Ars Contexta
A Claude Code plugin that builds a knowledge system of markdown files from a conversation with you: folder structure, note templates, a processing pipeline and
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Hooks write files and auto-commit to git on every write
- Generation reads and creates many files in the project
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/ars-contextaDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Run one by one in the Claude Code chat
/plugin marketplace add agenticnotetaking/arscontexta
/plugin install arscontexta@agenticnotetakingInstall the plugin: /plugin marketplace add agenticnotetaking/arscontexta, then /plugin install arscontexta@agenticnotetaking. Restart Claude Code and run /arscontexta:setup, answer the domain questions, and restart once more after generation.
Other ways from the author
/plugin marketplace add agenticnotetaking/arscontexta
/plugin install arscontexta@agenticnotetakingAfter installing, restart Claude Code and run /arscontexta:setup.
This is third-party code. Review the repository files before installing.
What it does
The plugin holds a conversation about how you think and work, and from your answers it builds a knowledge system for your domain: folders, context files, note templates with a schema, a processing pipeline and hooks. Everything lives in plain markdown files linked by wiki links, with no database and no cloud. Installation starts with a setup command that runs six phases from understanding your domain to generation and validation; it takes about twenty minutes and is token-heavy, but is done once. After generation you get pipeline commands: extracting insights from sources, finding connections, updating older notes, and quality and schema checks. Hooks enforce structure on every write, spot maintenance needs and auto-commit.
Who it is for. For people who keep a knowledge base or second brain and want the agent to remember context across sessions.
Good fit when
- You need a personal markdown note system shaped around how you work
- You want the agent not to start every session from a blank slate
- You need a pipeline that processes sources, finds connections and maintains notes
Not a fit when
- You need a ready knowledge base right now without spending time on setup
- You work outside Claude Code, in an agent without plugin and hook support
- You are not comfortable with generation being token-heavy
Example request
Set up a knowledge system for my research: I will describe how I work, and you build the structure and templatesLimitations
The plugin targets Claude Code: it relies on that environment's commands, skills and hooks. Initial setup takes about twenty minutes and is noticeably token-heavy, because the model reads research material and generates many files. Hooks auto-commit and validate writes, so the system writes to project files and to git. After generation you need another restart to activate the hooks and skills.
How to disable. Remove the arscontexta plugin via /plugin. The generated markdown files stay with you and can be kept or deleted by hand.
Security check
- Hooks write files and auto-commit to git on every write
- Generation reads and creates many files in the project
README in short
The README presents Ars Contexta as a Claude Code plugin that builds a knowledge system for the user through conversation rather than a template. Installation adds the marketplace and the plugin, then a setup command runs six phases to understand the domain, derive a configuration and generate folders, context files, templates, skills, hooks and a manual. The architecture splits content into three spaces: the agent's persistent memory, the note graph and operational coordination. The pipeline is built around six processing steps, each running in its own context via subagents. MIT licensed.
FAQ
Where is the data stored?
In plain markdown files on your disk, linked by wiki links. There is no database, no cloud and no service lock-in.
How is this different from a ready template?
The system is not taken from a template but derived from a conversation about your domain: the structure and decisions are chosen for how you work.
Related
A self-hosted knowledge base with block-level references and a built-in MCP server for connecting AI agents to your notes
A CLI for every Google Workspace API with JSON output and agent skills: Drive, Gmail, Calendar, Sheets and more
Local search over Markdown notes, docs and meeting transcripts: keywords, semantic search and reranking, with an MCP server
A task manager for AI-driven development: breaks a PRD into dependent tasks and guides the agent through them via MCP or CLI