watch-and-act

Pass

Audited by Gen Agent Trust Hub on Aug 11, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: An indirect prompt injection surface is present because the skill monitors and processes data from external, untrusted sources such as CI logs or health check responses.
  • Ingestion points: External state queried via tools like gh or curl in SKILL.md.
  • Boundary markers: Absent; the instructions do not specify the use of delimiters when processing external signal content.
  • Capability inventory: The skill uses Bash, Write, ScheduleWakeup, and Agent tools.
  • Sanitization: Absent; the skill records and acts on external evidence without specific filtering mentioned in the instructions.
  • [COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute shell commands for querying the state of external resources.
  • [EXTERNAL_DOWNLOADS]: The skill performs network operations (e.g., curl) to fetch status information from external endpoints as part of its primary polling functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 11, 2026, 06:20 PM
Security Audit — agent-trust-hub — watch-and-act