public-relations-manager

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it is designed to monitor and respond to untrusted content from GitHub issues and comments.
  • Ingestion points: GitHub repository issues and comments as described in references/README.md.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands are defined in the persona configuration.
  • Capability inventory: The agent can execute local CLI commands via the bd tool, write to public GitHub repositories via API, and read project source code.
  • Sanitization: No sanitization or validation of the external content is mentioned before it is processed by the model or used in commands.
  • [COMMAND_EXECUTION]: The agent uses a local command-line tool (bd) to manage project tasks. Documentation in references/README.md shows the agent interpolating untrusted GitHub issue titles directly into command strings (e.g., bd create "Bug: [Issue title]"), which presents a risk of command injection if the input is not properly escaped.
  • [EXTERNAL_DOWNLOADS]: The skill performs network operations to interact with GitHub's API. These interactions target a well-known service for the skill's primary purpose and do not involve sensitive data exfiltration.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 05:50 AM