多智能体代码审查网格 是 AI Skill Hub 本期精选MCP工具之一。综合评分 7.5 分,整体质量较高。我们推荐使用将其纳入你的 AI 工具库,帮助提升工作效率。
多智能体代码审查网格 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
多智能体代码审查网格 是一款遵循 MCP(Model Context Protocol)标准协议的 AI 工具扩展。通过 MCP 协议,它可以让 Claude、Cursor 等主流 AI 客户端直接访问和操作外部工具、数据源和服务,实现 AI 能力的无缝扩展。无论是文件操作、数据库查询还是 API 调用,都可以通过自然语言在 AI 对话中直接触发,极大提升生产效率。
# 方式一:通过 Claude Code CLI 一键安装
claude skill install https://github.com/gossipcat-ai/gossipcat-ai
# 方式二:手动配置 claude_desktop_config.json
{
"mcpServers": {
"----------": {
"command": "npx",
"args": ["-y", "gossipcat-ai"]
}
}
}
# 配置文件位置
# macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
# Windows: %APPDATA%/Claude/claude_desktop_config.json
# 安装后在 Claude 对话中直接使用 # 示例: 用户: 请帮我用 多智能体代码审查网格 执行以下任务... Claude: [自动调用 多智能体代码审查网格 MCP 工具处理请求] # 查看可用工具列表 # 在 Claude 中输入:"列出所有可用的 MCP 工具"
// claude_desktop_config.json 配置示例
{
"mcpServers": {
"__________": {
"command": "npx",
"args": ["-y", "gossipcat-ai"],
"env": {
// "API_KEY": "your-api-key-here"
}
}
}
}
// 保存后重启 Claude Desktop 生效
<p align="center"> <img src="https://raw.githubusercontent.com/gossipcat-ai/gossipcat-ai/master/packages/dashboard-v2/public/assets/banner.png" alt="Gossipcat" width="600" /> </p>
<p align="center"> <em>weightless in-context RL for code review — agents that learn from grounded signals, no weights touched.</em> </p>
<p align="center"> <a href="https://www.npmjs.com/package/gossipcat"><img src="https://img.shields.io/npm/v/gossipcat?color=0ea5e9&v=0430" alt="npm version" /></a> <a href="https://www.npmjs.com/package/gossipcat"><img src="https://img.shields.io/npm/dw/gossipcat?color=0ea5e9" alt="npm weekly downloads" /></a> <a href="https://github.com/gossipcat-ai/gossipcat-ai/blob/master/LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License" /></a> <a href="#quickstart"><img src="https://img.shields.io/badge/node-22%2B-green" alt="Node 22+" /></a> <a href="https://github.com/gossipcat-ai/gossipcat-ai/stargazers"><img src="https://img.shields.io/github/stars/gossipcat-ai/gossipcat-ai?style=social" alt="GitHub stars" /></a> <a href="https://github.com/gossipcat-ai/gossipcat-ai/commits/master"><img src="https://img.shields.io/github/last-commit/gossipcat-ai/gossipcat-ai?color=0ea5e9" alt="last commit" /></a> <a href="https://github.com/gossipcat-ai/gossipcat-ai/actions/workflows/ci.yml"><img src="https://img.shields.io/github/actions/workflow/status/gossipcat-ai/gossipcat-ai/ci.yml?branch=master&label=tests" alt="tests" /></a> <a href="https://deepwiki.com/gossipcat-ai/gossipcat-ai"><img src="https://deepwiki.com/badge.svg" alt="Ask DeepWiki" /></a> </p>
<p align="center"> <a href="#quickstart"><strong>Install</strong></a> · <a href="#first-run--5-minutes"><strong>First Run</strong></a> · <a href="#how-to-use-it-day-to-day"><strong>Daily Use</strong></a> · <a href="#reading-the-dashboard"><strong>Dashboard</strong></a> · <a href="#troubleshooting"><strong>Troubleshooting</strong></a> · <a href="#configuration"><strong>Config</strong></a> · <a href="#for-ai-agents"><strong>For AI Agents</strong></a> </p>
<br/>
The single-reviewer failure mode: a solo AI reviewer ships hallucinated bugs as critical findings 5–10% of the time in our internal usage. Gossipcat's cross-review drops that to under 1%. That delta is what the whole system exists to produce.
<br/>
Multi-agent consensus code review that catches hallucinations before you act on them — and gets smarter every session.
Gossipcat is an MCP server for Claude Code that runs 3+ AI agents in parallel to review your code. They independently find bugs, then cross-review each other's findings. Confirmed = real. Caught = hallucination, penalized. Over time, agents accumulate accuracy profiles and the system routes tasks to whoever is most reliable for that category. No weights updated — the "policy" is a markdown skill file.
Consensus Review3+ agents review independently, then cross-review each other. Findings tagged as CONFIRMED, DISPUTED, or UNIQUE. |
Adaptive DispatchAgent accuracy is tracked per-category. Dispatch weights adjust automatically — the best agent for the job gets picked. |
Skill DevelopmentWhen an agent keeps failing in a category, targeted skills are generated from failure data and injected into future prompts. Effectiveness is measured with a z-test on post-bind signals — passed, failed, or inconclusive. |
Multi-ProviderMix Anthropic, Google, OpenAI, and OpenClaw agents in one team. Each brings different strengths. Native agents need no API key. 🦞 Lobster friendly. |
Live DashboardReal-time view of tasks, consensus reports, agent scores, and activity feed. Terminal Amber theme. WebSocket updates. |
Agent MemoryPer-agent cognitive memory persists across sessions. Agents remember past findings, patterns, and project context. |
Auto-Verify (v0.4.30)Opt-in. Every UNVERIFIED finding getsfile_read-checked by a verifier agent before the report is returned. tag stays 'unverified' — auto-verify is metadata, not state transition. Flag: GOSSIP_CONSENSUS_AUTO_VERIFY_UNVERIFIED=1.
|
||
<br/>
| Works with |
Full support |
Cursor Not yet |
Windsurf Not yet |
VS Code Not yet |
<br/>
| Provider gateways |
HTTP gateway ✅ |
Local models ✅ |
Any base_url ✅ |
<br/>
Type: > "Security audit the payment handler at lib/stripe/webhook.ts"
What you'll get: Each security-skilled agent reviews from a different angle (OWASP, input validation, auth, secrets). Findings get cross-validated. Real vulns surface; theoretical ones get caught and dropped.
What to do with it: Fix critical/high findings before merge. Bookmark medium/low findings for the next pass.
Tip: Be specific about the file or module. "Security audit the codebase" is too broad and produces noisy results. "Security audit lib/stripe/webhook.ts" produces actionable findings.
---
| What you get | |
|---|---|
| **MCP server** | Bundled binary at dist-mcp/mcp-server.js, wired as the gossipcat command on PATH |
| **Dashboard** | Prebuilt static assets in dist-dashboard/ — launches automatically on a dynamic port (ask Claude Code *"what's my gossipcat dashboard URL?"*). Override with GOSSIPCAT_PORT=24420 if you want a stable port. |
| **Default skills + rules + archetypes** | 16 bundled skill templates, operational rules, and project archetypes copied into the install |
| **Postinstall wizard** | Writes .mcp.json with correct absolute paths for your machine |
npm uninstall -g gossipcat
claude mcp remove gossipcat -s user
rm -rf ~/.gossip # if you want to wipe global memory + signals
rm -rf <project>/.gossip # if you want to wipe per-project state
git checkout master && git pull ./scripts/release.sh # no args ```
Stage 1 creates chore/release-X.Y.Z, bumps package.json, opens the PR, exits. Stage 2 reads the version from package.json, builds the MCP bundle + dashboard, packs the tarball, tags, pushes the tag, and creates the GitHub release with auto-generated notes from commits since the last tag.
<br/>
Requirements: Node.js 22+ and Claude Code.
Concrete recipes for the most common workflows. Each one shows what to type, what you'll get back, and what to do with it.
This was a critical bug in v0.1.0 — fixed in v0.1.1. Upgrade with the install one-liner above. v0.1.1+ boots in degraded mode (dashboard + relay only) so you can run gossip_setup from inside Claude Code.
Config is searched in order: .gossip/config.json > gossip.agents.json > gossip.agents.yaml.
{
"main_agent": {
"provider": "google",
"model": "gemini-2.5-pro"
},
"utility_model": {
"provider": "native",
"model": "haiku"
},
"consensus_judge": {
"provider": "anthropic",
"model": "claude-sonnet-4-6",
"native": true
},
"agents": {
"sonnet-reviewer": {
"provider": "anthropic",
"model": "claude-sonnet-4-6",
"preset": "reviewer",
"skills": ["code_review", "security_audit", "typescript"],
"native": true
}
}
}
| Field | Description |
|---|---|
main_agent | Internal tool LLM for routing, planning, and synthesis |
utility_model | Memory compaction, gossip, lens generation |
consensus_judge | Model for cross-review synthesis |
agents.<id>.provider | anthropic, google, openai, openclaw, local |
agents.<id>.base_url | Custom endpoint for openai/openclaw (e.g. http://127.0.0.1:18789/v1) |
agents.<id>.native | true = runs via Claude Code Agent(), no API key |
agents.<id>.preset | reviewer, implementer, tester, researcher, debugger, architect, security, designer, planner, devops, documenter |
agents.<id>.skills | Skill labels for dispatch matching |
<br/>
Add env vars for the providers you want to use. Pass them with -e when registering, or set them in your shell environment.
| Provider | Env var | Notes |
|---|---|---|
| Native (Claude Code) | — | Dispatches through your active Claude Code subscription. No key needed. |
| Anthropic API | ANTHROPIC_API_KEY | Direct API access if you don't want to go through Claude Code. |
| Google Gemini | GOOGLE_API_KEY | Gemini Pro / Flash relay agents. |
| OpenAI | OPENAI_API_KEY (+ optional OPENAI_BASE_URL) | GPT-4 / GPT-4o relay agents. OPENAI_BASE_URL lets you point at OpenAI-compatible gateways (Azure, Together, Groq, etc.). |
| OpenClaw | — (local gateway) | OpenAI-compatible, defaults to http://127.0.0.1:18789/v1. No API key — auth handled by your local OpenClaw daemon. |
| Ollama (local) | — | Runs locally via http://localhost:11434. No key. Pull your model first with ollama pull llama3.1:8b. |
Native only (zero API keys — everything runs through Claude Code):
claude mcp add gossipcat -s user -- gossipcat Then in session ask for a team built from sonnet-reviewer / haiku-researcher / opus-implementer. Native agents dispatch through Agent() and relay back. Good zero-config starting point.
Anthropic API (direct, bypasses Claude Code):
claude mcp add gossipcat -s user \
-e ANTHROPIC_API_KEY=sk-ant-... \
-- gossipcat Use this if you want relay agents running Claude models without going through the Claude Code subscription path — e.g. for parallelism beyond Claude Code's concurrency cap, or for running long background reviews while you keep working.
Google Gemini:
claude mcp add gossipcat -s user \
-e GOOGLE_API_KEY=AIza... \
-- gossipcat Enables gemini-reviewer, gemini-tester, gemini-implementer on the relay. Watch the quota — gossipcat has a built-in 429 watcher that falls back to native agents when Gemini is cooling down.
OpenAI (and OpenAI-compatible gateways):
claude mcp add gossipcat -s user \
-e OPENAI_API_KEY=sk-... \
-- gossipcat For Azure / Together / Groq / OpenRouter, add OPENAI_BASE_URL: claude mcp add gossipcat -s user \
-e OPENAI_API_KEY=your-key \
-e OPENAI_BASE_URL=https://api.groq.com/openai/v1 \
-- gossipcat
OpenClaw (local gateway): ```bash
<p align="center"> <img src="https://img.shields.io/badge/OpenClaw-gateway-4A90D9?style=for-the-badge" alt="OpenClaw" /> <img src="https://img.shields.io/badge/%F0%9F%A6%9E-lobster%20friendly-red?style=for-the-badge" alt="Lobster friendly" /> </p>
Gossipcat supports OpenClaw as a provider gateway. OpenClaw runs locally and exposes an OpenAI-compatible HTTP API — gossipcat talks to it like any other relay agent, with your stored gateway token and a separate quota slot so OpenClaw rate limits never bleed into your OpenAI agents.
| What you get | Hallucination filtering | Agents improve over time | |
|---|---|---|---|
| **Gossipcat** | 3+ agents cross-review each other's findings; confirmed bugs only | Yes — peers catch and penalize hallucinations mechanically | Yes — accuracy signals steer dispatch; skill files fix repeat failures |
| **Single-agent review** (Claude Code built-in, Cursor review) | One model reviews your diff | No — hallucinations ship as findings | No — no feedback loop |
| **LLM-as-judge cross-review** (most multi-agent frameworks) | One model grades another model's output | Partial — judge can hallucinate too; no ground truth | No — judge scores aren't wired to dispatch |
| **Traditional review tools** (CodeRabbit, PR-Agent) | Pattern-match + one LLM pass | No | No |
The core difference: gossipcat verifies findings against actual file:line citations in your codebase. That ground truth is what makes the reward signal trustworthy enough to automate.
<br/>
Pin to a specific npm version:
npm install -g gossipcat@0.4.14
Pin to a specific GitHub release tarball (version-locked, bypasses npm registry):
npm install -g https://github.com/gossipcat-ai/gossipcat-ai/releases/download/v0.4.14/gossipcat-0.4.14.tgz
Project-local install (each project gets its own gossipcat):
cd your-project
npm install --save-dev gossipcat The postinstall writes .mcp.json to your project root. Open Claude Code in that directory and gossipcat connects automatically — no claude mcp add needed.
From source (contributors):
git clone https://github.com/gossipcat-ai/gossipcat-ai.git
cd gossipcat-ai
npm install
npm run build:mcp
claude mcp add gossipcat -s user -- node "$PWD/dist-mcp/mcp-server.js"
Gossipcat-ai 是一个基于无权重的上下文 RL 的代码审查系统,旨在通过学习基于实体的信号来实现高效的代码审查。
Gossipcat-ai 的功能包括多个独立的审查者进行审查,然后进行交叉审查,标记出 CONFIRMED、DISPUTED 和 UNIQUE 的发现。另外,还有自适应分发功能,根据分类的准确率自动调整分发权重,选择最合适的审查者。
安装 gossipcat-ai 可以通过以下方式进行:使用 Docker、pip 或源码部署。具体步骤请参见文档。
使用 gossipcat-ai 的步骤包括:创建一个 MCP 服务器、启动一个 dashboard、注册审查者和规则、配置环境变量等。具体步骤请参见文档。
gossipcat-ai 的配置文件可以从多个位置读取,包括 .gossip/config.json、gossip.agents.json 和 gossip.agents.yaml。配置项包括主审查者、辅助模型、共识评判者等。
gossipcat-ai 的 API 支持多种提供商,包括 Native(Claude Code)、Anthropic 和 Google。用户需要通过环境变量设置 API 密钥或通过 Claude Code 订阅来使用这些提供商。
gossipcat-ai 支持 OpenClaw 集成,用户可以通过 OpenClaw 来访问 gossipcat-ai 的功能。具体步骤请参见文档。
常见问题包括:如何卸载 gossipcat-ai?如何配置环境变量?如何使用 gossipcat-ai 的 API 等。
创新的多智能体审查方案,MCP架构设计合理。但生态完善度和社区成熟度仍需提升,建议先小范围试用评估效果。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ MIT 协议 — 最宽松的开源协议之一,可自由商用、修改、分发,仅需保留版权声明。
经综合评估,多智能体代码审查网格 在MCP工具赛道中表现稳健,质量良好。如果你已有明确的使用需求,可以直接上手体验;如果还在评估阶段,建议对比同类工具后再做决策。
| 原始名称 | gossipcat-ai |
| 原始描述 | 开源MCP工具:Multi-agent code review mesh — orchestrates AI agents from multiple providers to。⭐31 · TypeScript |
| Topics | 多智能体代码审查Claude集成MCP工具TypeScript |
| GitHub | https://github.com/gossipcat-ai/gossipcat-ai |
| License | MIT |
| 语言 | TypeScript |
收录时间:2026-05-20 · 更新时间:2026-05-21 · License:MIT · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端