Initial commit

This commit is contained in:
Zhongwei Li
2025-11-30 08:56:32 +08:00
commit 35aff94595
5 changed files with 196 additions and 0 deletions

View File

@@ -0,0 +1,15 @@
{
"name": "mcp-github",
"description": "GitHub repository, issue, and pull request management with PR review analysis",
"version": "1.1.5",
"author": {
"name": "shavakan",
"email": "cs.changwon.lee@gmail.com"
},
"commands": [
"./commands/pr-review-analyze.md"
],
"mcp": [
"./.mcp.json"
]
}

10
.mcp.json Normal file
View File

@@ -0,0 +1,10 @@
{
"github": {
"type": "stdio",
"command": "docker",
"args": ["run", "-i", "--rm", "-e", "GITHUB_PERSONAL_ACCESS_TOKEN", "ghcr.io/github/github-mcp-server"],
"env": {
"GITHUB_PERSONAL_ACCESS_TOKEN": "${GITHUB_PERSONAL_ACCESS_TOKEN}"
}
}
}

3
README.md Normal file
View File

@@ -0,0 +1,3 @@
# mcp-github
GitHub repository, issue, and pull request management with PR review analysis

View File

@@ -0,0 +1,119 @@
---
description: Analyze PR review comments and generate fix summary for another Claude instance
---
# PR Review Analysis
Fetch GitHub PR review comments, categorize by severity, generate actionable fix summary for another Claude Code instance.
## User Input
You MUST consider the user input in `$ARGUMENTS`: PR URL or number.
If empty, abort: "Usage: /pr-review-analyze <PR_URL or PR_NUMBER>"
## Severity Classification
**Blocking:** Security vulnerabilities, data corruption risks, breaking API changes, critical logic errors
**High Priority:** Performance bugs, race conditions, incomplete error handling, correctness issues
**Medium Priority:** Code quality improvements, refactoring opportunities, minor optimizations
**Low Priority:** Style suggestions, documentation improvements, code deduplication
## Execution
**Phase 1: Parse Input**
Parse PR identifier from $ARGUMENTS
**Gate:** Confirm PR identifier. Proceed to fetch data? (y/n)
**Phase 2: Fetch Data**
Fetch PR data (prefer GitHub MCP, fallback to gh CLI if unavailable):
- Extract review threads with: comments, file paths, line numbers, reviewer names, text
- For each comment, MUST capture: `path` (file path), `line` (specific line number), `body` (comment text)
- For each comment, get: `outdated` flag (code changed) and thread `isResolved` status (manually resolved)
- Store location data with each comment for Phase 5 reporting
- If both fail, abort: "Unable to fetch PR data. Install gh CLI or configure GitHub MCP"
**Phase 3: Report Outdated Bot Comments**
Report outdated bot comments (already fixed by code changes):
- Identify comments from Claude (username contains "claude") or Copilot (username "github-copilot")
- For bot comments where `outdated: true` AND `isResolved: false`, list them separately
- Report: "Found N outdated unresolved bot comments (already fixed by code changes but not manually resolved)"
- Note: GitHub API doesn't support auto-resolving threads programmatically
**Phase 4: Categorize**
Categorize active unresolved comments:
- Only categorize comments where `outdated: false` AND `isResolved: false`
- Skip outdated comments (already reported in Phase 3) and resolved comments
- Analyze content and context of active unresolved comments
- Apply severity definitions above based on actual impact
- Flag ambiguous comments as "Needs Severity Review"
**Gate:** Found N comments (X blocking, Y high priority). Generate summary? (y/n)
**Phase 5: Generate Summary**
Generate summary with blocking issues first, lower priority after
**Critical:** Every issue MUST include exact file location in `**file:line**` format:
- Extract `path` and `line` from comment metadata (captured in Phase 2)
- Format as `**{path}:{line}**` (e.g., `**auth.py:127**`)
- If line number unavailable, use `**{path}**` only
- Never omit location data - it's required for developers to locate issues
## Fix Summary Format
```markdown
## Context
PR: https://github.com/org/repo/pull/42
Branch: fix/auth-validation
## Outdated Comments (Already Fixed)
N bot comments are outdated (code changed) but not manually resolved. No action needed.
## Blocking Issues
1. **auth.py:127** - SQL injection vulnerability
- Problem: User input in f-string query
- Risk: Arbitrary SQL execution
- Fix: Use parameterized query with $1, $2 placeholders
2. **session.py:89** - API key logged in exception handler
- Problem: Exception traceback includes API key from config dict
- Risk: Credentials exposed in error logs
- Fix: Redact api_key before client initialization
## High Priority
1. **cache.py:203** - Redis KEYS blocks event loop [Copilot]
- Problem: KEYS is O(n) and blocks Redis
- Risk: Performance degradation on large keyspaces
- Fix: Replace `redis.keys(pattern)` with `redis.scan_iter(match=pattern)`
## Instructions
1. Fix all blocking issues
2. Run test suite to verify no regressions
3. Commit: "fix: address security and performance review findings"
4. Push to same branch
```
## Edge Cases
- **No review comments:** Output "No review comments found"
- **All comments outdated or resolved:** Output "No active unresolved comments requiring attention"
- **No blocking issues:** State explicitly in summary
- **Unclassifiable comments:** Separate "Needs Severity Review" section
- **Invalid PR:** Abort with tool error message
## Constraints
- Preserve exact technical terms from comments
- Don't fabricate fixes not mentioned by reviewers
- Keep each issue to 3-5 sentences max

49
plugin.lock.json Normal file
View File

@@ -0,0 +1,49 @@
{
"$schema": "internal://schemas/plugin.lock.v1.json",
"pluginId": "gh:Shavakan/claude-marketplace:mcps/github",
"normalized": {
"repo": null,
"ref": "refs/tags/v20251128.0",
"commit": "0466fa92727988ecf996f7f387326084a62f2e71",
"treeHash": "a2e686198dbdd6afdbf658d5ea40f4f74729d78af21691bd41901775981d34b9",
"generatedAt": "2025-11-28T10:12:48.299859Z",
"toolVersion": "publish_plugins.py@0.2.0"
},
"origin": {
"remote": "git@github.com:zhongweili/42plugin-data.git",
"branch": "master",
"commit": "aa1497ed0949fd50e99e70d6324a29c5b34f9390",
"repoRoot": "/Users/zhongweili/projects/openmind/42plugin-data"
},
"manifest": {
"name": "mcp-github",
"description": "GitHub repository, issue, and pull request management with PR review analysis",
"version": "1.1.5"
},
"content": {
"files": [
{
"path": ".mcp.json",
"sha256": "5f111950aba53f9e58b80bb10b36390b13a942ed1feafb020c1a251f7fe98940"
},
{
"path": "README.md",
"sha256": "fe80515d7634990837ecf6a4f36396971ad512231200791b4d8044236972a982"
},
{
"path": ".claude-plugin/plugin.json",
"sha256": "1d3fbb08f1762c167f1ff995b224a92a9c0f4df2ffddce0a75f8aa1a5888c04e"
},
{
"path": "commands/pr-review-analyze.md",
"sha256": "e5087e53adc7903444f2991d4f14f5f2a9579dd81b9c8156469c5dc4010c85a0"
}
],
"dirSha256": "a2e686198dbdd6afdbf658d5ea40f4f74729d78af21691bd41901775981d34b9"
},
"security": {
"scannedAt": null,
"scannerVersion": null,
"flags": []
}
}