Steadfast Repository Governance
LocalAbout Steadfast Repository Governance
Steadfast is a local-first CLI for developers and coding agents to check repository health and enforce governance policies. It combines read-only verification, audit reports, and optional MCP access without sending repository contents to a Steadfast-hosted service.
Key features
- 01Check repository scope and evidence
- 02Detect documentation and code drift
- 03Validate goals and release readiness
- 04Configure agent guidance and integrations
- 05Generate human-readable audit reports
Use cases
- 01Verify repository changes before review
- 02Audit evidence and documentation consistency
- 03Set up governance for a coding agent
MCP support
Steadfast can start a read-only MCP server for compatible agents. It exposes repository governance and verification capabilities.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMESteadfast
Repository checks and guardrails for coding agents.
Steadfast is a local-first CLI that helps developers and AI agents inspect repository health, enforce change scope, verify evidence, and spot documentation drift. It does not generate code, require an AI API key, or send repository contents to a Steadfast-hosted service.
Quick start
Requires Python 3.10+ and Git.
python -m pip install --upgrade steadfast-governance
cd /path/to/your/repository
steadfast onboard --agent codex
steadfast verify
Replace codex with claude, cursor, or all to configure another supported agent. Review generated files before trusting a repository. Existing files are preserved by default.
For the latest unreleased source, install from Git instead:
python -m pip install --upgrade "git+https://github.com/rllibo833-sudo/Steadfast.git"
The commands most people need
| Command | Purpose |
|---|---|
steadfast onboard | Set up repository guidance and selected agent integration. |
steadfast verify | Run the unified, read-only governance checks. |
steadfast report | Generate a human-readable audit report. |
steadfast audit | Check changed files, evidence, contracts, and documentation drift. |
steadfast discover | Print a machine-readable capability map for tools and agents. |
steadfast mcp-server | Start the read-only MCP server for compatible agents. |
See the quick-start guide for a short walkthrough, the MCP client setup guide to connect compatible AI clients, and the command reference for the full CLI list. For a copy-ready pull-request workflow and an adoption checklist, see Adopt in GitHub Actions.
What it checks—and what it cannot
- Scope: compare changed paths with repository policy.
- Evidence: check required proof files and evidence-bundle integrity.
- Security signals: flag selected risky source patterns and workflow trust-boundary issues.
- Documentation and code truth: detect stale command examples and package-entrypoint drift.
- Goals and releases: validate goal dependencies and evaluate release readiness.
- Agent support: discover skills, inspect MCP configuration, and install repo-local guidance.
These checks are guardrails, not a sandbox or proof of semantic correctness. Results depend on repository policy and configuration; review findings before treating them as release decisions. Steadfast does not silently rewrite plans or execute the configured pipeline as part of verify. Run steadfast pipeline separately only after reviewing its manifest.
Configuration
Run steadfast init to create starter governance files. Repository policy lives in .steadfast/policy.json; pipeline steps are configured in .steadfast/pipeline.json. Treat pipeline manifests as executable configuration and review their commands before running them.
Project links
- Source code and issues
- PyPI package
- Changelog
- Adoption and paid pilots
- Request an evaluation or collaboration
Development
python -m pip install -e .
python -m unittest discover -s tests -v
steadfast docs-check
steadfast code-truth
MIT licensed. See LICENSE.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.