jpmschweitzer/settled-reach
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(skills): add branch-type reviewer routing to review-pr

Browse Source 
Review-pr now selects reviewers based on branch type: server/client
get Hoshe+Tyre (code), copy gets Hoshe+Paula+Miri (narrative+lore),
visual gets Hoshe+Araminta (art direction), audio gets Hoshe+Ozzie
(player experience). Hoshe always present as QA baseline.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
jpmschweitzer
2026-02-12 18:15:55 +01:00
co-authored by Claude Opus 4.6
parent 1333926ac2
commit 941547082f
1 changed files with 127 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.claude/skills/review-pr/SKILL.md
+127 -33
View File
@@ -1,18 +1,18 @@
---
name: review-pr
description: >
Review a branch diff with dual agents before merge. Use when the user says
"review-pr", "review this PR", "review this branch", or invokes /review-pr.
Spawns Hoshe (code quality) and Tyre (architecture) in parallel to review
the diff against main. Reports approve/reject with inline comments.
Review a branch diff with team-appropriate agents before merge. Use when the
user says "review-pr", "review this PR", "review this branch", or invokes
/review-pr. Spawns reviewers matched to the branch type (code, copy, visual,
audio) in parallel. Reports approve/reject with inline comments.
user-invocable: true
allowed-tools: Bash, Read, Grep, Glob, Task
---
# PR Review Skill
Dual-agent review of a branch diff against main. Hoshe reviews code quality,
Tyre reviews architecture. Both must approve for a clean review.
Multi-agent review of a branch diff against main. Reviewer composition depends
on the branch type. All reviewers must approve for a clean review.
## Workflow
@@ -32,7 +32,21 @@ Fetch remote branches first:
git fetch --all
```
### 2. Generate the diff
### 2. Determine reviewer team
Map the branch name to a reviewer set. Use the branch prefix (before any `/`
or `-` suffix) to classify:
| Branch type | Branches | Reviewers |
|-------------|----------|-----------|
| **code** | `server`, `client`, `ci`, or unknown | Hoshe (code quality) + Tyre (architecture) |
| **copy** | `copy` | Hoshe (QA) + Paula (narrative depth) + Miri (world consistency) |
| **visual** | `visual` | Hoshe (QA) + Araminta (art direction) |
| **audio** | `audio` | Hoshe (QA) + Ozzie (player experience) |
If the branch name doesn't match any known type, default to **code** reviewers.
### 3. Generate the diff
```bash
git log --oneline main..<branch>
@@ -56,14 +70,17 @@ For large diffs (>1000 lines of source), provide **source files** rather than
raw diff to reviewers — cleaner context, better reviews. Read files with
`git show origin/<branch>:<path>` and include them in the prompt.
### 3. Spawn both reviewers in parallel
### 4. Spawn reviewers in parallel
Use the Task tool to spawn **two agents simultaneously** in a single message.
Use `model: sonnet` for both — sufficient for review, saves cost.
Use the Task tool to spawn **all reviewers simultaneously** in a single message.
Use `model: sonnet` for all — sufficient for review, saves cost.
**Agent 1 — Hoshe (Code Quality)**
- `subagent_type`: `hoshe`
- `model`: `sonnet`
---
#### Code reviews (`server`, `client`, `ci`)
**Hoshe (Code Quality)**
- `subagent_type`: `hoshe`, `model`: `sonnet`
- Prompt: Include source code and commit log. Ask Hoshe to review for:
- Correctness and bug risks
- Error handling gaps
@@ -71,11 +88,9 @@ Use `model: sonnet` for both — sufficient for review, saves cost.
- Code style and clarity
- Security concerns (OWASP top 10, injection risks)
- Performance issues
- Request structured verdict: APPROVE or REQUEST_CHANGES with file-specific comments
**Agent 2 — Tyre (Architecture)**
- `subagent_type`: `tyre`
- `model`: `sonnet`
**Tyre (Architecture)**
- `subagent_type`: `tyre`, `model`: `sonnet`
- Prompt: Include source code and commit log. Tell Tyre to read the relevant
`decisions/*.md` files first, then review for:
- Architectural consistency with project decisions
@@ -84,40 +99,119 @@ Use `model: sonnet` for both — sufficient for review, saves cost.
- Scalability implications
- Whether the change respects non-negotiable baselines (D-010, D-012)
- Tyre can read files directly from the branch using `git show origin/<branch>:<path>`
- Request structured verdict: APPROVE or REQUEST_CHANGES with file-specific comments
### 4. Present results
---
Format the combined review as a table per reviewer:
#### Copy reviews (`copy`)
**Hoshe (QA)**
- `subagent_type`: `hoshe`, `model`: `sonnet`
- Prompt: Include the changed files and commit log. Ask Hoshe to review for:
- Formatting consistency (markdown, file naming, frontmatter)
- Broken references or links
- Spelling and grammar
- File organization and structure
- Missing or orphaned files
**Paula (Narrative Depth)**
- `subagent_type`: `paula`, `model`: `sonnet`
- Prompt: Include the changed files and commit log. Tell Paula to read the
relevant `decisions/*.md` files first, then review for:
- Narrative quality and character voice consistency
- Whether dialogue and monologue feel authentic to the characters
- Consequences and stakes — do choices carry weight?
- Political and interpersonal depth
- Emotional resonance — does the text make you feel something?
**Miri (World Consistency)**
- `subagent_type`: `miri`, `model`: `sonnet`
- Prompt: Include the changed files and commit log. Tell Miri to read the
relevant `decisions/*.md` files first, then review for:
- Lore accuracy — do facts match established setting?
- Internal consistency across files
- IP originality — nothing should read as a copy from another franchise
- Faction, technology, and location details match the worldbuilding docs
- Setting serves gameplay mechanics (asymmetric information, perception)
---
#### Visual reviews (`visual`)
**Hoshe (QA)**
- `subagent_type`: `hoshe`, `model`: `sonnet`
- Prompt: Include the changed files and commit log. Ask Hoshe to review for:
- File format and naming conventions
- Asset organization and directory structure
- Missing or broken references in scene/resource files
- Import settings consistency
**Araminta (Art Direction)**
- `subagent_type`: `araminta`, `model`: `sonnet`
- Prompt: Include the changed files and commit log. Tell Araminta to read
the style guide and relevant design docs first, then review for:
- Visual consistency with the established style guide
- Color palette adherence
- UI pattern consistency (diegetic-first, clarity over beauty)
- Whether assets scale gracefully (boxes-with-labels to full-art)
- Mood and tone — sleek, advanced, subtle Commonwealth aesthetic
---
#### Audio reviews (`audio`)
**Hoshe (QA)**
- `subagent_type`: `hoshe`, `model`: `sonnet`
- Prompt: Include the changed files and commit log. Ask Hoshe to review for:
- File format and naming conventions
- Audio asset organization and directory structure
- Missing or broken references
- Import/bus configuration consistency
**Ozzie (Player Experience)**
- `subagent_type`: `ozzie`, `model`: `sonnet`
- Prompt: Include the changed files and commit log. Ask Ozzie to review for:
- Emotional impact — does the audio enhance the moment?
- Atmosphere and tone — does it feel like the Commonwealth?
- Player feedback clarity — can the player tell what just happened?
- Pacing — do sounds support or fight the gameplay rhythm?
- Memorable moments — will players remember these audio cues?
---
All reviewers: request structured verdict: APPROVE or REQUEST_CHANGES with
file-specific comments.
### 5. Present results
Format the combined review as a table per reviewer. Include one section per
reviewer that was spawned (2 for code/visual/audio, 3 for copy):
```
## Review: <branch> -> main
## Review: <branch> -> main (type: code|copy|visual|audio)
### Hoshe (Code Quality): [APPROVE | REQUEST_CHANGES]
### <Reviewer Name> (<Focus>): [APPROVE | REQUEST_CHANGES]
[Summary]
| # | File | Severity | Issue |
|---|------|----------|-------|
| 1 | path:line | critical/warning/suggestion | description |
### Tyre (Architecture): [APPROVE | REQUEST_CHANGES]
[Summary]
| # | File | Severity | Issue |
|---|------|----------|-------|
| 1 | path:line | critical/warning/suggestion | description |
### <Reviewer Name> (<Focus>): [APPROVE | REQUEST_CHANGES]
...
### Verdict: [APPROVED | CHANGES REQUESTED]
```
The overall verdict is APPROVED only if **both** reviewers approve.
The overall verdict is APPROVED only if **all** reviewers approve.
## Prompt template for reviewers
Use this structure when constructing the agent prompts (adapt as needed):
```
Review the following branch diff for merge into main.
Review the following {branch_type} branch diff for merge into main.
Branch: {branch}
Branch type: {branch_type} (code|copy|visual|audio)
Commits:
{commit_log}
@@ -126,7 +220,7 @@ Diff stats:
[Source files or diff here — exclude vendor/generated code]
Review focus: {focus_area}
Your review focus: {focus_area}
Respond with:
1. Verdict: APPROVE or REQUEST_CHANGES
@@ -138,7 +232,7 @@ Respond with:
If no issues found, say APPROVE with a brief positive summary.
```
## Posting results to Gitea
## 6. Posting results to Gitea
After presenting results to the user, post the review as a PR comment:
@@ -149,7 +243,7 @@ tea comment --login schweitz --repo jpmschweitzer/settled-reach <PR_NUMBER> "<re
Use a heredoc for multi-line review bodies:
```bash
tea comment --login schweitz --repo jpmschweitzer/settled-reach <PR_NUMBER> "$(cat <<'REVIEW'
## Dual-Agent Review: <branch> -> main
## Review: <branch> -> main
...review content...
REVIEW
)"
@@ -163,7 +257,7 @@ Note: `tea pr reject` does not work on your own PRs. Use `tea comment` instead.
tea comment --login schweitz --repo jpmschweitzer/settled-reach <PR_NUMBER> "$(cat /tmp/review.md)"
```
## Merging approved PRs
## 7. Merging approved PRs
`tea pr merge` fails (405) when branches have conflicts with main. Merge
locally instead: