Substrate Vulnerability Scanner reviews pallets and runtime code for arithmetic overflow, panics, incorrect weights, validate-first violations, unsigned transaction validation, bad randomness and bad origin checks.
The report opens with a coverage table that gives each of the seven patterns a verdict of found, clear or n/a with evidence, followed by the findings. A checklist and testing guidance for fuzzing, benchmarks and try-runtime come with it.
What You Get
- A coverage table with all seven patterns and a verdict with evidence for each
- Findings ordered by priority from critical to medium
- A dispatchable checklist covering arithmetic, panics, weights, origin and validation
- Fuzz test, benchmark and try-runtime examples
Common Use Cases
- Auditing a custom pallet before a parachain launch
- Reviewing weight functions for variable-cost operations
- Finding privileged calls guarded only by ensure_signed
- Checking a runtime upgrade path with try-runtime
What Makes It Unique
The skill targets how FRAME runtimes fail: a panic in a dispatchable stops block processing, release builds wrap arithmetic silently, and a benchmark covers only the path it measures. It requires a row for every pattern, scopes the verify-first pattern to the runtime version and rejects claims such as ensure_signed counting as an origin check for privileged calls.
Under The Hood
The skill is an instruction file with a linked pattern resource. The agent searches the project with ripgrep, reads pallets and runtime files, and reports in its response. Its checklist and examples mention commands the user can run, such as cargo test with runtime benchmarks, a node benchmark pallet command and try-runtime against a live node URI over a websocket, plus fuzz tests with test-fuzz. It does not edit pallet code.