bug-health-inline

Pass

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The workflow executes shell commands (mkdir, pnpm, cp, rm) to prepare working directories, run validation tests (type-check and build), and perform file backups and restorations. These commands are standard for development environments and essential for the skill's operational logic.
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection because it reads and processes the entire codebase through subagents.
  • Ingestion points: SKILL.md instructs the bug-hunter subagent to scan the repository and bug-fixer to read generated reports.
  • Boundary markers: The prompts for subagents do not define delimiters to isolate codebase content from instructions.
  • Capability inventory: The skill utilizes tools for file modification, directory creation, and command execution.
  • Sanitization: There is no evidence of sanitization for filenames or code content extracted from the repository before being used in shell commands or subsequent prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 20, 2026, 05:02 AM
Security Audit — agent-trust-hub — bug-health-inline