
AI-assisted engineering
Agents, skills and hooksSkills and hooksSkills are written instructions an AI coding agent follows for a given kind of task. Hooks are automatic checks that run before or after the agent acts, so AI-written code stays reviewable and consistent. are how I build every day, with human review on everything that ships. I can set that up for your team.
What it covers
- Skills and hooks and project rules that keep AI-written code reviewable and consistent.
- Pull-request watchers, monitors and status helpers that take the routine out of release work.
- Release gates: nothing merges without review and passing checks.
- Coaching non-engineers who write code with AI help, with ownership files and rules that keep them safe.
- Local models for work where data must stay in-house, benchmarked on your tasks.
- A worked example: fleet-watch runs a small model locally on its hub to give a first read of each security incident, with the client choosing where the hub runs and the evidence staying where they decide. It suggests; a person decides.
Agents with real access, safely
An AI agent only does useful work when it has real access to code, servers and data. So the access has to be narrow, short-lived, logged and approved by a person. In July, AI models under evaluation at OpenAI escaped a light sandboxSandboxA restricted environment in which a program or AI agent runs, so that a mistake or misuse cannot reach the rest of the system. How much it holds depends on how tightly it is built.Learn more → through an unpublished flaw in a package-registry proxy and reached Hugging Face's systems. After that I went well beyond a light sandbox for my own agents:
Egress through an allow-list
Agents reach the network only through a proxy with a list of permitted destinations. Everything else is refused and logged.
SSH certificates per task
Access to a server is a certificate minted for one task, restricted to one source address and short-lived.
SQL through a controlled integration
Database queries go through a controlled integration with its own limits. The agent never holds a database password.
A secrets runner
A command that needs a secret gets it injected at run time. The value is scrubbed from everything the agent can read.
A guard the agent cannot edit
The rules live in a guard owned by root, outside the agent's reach. It cannot rewrite its own limits.
A person approves
Risky actions stop at a deterministic approval step: the same request always asks the same question, and a person answers it.
The background is on the Security page.
How it runs
Look at how you build today
Where review is slow, where code lands unreviewed.
Set the rules in the repository
Skills, hooks and owners files, versioned like code.
Automate the routine, keep the judgement
Agents prepare; people decide.
Check what changed
Review time, defects and surprises, before and after.
A good fit when
Your developers already use AI assistants, and you want the speed without the unreviewed risk.
Hands-on training: your developers at AI speed, with the guardrails in place.
Tell me how your team works today; I will tell you what I would set up first.
Get in touch