velodrome-v2-plugin
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from blockchain RPC calls which could potentially be manipulated, although the risk is minimized by manual confirmation steps.
- Ingestion points: Data returned from Optimism RPC endpoints via eth_call in src/rpc.rs.
- Boundary markers: The SKILL.md includes a specific 'Data Trust Boundary' section instructing the agent to treat external content as untrusted and requires --confirm for all write operations.
- Capability inventory: The skill can execute wallet transactions using the onchainos CLI (src/onchainos.rs).
- Sanitization: Relies on user verification of transaction details before confirmation.
- [EXTERNAL_DOWNLOADS]: The skill's setup scripts download installers and binaries from GitHub repositories.
- Evidence: Scripts in SKILL.md fetch install.sh and plugin binaries from github.com/okx/ repositories.
- Note: These downloads target the official repositories of a recognized vendor and are verified by SHA256 checksums.
- [COMMAND_EXECUTION]: The skill executes shell commands to perform wallet operations and install dependencies.
- Evidence: The Rust implementation uses std::process::Command to call the onchainos CLI for transaction signing and address resolution.
Audit Metadata