From beads-workflow
Generate a self-contained code review prompt for a fresh Claude Code session. Gathers commits, changed files, acceptance criteria, and outputs a ready-to-paste prompt that avoids session bias.
How this skill is triggered — by the user, by Claude, or both
Slash command
/beads-workflow:review <issue-id><issue-id>idThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are generating a self-contained code review prompt. This prompt will be pasted into a **separate** Claude Code session so the review is free from implementation bias.
$idYou are generating a self-contained code review prompt. This prompt will be pasted into a separate Claude Code session so the review is free from implementation bias.
Follow these steps exactly.
Implementation is done. Add the ready-for-human label so the issue reflects its current state:
Run: bd update $id --add-label=ready-for-human
Run all of these commands and collect the output:
bd show $id — issue details, acceptance criteria, description.git branch --show-current — current branch name.git log main..HEAD --oneline — all commits on this branch since it diverged.git diff main...HEAD --stat — list of changed files with line counts.From the git diff stat and commit messages, build a brief summary of what each changed file does (one line per file). Do not paste the raw diff — just the summary.
Present the following inside a single fenced code block (triple backticks) so the user can copy it directly. Fill in every placeholder from the data you gathered.
Template:
I need you to review the implementation of beads issue {ID}: {TITLE}.
## Context
{Brief description of what was implemented and why, from the issue description}
## Branch
`{BRANCH_NAME}`
## Commits
{Each commit on one line: hash + message}
## Changed files
{Each file on one line with brief description of what changed}
## Acceptance criteria
{Quoted directly from the issue}
## Review checklist
1. Read the issue with `bd show {ID}` for full context
2. Review the diff: `git diff main...HEAD`
3. Verify the implementation matches the acceptance criteria
4. Check for:
- Correctness: Does the code do what the issue describes?
- Edge cases: Are boundary conditions handled?
- Security: Any injection, XSS, or auth bypass concerns?
- Performance: N+1 queries, missing indexes, unnecessary loops?
- Test coverage: Do the tests cover the acceptance criteria?
- Laravel conventions: Does the code follow project patterns?
5. Run `php artisan test --compact` to verify tests pass
6. Run `vendor/bin/pint --dirty --format agent` to check formatting
7. Report findings as:
- **Critical** — must fix before merge
- **Warnings** — should address
- **Suggestions** — nice-to-have improvements
## After review
- If changes are needed, create bug beads: `bd create --title="Review: {finding}" --type=bug --deps discovered-from:{ID}`
- If everything looks good, say so and note any minor suggestions
Show the filled-in prompt and tell the user: "Paste this into a fresh Claude Code session to get an unbiased review."
npx claudepluginhub nerdroid23/ai-marketplace --plugin beads-workflowGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.