# jigyasudham/veto [Health: Active]

**Category:** 🤖 Coding Agents  
**Repository:** https://github.com/jigyasudham/veto  
**GitHub Stars:** 3  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/jigyasudham-veto

## Description
A council of 49 specialist agents + 93 tools for every major AI CLI (Claude Code, Codex, Gemini, Cursor, Windsurf). Deterministic agents that optionally upgrade to LLM, a self-learning router, cross-CLI memory, and guards for dependency hallucinations and decision drift. No API keys; zero extra cost on subscriptions.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "veto": {
    "command": "npx",
    "args": ["-y","@jigyasudham/veto@latest"]
  }
}
```

## Documentation

## What jigyasudham/veto does

The jigyasudham/veto MCP server adds a broad developer-tool surface to AI coding clients. Its 49 specialists include a seven-agent Council and 42 worker agents covering development, security, memory, research, quality, and workflow tasks. The server exposes 93 MCP tools for code review, security and secrets scanning, task planning, dependency checks, session management, project mapping, documentation, CI gates, and related operations.

The server is intended for local use. It can provide persistent memory across supported clients, coordinate specialist agents, record decisions and outcomes, and route tasks to suitable capabilities. Compact mode reduces the number of schemas advertised initially while retaining access to the full catalog through discovery and invocation tools.

## How it works

Worker agents begin as deterministic expert modules. Analysis agents use techniques such as regex or AST detection, OWASP/CWE rules, and hash-based clone matching, while planning agents provide structured playbooks and checklists. They can optionally be upgraded with reasoning from the host AI through an agentic loop, or through MCP Sampling on clients that support `server.createMessage`.

The Council is designed for multi-agent discussion and can fall back to a deterministic verdict when no LLM path is available. Parallel agent execution is supported. A self-learning router can retune tier thresholds after recorded task outcomes when automatic application is enabled. The dependency verification tool checks proposed packages against a live registry before installation, addressing nonexistent package names and slopsquatting risks.

## Setup and configuration

Install the CLI globally with:

```bash
npm i -g @jigyasudham/veto
```

Then run `veto init` to register the server with the AI clients you use. An npx-based setup is also documented:

```bash
npx -y @jigyasudham/veto@latest init
```

The generated MCP configuration launches `veto-server` through the package registry. The global installation supplies the CLI, while the generated server command resolves the latest package on client restarts. Compact mode can be enabled with `VETO_COMPACT=1` in the MCP server environment or with `compact_tools: true` in `~/.veto/config.json`.

## Tools and capabilities

Notable tool groups include:

- Sessions, autosave, replay, status, and handoff
- Task routing, agent planning, delegation, composition, and parallel execution
- Council debates, ADRs, benchmarking, and decision tracking
- Code, diff, pull request, security, secrets, dependency, and full reviews
- CI gates, pre-commit checks, workflows, task parsing, and file watching
- Persistent memory, project maps, patterns, exports, imports, and search
- Local LLM, semantic search, SDD agents, Playwright, and IDE notification tools
- Documentation, OpenAPI generation, release notes, diagrams, RCA, and postmortems

## Limitations and notes

No separate API key is required for the documented subscription-based workflow, but LLM reasoning can consume tokens when the host uses API or pay-per-token billing. The server detects API-key environment variables during initialization and warns about this billing mode. Deterministic modules work offline, while LLM-enhanced behavior depends on the client path available.

The README notes that MCP Sampling is being deprecated in a future specification revision, so the agentic loop is the longer-term default. Compact mode changes the initially advertised tool surface; it does not remove the other catalog tools. The listed integrations include Claude Code, Codex CLI, Gemini CLI, Antigravity CLI, Cursor, Windsurf, Zed, and JetBrains.

## Getting started with this jigyasudham/veto MCP server
Always refer to the official documentation for the most accurate and up-to-date information.

_Full upstream README: https://allmcps.com/mcp/jigyasudham-veto/readme_