bitflow-hodlmm-zest-yield-loop
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill functions as a high-level orchestrator that executes local primitive scripts (such as
bitflow-hodlmm-withdraw,bitflow-hodlmm-deposit,hodlmm-move-liquidity, andzest-yield-manager) viachild_process.spawn. These executions are restricted to specific, hardcoded primitive names resolved within the local skill directory structure. - [EXTERNAL_DOWNLOADS]: The implementation performs network requests to fetch pool metadata and verify transaction statuses using official ecosystem endpoints:
https://bff.bitflowapis.finance/api/app/v1/poolsandhttps://api.mainnet.hiro.so. These references target trusted vendor and well-known service infrastructure. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external APIs that could potentially be used as an attack vector if the source data were compromised.
- Ingestion points: Fetches structured JSON data from Hiro and Bitflow APIs in
bitflow-hodlmm-zest-yield-loop.ts. - Boundary markers: Uses strict JSON parsing; the
AGENT.mdinstructions guide the agent to evaluate specific data fields likeeconomicCheckandfreshnessrather than raw text. - Capability inventory: The skill has the capability to execute local scripts and move funds via primitives in
bitflow-hodlmm-zest-yield-loop.ts. - Sanitization: External data is mapped to strictly typed TypeScript interfaces; the skill does not directly interpolate external strings into the agent's instructions, though the agent consumes the JSON output at runtime.
Audit Metadata