castai-core-workflow-a
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources. * Ingestion points: Fetches cluster policies and node status as JSON from api.cast.ai using curl in SKILL.md. * Boundary markers: None explicitly defined in the prompt, though the agent is instructed to use jq for parsing to isolate specific fields. * Capability inventory: Includes Bash (for curl and kubectl), Write, and Edit tools. * Sanitization: The skill utilizes jq to filter and structure the output from API responses, reducing the likelihood of processing arbitrary injected content.
- [DATA_EXFILTRATION]: The skill performs network operations to api.cast.ai. These requests are authenticated using CASTAI_API_KEY and CASTAI_CLUSTER_ID environment variables, which is a standard and safe practice for secret management. No sensitive local files are accessed or transmitted.
Audit Metadata