Savant Chat Documentation
Savant Chat is an AI code auditor. Smart contracts — Solidity, Vyper, Rust, or any other contract language — are its first and best-proven domain; the same language-agnostic engine audits any code, from ZK circuits and node clients to backends. It finds vulnerabilities before attackers do — in 10–30 minutes for a typical codebase, at a fraction of the cost of a manual audit.
How it works
Savant Chat is not a static analyzer. Instead of pattern-matching known bug shapes, a multi-agent stack runs thousands of specialized LLM calls over your codebase:
- 200+ vulnerability classes — distilled from a reference book of 20,000 real-world smart contract vulnerabilities.
- Semantic analysis — the agents trace value flows and model attacker incentives instead of matching syntax.
- PoC validation (Advanced and Pro tiers) — a separate critic subagent builds a proof-of-concept for every finding before it reaches your dashboard. If the PoC fails, the finding is discarded.
- Severity and confidence — every finding ships with a severity rating and a confidence score, so you can triage fast.
What Savant audits
Smart contracts are the flagship domain — where our benchmarks and contest results are published:
| Language | Ecosystem |
|---|---|
| Solidity | Ethereum and EVM L2s |
| Vyper | Curve-style DeFi |
| Rust | Solana, NEAR |
| Move, Cairo, FunC, or any other contract language | Aptos, Sui, Starknet, TON, … |
The same engine audits any other code — for example:
- ZK circuits & cryptography — Circom, Halo2, Noir, arkworks
- Node clients & infrastructure — Rust, Go, C++
- Off-chain code — TypeScript, Python, Go backends, keepers, and services
These are the domains with dedicated guides, not the engine's limits — if it's code, Savant can audit it. See the full supported languages and upload formats matrix.
Quick start
- Sign up — you get $75 in free credits, no credit card required.
- Upload your project — a ZIP archive of your repository or a GitHub import.
- Run the audit — severity-ranked, confidence-scored findings land in your dashboard in 10–30 minutes; on Advanced and Pro, every finding is PoC-validated first.
Follow the Getting Started guide for a detailed walkthrough.
Pricing at a glance
Pay as you go, priced per line of code. Credits never expire.
| Tier | Price | Best for |
|---|---|---|
| Lite | $0.07/line | Efficient models. CI-friendly quick pass. |
| Advanced | $0.12/line | Comprehensive analysis. The default for pre-audit runs. |
| Pro | $0.50/line | Highest-quality models for critical releases. |
See full audit pricing for billing mechanics and scoped quotes.
Guides
- Getting Started — run your first audit
- Supported languages and upload formats
- Prepare any codebase for audit
- HOWTO load non-standard Foundry / Solidity projects
Learn more
- Blog — engineering deep dives and benchmark reports.
- CTFBench — the open benchmark we use to measure smart contract detection accuracy.
- X (Twitter) — product news and announcements.