steer
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from external sources, specifically the Steer CLI output and multiple blockchain subgraph APIs via the
scripts/rebalance-context.mjshelper script. This external data is used to provide the agent with vault and market context, representing an indirect prompt injection surface. - Ingestion points: Data enters the agent's context through the output of
scripts/rebalance-context.mjs, which aggregates results from CLI commands (vaults inspect,markets history) and external GraphQL API calls to various blockchain indexers. - Boundary markers: The skill instructions in
SKILL.mdandreferences/bankr-execution.mdestablish security boundaries by requiring explicit human confirmation for all transactions and performing case-insensitive signer address and chain ID verification before submission. - Capability inventory: The skill utilizes
execute_clito run the Steer CLI and associated Node.js helper scripts, andsubmit_raw_transactionto broadcast reviewed and simulated blockchain transactions. - Sanitization: The
rebalance-context.mjsscript performs manual field extraction from JSON responses and includes aboundedDiagnosticfunction specifically designed to redact sensitive environment variables (likeSTEER_RPC_URLandSTEER_SUBGRAPH_STUDIO_KEY) from its output and error messages, preventing accidental credential exposure in logs.
Audit Metadata