autology
Preserve human understanding in AI-driven development. A living ontology that captures decisions, tracks reasoning, and compounds knowledge across Claude Code sessions.
claude mcp add --transport stdio curt-park-autology npx -y Curt-Park/autology
How to use
Autology is a team knowledge base and doc-code synchronization assistant designed to live in your Claude workflow. It stores knowledge as Obsidian-compatible Markdown files under docs/ with YAML frontmatter and wikilinks, and it can automatically triage, capture, and explore knowledge as you work. Use the autology plugin to classify existing knowledge, capture new decisions and conventions from conversations, and traverse the knowledge graph to understand relationships between concepts, decisions, and components. The workflow is tightly integrated with typical development actions: when you commit or decide something, Autology can triage what’s already documented, capture new items automatically, and help you explore related concepts to keep the team’s knowledge base up to date and linked to code.
How to install
Prerequisites:
- Node.js and npm installed on your system
- Access to Claude with plugin capability enabled (as this is a Claude plugin)
Installation steps:
- Ensure you have a functioning Node.js environment:
- Install Node.js from https://nodejs.org/
- Install the Autology plugin via npx (as supported by the MCP workflow):
npx -y Curt-Park/autology - Alternatively, if you prefer the plugin-marketplace flow described by the project:
- Open Claude and run:
- plugin marketplace add Curt-Park/autology
- plugin install autology@autology
- Open Claude and run:
- Verify installation by invoking a quick test command in Claude, for example:
- /autology:autology-tutorial
Notes:
- The plugin stores data in docs/ with a graph of wikilinks. Ensure your repository has or can create a docs/ directory.
- You may need appropriate permissions to install third-party plugins in your Claude environment.
Additional notes
Tips and considerations:
- Autology relies on wiki-style linking (wikilinks) and YAML frontmatter to build a navigable graph of knowledge. Keep docs/ well-structured to maximize triage and search capabilities.
- When capturing knowledge, Autology will deduplicate by grepping existing nodes before creating new ones. If you move or rename items, run /autology:sync-knowledge to align docs with code.
- Use /autology:explore-knowledge path <A> <B> to find short paths between concepts, which can help validate design decisions and capture rationale.
- For large repos, run full syncs periodically to detect drift between code and docs.
- Environment variables are typically not required for standard usage; if you customize storage paths or graph behavior, document those in env placeholders.
Related MCP Servers
ols4
The EMBL-EBI Ontology Lookup Service (OLS)
mem0 -selfhosted
Self-hosted mem0 MCP server for Claude Code. Run a complete memory server against self-hosted Qdrant + Neo4j + Ollama while using Claude as the main LLM.
MAMA
Always-on companion for Claude that remembers your decisions and their evolution. Local-first memory using SQLite + transformers.js embeddings.
bindly-claude-code
Knowledge completion layer for Claude Code - finish your thoughts and make them reusable across sessions and agents
obsidian-ai-curator
Smart note consolidation for Obsidian using Claude AI. Automatically identifies and merges scattered fragments into structured knowledge via MCP. Like Tetris for your thoughts - pieces fall into place.
obsidian-http
First HTTP-native MCP server for Obsidian - Compatible with any MCP client (Claude Code, Codex, Gemini, etc.) without stdio bugs