Conversation
Publish one canonical skill source with deterministic Claude and Codex projections. Validate staged provenance and resolve runtime skills through each harness manifest.
Owner
Author
|
Don't like the implementation of this. Way too much complexity for very little use-cases, as of now |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Ticket #7 requires harness-native plugin packaging without splitting logical identity or marketplace distribution.
This PR moves directly authored skills into canonical source trees, generates committed Claude and Codex projections with protected overlays and provenance locks, and updates native manifests to select those projections.
It adds deterministic generate/check commands, staged-index pre-commit enforcement, CI validation, manifest-driven catalog resolution and snapshots, and reliable Claude OAuth forwarding for isolated eval runs.
Verification includes 82 CLI tests, projection tests, shell tests on Bash 5 and 3.2, Claude manifest validation, and 5/5 scoped evals on both Codex and Claude.
Closes #7