scopechk
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core workflow relies on the ingestion of untrusted content from external documentation.
- Ingestion points: Step 2 in
SKILL.mdinstructs the agent to read criteria from files like README, AGENTS.md, and CONTRIBUTING files. - Boundary markers: There are no instructions to utilize delimiters or to ignore potential instructions embedded within the analyzed documentation.
- Capability inventory: The skill requires the agent to read files, access URLs, and verify the reachability of local and remote resources.
- Sanitization: The skill does not prescribe any sanitization or validation of the content retrieved from external sources before it is processed by the agent.
- [DATA_EXFILTRATION]: The reachability verification step (Step 5) requires probing various resources.
- The agent is instructed to check 'repositories, paths, services, accounts, tools,' specifically including 'a path on one machine.' If misused, this can be used to probe for the existence of sensitive local files or private network resources.
Audit Metadata