Fhe auditor
Skill Yue-Zhou1/zkcrypto-audit/plugins/post-quantum-auditors/skills/fhe-auditor
Audit FHE implementations for noise-budget accounting, bootstrapping correctness, modulus-switching safety, plaintext leakage, and key-switch parameter integrity.From its SKILL.md
npx -y skills add Yue-Zhou1/zkcrypto-audit --skill fhe-auditorAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.4 KB, 446 tokens by cl100k_base, as published. Nobody here has run it
fhe-auditor
Domain auditor for fully homomorphic encryption pipelines and transform safety.
When to Use
- Auditing ciphertext lifecycle across FHE transforms
- Reviewing noise growth calculations and claimed depth limits
- Checking bootstrapping and key-switch correctness assumptions
- Verifying modulus switching and plaintext/ciphertext boundary handling
When NOT to Use
- Classical encryption reviews without homomorphic transforms
- Parameter-level lattice audits that do not involve FHE lifecycle behavior
- Confirming suspected FHE issues without verification gates
Core Review Areas
- Ciphertext modulus transitions and switching semantics
- Noise growth tracking and budget accounting
- Bootstrapping parameter validity and correctness
- Plaintext/ciphertext boundary and slot-packing isolation
- Key-switch material provenance and application correctness
Workflow
Phase 1: Ciphertext lifecycle mapping
- Read
references/fhe-checklist.md - Execute
workflows/noise-budget-review.md - Map each ciphertext transform from encryption to final decrypt/consume step
Phase 2: Noise and transform review
- Validate noise-budget calculations at each transform stage
- Confirm bootstrapping refresh logic preserves security invariants
- Verify modulus and key-switch operations do not silently degrade correctness
Phase 3: Pattern hunt
- Read
references/finding-patterns.md - Prioritize exhausted noise budgets, precision drops, and leakage boundaries
Phase 4: Handoff
- Send surviving findings to
crypto-fp-check - Use
zkbugs-indexonly after the finding survives verification
Output Contract
Produce an FHE-specific handoff that includes:
- The ciphertext lifecycle stage and transform under review
- The measured/claimed noise budget and correctness invariant at risk
- Whether the issue is noise, bootstrapping, modulus switch, key-switch, or leakage related
- The next verification or reporting route
Reference Index
What ships with it: 4 files
3.4 KB alongside SKILL.md
agents/
- openai.yaml426 B
references/
- fhe-checklist.md1.2 KB
- finding-patterns.md659 B
workflows/
- noise-budget-review.md1.1 KB