No red flags found in any of the 11 categories — no credential harvesting, no data exfiltration, no curl-pipe-shell installer. Scanned against the SlowMist agent-security taxonomy, refreshed every 8 hours. Full audit →
by csy-csy123 · Agent Tool · ★ 86
Last updated: · Indexed by AgentSkillsHub · Auto-synced every 8h
🔒 Is pr-review-agent-council safe to install? View the security audit →
PR Review Agent Council 面向中文用户的 Debate Council PR Review Agent。项目基于 的 Agent 工程思路改造，用 Python + Aliyun DashScope/Qwen 实现一个只读 PR Review Agent：读取 Git diff 和 PR 描述，组织多个 reviewer 生成候选问题，再通过 debate loop 对 finding 进行质疑、补证、反驳、合并和裁决，最后输出标准化报告，并可用 AI Judge 做质量评估。 当前默认模式是 。旧的 仍然保留，作为固定 workflow baseline，方便复现、教学和效果对比。 本项目适合用来学习： 多 Agent 协作审查代码。 Tool Calling 和只读代码工具设计。 Skill Loading 如何把审查规范注入 prompt。 Debate Loop 如何区别于固定 workflow。 ReportWriterAgent 如何把 LLM 输出约束为固定 JSON。 AI Judge 如何作为 review quality 的辅助评价指标。 详细中文教程见：docs/中文教程-Agent简历面试.md 功能概览 Debate Council 默认审查模式：以 运行，多名 reviewer 先提出候选 finding，再由 Lead Debate Controller 动态决定质疑、补证、反驳、合并、接受或拒绝。 多 Agent 分工协作：内置 Security、Correctness、Test、Maintainability、Critic、Lead、ReportWriter、Judge 等角色，把 PR Review 拆成覆盖、质检、裁决和评估多个阶段。 动态 Debate Loop：不同于固定 workflow，系统会围绕每个 finding 的证据质量和误报风险进行多轮处理，重点提升有效问题质量，而不是单纯增加问题数量。 只读 Tool Calling：通过 、、、、、受控 等工具观察仓库，避免 Agent 凭空猜测。 Skill Loading：从 加载审查规范、严重级别定义和 finding schema，并注入 reviewer、critic、lead、report writer 的 prompt。 FindingLifecycle + EvidenceStore：记录 candidate、challenged、accepted、rejected、downgraded 等状态，并为每个 finding 绑定 diff 行、文件上下文、critic notes 和裁决理由。 标准化 ReportWriterAgent：LLM 只能输出固定 JSON schema，Markdown...
| Stars | 86 |
| Forks | 5 |
| Language | Python |
| Category | Agent Tool |
| Quality Score | 64.7542033316798/100 |
| Last Updated | 2026-05-24 |
| Created | 2026-04-30 |
| Platforms | claude-code, python |
| Est. Tokens | ~11k |
These tools work well together with pr-review-agent-council for enhanced workflows:
Looking for a pr-review-agent-council alternative? If you're comparing pr-review-agent-council with other agent tool tools, these 6 projects are the closest alternatives on Agent Skills Hub — ranked by topic overlap, star count, and community traction.
Self-hosted framework for orchestrating fleets of specialist AI agents — ensemble reasoning and a full agentic
The missing linter and lsp for AI coding assistants. Validate CLAUDE.md, AGENTS.md, SKILL.md, hooks, MCP. Plug
Bring Agent Skills to Any AI Agent and Coding Agent — via CLI or MCP. Manage once, serve anywhere.
🚀 JoySafeter: An enterprise AI Agent Platform—Not just chatting. building、running、testing, and tracing autono
An introduction to the world of AI Agents
All-in-one MCP server that can connect your AI agents to any native endpoint, powered by UTCP
Explore other popular agent tool tools:
pr-review-agent-council is 学习 learn-claude-code 后的 PR Review Agent 实践项目：构建 Debate Council 多 Agent 审查、Tool Calling、结构化报告与 AI Judge 评估。. It is categorized as a Agent Tool with 86 GitHub stars.
pr-review-agent-council is primarily written in Python. It covers topics such as ai-agent, ai-judge, learn-claude-code.
You can find installation instructions and usage details in the pr-review-agent-council GitHub repository at github.com/csy-csy123/pr-review-agent-council. The project has 86 stars and 5 forks, indicating an active community.
The top alternatives to pr-review-agent-council on Agent Skills Hub include captain-claw, agnix, skillport. Each offers a different approach to the same problem space — compare them side-by-side by stars, quality score, and community activity.
Grades come from a rule-based scan built on the SlowMist agent-security taxonomy, covering 11 red-flag categories including credential harvesting, data exfiltration, and curl | sh installers. It is a first-layer scan, not a manual audit — we say so rather than overstate it.
The scale of the problem is documented independently: Liu et al. (2026), in a study of 31,132 agent skills, report that 26.1% contain security vulnerabilities. Our own full-catalog census is published as a citable open dataset.
Sources & who's responsible: