GitMCP Universal Repository Processor Create a complete MCP-ready snapshot of any Git repository in one run. The actor clones the repository, filters content by scope (code, docs, issues, git history), and packages everything for downstream AI tooling. ## What you get after each run - Repository bundle — A ZIP archive stored as REPOSITORY_ZIP in the key-value store. - Manifest — MANIFEST.json summarising repository metadata, processed files, issues, and MCP integration hints. - Dataset records — Structured entries for each processed file and issue to drive analytics or custom exports. - Optional git history — Latest commits when includeGitHistory is true. ## Input parameters Configure inputs via the Apify UI or storage/key_value_stores/default/INPUT.json. See .actor/input_schema.json for defaults and accepted values. - repositoryUrl (string, required) — Git repository HTTPS/SSH URL. - accessToken (string, optional) — Personal access token for private repositories (use Secret Storage). - branch (string, optional) — Branch or tag to checkout after cloning. - scopes (array, optional) — Any combination of code, docs, issues (default includes all). - outputFormats (array, optional) — mcp, json, or both; controls dataset formatting. - maxFileSizeMb (number, optional) — Skip files larger than this threshold (default 25). - includeGitHistory (boolean, optional) — Collect last 100 commits when true. - runIdSuffix (string, optional) — Appended to dataset name for grouping runs. ### Example input json { "repositoryUrl": "https://github.com/apify/gitmcp-universal-repository-processor", "branch": "main", "scopes": ["code", "docs"], "includeGitHistory": true, "maxFileSizeMb": 20 } ## Output details - Dataset (default) — Records tagged as type="code", type="doc", or type="issue" with fields such as path, sizeBytes, lines, lang, url (for issues). - Key-value store (default) - REPOSITORY_ZIP — Downloadable ZIP of the processed repository snapshot. - MANIFEST.json — JSON manifest containing run metadata, file/issue counts, git history (if requested), and suggested MCP servers. - Logs — Checkout warnings, GitHub API throttling, and charging information. ## Pricing and charging When pay-per-event billing is active on Apify, the actor charges the repository_processed event after a successful run. Use test mode or local runs to avoid charges while prototyping. ## Tips - Provide an access token with repo scope for private GitHub repositories. - Limit scopes to speed up runs (e.g. ['docs'] for documentation-only dumps). - Use maxFileSizeMb to skip large binaries the MCP server does not need. ## Running locally npm install npm run build apify run Place INPUT.json under storage/key_value_stores/default/ and ensure Git and network access are available inside the Docker image or local environment.