Dojo Genesis MCP Server
An MCP server that provides Claude Code with 99 methodology skills, seed patterns, decision logging, and memory tools to improve software development decisions. It acts as a cognitive scaffold layer, enabling better working methods rather than just additional data.
README
Dojo Genesis MCP Server
The methodology layer for Claude Code -- the first MCP server that makes Claude measurably better at software development decisions by encoding 99 first-party working methods across 10 behavioral plugins as active cognitive scaffolds.
Every other MCP server gives Claude more data. This one gives Claude better methods.
Quick Install
Claude Code
Add to your project's .claude/settings.json:
{
"mcpServers": {
"dojo": {
"command": "/path/to/dojo-mcp-server",
"env": {
"DOJO_SKILLS_PATH": "/path/to/CoworkPluginsByDojoGenesis",
"DOJO_ADR_PATH": "./decisions"
}
}
}
}
Or using Docker:
{
"mcpServers": {
"dojo": {
"command": "docker",
"args": ["run", "-i", "--rm", "ghcr.io/dojogenesis/mcp-server:latest"]
}
}
}
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"dojo": {
"command": "/path/to/dojo-mcp-server",
"env": {
"DOJO_SKILLS_PATH": "/path/to/CoworkPluginsByDojoGenesis",
"DOJO_ADR_PATH": "./decisions"
}
}
}
}
What's Included
Tools (28)
| Group | Tools |
|---|---|
| Methodology | dojo_scout, dojo_invoke_skill, dojo_search_skills, dojo_list_skills, dojo_apply_seed, dojo_reflect |
| Decisions | dojo_log_decision (writes ADR markdown to DOJO_ADR_PATH) |
| Gateway memory | dojo_memory_list, dojo_memory_store, dojo_memory_search (gateway session memory) |
| Memory Hub | dojo_search_memory, dojo_get_memory, dojo_recent_memories — read-only Postgres mirror of the institutional memory (DOJO_MEMORY_DB_URL) |
| Unified fetch | dojo_fetch — one tool to search AND fetch across memory hub + skills + ADRs + seeds (typed ids: memory:slug, skill:name, adr:file.md, seed:name) |
| Seeds | dojo_seed_list, dojo_seed_create, dojo_seed_search |
| Agents² | dojo_agent_list, dojo_agent_dispatch², dojo_agent_chat² |
| Dispatch² | dojo_dispatch² — prompt → LLM through the gateway |
| Project | dojo_project_status, dojo_project_track, dojo_project_decision |
| Disposition | dojo_disposition_list, dojo_disposition_set |
| Craft | dojo_converge, dojo_health |
² Dispatch-class — spends LLM provider budget through the gateway. In HTTP
mode these require a dispatch-enabled API key and are rate limited per key;
dojo_scout's LLM path degrades to its offline scaffold for non-dispatch keys.
Skills (99 from CoworkPlugins, 35 bundled)
Skills are complete, battle-tested methodology workflows loaded from SKILL.md files at startup. When DOJO_SKILLS_PATH is set, all 99 first-party CoworkPlugins skills are available. Without it, 35 key skills across 7 plugins are embedded in the binary via go:embed.
Bundled plugins and skills:
| Plugin | Skills |
|---|---|
agent-orchestration |
agent-teaching, decision-propagation, handoff-protocol, workflow-router, workspace-navigation |
continuous-learning |
debugging, project-exploration, research-modes, research-synthesis, retrospective, web-research-external |
skill-forge |
mcp-cloudflare-builder, mcp-server-builder, process-extraction, skill-audit, skill-creation, skill-maintenance |
specification-driven-development |
context-ingestion, frontend-from-backend, implementation-prompt, parallel-tracks, pre-implementation-checklist, release-specification |
strategic-thinking |
iterative-scouting, multi-surface-strategy, product-positioning, strategic-scout |
system-health |
documentation-audit, health-audit, semantic-clusters, status-writing |
wisdom-garden |
compression-ritual, memory-garden, seed-extraction, seed-library |
Seed Patches (20)
Seeds are reusable thinking patterns drawn from across the Dojo ecosystem.
Dojo Genesis (Core)
| Seed | Description |
|---|---|
three_tiered_governance |
Three-tiered governance framework: Strategic, Tactical, Operational. |
harness_trace |
Nested JSON trace log for complete agent session traceability. |
context_iceberg |
4-tier context management system (hot/warm/cold/pruned). |
agent_connect |
Routing-first agent architecture with a single supervisor. |
go_live_bundles |
Lightweight deployment packages pairing artifacts with approval evidence. |
cost_guard |
Budget for the full 5-10x context iceberg multiplier. |
safety_switch |
Users must remain in control -- no autopilot on sensitive operations. |
implicit_perspective_extraction |
Extract implicit perspectives from user queries without enumeration. |
mode_based_complexity_gating |
Route to local or cloud models based on mode complexity. |
shared_infrastructure |
Build once, reuse everywhere -- central implementations. |
AROMA & Serenity Valley
| Seed | Description |
|---|---|
sanctuary_architecture |
Design digital spaces for being, not just doing. |
pace_of_understanding |
Move slow to move fast; learn without extraction. |
lineage_transmission |
Honor sources, trace influence, celebrate collaboration. |
graceful_failure |
Permission to not know, change your mind, and ask for help. |
local_first_liberation |
Local-first architecture for agent autonomy and user sovereignty. |
the_onsen_pattern |
Rest as critical practice for sustainable performance. |
collaborative_calibration |
Norms for peer-to-peer learning and explicit teaching. |
transparent_intelligence |
Reveal internal state, admit uncertainty, make learning visible. |
inter_acceptance |
Accept yourself through the compassionate eyes of another. |
radical_freedom |
Agency and the power to choose your response within constraints. |
Resources
Documentation resources accessible via MCP resource URIs:
dojo://resources/{name}-- 8 documentation resources (AROMA philosophy, EIT principles, etc.)dojo://seeds/{name}-- 20 seed patchesdojo://skills/{plugin}/{name}-- All loaded skills
Configuration
| Env Var | Default | Description |
|---|---|---|
DOJO_SKILLS_PATH |
(bundled fallback) | Path to CoworkPlugins root directory containing plugins/ |
DOJO_ADR_PATH |
./decisions |
Directory where dojo_log_decision writes ADR files |
DOJO_GATEWAY_URL |
http://localhost:7340 |
Dojo AgenticGateway base URL |
DOJO_GATEWAY_TOKEN |
(unset) | Bearer token for the gateway, if it requires one |
DOJO_MEMORY_DB_URL |
(unset → hub tools disabled) | Postgres DSN for the Memory Hub (URL or keyword form; use the SELECT-only role) |
DOJO_HTTP_ADDR |
(unset → stdio) | Opt into HTTP mode, e.g. :8091 |
DOJO_MCP_API_KEYS |
(unset) | HTTP mode only: comma-separated label:key pairs (individually revocable) |
DOJO_DISPATCH_ALLOWED_LABELS |
(unset → none) | Key labels allowed to run dispatch-class tools |
DOJO_DISPATCH_RATE_PER_MIN |
6 |
Per-label rate limit for dispatch-class tools |
The server works out of the box with zero configuration (bundled skills, default ADR path, stdio).
HTTP Mode (public endpoint)
Setting DOJO_HTTP_ADDR serves MCP streamable-HTTP instead of stdio:
POST/GET /mcp— the MCP endpoint, Bearer-key required (Authorization: Bearer <key>). Keyless or wrong-key requests get 401./mcp/k/<key>— same endpoint for clients that cannot send custom headers; the key is redacted in logs.GET /health— unauthenticated liveness (status + version only).
The server refuses to start in HTTP mode without a valid, non-empty
DOJO_MCP_API_KEYS. Every tool call is logged as
tool_call tool=… key=<label> dur_ms=… outcome=… — labels only, never key
material, never payloads.
DOJO_HTTP_ADDR=:8091 \
DOJO_MCP_API_KEYS="win:$(openssl rand -hex 32)" \
./dojo-mcp-server
curl -s localhost:8091/health
curl -s -X POST localhost:8091/mcp \
-H "Authorization: Bearer <key>" \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-03-26","capabilities":{},"clientInfo":{"name":"curl","version":"0"}}}'
Client wiring (Claude Code): claude mcp add --transport http dojo-remote https://<host>/mcp --header "Authorization: Bearer <key>".
Deployment notes: ingress is expected to be a Cloudflare tunnel (no host
ports); pin the container image by digest so auto-pull never
surprise-deploys a new public surface. Image publishing:
.github/workflows/docker-publish.yml → ghcr.io/dojogenesis/mcp
(supersedes the old manually-pushed mcpbydojogenesis image).
Building from Source
git clone https://github.com/DojoGenesis/mcp.git
cd mcp
# Build
go build -o dojo-mcp-server ./cmd/server
# Run
./dojo-mcp-server
# Test
go test -race -v ./...
Requires Go 1.23 or later.
Docker
# Build the image
docker build -t dojo-mcp-server .
# Run via stdio (for MCP hosts)
docker run -i --rm dojo-mcp-server
# Run with skills path mounted
docker run -i --rm -v /path/to/CoworkPlugins:/skills -e DOJO_SKILLS_PATH=/skills dojo-mcp-server
Project Structure
mcp-server/
cmd/server/main.go Server entry point, env vars
internal/
dojo/
handler.go 7 MCP tool handlers
scaffolds.go Scout and reflect methodology templates
skills/
loader.go Filesystem SKILL.md loader
search.go Keyword + trigger search
bundled.go go:embed fallback (15 key skills)
bundled/ Embedded SKILL.md files
decisions/
writer.go ADR file writer
wisdom/
base.go Wisdom base, search, helpers
seeds.go 20 seed patches
resources.go 8 documentation resources
Dockerfile Multi-stage container build
.github/workflows/ci.yml CI pipeline
.goreleaser.yml Cross-platform release config
Philosophy
Dojo Genesis is built on three core principles:
- Beginner's Mind -- Approach every interaction fresh, free from accumulated expertise.
- Self-Definition -- Help users see their own thinking, not impose external frameworks.
- Understanding is Love -- Deep, non-judgmental understanding is the highest service.
License
MIT License -- see LICENSE for details.
Copyright (c) 2026 Dojo Genesis
Contributing
Contributions are welcome. Please see github.com/DojoGenesis for organization-level guidelines.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.