T04:33:38.097Z", + "providerCalls": 0, + "live": "not_run", + "expectedCells": 6, + "expectedTurns": 6, + "maximumAttemptsPerCell": 1, + "setup": { + "passed": true, + "sdkExitCode": 0, + "runnerTypeScriptExitCode": 0, + "runnerdExitCode": 0, + "evidence": [ + { + "file": "setup-sdk.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + }, + { + "file": "setup-runner-typescript.txt", + "sha256": "c07e1077714b3088cee155d3ef3b6a568e15d04dc22dd8c4b1bc32bf4e5b4a95" + }, + { + "file": "setup-runnerd.txt", + "sha256": "659c4f2771e0c8565e747be57fe33e93669e7dd89e3c91adeac8990d2cb8f233" + } + ], + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "de11e42c462a3617ab8b8276f3f0e86e7f4a479e2d437eff2b3893e06dc39d0e", + "binaryBytes": 84259808, + "archiveMemberBytes": null, + "sourceSha": "0a9c5a75164cd0b02115ff12273aadb6a86c9464", + "sourceFingerprint": "977d88b7318da67938c3e1e4cff7b10481b739b34cd6a13e12b1be5f833506b2", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "buildOutputFingerprint": "d12136329d21552f65694670a8d7e41a4b821460c5e986dd7e1a8a471cfca451", + "runnerdSha256": "de11e42c462a3617ab8b8276f3f0e86e7f4a479e2d437eff2b3893e06dc39d0e", + "runnerdSelectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "runnerdSourceSha": "0a9c5a75164cd0b02115ff12273aadb6a86c9464", + "runnerdSourceFingerprint": "977d88b7318da67938c3e1e4cff7b10481b739b34cd6a13e12b1be5f833506b2" + }, + "gates": [ + { + "id": "NC-schemas", + "name": "Variant-native completion schemas", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/contracts/completion-result.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "distinguishes user-facing answer content", + "passed": true + }, + { + "name": "propagates answer and wait descriptions", + "passed": true + }, + { + "name": "requires a reason code for verification that was not run", + "passed": true + } + ], + "total": 13, + "passedTests": 13, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-schemas.json", + "sha256": "18e4ecbb2b3f87a003a40d8277be91ad894e53ad1f40495da12a8fd0b4100344" + } + }, + { + "id": "NC-tools", + "name": "Native tool delivery and retained provider catalogs", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/drivers/codex/codex-app-server-driver.lifecycle.test.ts", + "passed": true + }, + { + "file": "src/drivers/runner-tool-bridge.test.ts", + "passed": true + }, + { + "file": "src/drivers/opencode/mcp-bridge.test.ts", + "passed": true + }, + { + "file": "src/live/runnerd-codex-transport.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "runner semantic MCP bridge binds to loopback", + "passed": true + }, + { + "name": "OpenCode MCP bridge binds to loopback", + "passed": true + }, + { + "name": "preserves stock Codex instructions on task recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on prepared recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on direct recovery", + "passed": true + }, + { + "name": "includes ACPX terminal tools in the authenticated bridge catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native codex tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native opencode tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native claude_managed tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native aws_agentcore tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native acpx tool catalog", + "passed": true + } + ], + "total": 226, + "passedTests": 11, + "failedTests": 0, + "pendingTests": 215, + "evidence": { + "file": "NC-tools.json", + "sha256": "b606cd191fba836ce717a0c3100dbc3671bc72696d6cba757573b15875cff726" + } + }, + { + "id": "NC-resume", + "name": "Variant-native fingerprint and checkpoint refresh", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "server/src/services/native-runtime/native-session-resume.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "refreshes retained 'completion descriptions'", + "passed": true + }, + { + "name": "refreshes retained 'task-bound human-input description'", + "passed": true + }, + { + "name": "native completion tool guidance", + "passed": true + } + ], + "total": 45, + "passedTests": 3, + "failedTests": 0, + "pendingTests": 42, + "evidence": { + "file": "NC-resume.json", + "sha256": "02beffd1153172ba84ba7b68cfa18a4c5dd26f4e3ac1d4ee124bd54135d3c587" + } + }, + { + "id": "NC-eval", + "name": "Independent native oracle, defaults, admission and attempt policy", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "tests/runner-e2e/native-completion-scoring.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-defaults.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-admission.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/automatic-retry.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-evidence.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-flow-logs.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "allows toolchain paths and excludes every present or future credential", + "passed": true + }, + { + "name": "retains credential-free prerequisites inside the exact campaign root", + "passed": true + }, + { + "name": "retains the first transient_infrastructure failure without a second attempt", + "passed": true + }, + { + "name": "retains the first provider_variance failure without a second attempt", + "passed": true + } + ], + "total": 105, + "passedTests": 105, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-eval.json", + "sha256": "d6b90886fd1d0d3a9b105857cbf4e639c57f29a5f89376719f42ae2b356c0aaf" + } + }, + { + "id": "NC-node", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-node.txt", + "sha256": "e03c8e97cd351a9187c38b58dfacf8080d02ce7053a0c44624356941eef0bb06" + } + }, + { + "id": "NC-typecheck", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-typecheck.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + } + }, + { + "id": "NC-rust-carrier", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-rust-carrier.txt", + "sha256": "4b8f5043f996c48f8bc3e69735ef012f71c45981083d84bda3948c0ecacc5153" + } + }, + { + "id": "NC-discovery", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-discovery.txt", + "sha256": "3097ab7907c3b49d43920fab50d0b180b70cb6895c0bd1a58f72fd740fec6567" + } + }, + { + "id": "NC-manifest", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-manifest.txt", + "sha256": "7f5d044db6a7813857e842095c5e3aa749a98570a96ff9939483de9452322502" + } + } + ], + "passed": true +} diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-baseline.json b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-baseline.json new file mode 100644 index 00000000000..437c36d4e0c --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-baseline.json @@ -0,0 +1,312 @@ +{ + "schema": "paperclip.native-completion-preflight.v1", + "sourceSha": "e68a7edc569f73ba1eb26b5a47f6a22ce536babc", + "sourceFingerprint": "4bd265ca52d4e5a4a6080c292590419b2afe4a1280859d91507e1f6b8e0efaff", + "fixtureFingerprint": "bb270d66349eefc7b2d4abdcf6f058ebf2528e54b6daeda2037624e7dd575a23", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "variant": "historical", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "archiveSha": "9138f570c341c251a5727c32d6615ce238bc8e03", + "sourceErrors": [], + "immutable": true, + "layering": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "full_ancestry", + "shallow": true, + "sha": "e68a7edc569f73ba1eb26b5a47f6a22ce536babc", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": null + }, + "hydration": { + "present": false, + "verified": false, + "archiveSha256": null, + "checksumSha256": null + }, + "worktreeEntries": [], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "81c9bdacab93314f8cf2d6def32d9ae75ba519331cd6541079dc3a46b04d21a9", + "measuredAt": "2026-10-03T03:50:53.717Z", + "providerCalls": 0, + "live": "not_run", + "expectedCells": 6, + "expectedTurns": 6, + "maximumAttemptsPerCell": 1, + "setup": { + "passed": true, + "sdkExitCode": 0, + "runnerTypeScriptExitCode": 0, + "runnerdExitCode": 0, + "evidence": [ + { + "file": "setup-sdk.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + }, + { + "file": "setup-runner-typescript.txt", + "sha256": "0b439b2bd5b65d06783d0a61482269e16d6af4e6e5e3f98ec20138bb1982c5ed" + }, + { + "file": "setup-runnerd.txt", + "sha256": "02d95094500958649a21043a498c73e088159169f908f7ce1c927e90480fcae2" + } + ], + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "8e4d54800e3b4b829621b12a9d8ce434c89445492abe7584d4465922e3ff5cea", + "binaryBytes": 84271504, + "archiveMemberBytes": null, + "sourceSha": "e68a7edc569f73ba1eb26b5a47f6a22ce536babc", + "sourceFingerprint": "4bd265ca52d4e5a4a6080c292590419b2afe4a1280859d91507e1f6b8e0efaff", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "buildOutputFingerprint": "b131fe332da31bb4c1b3b8fbaf70f98e96f70de12c426dd087b47037f649b7d4", + "runnerdSha256": "8e4d54800e3b4b829621b12a9d8ce434c89445492abe7584d4465922e3ff5cea", + "runnerdSelectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "runnerdSourceSha": "e68a7edc569f73ba1eb26b5a47f6a22ce536babc", + "runnerdSourceFingerprint": "4bd265ca52d4e5a4a6080c292590419b2afe4a1280859d91507e1f6b8e0efaff" + }, + "gates": [ + { + "id": "NC-schemas", + "name": "Variant-native completion schemas", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/contracts/completion-result.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "distinguishes user-facing answer content", + "passed": true + }, + { + "name": "propagates answer and wait descriptions", + "passed": true + }, + { + "name": "requires a reason code for verification that was not run", + "passed": true + } + ], + "total": 13, + "passedTests": 13, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-schemas.json", + "sha256": "c3e291b0ef4edf10f531820a4c36a88199d85083f7ea08df50bfd1f1c146afe4" + } + }, + { + "id": "NC-tools", + "name": "Native tool delivery and retained provider catalogs", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/drivers/codex/codex-app-server-driver.lifecycle.test.ts", + "passed": true + }, + { + "file": "src/drivers/runner-tool-bridge.test.ts", + "passed": true + }, + { + "file": "src/drivers/opencode/mcp-bridge.test.ts", + "passed": true + }, + { + "file": "src/live/runnerd-codex-transport.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "runner semantic MCP bridge binds to loopback", + "passed": true + }, + { + "name": "OpenCode MCP bridge binds to loopback", + "passed": true + }, + { + "name": "preserves stock Codex instructions on task recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on prepared recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on direct recovery", + "passed": true + }, + { + "name": "includes ACPX terminal tools in the authenticated bridge catalog", + "passed": true + } + ], + "total": 222, + "passedTests": 7, + "failedTests": 0, + "pendingTests": 215, + "evidence": { + "file": "NC-tools.json", + "sha256": "d1e882914075f6afaaaf9958c64131c30864eef4634510f64aa76986d9d49739" + } + }, + { + "id": "NC-resume", + "name": "Variant-native fingerprint and checkpoint refresh", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "server/src/services/native-runtime/native-session-resume.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "refreshes retained 'completion descriptions'", + "passed": true + }, + { + "name": "refreshes retained 'task-bound human-input description'", + "passed": true + } + ], + "total": 44, + "passedTests": 2, + "failedTests": 0, + "pendingTests": 42, + "evidence": { + "file": "NC-resume.json", + "sha256": "5d00901330b17f8ddf0be35c27c1a8c82e69c337abd3ace9dadb4a4174825954" + } + }, + { + "id": "NC-eval", + "name": "Independent native oracle, defaults, admission and attempt policy", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "tests/runner-e2e/native-completion-scoring.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-defaults.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-admission.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/automatic-retry.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-evidence.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-flow-logs.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "allows toolchain paths and excludes every present or future credential", + "passed": true + }, + { + "name": "retains credential-free prerequisites inside the exact campaign root", + "passed": true + }, + { + "name": "retains the first transient_infrastructure failure without a second attempt", + "passed": true + }, + { + "name": "retains the first provider_variance failure without a second attempt", + "passed": true + } + ], + "total": 105, + "passedTests": 105, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-eval.json", + "sha256": "de290cd6636e5ed9e0dad6c700454f2028ab3dda9f1322357f7266d70a7e08ca" + } + }, + { + "id": "NC-node", + "passed": false, + "exitCode": 1, + "evidence": { + "file": "NC-node.txt", + "sha256": "a605ba3d994a87b4af0946331d9ec7bff75b1d38217529cccfe7d71bab3a3d19" + } + }, + { + "id": "NC-typecheck", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-typecheck.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + } + }, + { + "id": "NC-rust-carrier", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-rust-carrier.txt", + "sha256": "0eb23e2b890e8bdb5dda4a5a7f65358b1e38847839a4bf0d8fa49a28f4237f91" + } + }, + { + "id": "NC-discovery", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-discovery.txt", + "sha256": "3097ab7907c3b49d43920fab50d0b180b70cb6895c0bd1a58f72fd740fec6567" + } + }, + { + "id": "NC-manifest", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-manifest.txt", + "sha256": "7f5d044db6a7813857e842095c5e3aa749a98570a96ff9939483de9452322502" + } + } + ], + "passed": false +} diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-candidate.json b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-candidate.json new file mode 100644 index 00000000000..75c2da53ef9 --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-anchor-candidate.json @@ -0,0 +1,336 @@ +{ + "schema": "paperclip.native-completion-preflight.v1", + "sourceSha": "b603222b8496c234b64c0642ca3ca4d5d8990318", + "sourceFingerprint": "4c804ad3b02100bbd365740df239448d9a258648e1cc6b15b5cccde15717638c", + "fixtureFingerprint": "bb270d66349eefc7b2d4abdcf6f058ebf2528e54b6daeda2037624e7dd575a23", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "variant": "candidate", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "archiveSha": "9138f570c341c251a5727c32d6615ce238bc8e03", + "sourceErrors": [], + "immutable": true, + "layering": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "full_ancestry", + "shallow": true, + "sha": "b603222b8496c234b64c0642ca3ca4d5d8990318", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": null + }, + "hydration": { + "present": false, + "verified": false, + "archiveSha256": null, + "checksumSha256": null + }, + "worktreeEntries": [], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "ddb4ede8b03b820210f28cce91448f238b09bdd990552ac8a86bd00914f03faa", + "measuredAt": "2026-10-03T03:50:54.046Z", + "providerCalls": 0, + "live": "not_run", + "expectedCells": 6, + "expectedTurns": 6, + "maximumAttemptsPerCell": 1, + "setup": { + "passed": true, + "sdkExitCode": 0, + "runnerTypeScriptExitCode": 0, + "runnerdExitCode": 0, + "evidence": [ + { + "file": "setup-sdk.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + }, + { + "file": "setup-runner-typescript.txt", + "sha256": "c07e1077714b3088cee155d3ef3b6a568e15d04dc22dd8c4b1bc32bf4e5b4a95" + }, + { + "file": "setup-runnerd.txt", + "sha256": "b1a40608beb64df47ebb9cd0d2ccb9e4b79a837e8f4c2accf74a35e704fee05f" + } + ], + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "de11e42c462a3617ab8b8276f3f0e86e7f4a479e2d437eff2b3893e06dc39d0e", + "binaryBytes": 84259808, + "archiveMemberBytes": null, + "sourceSha": "b603222b8496c234b64c0642ca3ca4d5d8990318", + "sourceFingerprint": "4c804ad3b02100bbd365740df239448d9a258648e1cc6b15b5cccde15717638c", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "buildOutputFingerprint": "d12136329d21552f65694670a8d7e41a4b821460c5e986dd7e1a8a471cfca451", + "runnerdSha256": "de11e42c462a3617ab8b8276f3f0e86e7f4a479e2d437eff2b3893e06dc39d0e", + "runnerdSelectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "runnerdSourceSha": "b603222b8496c234b64c0642ca3ca4d5d8990318", + "runnerdSourceFingerprint": "4c804ad3b02100bbd365740df239448d9a258648e1cc6b15b5cccde15717638c" + }, + "gates": [ + { + "id": "NC-schemas", + "name": "Variant-native completion schemas", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/contracts/completion-result.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "distinguishes user-facing answer content", + "passed": true + }, + { + "name": "propagates answer and wait descriptions", + "passed": true + }, + { + "name": "requires a reason code for verification that was not run", + "passed": true + } + ], + "total": 13, + "passedTests": 13, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-schemas.json", + "sha256": "0b94b16c11911c76c8f00131ae5baaf317d8028afe0af4e61a0b958be8732961" + } + }, + { + "id": "NC-tools", + "name": "Native tool delivery and retained provider catalogs", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/drivers/codex/codex-app-server-driver.lifecycle.test.ts", + "passed": true + }, + { + "file": "src/drivers/runner-tool-bridge.test.ts", + "passed": true + }, + { + "file": "src/drivers/opencode/mcp-bridge.test.ts", + "passed": true + }, + { + "file": "src/live/runnerd-codex-transport.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "runner semantic MCP bridge binds to loopback", + "passed": true + }, + { + "name": "OpenCode MCP bridge binds to loopback", + "passed": true + }, + { + "name": "preserves stock Codex instructions on task recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on prepared recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on direct recovery", + "passed": true + }, + { + "name": "includes ACPX terminal tools in the authenticated bridge catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native codex tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native opencode tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native claude_managed tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native aws_agentcore tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native acpx tool catalog", + "passed": true + } + ], + "total": 226, + "passedTests": 11, + "failedTests": 0, + "pendingTests": 215, + "evidence": { + "file": "NC-tools.json", + "sha256": "cba2682051cf5d59e94cdf877a12f48419eed129efd59478bc4e7141e97e4b48" + } + }, + { + "id": "NC-resume", + "name": "Variant-native fingerprint and checkpoint refresh", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "server/src/services/native-runtime/native-session-resume.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "refreshes retained 'completion descriptions'", + "passed": true + }, + { + "name": "refreshes retained 'task-bound human-input description'", + "passed": true + }, + { + "name": "native completion tool guidance", + "passed": true + } + ], + "total": 45, + "passedTests": 3, + "failedTests": 0, + "pendingTests": 42, + "evidence": { + "file": "NC-resume.json", + "sha256": "bc489f310ff0f112588353ca18aad2a54cf358983748192427c18ec13846d5b5" + } + }, + { + "id": "NC-eval", + "name": "Independent native oracle, defaults, admission and attempt policy", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "tests/runner-e2e/native-completion-scoring.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-defaults.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-admission.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/automatic-retry.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-evidence.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-flow-logs.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "allows toolchain paths and excludes every present or future credential", + "passed": true + }, + { + "name": "retains credential-free prerequisites inside the exact campaign root", + "passed": true + }, + { + "name": "retains the first transient_infrastructure failure without a second attempt", + "passed": true + }, + { + "name": "retains the first provider_variance failure without a second attempt", + "passed": true + } + ], + "total": 105, + "passedTests": 105, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-eval.json", + "sha256": "91ebe00f1009cc4e75457a238cda556104b4e4481c60b5247b1c2321e8a7cb02" + } + }, + { + "id": "NC-node", + "passed": false, + "exitCode": 1, + "evidence": { + "file": "NC-node.txt", + "sha256": "1415d1b2fc4235811c6b8081f8fb47b7b6b368749d9568f10fa3e3c208210387" + } + }, + { + "id": "NC-typecheck", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-typecheck.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + } + }, + { + "id": "NC-rust-carrier", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-rust-carrier.txt", + "sha256": "9193315e28c2dac2b754fb4ef37ad9522dc1909363849a29a7b9a26a672782de" + } + }, + { + "id": "NC-discovery", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-discovery.txt", + "sha256": "3097ab7907c3b49d43920fab50d0b180b70cb6895c0bd1a58f72fd740fec6567" + } + }, + { + "id": "NC-manifest", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-manifest.txt", + "sha256": "7f5d044db6a7813857e842095c5e3aa749a98570a96ff9939483de9452322502" + } + } + ], + "passed": false +} diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-baseline.json b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-baseline.json new file mode 100644 index 00000000000..34f6f8c44da --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-baseline.json @@ -0,0 +1,312 @@ +{ + "schema": "paperclip.native-completion-preflight.v1", + "sourceSha": "5cafc2cbc48f86ad95dfbfb7dc71a9a588238083", + "sourceFingerprint": "95dadaf66ec0f30da2f2822b2ff7392615740814a1ab485183a6eab4000e80d8", + "fixtureFingerprint": "81145a336d588a289ffed719fd3f71ecdfad07011a20bbb9e40ed97d434f4ed9", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "variant": "historical", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "archiveSha": "9138f570c341c251a5727c32d6615ce238bc8e03", + "sourceErrors": [], + "immutable": true, + "layering": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "full_ancestry", + "shallow": true, + "sha": "5cafc2cbc48f86ad95dfbfb7dc71a9a588238083", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": null + }, + "hydration": { + "present": false, + "verified": false, + "archiveSha256": null, + "checksumSha256": null + }, + "worktreeEntries": [], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "e70513f47284d0ed22f284e421a289de010aef98843cf0d5ab31e543c9add4b8", + "measuredAt": "2026-10-03T03:42:55.570Z", + "providerCalls": 0, + "live": "not_run", + "expectedCells": 6, + "expectedTurns": 6, + "maximumAttemptsPerCell": 1, + "setup": { + "passed": true, + "sdkExitCode": 0, + "runnerTypeScriptExitCode": 0, + "runnerdExitCode": 0, + "evidence": [ + { + "file": "setup-sdk.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + }, + { + "file": "setup-runner-typescript.txt", + "sha256": "0b439b2bd5b65d06783d0a61482269e16d6af4e6e5e3f98ec20138bb1982c5ed" + }, + { + "file": "setup-runnerd.txt", + "sha256": "767b1a5ba8531791dfa4c0850f175c4a28f4cf654238ded7f41537509119c9f6" + } + ], + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "8e4d54800e3b4b829621b12a9d8ce434c89445492abe7584d4465922e3ff5cea", + "binaryBytes": 84271504, + "archiveMemberBytes": null, + "sourceSha": "5cafc2cbc48f86ad95dfbfb7dc71a9a588238083", + "sourceFingerprint": "95dadaf66ec0f30da2f2822b2ff7392615740814a1ab485183a6eab4000e80d8", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "buildOutputFingerprint": "b131fe332da31bb4c1b3b8fbaf70f98e96f70de12c426dd087b47037f649b7d4", + "runnerdSha256": "8e4d54800e3b4b829621b12a9d8ce434c89445492abe7584d4465922e3ff5cea", + "runnerdSelectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "runnerdSourceSha": "5cafc2cbc48f86ad95dfbfb7dc71a9a588238083", + "runnerdSourceFingerprint": "95dadaf66ec0f30da2f2822b2ff7392615740814a1ab485183a6eab4000e80d8" + }, + "gates": [ + { + "id": "NC-schemas", + "name": "Variant-native completion schemas", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/contracts/completion-result.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "distinguishes user-facing answer content", + "passed": true + }, + { + "name": "propagates answer and wait descriptions", + "passed": true + }, + { + "name": "requires a reason code for verification that was not run", + "passed": true + } + ], + "total": 13, + "passedTests": 13, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-schemas.json", + "sha256": "5740137d1a8fa4c083bcb0c28717b1b90676d0d95f73b0b105fb47b1af8c4a3d" + } + }, + { + "id": "NC-tools", + "name": "Native tool delivery and retained provider catalogs", + "passed": false, + "exitCode": 1, + "files": [ + { + "file": "src/drivers/codex/codex-app-server-driver.lifecycle.test.ts", + "passed": true + }, + { + "file": "src/drivers/runner-tool-bridge.test.ts", + "passed": false + }, + { + "file": "src/drivers/opencode/mcp-bridge.test.ts", + "passed": false + }, + { + "file": "src/live/runnerd-codex-transport.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "runner semantic MCP bridge binds to loopback", + "passed": false + }, + { + "name": "OpenCode MCP bridge binds to loopback", + "passed": false + }, + { + "name": "preserves stock Codex instructions on task recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on prepared recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on direct recovery", + "passed": true + }, + { + "name": "includes ACPX terminal tools in the authenticated bridge catalog", + "passed": true + } + ], + "total": 222, + "passedTests": 5, + "failedTests": 2, + "pendingTests": 215, + "evidence": { + "file": "NC-tools.json", + "sha256": "b7db5ab171e744522acd8bb7f22026cd2cd5f0f6a08af1ec36de22925c443265" + } + }, + { + "id": "NC-resume", + "name": "Variant-native fingerprint and checkpoint refresh", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "server/src/services/native-runtime/native-session-resume.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "refreshes retained 'completion descriptions'", + "passed": true + }, + { + "name": "refreshes retained 'task-bound human-input description'", + "passed": true + } + ], + "total": 44, + "passedTests": 2, + "failedTests": 0, + "pendingTests": 42, + "evidence": { + "file": "NC-resume.json", + "sha256": "269ae8297d698fc98af246dafaad28077c16297ae46a4ab5d4bd901a4b444e44" + } + }, + { + "id": "NC-eval", + "name": "Independent native oracle, defaults, admission and attempt policy", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "tests/runner-e2e/native-completion-scoring.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-defaults.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-admission.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/automatic-retry.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-evidence.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-flow-logs.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "allows toolchain paths and excludes every present or future credential", + "passed": true + }, + { + "name": "retains credential-free prerequisites inside the exact campaign root", + "passed": true + }, + { + "name": "retains the first transient_infrastructure failure without a second attempt", + "passed": true + }, + { + "name": "retains the first provider_variance failure without a second attempt", + "passed": true + } + ], + "total": 104, + "passedTests": 104, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-eval.json", + "sha256": "2ce7cad255715aaf4bea36a9b15f6647ddd0c3a2595b63f6cbfc15ec36931718" + } + }, + { + "id": "NC-node", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-node.txt", + "sha256": "213a7bcc65f7a702405a5a0597accaba50ca6a432209045d2553962afa733e4e" + } + }, + { + "id": "NC-typecheck", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-typecheck.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + } + }, + { + "id": "NC-rust-carrier", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-rust-carrier.txt", + "sha256": "3de3aab3e33b5bb45c1e9a0737e69d75f9249ed593d8dae6d78637df599a5d8e" + } + }, + { + "id": "NC-discovery", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-discovery.txt", + "sha256": "3097ab7907c3b49d43920fab50d0b180b70cb6895c0bd1a58f72fd740fec6567" + } + }, + { + "id": "NC-manifest", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-manifest.txt", + "sha256": "7f5d044db6a7813857e842095c5e3aa749a98570a96ff9939483de9452322502" + } + } + ], + "passed": false +} diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-candidate.json b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-candidate.json new file mode 100644 index 00000000000..15d75f45efc --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-admission/failed-sandbox-candidate.json @@ -0,0 +1,336 @@ +{ + "schema": "paperclip.native-completion-preflight.v1", + "sourceSha": "76a247643869efda6f2adec300c0f46c88e79cae", + "sourceFingerprint": "4719959c5a354cf99a0f99bd603135e5aba2b2bdf69999d992e6ad06ef3e58a8", + "fixtureFingerprint": "81145a336d588a289ffed719fd3f71ecdfad07011a20bbb9e40ed97d434f4ed9", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "variant": "candidate", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "archiveSha": "9138f570c341c251a5727c32d6615ce238bc8e03", + "sourceErrors": [], + "immutable": true, + "layering": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "full_ancestry", + "shallow": true, + "sha": "76a247643869efda6f2adec300c0f46c88e79cae", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": null + }, + "hydration": { + "present": false, + "verified": false, + "archiveSha256": null, + "checksumSha256": null + }, + "worktreeEntries": [], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "801c1ab2bc532b330d88a674ef1b1856c4ba7f23b02922feb2d0d99e47063f26", + "measuredAt": "2026-10-03T03:42:55.587Z", + "providerCalls": 0, + "live": "not_run", + "expectedCells": 6, + "expectedTurns": 6, + "maximumAttemptsPerCell": 1, + "setup": { + "passed": true, + "sdkExitCode": 0, + "runnerTypeScriptExitCode": 0, + "runnerdExitCode": 0, + "evidence": [ + { + "file": "setup-sdk.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + }, + { + "file": "setup-runner-typescript.txt", + "sha256": "c07e1077714b3088cee155d3ef3b6a568e15d04dc22dd8c4b1bc32bf4e5b4a95" + }, + { + "file": "setup-runnerd.txt", + "sha256": "73997f6a1ede8abd3a236173e34ccaae621cba937cf4733f9393172443409b6e" + } + ], + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "de11e42c462a3617ab8b8276f3f0e86e7f4a479e2d437eff2b3893e06dc39d0e", + "binaryBytes": 84259808, + "archiveMemberBytes": null, + "sourceSha": "76a247643869efda6f2adec300c0f46c88e79cae", + "sourceFingerprint": "4719959c5a354cf99a0f99bd603135e5aba2b2bdf69999d992e6ad06ef3e58a8", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "buildOutputFingerprint": "d12136329d21552f65694670a8d7e41a4b821460c5e986dd7e1a8a471cfca451", + "runnerdSha256": "de11e42c462a3617ab8b8276f3f0e86e7f4a479e2d437eff2b3893e06dc39d0e", + "runnerdSelectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "runnerdSourceSha": "76a247643869efda6f2adec300c0f46c88e79cae", + "runnerdSourceFingerprint": "4719959c5a354cf99a0f99bd603135e5aba2b2bdf69999d992e6ad06ef3e58a8" + }, + "gates": [ + { + "id": "NC-schemas", + "name": "Variant-native completion schemas", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "src/contracts/completion-result.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "distinguishes user-facing answer content", + "passed": true + }, + { + "name": "propagates answer and wait descriptions", + "passed": true + }, + { + "name": "requires a reason code for verification that was not run", + "passed": true + } + ], + "total": 13, + "passedTests": 13, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-schemas.json", + "sha256": "12cf714c82c44cdecbec0d5c86a6b4667aa209811e70a5f20c0c29c7e0b7f33c" + } + }, + { + "id": "NC-tools", + "name": "Native tool delivery and retained provider catalogs", + "passed": false, + "exitCode": 1, + "files": [ + { + "file": "src/drivers/codex/codex-app-server-driver.lifecycle.test.ts", + "passed": true + }, + { + "file": "src/drivers/runner-tool-bridge.test.ts", + "passed": false + }, + { + "file": "src/drivers/opencode/mcp-bridge.test.ts", + "passed": false + }, + { + "file": "src/live/runnerd-codex-transport.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "runner semantic MCP bridge binds to loopback", + "passed": false + }, + { + "name": "OpenCode MCP bridge binds to loopback", + "passed": false + }, + { + "name": "preserves stock Codex instructions on task recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on prepared recovery", + "passed": true + }, + { + "name": "preserves stock Codex instructions on direct recovery", + "passed": true + }, + { + "name": "includes ACPX terminal tools in the authenticated bridge catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native codex tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native opencode tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native claude_managed tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native aws_agentcore tool catalog", + "passed": true + }, + { + "name": "preserves answer and internal wait descriptions in the serialized native acpx tool catalog", + "passed": true + } + ], + "total": 226, + "passedTests": 9, + "failedTests": 2, + "pendingTests": 215, + "evidence": { + "file": "NC-tools.json", + "sha256": "1626b969efcb20658bbc35379327c40b567e46cde4c3ef61d72ee1b9edc630e0" + } + }, + { + "id": "NC-resume", + "name": "Variant-native fingerprint and checkpoint refresh", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "server/src/services/native-runtime/native-session-resume.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "refreshes retained 'completion descriptions'", + "passed": true + }, + { + "name": "refreshes retained 'task-bound human-input description'", + "passed": true + }, + { + "name": "native completion tool guidance", + "passed": true + } + ], + "total": 45, + "passedTests": 3, + "failedTests": 0, + "pendingTests": 42, + "evidence": { + "file": "NC-resume.json", + "sha256": "07801e7c62b5d9a415112832b4ff4a9fc84767961c72bc9b6e5b37525c85ed85" + } + }, + { + "id": "NC-eval", + "name": "Independent native oracle, defaults, admission and attempt policy", + "passed": true, + "exitCode": 0, + "files": [ + { + "file": "tests/runner-e2e/native-completion-scoring.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-defaults.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/native-completion-admission.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/automatic-retry.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-evidence.test.ts", + "passed": true + }, + { + "file": "tests/runner-e2e/context-integrity-flow-logs.test.ts", + "passed": true + } + ], + "requirements": [ + { + "name": "allows toolchain paths and excludes every present or future credential", + "passed": true + }, + { + "name": "retains credential-free prerequisites inside the exact campaign root", + "passed": true + }, + { + "name": "retains the first transient_infrastructure failure without a second attempt", + "passed": true + }, + { + "name": "retains the first provider_variance failure without a second attempt", + "passed": true + } + ], + "total": 104, + "passedTests": 104, + "failedTests": 0, + "pendingTests": 0, + "evidence": { + "file": "NC-eval.json", + "sha256": "86ce7470df140b9c2d6a6fb9e77817a3841924b2c5d7eea61ec6509bc8908d66" + } + }, + { + "id": "NC-node", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-node.txt", + "sha256": "0711e7a8af3e145ae6621624d3db36c7d7effa142dbe2e5e4acd1c5cc1d47b28" + } + }, + { + "id": "NC-typecheck", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-typecheck.txt", + "sha256": "01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b" + } + }, + { + "id": "NC-rust-carrier", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-rust-carrier.txt", + "sha256": "6b58a4e153d12a5936cfbca164b324a549c86bc4e307c5b23e4ba28b372a82c2" + } + }, + { + "id": "NC-discovery", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-discovery.txt", + "sha256": "3097ab7907c3b49d43920fab50d0b180b70cb6895c0bd1a58f72fd740fec6567" + } + }, + { + "id": "NC-manifest", + "passed": true, + "exitCode": 0, + "evidence": { + "file": "NC-manifest.txt", + "sha256": "7f5d044db6a7813857e842095c5e3aa749a98570a96ff9939483de9452322502" + } + } + ], + "passed": false +} diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-admission/hosted-source-reproduction.json b/doc/plans/2026-10-02-native-completion-master-recovery-admission/hosted-source-reproduction.json new file mode 100644 index 00000000000..a0600b4c4ca --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-admission/hosted-source-reproduction.json @@ -0,0 +1,75 @@ +{ + "schema": "paperclip.native-completion-metadata-reproduction.v1", + "providerCalls": 0, + "setupChildren": 0, + "qualification": "metadata_only_fixture_not_full_gate_admission", + "root": "/private/tmp/paperclip-native-real-shallow-RCnC9H", + "receipts": [ + { + "variant": "candidate", + "checkout": "/private/tmp/paperclip-native-real-shallow-RCnC9H/candidate", + "sha": "0931c6d0b58499639687b188e6080a658bac3efc", + "actualAdmittedParent": "e18c2cf9e96a4d31acb6d03ce918dfe223107d3f", + "productionAndFixedContextDigests": 22, + "wrongVariantRejected": true, + "layering": true, + "immutable": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "shallow_immediate_parent_anchor", + "shallow": true, + "sha": "0931c6d0b58499639687b188e6080a658bac3efc", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": "e18c2cf9e96a4d31acb6d03ce918dfe223107d3f" + }, + "hydration": { + "present": true, + "verified": true, + "archiveSha256": "3e91968477b4ab0b79ed0a2e6c83d22a7a18c5963fbfb20291e362678b3e3f03", + "checksumSha256": "4b18feed8f40b2d70bf40538357b42471f72f1f30a70e23edf87413577e943a7" + }, + "worktreeEntries": [ + "?? runner-e2e-build/runner-e2e-build-bundle.tar.gz", + "?? runner-e2e-build/runner-e2e-build-bundle.tar.gz.sha256" + ], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "75e15912fa5ec70d48a5c947db8acc5321544bce31339875c9e212a59be182cb" + }, + { + "variant": "historical", + "checkout": "/private/tmp/paperclip-native-real-shallow-RCnC9H/historical", + "sha": "b9b1d8a66a05d0a25401d040a48e67aad32a7e25", + "actualAdmittedParent": "459455acb11a012a97ad1b4afb77a1dc024a88bc", + "productionAndFixedContextDigests": 22, + "wrongVariantRejected": true, + "layering": true, + "immutable": true, + "sourceMetadata": { + "schema": "paperclip.native-completion-source-metadata.v1", + "lineage": { + "strategy": "shallow_immediate_parent_anchor", + "shallow": true, + "sha": "b9b1d8a66a05d0a25401d040a48e67aad32a7e25", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "anchor": "459455acb11a012a97ad1b4afb77a1dc024a88bc" + }, + "hydration": { + "present": true, + "verified": true, + "archiveSha256": "2b322686b0299c748194d7e4deb54c12f5b8074fb13dd38ad6135406f2311c05", + "checksumSha256": "302a7c4e1dfa45b13cc43f8b1e6ee1a961f94298e32cef08f0fa8adef576e21f" + }, + "worktreeEntries": [ + "?? runner-e2e-build/runner-e2e-build-bundle.tar.gz", + "?? runner-e2e-build/runner-e2e-build-bundle.tar.gz.sha256" + ], + "errors": [] + }, + "sourceMetadataErrors": [], + "sourceMetadataFingerprint": "5a1f3a301092ee7c4be8c3c9491c5f4963e5c0a84cd101110070f65ec4056604" + } + ] +} diff --git a/doc/plans/2026-10-02-native-completion-master-recovery-manifest.json b/doc/plans/2026-10-02-native-completion-master-recovery-manifest.json new file mode 100644 index 00000000000..059a10353a3 --- /dev/null +++ b/doc/plans/2026-10-02-native-completion-master-recovery-manifest.json @@ -0,0 +1,218 @@ +{ + "schema": "paperclip.native-completion-master-comparison.v2", + "baseSha": "59c07ede72dc08b8aba149a01cc11e0b7a204621", + "archiveSha": "9138f570c341c251a5727c32d6615ce238bc8e03", + "admission": { + "candidate": { + "sourceSha": "0a9c5a75164cd0b02115ff12273aadb6a86c9464", + "variant": "candidate", + "sourceFingerprint": "977d88b7318da67938c3e1e4cff7b10481b739b34cd6a13e12b1be5f833506b2", + "fixtureFingerprint": "6568138fe6b1abb1d44a48f9678140c2e63adb2edb09e07047492250cf5e40d6", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "passed": true, + "providerCalls": 0, + "receiptSha256": "03c7d12d0511ef0a658f0f768887f5524b85a86b56f759f1275c6098299136ff", + "publicEntrypoint": "prepareNativeCompletionPreflight -> verifyNativeCompletionPreflight", + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "de11e42c462a3617ab8b8276f3f0e86e7f4a479e2d437eff2b3893e06dc39d0e", + "binaryBytes": 84259808, + "archiveMemberBytes": null, + "sourceSha": "0a9c5a75164cd0b02115ff12273aadb6a86c9464", + "sourceFingerprint": "977d88b7318da67938c3e1e4cff7b10481b739b34cd6a13e12b1be5f833506b2", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "selectedTypeScriptAssertions": 132, + "explicitlySkippedTypeScriptAssertions": 257, + "nodeCalibrationTests": 127, + "rustCalibrationTests": 1 + }, + "baseline": { + "sourceSha": "00a761b967f9f73b0f45069c0ba828fae277a76e", + "variant": "historical", + "sourceFingerprint": "07d39fd1b0e9fc15455193e0c77550b5da21fe67d4cdf4abb1599ecea786799b", + "fixtureFingerprint": "6568138fe6b1abb1d44a48f9678140c2e63adb2edb09e07047492250cf5e40d6", + "manifestFingerprint": "b92f3ab7d561188bc09214433b264f2e7810e8ccdc237f68caa82e1aecc1811e", + "passed": true, + "providerCalls": 0, + "receiptSha256": "1903c8ef958db2f0f1e704363fe53d00696412e245742817f41ea6ecb31ce6b6", + "publicEntrypoint": "prepareNativeCompletionPreflight -> verifyNativeCompletionPreflight", + "runnerdProvenance": { + "schema": "paperclip.native-completion-runnerd-provenance.v1", + "mode": "fresh_local_build", + "passed": true, + "selectedPath": "packages/paperclip-runner/runner/target/debug/paperclip-runnerd", + "binarySha256": "8e4d54800e3b4b829621b12a9d8ce434c89445492abe7584d4465922e3ff5cea", + "binaryBytes": 84271504, + "archiveMemberBytes": null, + "sourceSha": "00a761b967f9f73b0f45069c0ba828fae277a76e", + "sourceFingerprint": "07d39fd1b0e9fc15455193e0c77550b5da21fe67d4cdf4abb1599ecea786799b", + "archiveSha256": null, + "workflowRunId": null, + "workflowRunAttempt": null, + "artifactName": null, + "errors": [] + }, + "selectedTypeScriptAssertions": 127, + "explicitlySkippedTypeScriptAssertions": 257, + "nodeCalibrationTests": 127, + "rustCalibrationTests": 1 + } + }, + "targetBranches": { + "candidate": "codex/native-completion-master-qualified-candidate", + "baseline": "codex/native-completion-master-qualified-baseline" + }, + "workflow": ".github/workflows/runner-full-stack-e2e.yml", + "workflowDispatchRef": "master", + "maximumCells": 12, + "expectedProviderTurns": 12, + "maximumAttemptsPerCell": 1, + "automaticRetryPolicy": "single_attempt", + "budgetMonthlyCentsPerCompanyAndAgent": 1000, + "cellsPerVariant": [ + { + "executionId": "native-completion.runner-codex.local.assigned-skill-explicit-invocation", + "suiteDefinitionHash": "0e034f033ab85468bb0b72ef7f3da2de297e660ea916112ab8459ca4668a03cc", + "profileId": "runner-codex", + "provider": "codex", + "model": "gpt-5.6-sol", + "credential": "OPENAI_API_KEY", + "environment": "local", + "caseId": "assigned-skill-explicit-invocation", + "expectedRunCount": 1, + "automaticRetryPolicy": "single_attempt", + "deadlineMs": 720000, + "modelQualification": { + "source": "adapter_constant", + "qualificationId": "codex:qualified-model" + } + }, + { + "executionId": "native-completion.runner-codex.local.native-blocked-report", + "suiteDefinitionHash": "0e034f033ab85468bb0b72ef7f3da2de297e660ea916112ab8459ca4668a03cc", + "profileId": "runner-codex", + "provider": "codex", + "model": "gpt-5.6-sol", + "credential": "OPENAI_API_KEY", + "environment": "local", + "caseId": "native-blocked-report", + "expectedRunCount": 1, + "automaticRetryPolicy": "single_attempt", + "deadlineMs": 300000, + "modelQualification": { + "source": "adapter_constant", + "qualificationId": "codex:qualified-model" + } + }, + { + "executionId": "native-completion.runner-opencode.local.assigned-skill-explicit-invocation", + "suiteDefinitionHash": "0e034f033ab85468bb0b72ef7f3da2de297e660ea916112ab8459ca4668a03cc", + "profileId": "runner-opencode", + "provider": "opencode", + "model": "openrouter/deepseek/deepseek-v4-flash-0731", + "credential": "OPENROUTER_API_KEY", + "environment": "local", + "caseId": "assigned-skill-explicit-invocation", + "expectedRunCount": 1, + "automaticRetryPolicy": "single_attempt", + "deadlineMs": 720000, + "modelQualification": { + "source": "adapter_constant", + "qualificationId": "opencode:qualified-model" + } + }, + { + "executionId": "native-completion.runner-opencode.local.native-blocked-report", + "suiteDefinitionHash": "0e034f033ab85468bb0b72ef7f3da2de297e660ea916112ab8459ca4668a03cc", + "profileId": "runner-opencode", + "provider": "opencode", + "model": "openrouter/deepseek/deepseek-v4-flash-0731", + "credential": "OPENROUTER_API_KEY", + "environment": "local", + "caseId": "native-blocked-report", + "expectedRunCount": 1, + "automaticRetryPolicy": "single_attempt", + "deadlineMs": 300000, + "modelQualification": { + "source": "adapter_constant", + "qualificationId": "opencode:qualified-model" + } + }, + { + "executionId": "native-completion.runner-acpx-claude.local.assigned-skill-explicit-invocation", + "suiteDefinitionHash": "0e034f033ab85468bb0b72ef7f3da2de297e660ea916112ab8459ca4668a03cc", + "profileId": "runner-acpx-claude", + "provider": "acpx", + "model": "claude-sonnet-5", + "credential": "ANTHROPIC_API_KEY", + "environment": "local", + "caseId": "assigned-skill-explicit-invocation", + "expectedRunCount": 1, + "automaticRetryPolicy": "single_attempt", + "deadlineMs": 720000, + "modelQualification": { + "source": "qualified_runner_profile", + "qualificationId": "acpx:claude" + } + }, + { + "executionId": "native-completion.runner-acpx-claude.local.native-blocked-report", + "suiteDefinitionHash": "0e034f033ab85468bb0b72ef7f3da2de297e660ea916112ab8459ca4668a03cc", + "profileId": "runner-acpx-claude", + "provider": "acpx", + "model": "claude-sonnet-5", + "credential": "ANTHROPIC_API_KEY", + "environment": "local", + "caseId": "native-blocked-report", + "expectedRunCount": 1, + "automaticRetryPolicy": "single_attempt", + "deadlineMs": 300000, + "modelQualification": { + "source": "qualified_runner_profile", + "qualificationId": "acpx:claude" + } + } + ], + "fixtureIdentityVerified": true, + "fixedMasterContextVerified": true, + "onlyDeclaredElevenNativeSourcesDiffer": true, + "providerCallsSoFar": 0, + "liveStatus": "admitted_pending_dispatch", + "preservedInitialSetupCohort": { + "manifestSource": "7dbc3c4594e13aebd55f7c9bf605fd332ce37544", + "reportSource": "27653eb1a8f8ce839776d760f4563f672e5a706c", + "candidate": "e18c2cf9e96a4d31acb6d03ce918dfe223107d3f", + "baseline": "459455acb11a012a97ad1b4afb77a1dc024a88bc", + "workflowRuns": { + "candidate": { + "id": 37091690197, + "url": "https://github.com/paperclipai/paperclip/actions/runs/37091690197" + }, + "baseline": { + "id": 37091691408, + "url": "https://github.com/paperclipai/paperclip/actions/runs/37091691408" + } + }, + "cells": 12, + "providerCalls": 0, + "classification": "pre_provider_source_admission" + }, + "commonRepairs": { + "strictScorerSchema": "paperclip.native-completion-observation.v2", + "closedTerminalNameCarrierSha256": "132ef0ad3555d7cdb705946d879e5c8f5264da6656267cf0ed2bd6f348dfffca", + "publicIdentityNames": [ + "PAPERCLIP_RUNNER_E2E_SOURCE_SHA", + "GITHUB_RUN_ID", + "GITHUB_RUN_ATTEMPT" + ], + "hostedRustCalibration": "not_executed; trusted same-run source/archive/selected binary proof; exact local calibration and normal CI retained separately" + } +} diff --git a/doc/plans/2026-10-02-stock-harness-paperclip-checklist.md b/doc/plans/2026-10-02-stock-harness-paperclip-checklist.md index d333e76d03c..72a03f10f7e 100644 --- a/doc/plans/2026-10-02-stock-harness-paperclip-checklist.md +++ b/doc/plans/2026-10-02-stock-harness-paperclip-checklist.md @@ -499,3 +499,8 @@ results, remaining exceptions, and follow-ups here before checking it off. ### Native completion isolation: current-master qualification PR #14961 is now independent of #14948 and based on master. Candidate `e18c2cf9e96a4d31acb6d03ce918dfe223107d3f` and historical baseline `459455acb11a012a97ad1b4afb77a1dc024a88bc` pass exact-source native-only admission with identical fixture/context receipts. The original assigned-skill/document journey is preserved; corrected blocker grading and enforced single attempts cover six paired native cells. Live results remain pending. [Manifest and current report](2026-10-02-native-completion-master-qualification.md) retain setup failures and the earlier native report separately. #14948's behavioral hold and #15007's attribution limits remain unchanged. No merge is authorized for #14961 yet. + + +### Selected native qualification recovery — 2026-10-02 + +PR #14961 remains draft and unmerged. Corrected standalone candidate 0a9c5a751 / historical 00a761b96 preserve master 59 defaults/manual/shared/skills, differing only in the original eleven native description/refresh implementation and unit-test paths. Actual launcher prepare+verify admission passes on both. Six paired live outcomes are pending; the original twelve pre-provider failures and all local setup/calibration failures remain inspectable in the native report/recovery manifest. Closed terminal names and strict same-call grading are common to both variants; anonymous historical OpenCode evidence is not regraded. Hosted Rust calibration is explicitly not executed, with trusted same-run selected-binary proof; exact local Rust calibration and normal CI are separate evidence. No behavior rerolls or additional paid breadth are authorized.