Repository navigation
feat: share the PR skill across GitHub and GitLab with a cold read - #45
Merged
Merged
Conversation
The work-grown PR skill had matured past the GitHub-only one: it checks the diff with blind reader agents before writing the description, and it handles stacked PRs by targeting the parent branch. Bring that in as one shared skill with per-platform references for GitHub and public GitLab, leaving out all employer-specific tickets, hosts, and URLs. Add the two cold-reader agents the skill relies on, for both Claude Code and OpenCode.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
📝 Summary
The PR skill only covered GitHub and wrote descriptions from the author's memory of the session. It is now one shared
creating-pull-requestsskill for GitHub and GitLab, with platform syntax in per-platform reference files. Before writing a description, it has blind reader agents read the diff, so Key Decisions follow what a fresh reviewer finds significant and rationale goes where the diff alone is unclear.Key Decisions
🧭 Shared Skill with Platform References
The workflow, title format, description sections, and linking rules live in one
SKILL.md. Auto-link tokens, emoji syntax, length limits, stacked-PR flags, and the create command live inreferences/github.mdandreferences/gitlab.md. GitLab targets the publicgitlab.cominstance.👀 Cold Read by Blind Agents
The skill stages the branch diff in a scratch file and hands only its path to
cheapest-pr-cold-reader, which has just theReadtool and never sees commits, the draft, or session context. Items it marksunclear, or a diff too large for it, go tomid-tier-pr-cold-reader. The two agents share identical instructions but stay separate templates, following the one-template-per-agent pattern in.chezmoitemplates/. Their models follow the repo's existing tier mapping: Haiku and Sonnet on Claude Code, luna and terra on OpenCode.🥞 Stacked PRs Target Their Parent
A stacked PR now targets the branch directly below it in the stack on both platforms, so its diff shows only its own change. A bundled script infers the nearest parent branch from commit topology when the user does not name it. On GitHub, a closing keyword has no effect until the PR targets the default branch, and the reference says so.
🏷️ Per-Decision Emoji
## Key Decisionsstays plain, and each decision heading carries an emoji chosen for that decision instead of one fixed emoji for the whole section.🔄 Before & After