# timps-swarm [Health: Active]

**Category:** 🤖 Coding Agents  
**Repository:** https://github.com/Sandeeprdy1729/timps-swarm  
**GitHub Stars:** 1  
**npm Downloads (last month):** 191  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/timps-swarm

## Description
161 AI agents for code gen, health, AI/ML, infra, India-stack. One command.

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

```json
"mcpServers": {
  "timps-swarm": {
    "command": "npx",
    "args": ["-y","timps-swarm"]
  }
}
```

## Documentation

## What timps-swarm MCP server does

The timps-swarm MCP server makes 161 specialist agents available to coding tools through MCP. The agents cover software delivery, security, infrastructure, AI/ML, health, and India-stack scenarios. Each specialist is exposed as an MCP tool and can also be registered as a native sub-agent for supported clients that can dispatch work in parallel.

The package includes task-oriented CLI commands. `audit` scans a repository for CVEs, hardcoded secrets, and OWASP issues without requiring the Python backend or an API key. `fix` runs a 10-agent software development lifecycle pipeline, while `api-design`, `db-design`, and `health` generate API specifications, database designs, and machine diagnostics respectively.

## How it works

Install the npm package and start the MCP entry point with `npx timps-swarm mcp`. The Node.js proxy communicates with a running FastAPI service, either locally or through `TIMPS_API_URL`; the Python repository is optional for the npm-based setup. The server uses stdio transport.

Provider selection checks MCP sampling first, then Gemini, Anthropic, OpenAI, Groq, Ollama, and the built-in TIMPS-Coder model in that order. Ollama provides a local option, while the hosted providers use their corresponding environment variables when available.

The `install-mcp` command detects supported developer tools, writes an MCP configuration entry, and creates agent definition files for Claude Code, Cursor, and Codex workflows. Configuration updates are idempotent. `--dry-run`, `--tool`, and `--no-sub-agents` alter installation behavior, and `uninstall-mcp` removes the generated entries and TIMPS agent files without removing unrelated configuration.

## Setup and configuration

Install the package globally or invoke it directly:

```bash
npm install -g timps-swarm
npx timps-swarm install-mcp
```

The installer can configure detected tools or one selected tool. It writes entries for clients such as Claude Code, Cursor, Windsurf, Continue, Zed, Gemini CLI, Codex CLI, VS Code, and Cline. Restart the client after configuration so the tools and sub-agents become available.

Optional provider variables include `GEMINI_API_KEY`, `ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, and `GROQ_API_KEY`. `OLLAMA_HOST` can configure Ollama, while `REDIS_URL` is forwarded in generated environment blocks. `TIMPS_API_URL` selects a remote FastAPI service. No provider key is needed for the standalone repository audit command.

## Tools and capabilities

- Run repository security audits with `npx timps-swarm audit ./`.
- Apply the multi-agent code-fixing workflow with `fix`.
- Generate OpenAPI specifications from natural-language requirements.
- Produce database schemas, DDL, ER diagrams, and migrations from a description.
- Diagnose a development machine with the health workflow.
- Dispatch specialist agents as parallel sub-agents where the client supports them.

The timps-swarm MCP server is suited to developers who want a single MCP registration for many focused workflows rather than configuring each specialist independently.

## Limitations and notes

The README documents MCP integrations for specific tools and does not establish compatibility with every MCP client. The Python backend is not required for the npm proxy, but backend-dependent operations need a reachable FastAPI service. Hosted model use depends on the relevant provider credentials; local execution depends on Ollama availability when that provider is selected. The README's parallel execution speed comparison is an example, not a guaranteed performance level.

_Full upstream README: https://allmcps.com/mcp/timps-swarm/readme_