feat(cli): add pickled build --verify-only - #64
Merged
Conversation
Prove a build's harness without running an agent: per build, run the preflight (untouched fixture must fail failToPass, pass passToPass) and the optional reference control (referenceSolution must apply and clear the verifier), reporting proven / unproven / broken. Exits non-zero on a broken harness. The proof is factored into a shared proveBuild that runBuildCell also uses, so cell behavior is unchanged. Also surface 'verifier proven' on normal build cells (proven was previously silent) and enrich the init build template with passToPass + referenceSolution.
|
🎉 This PR is included in version 0.38.0 🎉 The release is available on:
Your semantic-release bot 📦🚀 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Prove a build's harness before spending tokens.
pickled build --verify-onlyruns each build's preflight and reference control with no agent: per build it checks that the untouched fixture failsfailToPassand passespassToPass, and that a declaredreferenceSolutionapplies and clears the verifier. It prints one line per build (proven, unproven, or broken) and exits non-zero on a broken harness, so it works as a cheap deterministic gate before the real paid run.proveBuildthatrunBuildCellalso calls, so cell scoring is unchanged.verifier proven(previously only unproven and broken were shown).--outputwrites the proof JSON; flags that do not apply (--agent,--sample,--threshold, and the rest) are rejected rather than silently ignored.pickled initnow includespassToPassandreferenceSolutionin the build example.