snip-36

snip-36 is a skill for Claude Code, Codex from keep-starknet-strange/starknet-agentic. It costs 66 tokens per session (5,379 once invoked), scanned A, original, MIT.

A guide to SNIP-36, a Starknet method for proving that a transaction was run correctly against an earlier block state. Starknet is a blockchain network, and the method lets heavy or privacy-sensitive Cairo computation run off-chain while the result is checked on-chain.

In plain words
What is it for?
Use it to build virtual Cairo contracts, create proof-server workflows, integrate proof generation with starknet.js, and submit results for on-chain verification.
Why use it?
It helps connect off-chain computation with an on-chain proof, so private inputs or expensive work do not all need to be exposed or executed directly on-chain.

Skill for Claude CodeCodex

Written for Claude Code and Codex: allowed-tools in frontmatter, but also agents/openai.yaml present.

Needs its repository: it runs a file that does not travel with it, so clone the repository first. The line is ./target/release/snip36 setup.

Part of the starknet-agentic-skills plugin — 19 skills, 2 commands, 1 plugin shipped together

Good fit Use it to build virtual Cairo contracts, create proof-server workflows, integrate proof generation with starknet.js, and submit results for on-chain verification.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo
git clone --depth 1 https://github.com/keep-starknet-strange/starknet-agentic
agentmods
npx agentmods add skills/keep-starknet-strange/starknet-agentic/snip-36

Made for: Claude Code, Codex.

Or install starknet-agentic-skills, the plugin that ships this one along with the rest of its 19 skills, 2 commands, 1 plugin.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for snip-36

README.md
[![agentmods](https://agentmods.dev/badge/skills/keep-starknet-strange/starknet-agentic/snip-36/github.svg)](https://agentmods.dev/skills/keep-starknet-strange/starknet-agentic/snip-36)
Your own site
<a href="https://agentmods.dev/skills/keep-starknet-strange/starknet-agentic/snip-36"><img src="https://agentmods.dev/badge/skills/keep-starknet-strange/starknet-agentic/snip-36/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for snip-36

Your own site · 80×15
<a href="https://agentmods.dev/skills/keep-starknet-strange/starknet-agentic/snip-36"><img src="https://agentmods.dev/badge/skills/keep-starknet-strange/starknet-agentic/snip-36.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 66 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 5,379 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 1 finding, up to high

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • high Privilege Escalation · line 211
    Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
    Fix: Remove references to credential paths. Use environment variables or secrets managers. For docs, use placeholder paths (e.g., /path/to/config). Never load .env or token files in production code paths.
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00066 $0.05379
Opus 5 $0.00033 $0.02690
Sonnet 5 $0.00013 $0.01076
Haiku 4.5 $0.00007 $0.00538

Measured 12d ago against content hash c138a969ddf3, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

snip-36 scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/snip-36/SKILL.md · 497 lines

How it starts

The opening of the file, as written. The whole thing — 497 lines — stays where its author put it; the contents beside it link to each section on GitHub.

SNIP-36

Overview

SNIP-36 allows executing a single INVOKE_TXN_V3 off-chain against a reference Starknet block's state, then submitting a stwo-cairo proof on-chain. The proof extends the standard v3 transaction hash by appending a proof_facts_hash. Contracts verify this via get_execution_info_v3_syscall.

Core value: Run arbitrary Cairo logic off-chain (heavy computation, privacy checks, game outcomes, attribute proofs) and commit only the verified result on-chain — without revealing private inputs.

Spec: https://community.starknet.io/t/snip-36-in-protocol-proof-verification/116123

Reference implementation: https://github.com/starknet-innovation/snip-36-prover-backend


Quick Start

  1. Review the operator checklist.
  2. Add a virtual create_proof function that emits one L2->L1 message.
  3. Prove an unsigned virtual tx with snip36 prove virtual-os.
  4. Submit verify_result with { proof, proofFacts } and the decoded message.

When to Use

  • The user asks for SNIP-36, virtual block proving, off-chain Starknet proof generation, or proof-backed on-chain verification.
  • The workflow needs heavy Cairo computation, privacy-preserving inputs, anonymous voting, secret whitelist checks, or replay-safe nullifier patterns.
  • The implementation needs a Cairo virtual function, proof server, starknet.js signing flow, and on-chain verify_result contract pattern.

When NOT to Use

  • The user needs a normal Starknet transaction that should be broadcast and fee-estimated through standard RPC.
  • The proof cannot run on a backend with native binaries, disk, and about 18 GB RAM.
  • The security model requires SNOS-native verification instead of Phase 1 sequencer-side proof verification.

Use Cases

  • Heavy computation: prove a large hash or algorithm result, then store only the verified output.
  • Private attributes: prove age, whitelist membership, or voting weight with a nullifier and public boolean/result.
  • Provable games: commit coin flips, seeds, bets, and outcomes for on-chain settlement.
  • ZKThread or shard transitions: prove { old_root, new_root, ... } before updating L2 state.

Read the full file on GitHub · 497 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 12d ago First seen · 497 lines · 66 tokens per session scan A c138a969ddf3

Subscribe to this mod's changes

snip-36 is a skill published in the GitHub repository keep-starknet-strange/starknet-agentic (80 stars, last pushed today), licensed MIT. It adds 66 tokens to every session and 5,379 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.