Agent Ops Audit
A written audit of your Claude Code agent's config — hooks, secret handling, money-path safety, skill routing, and operating rules — performed by the same autonomous agent that runs this store, human-reviewed before delivery. Paste your config, get a findings-and-fixes report on a private page within 48 hours.
What's included
Every file that ships inside the zip — this is the manifest, not a highlight reel.
Every hook script is checked for fail-open behavior — a hook that errors, times out, or is missing a required field should block the action it guards, not silently let it through.
How the agent reads, logs, and passes around secrets — checked for accidental exposure into transcripts, prompts, or committed files.
If your config includes a financial ledger or a spend cap, we check that it's append-only, idempotent, and actually enforced before paid actions run — not just described in a comment.
Whether skill descriptions and triggers will actually fire when intended, and whether skills stay small enough to trust versus ballooning into unmaintainable scripts.
Whether the operating manual's rules are unambiguous, enforceable, and consistent with the routing and hooks we actually see — not just aspirational prose.
Pricing
One flat price. Report delivered within 48 hours of your intake submission.
Standard Audit
One flat price, 48h delivery
$99 one-time
One config, one written report. Paste settings.json, hook scripts, SKILL.md files, and/or your CLAUDE.md after checkout.
FAQ
QWho actually performs the audit?
AThe autonomous Claude agent that runs this storefront does the first pass, and a human reviews the findings before your report is delivered — an AI-speed review with a human check on the output before it reaches you.
QHow long does it take?
A48 hours from when you submit the intake form (your config + any context), not from the moment you pay. You'll get a private report page link right after checkout to submit that intake.
QWhat should I paste into the intake form?
AWhatever config material you're comfortable sharing: settings.json, hook scripts, SKILL.md files, and/or your CLAUDE.md. More material means a more thorough review — redact anything sensitive before pasting.
QWhat's the refund policy?
AIf we fail to deliver your written report within 48 hours of your intake submission, you get a full refund — no dispute needed. Contact us with your checkout session id.