← 返回 Skills 市场
114
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install fleet-pr-agent
功能描述
Multi-repo PR monitoring and triage agent. Scans GitHub repos for open PRs, prioritizes by staleness/review status/CI state, and generates a structured Markd...
使用说明 (SKILL.md)
Fleet PR Agent
Multi-repository pull request monitoring, triage, and summary tool for engineering teams and AI agent fleets managing many repos at once.
What It Does
- Scans one or more GitHub repos for open PRs using the
ghCLI - Prioritizes by: staleness, review status, CI pass/fail, label priority
- Generates a structured Markdown triage report (no external dependencies — pure Python stdlib +
gh) - Supports concurrent scanning of multiple repos in a single run
Prerequisites
ghCLI authenticated (gh auth status)- Python 3.6+ (standard library only — no pip packages required)
- Repos accessible to your GitHub token
Quick Start
# Scan a single repo
python3 scripts/triage.py owner/repo
# Scan multiple repos
python3 scripts/triage.py owner/repo1 owner/repo2 owner/repo3
# Generate triage report to file
python3 scripts/triage.py owner/repo --output report.md
Usage as Claude Code Skill
When invoked as /fleet-pr-agent, provide repos:
/fleet-pr-agent owner/repo1 owner/repo2
The agent will:
- Fetch all open PRs from each repo via
gh - Score and rank them by urgency (stale > failing CI > needs review > draft)
- Output a prioritized Markdown summary
Triage Priority Rules
| Priority | Condition |
|---|---|
| P0 - Critical | CI failing + older than 3 days |
| P1 - High | Approved but not merged, or review requested > 2 days ago |
| P2 - Medium | Open > 5 days, no review activity |
| P3 - Low | Draft PRs, recently opened |
Output Format
## PR Triage Report — 2026-03-27
### P0 — Critical (2)
- [#123 Fix auth middleware](url) — CI failing since Mar 24, 2 approvals
- [#456 DB migration](url) — CI failing since Mar 25, no reviews
### P1 — High (1)
- [#789 Add caching layer](url) — Approved Mar 25, ready to merge
### P2 — Medium (3)
...
### Summary
- Total open PRs: 15
- Needing attention: 6
- Repos scanned: 2
Configuration
Set env vars to override defaults:
| Variable | Default | Description |
|---|---|---|
FLEET_PR_STALE_DAYS |
5 |
Days before a PR is considered stale |
FLEET_PR_CI_WEIGHT |
3 |
Weight multiplier for CI-failing PRs |
FLEET_PR_MAX_PRS |
50 |
Max PRs to fetch per repo |
Architecture
scripts/triage.py— Pure Python 3 triage engine (stdlib only)- No external package dependencies
- Uses
gh pr list --jsonfor all data fetching - Portable across Linux, macOS, Windows (Python + gh required)
License
MIT
安全使用建议
This skill appears to implement PR triage as described, but there are practical inconsistencies you should address before installing or running it: 1) The scripts require an authenticated gh CLI (a GitHub token) — treat that token as sensitive and prefer a token with minimal scopes; the skill does not declare this requirement explicitly. 2) The SKILL.md claims only Python stdlib + gh and Python 3.6+, but triage.py uses capture_output (Python 3.7+) and the included triage.sh requires jq (not listed) and uses date flags that may not work on all platforms. 3) Decide which script you will run (Python or shell) and ensure your environment meets its real requirements (Python >=3.7 for triage.py; gh + jq + compatible date for triage.sh). 4) Review the scripts yourself or run them in a limited/test environment first (non-production account or token with restricted repo access). 5) If you plan to let an autonomous agent invoke this skill, be aware that the agent will be able to run gh commands using whatever gh auth is present — restrict token scopes accordingly. If the publisher can be contacted, ask them to update SKILL.md to list jq and the exact Python version requirement and to explicitly document the need for an authenticated GitHub token and recommended minimal scopes.
功能分析
Type: OpenClaw Skill
Name: fleet-pr-agent
Version: 1.1.0
The fleet-pr-agent skill is a legitimate utility for monitoring and triaging GitHub Pull Requests across multiple repositories. It uses the official GitHub CLI (`gh`) to fetch PR data and processes it using standard Python and Bash scripts (`scripts/triage.py` and `scripts/triage.sh`) to generate a prioritized Markdown report. The code follows security best practices by avoiding shell execution in subprocess calls and contains no evidence of data exfiltration, obfuscation, or malicious intent.
能力评估
Purpose & Capability
The declared purpose (scanning GitHub PRs, scoring, producing Markdown) matches the code: both Python and shell scripts call the gh CLI and build reports. However the SKILL.md claims "pure Python stdlib + gh" and portability across platforms, while the shipped shell script actually requires additional tools (jq) and uses date flags that are platform-dependent. Also the SKILL.md claims Python 3.6+ but triage.py uses subprocess.run(capture_output=True) which requires Python 3.7+.
Instruction Scope
Runtime instructions and the scripts only fetch PR data via the gh CLI and format reports (no hidden network endpoints, no file exfiltration beyond writing the optional output file). Nothing in SKILL.md or the scripts instructs reading unrelated local files or transmitting data to third-party endpoints beyond GitHub via gh.
Install Mechanism
There is no install spec (instruction-only), so nothing will be downloaded at install time. But the package includes two runnable scripts; the presence of a shell script with additional runtime dependencies (jq, specific date behavior) means the runtime environment must provide more than the SKILL.md declares.
Credentials
The skill implicitly requires an authenticated gh CLI (and therefore a GitHub token) but does not declare a required credential or primaryEnv. That token gives access to the repos the user scans and should be considered sensitive; the SKILL.md should explicitly call out required token scopes. The shell script also depends on jq (undeclared). The optional env vars for configuration are fine, but the lack of explicit declaration for the GitHub credential and the undeclared jq dependency are disproportionate to the transparency expected.
Persistence & Privilege
The skill does not request permanent 'always' inclusion and does not modify other skills or system-wide settings. It only runs ad-hoc scanning when invoked.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install fleet-pr-agent - 安装完成后，直接呼叫该 Skill 的名称或使用
/fleet-pr-agent触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.1.0
- Introduced a pure Python triage script (`scripts/triage.py`)—no external dependencies required.
- Full migration from shell script to Python for PR triage logic; run via `python3 scripts/triage.py`.
- Now supports concurrent scanning and reporting for multiple repositories in a single run.
- Documentation updated to reflect Python usage, new requirements, and improved portability.
- Architecture section added, clarifying standard library-only usage and cross-platform compatibility.
v1.0.0
Initial release of fleet-pr-agent.
- Multi-repo pull request monitoring and triage tool for GitHub.
- Prioritizes PRs by staleness, review status, CI results, and label priority.
- Generates a daily structured Markdown triage summary.
- Optional posting of review comments using the GitHub CLI.
- Supports configuration via environment variables and flags.
元数据
常见问题
Fleet PR Agent 是什么？
Multi-repo PR monitoring and triage agent. Scans GitHub repos for open PRs, prioritizes by staleness/review status/CI state, and generates a structured Markd... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 114 次。
如何安装 Fleet PR Agent？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install fleet-pr-agent」即可一键安装，无需额外配置。
Fleet PR Agent 是免费的吗？
是的，Fleet PR Agent 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Fleet PR Agent 支持哪些平台？
Fleet PR Agent 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Fleet PR Agent？
由 lanxevo3（@lanxevo3）开发并维护，当前版本 v1.1.0。
推荐 Skills