Archon MCP Server

MCP (Model Context Protocol) server for Archon — a 950+ page engineering knowledge base covering system design, algorithms, security, AI/ML, EDA, DevOps, and more. Lets any MCP-compatible client query the knowledge base directly.

21 tools | 3 resource templates | 5 prompts

Installation

cd mcp-server
npm install

Global install (npm)

npm install -g .
archon-mcp          # stdio mode
archon-mcp --http   # HTTP/SSE mode

Running

Stdio mode (default)

Used by Claude Code, Cursor, Claude Desktop, and other local MCP clients.

npm start
# or
node index.js

HTTP/SSE mode

For remote access, browser-based clients, or shared servers.

node index.js --http
# Starts on http://localhost:3456
# Override port: PORT=8080 node index.js --http

Environment variables

| Variable | Default | Description | |-------------|---------|-------------------------| | PORT | 3456 | HTTP/SSE server port | | LOG_LEVEL | info | debug, info, warn, error |

Adding to MCP Clients

Claude Code

claude mcp add archon -- node /path/to/knowledge-vault/mcp-server/index.js

Cursor

Add to .cursor/mcp.json:

{
  "mcpServers": {
    "archon": {
      "command": "node",
      "args": ["/path/to/knowledge-vault/mcp-server/index.js"]
    }
  }
}

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "archon": {
      "command": "node",
      "args": ["/path/to/knowledge-vault/mcp-server/index.js"]
    }
  }
}

Remote / SSE clients

Start in HTTP mode (node index.js --http), then connect to:

• SSE endpoint: http://localhost:3456/sse
• Health check: http://localhost:3456/health

Tools (21)

1. search_archon

Full-text multi-keyword search with relevance scoring (title > description > tags > body).

| Parameter | Type | Default | Description | |-----------|--------|---------|------------------------| | query | string | -- | One or more keywords | | limit | number | 10 | Max results (max 50) |

search_archon({ query: "circuit breaker retry", limit: 5 })

2. get_page

Get the full markdown content of a page by its path.

| Parameter | Type | Description | |-----------|--------|-------------------------------------| | path | string | Page path (without .md extension) |

get_page({ path: "system-design/api-design/api-gateway-patterns" })

3. list_sections

List all top-level sections with their page counts. No parameters.

4. get_related

Find the top 5 pages related to a given page based on shared tags.

| Parameter | Type | Description | |-----------|--------|--------------------------| | path | string | Page path to find related pages for |

get_related({ path: "ai-ml-engineering/ai-agents" })

5. list_pages

List all pages in a section with title, path, description, and difficulty.

| Parameter | Type | Description | |-----------|--------|------------------------------------| | section | string | Section name (e.g. "eda", "devops") |

list_pages({ section: "cheat-sheets" })

6. get_cheat_sheet

Get a cheat sheet by topic name. Lists available cheat sheets if not found.

| Parameter | Type | Description | |-----------|--------|------------------------------------------| | topic | string | Topic (e.g. "docker", "kubernetes") |

get_cheat_sheet({ topic: "docker" })

7. search_by_tag

Find all pages with a specific tag.

| Parameter | Type | Default | Description | |-----------|--------|---------|---------------------------------| | tag | string | -- | Tag name (e.g. "redis", "react") | | limit | number | 20 | Max results |

search_by_tag({ tag: "distributed-systems" })

8. search_by_difficulty

Find pages by difficulty level, optionally filtered by section.

| Parameter | Type | Default | Description | |--------------|--------|---------|------------------------------------------------| | difficulty | enum | -- | beginner, intermediate, advanced, expert | | section | string | -- | Optional: limit to a section | | limit | number | 20 | Max results |

search_by_difficulty({ difficulty: "beginner", section: "system-design" })

9. get_comparison

Get a comparison page (X vs Y). Matches by technology names.

| Parameter | Type | Description | |-----------|--------|--------------------| | tech1 | string | First technology | | tech2 | string | Second technology |

get_comparison({ tech1: "react", tech2: "vue" })

10. get_interview

Get a system design interview walkthrough.

| Parameter | Type | Description | |-----------|--------|--------------------------------------------------| | topic | string | Topic (e.g. "instagram", "uber", "chat-system") |

get_interview({ topic: "instagram" })

11. get_lld

Get a low-level design interview problem.

| Parameter | Type | Description | |-----------|--------|----------------------------------------------------| | topic | string | Topic (e.g. "parking-lot", "elevator", "chess") |

get_lld({ topic: "parking-lot" })

12. get_war_room

Get a real incident case study from the War Room.

| Parameter | Type | Description | |-----------|--------|---------------------------------------------------------| | topic | string | Incident (e.g. "cloudflare", "crowdstrike", "knight-capital") |

get_war_room({ topic: "cloudflare" })

13. get_checklist

Get a production readiness checklist.

| Parameter | Type | Description | |-----------|--------|--------------------------------------------------------------| | topic | string | Topic (e.g. "pre-launch", "security-review", "observability-readiness") |

get_checklist({ topic: "pre-launch" })

14. get_runbook

Get an operational runbook or debugging playbook.

| Parameter | Type | Description | |-----------|--------|-------------------------------------------------------| | topic | string | Topic (e.g. "memory-leak", "high-error-rate", "api-slow") |

get_runbook({ topic: "memory-leak" })

15. get_learning_path

Get a structured learning path for a role or topic.

| Parameter | Type | Description | |-------------|--------|-----------------------------------------------------------------| | path_name | string | Path name (e.g. "backend-engineer", "frontend-engineer", "devops-engineer") |

get_learning_path({ path_name: "backend-engineer" })

16. get_glossary_term

Look up an engineering term from the glossary. Falls back to search if not found.

| Parameter | Type | Description | |-----------|--------|-------------------------------------------| | term | string | Term (e.g. "CQRS", "CAP theorem", "idempotency") |

get_glossary_term({ term: "CQRS" })

17. get_code_examples

Extract all code blocks from a page, optionally filtered by programming language.

| Parameter | Type | Description | |------------|--------|----------------------------------------------| | path | string | Page path | | language | string | Optional: filter by language (e.g. "typescript") |

get_code_examples({ path: "build-from-scratch/redis", language: "typescript" })

18. random_page

Get a random page. Great for learning something new.

| Parameter | Type | Description | |-----------|--------|------------------------------------| | section | string | Optional: limit to a section |

random_page({ section: "war-room" })

19. get_build_guide

Get a "build from scratch" tutorial.

| Parameter | Type | Description | |-----------|--------|-----------------------------------------------------| | topic | string | Topic (e.g. "redis", "load-balancer", "rate-limiter") |

get_build_guide({ topic: "redis" })

20. get_eda_guide

Get an Exploratory Data Analysis (EDA) page.

| Parameter | Type | Description | |-----------|--------|----------------------------------------------------------| | topic | string | Topic (e.g. "missing-data", "outlier-analysis", "pandas-fundamentals") |

get_eda_guide({ topic: "missing-data" })

21. list_all_tags

List all unique tags across the knowledge base with counts. No parameters.

Resources

MCP Resources expose knowledge base content as browsable URIs.

| URI Pattern | Description | |----------------------------------------|--------------------------------------| | archon://sections/{section_name} | All pages in a section (JSON) | | archon://cheat-sheets/{topic} | Cheat sheet content (Markdown) | | archon://learning-paths/{path_name} | Learning path content (Markdown) |

Examples

archon://sections/system-design
archon://cheat-sheets/docker
archon://learning-paths/backend-engineer

Prompts

MCP Prompts generate structured prompts with relevant knowledge base context.

explain-concept

Pulls relevant pages and creates a structured explanation.

| Argument | Type | Description | |----------|------|------------------------------------| | topic | string | Concept to explain | | level | enum | beginner, intermediate, advanced, expert |

interview-prep

Pulls interview pages and creates a preparation guide.

| Argument | Type | Description | |----------|------|------------------------------------| | type | enum | system-design, lld, behavioral | | topic | string | Specific topic |

debug-issue

Pulls debugging playbooks and creates a structured debugging guide.

| Argument | Type | Description | |-----------|------|------------------------------------| | problem | string | Description of the problem |

compare-tech

Pulls comparison pages and creates a structured comparison.

| Argument | Type | Description | |----------|------|--------------------| | tech1 | string | First technology | | tech2 | string | Second technology |

eda-workflow

Pulls EDA workflow + checklist and creates a guided analysis plan.

| Argument | Type | Description | |-----------------------|------|--------------------------------| | dataset_description | string | Brief description of dataset |

Architecture

• Startup: All markdown files under docs/ are read, frontmatter is parsed with regex, and everything is cached in memory
• Search: Case-insensitive multi-word matching with relevance scoring (title 10pts > description 5pts > tags 3pts > body 1pt)
• Transport: stdio by default; HTTP/SSE on port 3456 with --http flag
• Logging: Structured JSON to stderr (never pollutes stdout/MCP transport)
• Shutdown: Graceful handling of SIGINT/SIGTERM with connection cleanup

Smithery

This server includes a smithery.yaml for publishing to the Smithery MCP registry.