Hw Security Signoff

HW Security Sign-Off

Purpose

Define the handoff contract between Foundry (builder) and Forge (security auditor) for hardware security sign-off. Ensure all security-critical design artifacts are delivered, reviewed, and approved before tape-out commitment.

Scope Constraints

Coordinates the handoff process between builder and auditor roles. Does not perform the security review itself (delegates to rtl-security-review, microarch-analysis, physical-design-security). Does not modify design files.

Inputs

• Design name and tape-out target date
• Security-critical modules identified by Foundry
• Trust boundary definitions
• Threat model (if available) or threat categories in scope
• Any prior security review findings

Input Sanitization

No user-provided values are used in commands or file paths. All inputs are treated as read-only analysis targets.

Procedure

Progress Checklist

• [ ] Step 1: Verify builder artifact delivery
• [ ] Step 2: Validate security scope agreement
• [ ] Step 3: Coordinate security reviews
• [ ] Step 4: Track finding resolution
• [ ] Step 5: Issue sign-off decision

Step 1: Verify Builder Artifact Delivery

Foundry must deliver the following before security review begins:

• [ ] RTL source for all security-critical modules (final, synthesis-ready)
• [ ] Security-critical module list with trust boundary annotations
• [ ] Register map with access control policy per register
• [ ] FSM state diagrams for security-relevant state machines
• [ ] Clock domain crossing report
• [ ] DFT/scan chain documentation (which chains touch security logic)
• [ ] Debug interface specification (JTAG, trace ports)
• [ ] Known limitations or accepted risks documented

Reject handoff if any artifact is missing. Document gaps and return to Foundry.

Step 2: Validate Security Scope Agreement

• Confirm threat categories in scope (side-channel, fault injection, logical bypass, debug access).
• Agree on security-critical module boundaries (which modules are in-scope for full review).
• Define review depth per module: full RTL review, interface-only, or documentation review.
• Set severity classification: critical (blocks tape-out), high (requires mitigation plan), medium (accept with justification), low (track).
• Confirm timeline and review capacity.

Step 3: Coordinate Security Reviews

Dispatch to Forge specialist skills based on scope:

• rtl-security-review: Access control gates, FSM transitions, information leakage for all in-scope modules.
• microarch-analysis: Speculative execution, cache timing, branch predictor attacks if CPU/core is in scope.
• physical-design-security: Power domain isolation, clock domain security, layout-level leakage if physical design is available.

Track review progress per module and per skill.

Step 4: Track Finding Resolution

• Maintain finding tracker with: ID, module, category, severity, status, owner, target date.
• For each critical finding: Foundry implements fix, Forge re-reviews.
• For each high finding: Foundry provides mitigation plan, Forge approves plan.
• For each medium finding: Foundry documents accepted risk with justification.
• Verify all critical findings are resolved before sign-off.

Step 5: Issue Sign-Off Decision

• Approved: All critical/high findings resolved, no open blockers.
• Conditional approval: All critical resolved, high findings have approved mitigation plans with committed timelines.
• Blocked: Open critical findings or unresolved high findings without mitigation plans.

Document decision with: reviewer, date, scope covered, open items (if conditional), and next review trigger.

> Compaction resilience: If context was lost, re-read the Inputs section for the design under review, check the Progress Checklist, then resume from the earliest incomplete step.

Output Format

Sign-Off Summary

| Field | Value | |-------|-------| | Design | ... | | Reviewer | Forge | | Date | ... | | Decision | Approved / Conditional / Blocked | | Scope | Modules A, B, C | | Open items | ... |

Finding Tracker

| ID | Module | Category | Severity | Status | Owner | |----|--------|----------|----------|--------|-------| | F1 | accessctrl | Bypass | Critical | Fixed | Foundry | | F2 | debugif | Leakage | High | Mitigated | Foundry |

Handoff

• Hand off to forge/rtl-security-review for detailed RTL security analysis.
• Hand off to forge/microarch-analysis for microarchitectural attack surface review.
• Hand off to foundry/chip-design-flow for design fixes required by findings.

Quality Checks

• [ ] All builder artifacts delivered and complete
• [ ] Security scope and threat categories agreed
• [ ] All security-critical modules reviewed by appropriate Forge skill
• [ ] All critical findings resolved
• [ ] All high findings resolved or have approved mitigation plans
• [ ] Sign-off decision documented with scope and conditions

Evolution Notes

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: dtsong
• Source: dtsong/my-claude-setup
• License: MIT

Install and usage instructions live in the source repository linked above.