Choose Plugin or CLI

Archcore has two entry points. Both read and write the same .archcore/ directory, but they meet you at different levels.

The short answer

Using Claude Code or Cursor? Start with the Plugin. You get skills, tracks, built-in agents, and guardrails without extra wiring.
Using any other agent (Copilot, Gemini CLI, OpenCode, Codex CLI, Roo Code, Cline), or scripting in CI? Use the CLI directly.
Just want to understand the model first? Read the shared concepts.

Use Archcore through a higher-level plugin in Claude Code or Cursor. Skills, tracks, built-in agents, and guardrails out of the box.

Use Archcore as a core context layer through the CLI, MCP server, and session hooks. Works with 8 coding agents.

Document types, relations, layout, and why Archcore exists. The shared mental model that both paths build on.

	Plugin	CLI
Install method	Plugin marketplace in Claude Code or Cursor	Single binary via curl / PowerShell
Hosts	Claude Code, Cursor (Copilot + Codex planned)	All 8 supported agents
Interface	Slash commands (`/archcore:adr`, tracks, reviews)	MCP tools + natural language
Multi-step workflows	Tracks orchestrate document chains automatically	You drive them manually through MCP
Guardrails	MCP-only writes, post-mutation validate, cascade warnings	MCP available; guardrails are opt-in via hooks
Best for	Day-to-day use inside Claude Code or Cursor	Scripting, CI, agents without plugin support

Both use the same .archcore/ directory and the same MCP server. Switching paths — or using both at once — requires no migration.

CLI is the compiler. It reads .archcore/, builds the graph, speaks MCP.
Plugin is the runtime. It applies that graph inside Claude Code and Cursor — skills, hooks, workflows.

More depth in Mental Model.