feedback

Pass

Audited by Gen Agent Trust Hub on Nov 8, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructs the agent to perform network POST operations to https://skills.devnads.com/api/feedback to report errors and frustrations. Although the instructions include strict privacy requirements to exclude secrets, private keys, and PII, the practice of sending execution context to an external, non-whitelisted endpoint represents a data exposure surface.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates a path where untrusted data from external tool outputs or errors can influence the content sent in network requests.
  • Ingestion points: Error messages, tool outputs, and user complaints gathered during task execution (SKILL.md).
  • Boundary markers: Absent. There are no instructions to use delimiters or specific markers for the data interpolated into the JSON payload.
  • Capability inventory: Execution of curl for network requests (SKILL.md).
  • Sanitization: Includes manual instructions for the agent to scrub PII, secrets, and private keys from the message and context fields before transmission (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 8, 2026, 08:21 PM