GUARD2–4 weeks total · fixed scope
Your AI surface lacks a real threat model
security that scales with your internal-tool surface.
The problem
You've shipped internal tools and tool servers faster than you've threat-modelled them. Agents hold credentials, reach production, and consume untrusted input from pages and APIs. Nobody has written down what the blast radius actually is.
What ships
- 01
internal-tool + MCP server hardening
- 02
CI/CD supply-chain review
- 03
secret management and rotation playbooks
- 04
handoff: remediation PRs plus reviewer checklist you own
Acceptance criteria
- A written threat model for the AI surface, with findings ranked by blast radius.
- Remediation delivered as pull requests your team reviews and merges.
- A reviewer checklist your engineers own and can apply without us.
- Secret management and rotation playbooks in your runbook, not ours.
Engagement shape
scope call → audit (2 wk, fixed) → remediation sprint (optional)
Ideal for
Teams shipping AI features whose security review is still ad-hoc.
Quoted per team, surface, system or pipeline. No hourly billing, no open-ended retainer.