Archcore plugin
Git-native project context for agents: specs, decisions and rules live in Markdown beside the code and are read before edits
- 75 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- The plugin installs hooks that fire at session start and around tool calls
- The installer is downloaded and run through `curl | bash`
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/archcoreDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The plugin installs hooks that fire at session start and around tool calls
- The installer is downloaded and run through `curl | bash`
This is third-party code. Review the repository files before installing.
What it does
Archcore keeps specs, decisions, rules and plans in Git next to the code, and the agent reads them before changing code so a change fits the architecture and past decisions. Without that an agent does not know what a feature must do, where code belongs and which rules already apply, so it guesses and you explain the same things every session. Documents are plain Markdown in `.archcore/`, with the type in the filename, for example `rate-limiting.spec.md`. Requests are simple: "plan rate limiting for the public API", "record why we chose Redis for the token bucket", "document how the billing module works". The agent picks the document type and asks only what it cannot find in the repo. Three commands cover the loop: `/archcore:plan` before work makes a spec, examples and tasks sized to the change, `/archcore:document` turns a sentence into a finished linked document, and `/archcore:review` before merge checks the branch against the documents and shows which side is wrong. `/archcore:init` reads the repo and proposes a first document set, and `/archcore:init import` turns `CLAUDE.md`, `AGENTS.md` and rule files into typed documents. The installer adds the CLI and plugins for Claude Code, Codex CLI, GitHub Copilot CLI and OpenCode, and `archcore init` connects chosen agents to the project. A document change shows as a diff in a pull request.
Who it is for. Teams where agents write code in a shared repo and forget architecture decisions between sessions, with project rules scattered across `CLAUDE.md`, `AGENTS.md` and chats.
Good fit when
- A big feature needs a spec with tasks that the agent reads before working
- A decision such as choosing Redis should stay recorded with its reason
- Old `CLAUDE.md` and `AGENTS.md` files should become typed documents
- Before merge a branch is checked against accepted rules and specs
- Module docs should come from one sentence and live in Git
Not a fit when
- The project is small and one rules file is enough
- You need to keep knowledge outside the repo or in a cloud wiki
- The team is not ready to keep documents in Git with the code
Example request
Plan rate limiting for the public API: a spec, examples and tasks, then record the Redis decision.Limitations
The installer runs through `curl | bash` on macOS, Linux and WSL and through `irm | iex` on Windows, so read the script before running it. Plugins install automatically for the Claude Code, Codex CLI, GitHub Copilot CLI and OpenCode it finds, while `archcore init` connects Gemini CLI, Roo Code and Cline. Cursor needs one extra step. Documents live in `.archcore/` inside the repo, and privacy details are on the project site. The plugin hooks fire at session start and before and after tool calls.
How to disable. Remove the plugin from the agent (for example `/plugin uninstall archcore`), the `archcore` binary and, if you wish, the `.archcore/` directory in the repo. The documents are plain Markdown and can stay.
FAQ
How do I install Archcore?
With `curl -fsSL https://archcore.ai/install.sh | bash` on macOS, Linux and WSL or `irm https://archcore.ai/install.ps1 | iex` in PowerShell on Windows.
Which agents are supported?
Claude Code, Codex CLI and GitHub Copilot CLI get the plugin from the installer, OpenCode installs through `archcore init`, Gemini CLI, Roo Code and Cline connect through it too, and Cursor needs one extra step.
What do the three main commands do?
`/archcore:plan` prepares a spec and tasks, `/archcore:document` turns a sentence into a document and `/archcore:review` checks the branch against documents.
Where are documents stored?
In the `.archcore/` directory of the repo as Markdown, versioned with the code, and changes show in a pull request.
How do I migrate existing rules?
Run `/archcore:init import` in the agent: Archcore turns `CLAUDE.md`, `AGENTS.md` and rule files into typed documents.
Discussion
No questions or reviews yet.
Related
PluginsArchgate turns architecture decisions into executable rules and blocks violations in CI, hooks and AI agent work
A Claude Code working configuration: a planner with per-milestone review, sub-agents and skills for analysis, refactoring and docs sync
Tasks, docs, specs and decisions live as markdown files in your repository, and the agent reads them through the CLI and MCP
A Claude Code plugin for project architecture: ADRs, multi-role reviews, findings turned into plans, a pragmatic mode.
A CLI that analyses a repository in six model phases and writes AGENTS.md or CLAUDE.md, with ExecPlan planning on top.
A .gaai/ folder that splits agent work into discovery and delivery: a story with criteria, then isolated implementation and QA.
README in short
Covers install and project connection, the three commands, the `.archcore/` structure, supported agents, privacy, a comparison with other approaches and doc links.