# gitworthy [Health: Active]

**Category:** 🔄 Version Control  
**Repository:** https://github.com/trippyogi/gitworthy  
**GitHub Stars:** 2  
**npm Downloads (last month):** 259  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/gitworthy

## Description
is it worth your commit? Pre-flight triage for OSS contribution targets.

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

```json
"mcpServers": {
  "gitworthy": {
    "command": "npx",
    "args": ["-y","gitworthy@latest"],
    "env": {
      "GITHUB_TOKEN": "",
      "GITWORTHY_MCP_TOKEN": ""
    }
  }
}
```

**Requires environment variables:** `GITHUB_TOKEN`, `GITWORTHY_MCP_TOKEN` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What gitworthy MCP server does

The gitworthy MCP server helps coding agents decide whether an open-source contribution target is worth pursuing before implementation begins. It works with GitHub repositories and organizations, using repository and issue evidence to identify work that may already be underway, completed, assigned, duplicated, or blocked by contribution policy.

Results distinguish between three actions. `ACT` adds a candidate to the work queue after its evidence is reviewed. `VERIFY` identifies unresolved checks that must be investigated. `SKIP` removes a target from consideration when the available evidence shows that starting work would be inappropriate. These decisions apply to individual candidates rather than serving as one verdict for an entire scan.

## How it works

The workflow starts with a policy and scan phase, then narrows the results before running more expensive checks. `hunt` scans a repository or organization, filters weak or clearly unsuitable rows, ranks the remaining candidates, and runs preflight checks on only a bounded shortlist. Each returned candidate includes its own `worth_check` result and evidence.

The checks cover linked pull requests, assignments, referenced commits, previous attempts, work on remote branches, fixes merged into `main`, changes missing from a published package, contribution requirements, and related issues. The server reports checks it could not complete, rather than treating missing visibility as proof that no related work exists.

The gitworthy MCP server is intended for a scout-to-execute workflow. After selecting a candidate, the agent should rerun `worth_check`, reproduce the issue or request, follow the repository’s contribution rules, and only then implement and submit work. Results can become stale; stored decisions are considered stale after 24 hours.

## Setup and configuration

Node.js 22 or newer is required. For local use, configure an MCP client to run `npx -y gitworthy@latest mcp` over stdio. The server needs `GITHUB_TOKEN`; read access to public repositories is sufficient, while a classic token or a fine-grained token with Issues: Read provides better visibility into timeline cross-references.

For cloud agents or remote clients, start the same engine with `mcp --http`, then connect the client to its MCP URL. Non-loopback HTTP binds require `GITWORTHY_MCP_TOKEN`; keep `GITHUB_TOKEN` on the server rather than sending it to the client. Pin a package version for repeatable workflows and CI instead of relying on `@latest`.

## Tools and capabilities

The gitworthy MCP server exposes the repository-triage workflow to MCP clients. The documented operations include:

- `hunt` for bounded discovery, ranking, filtering, and candidate preflight.
- `worth_check` for a fresh evaluation of a specific target.
- `portfolio` for mode and dispatch around a candidate set; it does not create a new verdict.

The command-line interface also provides `doctor` for environment checks, `check` for evaluating a target such as `owner/repo#123`, and `brief` for rendering a stored decision as Markdown. Brief generation reads the local store and does not call GitHub, npm, or an LLM.

## Limitations and notes

An `ACT` result means a target is worth considering, not that implementation should begin without another check. Broad scans do not produce a full verdict for every issue, and weak GitHub token permissions can leave linked-work checks incomplete. Heuristic evidence cannot independently create a hard `SKIP`; definitive skip decisions require conclusive evidence.

The gitworthy MCP server does not replace repository-specific contribution rules or the agent’s execution process. It helps select and validate work, but the agent remains responsible for rechecking state, reproducing the issue, implementing the change, reviewing it, and using the repository’s permitted submission path.

_Full upstream README: https://allmcps.com/mcp/gitworthy/readme_