{
  "markdown": "# atlas\n\nTurn a stated need into a full system design by mining the Atlas knowledge graph.\n\nAtlas is a knowledge graph of AI-agent stacks, products, capabilities, processes,\ndata models, and the relationships between them. This plugin makes that graph a\nfirst-class design tool inside Gemini CLI: ask for a system and Atlas\nmines the graph to turn the stated need into a layered, evidence-backed design.\n\n## Installation — Gemini CLI\n\n```bash\nnpm install -g @a5c-ai/atlas-gemini-cli\natlas-gemini-cli install --global\n```\n\nRestart Gemini CLI to pick up the installed plugin.\n\n## What's Included\n\n- **Skills**: atlas, atlas-graph-query\n- **Commands**: (directory)\n- **MCP**: the Atlas knowledge-graph server (`atlas`), wired natively into\n  Gemini CLI's own MCP config — no manual setup.\n\n## Atlas MCP\n\nThe plugin injects the Atlas MCP server into Gemini CLI's native MCP\nconfig format automatically. By default it points at:\n\n```\nhttps://atlas-staging.a5c.ai/api/mcp\n```\n\nOverride the endpoint at runtime with the `ATLAS_MCP_URL` environment variable.\nOnce the harness is running, the `mcp__atlas__atlas_public_*` tools (search,\nrecord, neighbors, kinds, clusters, stats, wiki pages) are available to the\nskills and commands.\n\n## How it works\n\n- The **`atlas` skill** turns a stated need into a layered system design by\n  mining the graph — when to use Atlas, how to anchor a search, expand\n  neighbors, and synthesize a design.\n- The **`atlas-graph-query` skill** documents the MCP tool surface so any agent\n  (sub-agents included) can query the graph without re-deriving conventions.\n- The **commands** (`(directory)`) run interactive, graph-driven workflows —\n  systems discovery, process mining, data mining, and nuance collection. They\n  drive iterative, TDD-style runs whose orchestration is delegated to the\n  Babysitter run lifecycle, so each phase is checkable and resumable.\n\n> The commands use Babysitter purely as the orchestration runtime; the design\n> work itself is driven by the Atlas graph via the MCP tools above.\n\n## Verification\n\n```bash\nbabysitter harness:discover --json | grep gemini\n```\n",
  "bytes": 2110,
  "sha": "b36889dea384bc662147fa930c966572d0fe6dc5adf01a913d441f03e0fda107",
  "repo_slug": "a5c-ai/atlas-gemini",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_a5c_ai_atlas_gemini_de15082c/readme"
}