Governed AI delivery

Turn AI‑assisted delivery into a governed local workflow — in any repo, any IDE.

{ai} engineering installs a deterministic governance layer into any repository — specs, decisions, skills, agents, hooks, and an audit trail, all as versioned local files. No hosted control plane. No provider lock-in.

Install in a minuteView on GitHub

54 skills · 9 agents · 6 surfaces · 1 governed flow

ai-eng — zsh
uv-first · under a minute
$ uv tool install ai-engineering   # install the CLI
$ ai-eng install .                 # add governance to your repo
$ ai-eng doctor
  [PASS] hooks · mirrors · manifest · required tools
$

See it run

From install to a governed session

One command installs the CLI; ai-eng doctor reports a[PASS] and you are working under the governed flow.

demo — ai-eng
Static by default — click to play. Honours reduced-motion preferences.

Not a prompt pack. A governed delivery system.

The governed flow

One chain, from intent to merge

You drive the intent and approve each step; the gates catch the rest — no secrets, broken docs, or untested changes reach a merge.

  1. /ai-brainstorm
  2. /ai-plan
  3. /ai-build
  4. /ai-pr

/ai-build swaps to /ai-autopilot for large multi-concern specs.

the governed workflow
The governed workflow: /ai-brainstorm to /ai-plan to /ai-build or /ai-autopilot to /ai-pr, with you approving each step and automatic checks — clean diff, tests, docs, review — gating the merge

Why it holds

Governance that is enforced, not suggested

  • Ship a whole spec in one run

    /ai-autopilot decomposes it, builds a dependency DAG, runs parallel waves, and converges on a reviewed PR.

  • What you approved is what shipped

    a brainstorm hard-gate plus a spec-lifecycle state machine keep every change anchored to the approved spec (Rung 2 SDD — spec and code stay in sync, not just spec-first that drifts).

  • An audit trail you own

    every AI action lands in a hash-chained NDJSON log you can verify offline, with no telemetry by default.

  • Every bypass has an owner and an expiry

    no # noqa or @ts-ignore; findings are refactored or formally risk-accepted with a severity-based TTL.

  • Every tool call is screened before it runs

    a deterministic guard checks each edit, write, and shell command and stops risky ones.

  • AI quality is a tested property

    skills are measured with pass@k, and a regression beyond five points blocks the pull request.

The toolkit

Fifty-four skills and nine agents, one flow

  • 54skills
  • 9agents
  • 6surfaces
  • 1governed flow
skills · agents — taxonomy
Toolkit taxonomy: 54 skills and 9 agents grouped by activity — plan and build, ship safely, design and docs, research and learn — across 6 IDE surfaces

Same flow, every surface

  • Claude Code
  • GitHub Copilot
  • Codex
  • Antigravity
  • OpenCode
  • Cursor

Why governance

A deterministic plane that gates the probabilistic one

LLMs are fast and fallible. {ai}engineering puts a deterministic layer underneath them — hooks, gates, a hash-chained audit log, and a spec-lifecycle state machine — so the creative work stays probabilistic while everything that reaches a merge is checked, owned, and reproducible.

The judgment is yours; the guarantees are the framework's.

  • PyPI version
  • Python 3.11+
  • CI
  • Quality Gate
  • Coverage
  • Snyk security
  • License: MIT
contributors
Contributors to arcasilesgroup/ai-engineering
star history
GitHub star history for arcasilesgroup/ai-engineering

Start now

Install once. Approve each step. Ship governed.

one command
uv-first
$ uv tool install ai-engineering

Then open your editor and type /ai-start. Prefer to ease in? Start in observe mode and enforce only what proves useful. Update any time with ai-eng update.

Get it on PyPIRead the docs