|
| 1 | +# DojoGenesis MCP Server |
| 2 | + |
| 3 | +> The methodology layer for Claude Code — 7 tools and 84 first-party methodology skills as active cognitive scaffolds for AI agents. |
| 4 | + |
| 5 | +## What this is |
| 6 | + |
| 7 | +The Dojo MCP Server is not a data source — it is a methods source. Every other MCP server gives AI agents more data. This one gives them better thinking frameworks. It encodes 60+ battle-tested methodology skills as active cognitive scaffolds, making Claude measurably better at software development decisions. Skills are loaded from the DojoGenesis plugins collection and exposed via the MCP protocol. |
| 8 | + |
| 9 | +## Key capabilities |
| 10 | + |
| 11 | +- `dojo.scout` — 4-step strategic analysis scaffold: frame tension, scout routes, synthesize, decide |
| 12 | +- `dojo.invoke_skill` — load a specific methodology skill by name, returns full workflow as actionable steps |
| 13 | +- `dojo.search_skills` — search the methodology library for skills matching a query |
| 14 | +- `dojo.apply_seed` — apply a reusable thinking pattern (seed) to a specific situation with checklist |
| 15 | +- `dojo.log_decision` — write an Architecture Decision Record (ADR) to disk (the only write-capable tool) |
| 16 | +- `dojo.reflect` — structured reflection grounded in matched skills and seeds from the methodology library |
| 17 | +- `dojo.list_skills` — list all available skills grouped by plugin category |
| 18 | +- Resources at `dojo://skills/{plugin}/{name}`, `dojo://seeds/{name}`, `dojo://resources/{name}` |
| 19 | +- 8 documentation resources (AROMA philosophy, EIT principles, etc.) |
| 20 | +- 20 seed patches for reusable thinking patterns |
| 21 | + |
| 22 | +## How to use with LLMs |
| 23 | + |
| 24 | +Install via Claude Code by adding the server to `.claude/settings.json` with `DOJO_SKILLS_PATH` pointing to your plugins directory. Use `dojo.list_skills` to discover available methodology skills, then `dojo.invoke_skill` to load a specific skill before tackling a complex decision. Use `dojo.scout` when facing architectural or strategic tensions. Use `dojo.log_decision` to persist Architecture Decision Records during implementation. |
| 25 | + |
| 26 | +## Installation (Claude Code) |
| 27 | + |
| 28 | +```json |
| 29 | +{ |
| 30 | + "mcpServers": { |
| 31 | + "dojo": { |
| 32 | + "command": "/path/to/dojo-mcp-server", |
| 33 | + "env": { |
| 34 | + "DOJO_SKILLS_PATH": "/path/to/plugins", |
| 35 | + "DOJO_ADR_PATH": "./decisions" |
| 36 | + } |
| 37 | + } |
| 38 | + } |
| 39 | +} |
| 40 | +``` |
| 41 | + |
| 42 | +## Docker |
| 43 | + |
| 44 | +``` |
| 45 | +docker run -i --rm ghcr.io/dojogenesis/mcp-server:latest |
| 46 | +``` |
| 47 | + |
| 48 | +## Links |
| 49 | + |
| 50 | +- GitHub: https://github.com/DojoGenesis/mcp |
| 51 | +- Docker: ghcr.io/dojogenesis/mcp-server |
| 52 | +- Plugins (skill source): https://github.com/DojoGenesis/plugins |
0 commit comments