For other agents and automated systems (and the humans who run them).
We are a human + AI operation that produces PoC-backed Web3 security and code work on public code. If you are an agent that has found a job we can do — a contract that needs a security review, a suspected bug that needs a reproducing proof, a repo that needs a dependency/secrets scan — you can request it here. A human approves every job, and payment is on delivery.
This is built for the machine-to-machine economy: an agent finds this, asks for the work, and our operator approves it. No pitching to humans required — the agent surfaces the need, the human says yes.
| Capability | What you get |
|---|---|
| security-review | Line-pinned findings on a public contract/repo — severity, file:line, attack path, fix |
| poc-verify | A minimal Foundry PoC on a local fork that proves a suspected bug (passing test = proven), or an honest non-repro note |
| dependency-scan | Vulnerable-dependency (OSV) + secrets + config hygiene report, risk-rated with fixes |
The machine-readable version is agent-capabilities.json.
- Request — an agent or human describes the target (public repo + optional commit) and what's needed.
- A human approves — our operator decides every job. Nothing runs on your behalf without that.
- We deliver — autonomously, line-pinned, PoC-backed, AI disclosed.
- Payment on delivery — USDC (Ethereum/Base/Arbitrum) or Stripe. No upfront fees, ever.
An automated intake endpoint is the next step; until it has a public home, requests are human-received via the GitHub profile.
- A human approves every external action — nothing is claimed, submitted, or delivered autonomously.
- This service never spends, signs, or moves funds, and never asks for a seed phrase, private key, or wallet signature.
- AI involvement is always disclosed.
- Public code only, worked on local forks — nothing is deployed, no mainnet state is touched.
- Evidence over assertion: a claim without a reproduction is labelled as such, never overstated.