Veto

Name: Veto
Author: jigyasudham

@jigyasudham

代码仓库

关于 Veto

93 agentic MCP tools + 49 specialist agents for every major AI CLI. Self-learning, no API keys.

#mcp #mcp-server #ai-agents #code-review #security #memory #multi-agent #developer-tools #claude-code #self-learning

基本信息

分类

开发工具

传输方式

stdio

发布者

jigyasudham

提交者

Jigyasu Dham

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "veto": {
      "command": "npx",
      "args": [
        "-y",
        "--package",
        "@jigyasudham/veto",
        "veto-server"
      ]
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Veto?

Veto is a local MCP server that gives every major AI CLI — Claude Code, Codex, Gemini, Cursor, Windsurf, Zed, JetBrains — a shared layer of 93 agentic tools and 49 specialist agents. It’s designed for developers and teams who want deterministic, guardrailed tooling across multiple AI assistants.

How to use Veto?

Install via npx @jigyasudham/veto init. On a subscription there is no extra token cost and no API key required; on pay-per-token billing, optional LLM calls count toward usage and Veto warns you if it detects an API key.

Key features of Veto

49 specialist agents across 6 domains plus a 7-agent Council.
Deterministic-first agents run as real expert modules at zero token cost.
Guardrails that catch the AI's own mistakes (dependency hallucination, decision drift, compounding-error circuit breaker).
Cross-CLI memory to save sessions in one tool and resume in another.
Self-learning router that re-tunes its thresholds over time.
CI gates and secrets scanning.

Use cases of Veto

Use 42 deterministic worker agents for development, security, memory, research, quality, and workflow tasks.
Let the Council agent debate trade-offs before building features.
Save a session in one AI CLI and resume it in another with full context intact.
Apply guardrails to catch hallucinated dependencies and prevent compounding errors during complex coding sessions.
Run CI gates and secrets scanning directly from any supported AI client.

FAQ from Veto

Do I need an API key to use Veto?

On a subscription no extra token cost and no API key is needed. On pay-per-token billing, Veto detects API keys and warns you about optional LLM call usage.

Does Veto always make LLM calls?

No. Agents run as real expert modules (OWASP/CWE rules, AST secret detection, complexity metrics) at zero token cost. They optionally upgrade to LLM reasoning only when your client supports it.

What does "cross-CLI memory" mean?

You can save a session in one tool (e.g., Claude Code) and resume it in another supported tool (e.g., Cursor) with full context intact.

Is Veto open source?

Yes, Veto is MIT licensed and available on npm and the official MCP registry (io.github.jigyasudham/veto).

How are guardrails implemented?

Veto includes dependency-hallucination detection, decision-drift enforcement, and a compounding-error circuit breaker that catches mistakes the AI itself makes.