bee-collect

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill constructs and executes shell commands using user-provided text to populate arguments for the GitHub CLI (gh) and custom project tools (gw). Although the skill provides templates designed to mitigate command injection, such as the use of quoted heredocs (cat <<'EOF'), the dynamic assembly of shell commands remains a surface for potential injection if the agent deviates from these best practices.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to generate issue content.
  • Ingestion points: User-provided 'brain dumps', task lists, and TODOs are parsed in Phase 1 to create issue content.
  • Boundary markers: The skill does not define specific delimiters or instructions for the agent to ignore potentially malicious instructions embedded within the user's task descriptions.
  • Capability inventory: The skill possesses the ability to search the codebase, read issue history, and create new issues on GitHub.
  • Sanitization: There are no explicit instructions for sanitizing or escaping user-provided strings before they are interpolated into issue titles and bodies.
  • [DATA_EXFILTRATION]: The skill is designed to search the codebase for technical context (such as function definitions and TODO/FIXME comments) and include these findings in the 'Context' section of GitHub issues. This flow represents the export of internal technical information to an external platform, which could lead to the unintentional exposure of sensitive implementation details or developer notes.
  • [DYNAMIC_EXECUTION]: The skill generates data files at runtime, specifically JSON files in /tmp/, which are then used as input for batch processing commands (gw gh issue batch). The generation of executable data patterns at runtime is a potential vector for manipulation if the input source is compromised or inadequately validated.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 01:31 AM