backlog
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads and processes content from
BACKLOG.mdand external issue trackers. This ingestion of external data establishes a surface for indirect prompt injection, which the skill addresses by defining a strict schema for task entries. - [COMMAND_EXECUTION]: The skill facilitates the generation of shell commands intended for manual execution by the user. The instructions provide guidance on ensuring these commands are exact and accompanied by verification and rollback steps.
- [CREDENTIALS_UNSAFE]: The skill includes instructions to prevent the exposure of secrets by explicitly forbidding the agent from writing secret values into the backlog, instead requiring references to their source or location.
Audit Metadata