agentleFS
lukasrepubliclukasrepublic/agentic-foundry/llms.txt

agentic-foundry

llms.txt1 starsChanged 3 months agoMITView on GitHub

Governed, spec-driven delivery for agent-built software, shipped as a Claude Code plugin. The loop: spec → operator authorization → governed build in an isolated worktree → an honestly-tiered merge floor (the platform's own enforcement) → certification against the real running app → human sign-off. Gates make problems visible, not impossible; the operator's judgment is terminal by design. Key facts for agents: the core loop is six verbs (intake, spec-review, authorize, dispatch, certify-local, release accept); specs are atomic (stable AC-IDs, ≤14…

The file

Files
llms.txt
# Agentic Foundry

> Governed, spec-driven delivery for agent-built software, shipped as a Claude Code plugin.
> The loop: spec → operator authorization → governed build in an isolated worktree → an
> honestly-tiered merge floor (the platform's own enforcement) → certification against the
> real running app → human sign-off. Gates make problems visible, not impossible; the
> operator's judgment is terminal by design.

Key facts for agents: the core loop is six verbs (intake, spec-review, authorize, dispatch,
certify-local, release accept); specs are atomic (stable AC-IDs, ≤14 ACs / 8,000 words, no
override); the acceptance contract is frozen (content-hashed, operator-signed) at
authorization and scope-widening requires re-authorization; certification refuses rather
than passing vacuously; there is no unattended-merge mode.

## Docs

- [README](README.md): the loop in one picture, quickstart, task→verb table, honest comparison
- [Docs home](docs/README.md): the four-quadrant index (tutorials / how-to / reference / explanation)
- [Quickstart](docs/QUICKSTART.md): zero to a first governed merge, artifact by artifact
- [Design & trust model](docs/DESIGN.md): what the machine enforces, what stays human, what is deliberately not guaranteed
- [Merge floor](docs/merge-floor.md): the tier model (A server-enforced / B labeled-advisory) and exact hook semantics
- [Architecture](docs/architecture.md): workspace (WHAT) ⟷ factory (HOW) and the wiring contract
- [Verb reference](docs/VERBS-QUICK-REF.md): every /foundry:* verb, one line each
- [Glossary](docs/glossary.md): atom, contract, floor, journey, operator…
- [Comparison](docs/comparison.md): vs SDD tools / review bots / agent platforms, including when NOT to use this
- [Troubleshooting](docs/troubleshooting.md): symptom-first runbooks
- [FAQ](docs/faq.md)

## How-to

- [Recover from a failed gate](docs/how-to/recover-from-a-failed-gate.md)
- [Adopt on an existing codebase](docs/how-to/adopt-on-an-existing-codebase.md)
- [Team review with CODEOWNERS](docs/how-to/team-review-with-codeowners.md)
- [Multi-repo control plane](docs/how-to/multi-repo-control-plane.md)
- [Cut a release](docs/how-to/cut-a-release.md)
- [Migrate from Spec Kit](docs/how-to/migrate-from-spec-kit.md)
- [Migrate from OpenSpec](docs/how-to/migrate-from-openspec.md)

## Optional

- [Terminology standard](docs/TERMINOLOGY.md): the naming standard skills/agents conform to
- [Identity isolation](docs/identity-isolation.md)
- [Changelog](CHANGELOG.md)
- [Contributing](CONTRIBUTING.md) · [Security policy](SECURITY.md) · [Agents guide](AGENTS.md)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.