From address-review
Parse a review document for actionable feedback items, work through them systematically, and track resolution progress. Use when the user says "address review", "address the review", "address @docs/reviews/...", "work through the review feedback", "resolve review items", "address the issues in @docs/...", or references addressing a review document by path.
How this skill is triggered — by the user, by Claude, or both
Slash command
/address-review:address-reviewThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Parse a review document, extract actionable items, and work through them systematically.
Parse a review document, extract actionable items, and work through them systematically.
The user may provide these options inline:
--skip 2,5,8)The user provides a path to a review document, typically as an @-reference or a file path argument. Common locations include docs/reviews/, docs/plans/reviews/, or any markdown file the user references.
Read the file. If the path does not exist, report that and stop.
Extract every actionable item from the review document. Items may appear as:
- [ ] Fix the error handling in auth.go- Fix the error handling in auth.go1. Fix the error handling in auth.goFor each extracted item, record:
| Field | Description |
|---|---|
| Number | Sequential index (1, 2, 3, ...) |
| Summary | One-line description of the item |
| Type | code-change, documentation, question, or style |
| File hints | Any file paths, function names, or line numbers mentioned |
| Context | The full original text from the review document |
Type classification rules:
When in doubt, classify as code-change.
Display a numbered summary of all extracted items:
## Review Items (N total)
| # | Type | Summary | File hints |
|---|------|---------|------------|
| 1 | code-change | Fix error handling in auth.go | auth.go:42 |
| 2 | documentation | Update README install section | README.md |
| 3 | question | Should we support Windows paths? | - |
| 4 | style | Rename helper to follow convention | utils/helper.go |
If --dry-run was specified: Stop here after displaying the table. Do not make any changes.
Ask the user to confirm the list. The user may:
Merge any --skip numbers from the command line with any skips the user specifies interactively.
Process each item in order (or user-specified order). For each item:
### Addressing item N of M: <summary>
Type: <type> | Files: <file hints>
Use the file hints from the review item to find the relevant code:
After addressing an item, report:
Item N: Resolved - <brief description of what was done>
If the item cannot be resolved (ambiguous, requires external input, or references code that does not exist):
Item N: Skipped - <reason>
If --commit-per-item was specified: Commit the changes for this item immediately before moving to the next. Use a conventional commit message referencing the review:
fix: <description of the change>
If --commit-per-item was NOT specified (the default behavior):
After all items are addressed, group the changes into the smallest logical commits that are appropriate. Each commit should represent a single coherent change:
For each commit group, generate a conventional commit message:
fix: address review feedback for <area>
Resolves items N, M from <review-file-name>.
After all items are processed, display a summary:
## Review Resolution Summary
**Source**: <review file path>
**Total items**: N
**Resolved**: X
**Skipped**: Y
| # | Type | Summary | Status | Commit |
|---|------|---------|--------|--------|
| 1 | code-change | Fix error handling in auth.go | Resolved | abc1234 |
| 2 | documentation | Update README install section | Resolved | def5678 |
| 3 | question | Should we support Windows paths? | Skipped (needs discussion) | - |
| 4 | style | Rename helper to follow convention | Resolved | abc1234 |
If any items were skipped, list them with explanations at the end:
### Items Needing Discussion
- **#3**: Should we support Windows paths? - This is an open design question.
Recommendation: <your recommendation based on codebase analysis>
This skill handles various markdown formats commonly used in review documents. The expected structure is flexible, but works best with:
# Review: <title>
## Section Name (optional)
- [ ] First actionable item
- [ ] Second actionable item with reference to `src/file.go:42`
- A bullet without checkbox is also treated as actionable
- [x] Already-completed items are skipped
### Subsection (optional)
1. Numbered items work too
2. Another numbered item
> Quoted text is treated as context, not as an actionable item
General prose paragraphs are treated as context for the items that follow them,
not as actionable items themselves.
Skipped patterns (not treated as actionable items):
- [x] Done item> This is contextGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub cboone/agent-harness-plugins --plugin address-review