py-async

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for fetching data from external network sources, creating a potential attack surface where instructions embedded in external content could influence the agent's behavior.
  • Ingestion points: references/examples.md (Example 2) demonstrates fetching data from external URLs using aiohttp, and references/examples.md (Example 4) illustrates processing streams of external data.
  • Boundary markers: The instructions do not provide explicit delimiters or "ignore instructions" directives when processing or interpolating data retrieved from external sources.
  • Capability inventory: The skill demonstrates network access via aiohttp and httpx along with complex asynchronous task management.
  • Sanitization: No evidence of data sanitization or input validation is present in the examples provided for processing network response content.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 06:31 PM