agentsclimarketplace
Sign in

Pqc signature auditor

Skill Yue-Zhou1/zkcrypto-audit/plugins/post-quantum-auditors/skills/pqc-signature-auditor

Audit post-quantum signature implementations — ML-DSA (FIPS 204), SLH-DSA (FIPS 205), FN-DSA/Falcon (pending standardization), and stateful hash signatures XMSS/LMS/HSS (NIST SP 800-208) — for rejection-sampling correctness, hedged/deterministic signing modes, verification bound enforcement, and one-time-signature state management. Use when reviewing PQ signing, verification, or OTS index/state persistence.From its SKILL.md

Install
npx -y skills add Yue-Zhou1/zkcrypto-audit --skill pqc-signature-auditor

Assembled 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

4.1 KB, 950 tokens by cl100k_base, as published. Nobody here has run it

pqc-signature-auditor

Domain auditor for standardized post-quantum signatures, in two families with different failure modes: stateless schemes (ML-DSA, SLH-DSA, FN-DSA) where the risks are sampling and bound enforcement, and stateful hash-based schemes (XMSS/LMS/HSS) where the dominant risk is one-time-key state reuse.

Do not fold ML-KEM/Kyber findings into this skill — KEM review belongs to pqc-kem-auditor.

When to Use

  • Auditing ML-DSA (FIPS 204) signing/verification: rejection sampling, hint computation, deterministic vs hedged modes
  • Auditing SLH-DSA (FIPS 205): FORS/WOTS+ structure, addressing, and randomizer handling
  • Auditing FN-DSA/Falcon implementations (with its standardization status pinned in source notes — FIPS 206 not yet final)
  • Auditing XMSS/LMS/HSS under NIST SP 800-208: OTS index monotonicity, crash recovery, backup/restore, cloning, and hardware-binding requirements

When NOT to Use

  • ML-KEM/Kyber or any KEM decapsulation review -> pqc-kem-auditor
  • Classical signatures (ECDSA, Schnorr, EdDSA, RSA) -> signature-scheme-auditor
  • Generic lattice parameter/noise design for custom schemes -> lattice-auditor
  • Timing measurement methodology for samplers -> side-channel-auditor

Core Review Areas

  1. Parameter-set and version provenance: FIPS 204/205 final vs round-3 Dilithium/SPHINCS+ differences; Falcon flagged as pre-standard
  2. ML-DSA signing: rejection-sampling loop exactness (norm checks on z, r0, hint count), iteration bounds, and no leakage of rejected candidates
  3. Signing modes: deterministic vs hedged (rnd) handling per FIPS 204; SLH-DSA opt_rand per FIPS 205; fault posture of deterministic modes
  4. Verification bounds: every norm/weight/count bound checked (z bound, hint weight, FORS/WOTS checksum rules) — skipping any bound admits forgery
  5. Stateful OTS index management (XMSS/LMS/HSS): monotonic persistent counter committed BEFORE signature release, crash recovery, backups, replication, and SP 800-208 hardware-binding requirements
  6. Encoding/serialization: canonical encodings, length checks, context strings (FIPS 204/205 ctx parameter) bound correctly

Workflow

Phase 1: Family and version mapping

  • Read references/pqc-signature-checklist.md
  • Pin the scheme, parameter set, and spec version; queue spec-delta-checker for conformance claims

Phase 2: Family-specific review

  • Stateless (ML-DSA, SLH-DSA, FN-DSA): execute workflows/pqc-signature-review.md
  • Stateful (XMSS/LMS/HSS): execute workflows/stateful-hash-signature-review.md

Phase 3: Pattern hunt

  • Read references/finding-patterns.md
  • Prioritize skipped verification bounds, rejection-loop shortcuts, and OTS index rollback paths

Phase 4: Handoff

  • Send surviving findings to crypto-fp-check
  • Cross-route sampler leakage to side-channel-auditor, RNG lifecycle to randomness-auditor, KEM questions to pqc-kem-auditor

Output Contract

Produce a PQ-signature handoff that includes:

  • signature_family_and_parameter_set
  • sign_or_verify_path
  • state_or_sampling_invariant
  • evidence
  • disposition (one of verified, false_positive, unverified, observation, residual_risk)
  • next_route

Reference Index

What ships with it: 6 files

13.0 KB alongside SKILL.md

agents/

references/

workflows/

Keep looking

Skills are one crate of 326,852. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.