[JoyboyBrian]

The replay planner repeatedly validates identical declared input paths across thousands of packages. Reuse successful validation only within one plan_details invocation; keep every manifest read/hash, exact HEAD and changed-path check, consumer and selection cause. The production change is five added lines.

On the pinned 3,678-package checkout, the existing complete-plan benchmark reports a local shared-runner planning median of 1.203 s → 0.640 s (1.88x). All 40 before/after complete plan results match. These are warm, in-process macOS measurements on a busy host; raw samples and source hashes are retained, with no cold-cache or whole-CI speedup claim.

Validation:

• Nine focused modules, 92 tests per mode, pass normally and optimized on CPython 3.12.2 and 3.14.6; existing platform/tool skips are recorded.
• Six new tests cover shared consumers/causes/hashes, filesystem and HEAD/diff freshness, exact spelling and malformed inputs. Four predecessor/fault controls are rejected.
• Fresh read-only, Git-free export passes the six tests normally and optimized, with unchanged source hashes. Final committed bytes match the measured implementation.
• Selected PR6030 source-binding baseline passes; no scientific replay or frontier change.

The cache is scoped to the existing stable-checkout contract and retains O(U) distinct input names until the call ends; peak memory was not measured. Native Windows execution and hosted CI remain unrun. Since this touches the shared runner, the existing CI planner will select all registered packages. Package export/source-history hooks do not apply to this shared-code change.

Claim: #6077

Credit Douglas Colkitt for the runner, Ali Romman for the reused benchmark, and the retained path/sparse-input guard contributors. Prepared by JoyboyBrian with substantial OpenAI Codex assistance; Apache-2.0. No independent human review is claimed.

[JoyboyBrian]

CI follow-up for f4f73e14604fa8f738ae3e5caa2381551e20ec0d.

The plan failure reproduces on clean canonical main 721c03822c421b7f7fcbb2598993b947c4d88419: the two current-status navigation tests fail before this PR’s scientific replay. Existing PR #6050 (41ac3a47b91b4184e6c54af417bf576bfb429926) fixes both documentation mismatches; its exact two-file change passes all six tests locally, normally and with -O. Awaiting the shared upstream repair and a fresh CI run; this is not a successful scientific-replay claim.