agentleFS
ProvableHQProvableHQ/snarkVM/.claude/skills/snarkvm-review/SKILL.md

snarkvm-review

Skill1.2k starsChanged 21 days agoApache-2.0View on GitHub

Security-focused PR review for snarkVM codebase. WHEN: User says "review PR", "audit PR", "security review", "check PR changes", or wants thorough analysis of PR changes for bugs/vulnerabilities. WHEN NOT: Fixing review feedback (use snarkvm-fix pr), fetching context only (use snarkvm-github), or fixing issues (use snarkvm-fix).

Trending

More talk lately.about the whole repository

Dec 2025: 2 mentionsJan 2026: 2 mentionsFeb 2026: 1 mentionMar 2026: 1 mentionApr 2026: 1 mentionMay 2026: 3 mentionsJun 2026: 4 mentionsJul 2026: 3 mentionsAug 2026: 7 mentionsSep 2026: 7 mentionsOct 2026: 8 mentionsNov 2026: 0 mentions
Dec 2025Jun 2026Nov 2026
Mentions a month of the ProvableHQ/snarkVM repository that our weekly crawl found. There is no adoption data for this file yet.

Who's talking about ProvableHQ/snarkVM

The biggest 5 of 50 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 50

Talked about on GitHub (50 times).

On GitHub · 50

45 more on GitHub

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 1 is distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Update $WS/state-pr-$PR.md with risk assessment.

    line 54
  2. Distinctive:

    Before diving into code:

    line 60
  3. Very specific:

    Read PR description: jq -r .body "$WS/context-pr-$PR.json"

    line 61
  4. Very specific:

    Check linked issues: cat "$WS/linked-issues-pr-$PR.txt"

    line 62
  5. Very specific:

    Update $WS/state-pr-$PR.md with findings:

    line 80
  6. Very specific:

    If requesting changes, create handoff for /snarkvm-fix pr:

    line 91

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Security-Focused PR Review

Mindset: Assume there is a bug. Find it. Ultrathink.

This skill runs in a forked context to keep exploration out of your main conversation.

Usage

/snarkvm-review <pr_number>

Setup

PR=$ARGUMENTS
WS=".claude/workspace"
SKILL_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/../snarkvm-github" && pwd)"

1. Load Context

if [[ ! -f "$WS/state-pr-$PR.md" ]]; then
  "$SKILL_DIR/scripts/fetch-pr.sh" "$PR"
fi

Read: $WS/state-pr-$PR.md, $WS/files-pr-$PR.txt

2. Triage by Risk

Categorize changed files by risk:

  • CRITICAL: algorithms/ (crypto primitives), circuit/ (constraint generation), synthesizer/src/vm/ (consensus logic)
  • HIGH: console/ (types, serialization), synthesizer/process/ (execution), ledger/ (state management)
  • MEDIUM: synthesizer/program/ (instructions, logic), fields/, curves/
  • LOW: utilities/, test files, docs, CI config

Update $WS/state-pr-$PR.md with risk assessment.

For large PRs (30+ files): Focus CRITICAL/HIGH risk areas first. Use parallel subagents by category (crypto/circuit, synthesizer, other).

3. Understand Intent

Before diving into code:

  1. Read PR description: jq -r .body "$WS/context-pr-$PR.json"
  2. Check linked issues: cat "$WS/linked-issues-pr-$PR.txt"
  3. What consensus invariants must hold?
  4. What could go wrong?

4. Analyze Code

Small/medium PRs (< 30 files): Sequential. For each CRITICAL/HIGH risk file — read full file, trace logic, check boundaries, write findings to state file, release from memory.

Large PRs (30+ CRITICAL/HIGH files): Spawn parallel Task subagents by category (crypto/circuit, synthesizer, other). Each reads assigned files, applies AGENTS.md checklists, returns findings table.

Apply AGENTS.md Review Checklist (Correctness, Crypto-Specific, Memory & Performance, Security) to each file.

5. Validate

Validate affected crates per AGENTS.md.

6. Report

Update $WS/state-pr-$PR.md with findings:

SevLocationIssueSuggested Fix

Severity: BLOCKER (must fix), BUG (should fix), ISSUE (consider), NIT (optional).

Recommendation: Approve, Request changes, or Needs discussion.

7. Handoff

If requesting changes, create handoff for /snarkvm-fix pr:

sed -e "s/{{NUM}}/$PR/g" "$SKILL_DIR/templates/handoff.md" > "$WS/handoff-pr-$PR.md"

Common Vulnerability Patterns in snarkVM

  • Consensus divergence: Changes that produce different outputs for same inputs across versions
  • Circuit/console mismatch: Circuit constraints don't match console logic, producing invalid proofs
  • Serialization breaking: Format changes that break backwards compatibility
  • Field overflow: Arithmetic on field elements without proper bounds checking
  • Randomness misuse: Non-cryptographic or predictable randomness in proof generation
  • Unchecked deserialization: Accepting malformed data that violates type invariants

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.