Plugin Overview
Plugin
The Archcore plugin puts your repository’s architecture, rules, and past decisions into your coding agent before it edits a file. It runs on top of the Archcore CLI and reads the same .archcore/ directory.
What you get
Section titled “What you get”- Four slash commands:
/archcore:init,/archcore:plan,/archcore:document, and/archcore:review. Each command does one thing and states its outcome first. - Two built-in agents: a read/write assistant and a read-only auditor.
- Automatic context: the plugin adds the relevant rules and ADRs to the conversation before the agent edits a source file.
- Guardrails: writes to
.archcore/go through MCP only, the CLI validates every mutation, and a cascade check flags dependents that may be stale.
When to use the plugin
Section titled “When to use the plugin”- You work in Claude Code, Cursor, Codex CLI, or GitHub Copilot CLI.
- You want the plugin to work out which documents a request needs, instead of explaining document types to the agent.
- You want context injection and validation to run without being asked.
If you use a different agent, or you want a minimal scriptable install, start with the CLI instead.
Next steps
Section titled “Next steps”- Install the plugin in Claude Code, Cursor, Codex CLI, or GitHub Copilot CLI.
- Supported AI agents lists what each host supports.
- How the plugin works covers hooks, injection, MCP-only writes, and validation.
- Plugin skills documents all four commands.