A Claude Code plugin, Ars Contexta generates personalized, persistent knowledge systems from conversational input, acting as a "second brain" for agents. It targets users seeking to move beyond ephemeral AI sessions, offering a derived, markdown-based knowledge graph tailored to individual workflows, backed by extensive research claims, and free from vendor lock-in.

How It Works

Ars Contexta employs a unique "derivation, not templating" approach. Users describe their domain and thinking process via a conversational setup (/arscontexta:setup). An engine then maps these signals to a cognitive architecture—including folder structures, context files, processing pipelines, hooks, and navigation maps—grounded in 249 research claims. The system utilizes a "Three-Space Architecture" (self/, notes/, ops/) to organize agent identity, the knowledge graph, and operational state, ensuring a structured and persistent output.

Quick Start & Requirements

Installation involves adding the plugin marketplace (/plugin marketplace add agenticnotetaking/arscontexta), installing the plugin (/plugin install arscontexta@agenticnotetaking), restarting Claude Code, and running the conversational setup (/arscontexta:setup). This setup phase takes approximately 20 minutes and is token-intensive. A second restart activates generated hooks.

• Prerequisites: Claude Code v1.0.33+, tree, ripgrep (rg). qmd is optional for semantic search.
• Setup Time: ~20 minutes (one-time).

Highlighted Details

• Markdown Knowledge Graph: Generates a plain markdown vault connected by wiki links, forming a traversable knowledge graph without databases or cloud lock-in.
• 6 Rs Processing Pipeline: Implements a meta-cognitive pipeline (Record, Reduce, Reflect, Reweave, Verify, Rethink) using subagents for fresh context per phase, optimizing LLM attention.
• Automated Quality Hooks: Four hooks (Session Orient, Write Validate, Auto Commit, Session Capture) enforce structure, schema compliance, and state persistence on every write and session.
• Research-Backed Design: Every system component is linked to specific research claims in cognitive science, knowledge management, and agent architecture, ensuring principled design.

Maintenance & Community

The README does not detail specific contributors, sponsorships, or community channels like Discord or Slack. The roadmap indicates "Multi-agent processing" is currently in progress.

Licensing & Compatibility

The project is released under the MIT License, which is permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

The project is at version v0.8.0, suggesting ongoing development. Key features like multi-agent processing are still in progress. The initial setup is token-intensive and requires approximately 20 minutes, which may be a barrier for some users. Specific command-line tools (tree, ripgrep) are mandatory prerequisites.