Prepare v0.14.0-beta.0 public repository - #16
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e5112f4f94
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| let reason = null; | ||
| if (found.run.status !== "completed") reason = `supervised run is not completed (status ${found.run.status})`; | ||
| else if (found.run.receipt?.status !== "finalized") reason = "supervised run receipt is not finalized"; | ||
| else if (!receiptValue || receiptValue.schemaVersion !== "supervised-receipt/v1-beta" || receiptValue.runId !== supervisedRunId || receiptValue.finalized !== true) reason = "supervised receipt is missing, incomplete, or incompatible"; |
There was a problem hiding this comment.
When a supervised run is linked, this condition accepts any receipt containing only the expected schema, matching run ID, and finalized: true; unlike workflow receipts, it never rebuilds the receipt or verifies its canonical event and evidence files. The fixture in tests/contracts/pilot-receipt-link.js consequently qualifies entirely hand-written run, receipt, and ownership JSON, so fabricated or corrupted local state can satisfy the full-reviewed-run gate without passing finalizeSupervisedRun. Require canonical receipt and evidence consistency before returning qualified.
AGENTS.md reference: AGENTS.md:L5-L5
Useful? React with 👍 / 👎.
| const eligible = !evidence || evidence.status === "qualified"; | ||
| const maintainerEvidence = found.record.participant.classification === "maintainer-dogfood"; |
There was a problem hiding this comment.
When a maintainer records a qualifying observation without --json, this new eligibility rule makes it eligible, but src/pilot/cli.js:42 still prints Independent evidence eligible: yes. That directly relabels maintainer technical evidence as independent evidence in the user-facing projection, despite the stored classification correctly distinguishing them. Update the CLI label to reflect the qualification classification or validation model.
AGENTS.md reference: AGENTS.md:L5-L5
Useful? React with 👍 / 👎.
| validationModel: "maintainer-technical-acceptance", | ||
| independentUserValidationRequired: false, | ||
| complete: inspected.invalid.length === 0 && gates.every((gate) => gate.status === "met"), |
There was a problem hiding this comment.
For every non-JSON cewp pilot status invocation, src/pilot/cli.js:52 still prints Independent participants: <n>/3, even though this change declares independent validation unnecessary and removes the participant gate. A technically complete pilot can therefore be displayed alongside an apparently unmet 0/3 requirement, making the CLI projection contradict Core status. Update the human-readable output to describe the maintainer technical acceptance model instead.
AGENTS.md reference: AGENTS.md:L5-L5
Useful? React with 👍 / 👎.
Summary
Validation
Release boundary
This PR prepares v0.14.0-beta.0. npm publication remains blocked until registry authentication is restored.