agentleFS
Sign inStart free
LucaDominiciLucaDominici/arbiter/llms.txt

arbiter

llms.txt0 starsChanged 54 days agoApache-2.0View on GitHub

AI governance that installs itself. A parameterized source of truth (AGENTS.md + invariants + hooks + gates + CI) generated across stacks and tools, enforced mechanically at edit/commit/push/CI. Optional orchestration layer (/ship, /drain, sub-agents) drives issues to merged PRs. Zero telemetry. Expert-only. arbiter tells agents WHAT rules apply, then enforces them mechanically — personas drift, gate scripts don't. Read AGENTS.md first; it is the canonical governance file all tools read. Authority hierarchy: AGENTS.md > ADRs > CANON.md > implementation decisions.…

The file

Create a free account to mount
Files
llms.txt
# arbiter

> AI governance that installs itself. A parameterized source of truth (AGENTS.md + invariants + hooks + gates + CI) generated across stacks and tools, enforced mechanically at edit/commit/push/CI. Optional orchestration layer (/ship, /drain, sub-agents) drives issues to merged PRs. Zero telemetry. Expert-only.

arbiter tells agents WHAT rules apply, then enforces them mechanically — personas drift, gate scripts don't. Read AGENTS.md first; it is the canonical governance file all tools read. Authority hierarchy: AGENTS.md > ADRs > CANON.md > implementation decisions. Every task runs the gate before commit. Never commit to main; never add `any` (TS) or `.unwrap()` (Rust); never leave orphan TODOs.

## Core governance

- [AGENTS.md](AGENTS.md): Canonical governance + invariants (INV-NN). Read first.
- [GLOBAL_INVARIANTS.md](GLOBAL_INVARIANTS.md): Hand-maintained invariant mosaic (INV-01..INV-150, catalog max). Authoritative when present.
- [docs/internal/SYSTEM/CANON.md](docs/internal/SYSTEM/CANON.md): Process rules CANON-NN from audit waves #151-#186. Cite on violation.
- [docs/INDEX.md](docs/INDEX.md): Generated index of 257 docs with canonical_id/status/kind. Navigation entry point.
- [.claude/knowledge-map.json](.claude/knowledge-map.json): Machine-readable track routing (frontend/backend/docs).
- [.claude/AGENT_REGISTRY.md](.claude/AGENT_REGISTRY.md): Sub-agents (red-team, bridge-reviewer, context-checker, codebase-scanner, ai-pr-gate) + interaction chains.

## CLI and gates

- [website/reference/cli.md](website/reference/cli.md): Generated CLI reference (all subcommands, exit codes 0/1/2/78, --json envelopes). Source of truth: src/cli.ts.
- [scripts/check-all.mjs](scripts/check-all.mjs): The gate. `node scripts/check-all.mjs L1` (pre-commit, ~2 min) / `L2` (pre-push, ~10 min) / `full` (~35 min).
- [.claude/commands/](.claude/commands/): Slash-command runbooks for agents — [audit.md](.claude/commands/audit.md), [drain.md](.claude/commands/drain.md), [impact.md](.claude/commands/impact.md), [review.md](.claude/commands/review.md), [ship.md](.claude/commands/ship.md), [tabletop.md](.claude/commands/tabletop.md) ((self-only) = arbiter's own dogfooding process, never emitted to a generated project; see scripts/data/self-only-surfaces.json).

## Method and architecture

- [docs/internal/METHOD/PROCESS.md](docs/internal/METHOD/PROCESS.md): Track model, post-commit classification, doc semver, RFC process.
- [docs/internal/METHOD/TESTING.md](docs/internal/METHOD/TESTING.md): Testing method.
- [docs/internal/architecture/ARCHITECTURE.md](docs/internal/architecture/ARCHITECTURE.md): Architecture spine.
- [docs/internal/SYSTEM/HOOK-CONTRACTS.md](docs/internal/SYSTEM/HOOK-CONTRACTS.md): Stable hook env-var/stdin/exit shapes.
- [docs/api/README.md](docs/api/README.md): Public API surface + generated api-extractor snapshots in api/.
- [docs/PRODUCT/PRD.md](docs/PRODUCT/PRD.md): What arbiter is and is not.
- [docs/PRODUCT/GOLD-CONTROLS.md](docs/PRODUCT/GOLD-CONTROLS.md): Gold controls: every family, its tools per stack, how to apply it and its audited status.

## Optional

- [docs/internal/ADR/](docs/internal/ADR/): Architecture decision records. Digest in [docs/internal/SYSTEM/DECISIONS.md](docs/internal/SYSTEM/DECISIONS.md).
- [docs/internal/PRODUCT/STATUS.md](docs/internal/PRODUCT/STATUS.md): Living dashboard. [docs/internal/PRODUCT/GAP.md](docs/internal/PRODUCT/GAP.md): Gap register.
- [docs/internal/runbooks/dependabot-handling.md](docs/internal/runbooks/dependabot-handling.md): Runbook sample (agent-native, gh-CLI steps).
- [.claude/rules/](.claude/rules/): Always-loaded rules (canon enforcement, incidental capture, batch execution, exec protocol).
- [CHANGELOG.md](CHANGELOG.md), [ROADMAP.md](ROADMAP.md), [CONTRIBUTING.md](CONTRIBUTING.md): Standard project metadata.

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.

Check your setup, free