Cortex TMS CLI
A CLI that scaffolds and validates governance docs for coding agents, such as PATTERNS.md and CLAUDE.md, and flags staleness.
- 180 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The CLI creates and checks documentation files in the project and reads git history; no external services are called.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/cortex-tmsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The CLI creates and checks documentation files in the project and reads git history; no external services are called.
This is third-party code. Review the repository files before installing.
What it does
Cortex TMS creates the set of documents a coding agent relies on and checks that they are in order. `cortex-tms init` lays out templates: `PATTERNS.md` with accepted code patterns, `CLAUDE.md` with agent working rules, `ARCHITECTURE.md` with the system design and `DOMAIN-LOGIC.md` with business rules. Presets exist for Node, Python and Go. `cortex-tms validate` checks that mandatory files exist, that hot files stay under 200 lines, that no placeholders such as `[Project Name]` remain and that finished tasks moved to the archive. Since version 4.0 it compares the doc modification date with commits in the code and warns that a doc is stale. The `archive` command keeps the task list short. `cortex-tms mcp` starts a read-only MCP server that exposes these docs as resources to Claude Desktop, Cursor, Windsurf and other clients.
Who it is for. Developers and teams who work on a project with Claude Code or Copilot and are tired of the agent ignoring agreements and drifting.
Good fit when
- A new repository needs a ready set of `CLAUDE.md`, `PATTERNS.md` and `ARCHITECTURE.md` for Node or Python.
- CI should check that agent docs have not fallen dozens of commits behind the code.
- The task list has grown and finished items should be archived.
- Project rules must be available to Cursor or Claude Desktop as MCP resources.
Not a fit when
- You need a tool that writes code itself: the CLI handles only documents.
- You need a semantic check of meaning: staleness is computed from commit dates.
- The project is tiny and one `README.md` is enough.
Example request
Scaffold a governance doc set for this Python project with cortex-tms init and show what validate reports.Limitations
Staleness detection rests on file modification dates and the number of meaningful commits and cannot see semantic divergence between a doc and code. Templates must be filled in: `cortex-tms validate` complains about leftover placeholders. Rules in `CLAUDE.md` are followed only by an agent that reads the file and are not enforced. Node.js is required and commands run through `npx cortex-tms@latest` or after `npm install -g cortex-tms@latest`. The MCP server exposes documents read-only.
How to disable. Remove the global package with `npm uninstall -g cortex-tms`. The created documents and the `.cortex/` directory in the project are deleted by hand.
FAQ
How do I try it without installing?
Run `npx cortex-tms@latest init`. For regular use install `npm install -g cortex-tms@latest`.
Which documents are created?
`PATTERNS.md`, `CLAUDE.md`, `ARCHITECTURE.md` and `DOMAIN-LOGIC.md`, plus `NEXT-TASKS.md` and `copilot-instructions.md`. The set depends on `--scope` and the preset.
How do I validate docs in CI?
Run `cortex-tms validate --strict`: warnings become errors. The `--fix` flag restores missing files.
How does the staleness check work?
Doc modification dates are compared with meaningful code commits. If a doc lags, `validate` warns with the number of days and commits.
How do I connect the MCP server?
Run `cortex-tms mcp --print-config`: it prints ready config snippets for clients. The server itself starts with `cortex-tms mcp`.
Discussion
No questions or reviews yet.
Related
CLI toolsTasks, docs, specs and decisions live as markdown files in your repository, and the agent reads them through the CLI and MCP
Ctxo builds a repository dependency graph and gives an agent the blast radius of a change, commit history and code health over MCP
A Claude Code working configuration: a planner with per-milestone review, sub-agents and skills for analysis, refactoring and docs sync
An MCP server in one binary: AI reads and edits files across several projects and runs your commands
Git-native project context for agents: specs, decisions and rules live in Markdown beside the code and are read before edits
An MCP server that gives Claude Desktop and other clients file tools, a shell and sub-agents in the style of Claude Code
README in short
The project builds a document governance discipline for agents: consistency, freshness and human oversight. It describes the `init`, `validate`, `status`, `archive` and `mcp` commands, document templates and the limits of staleness detection. Version 4.0 added git-based staleness detection.