technical-implementation

Warn

Audited by Gen Agent Trust Hub on Feb 22, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION] (MEDIUM): The skill is designed to run shell commands discovered in workspace files like docs/workflow/environment-setup.md and tracking.md. This mechanism allows for execution of commands that are not hardcoded in the skill itself.
  • [EXTERNAL_DOWNLOADS] (LOW): The setup instructions provide examples of system package installation (e.g., sudo apt-get install), which involve downloading external software with elevated privileges.
  • [PROMPT_INJECTION] (LOW): The skill possesses a significant surface for indirect prompt injection (Category 8). 1. Ingestion points: Reads instructions from docs/workflow/environment-setup.md, .claude/skills/, and implementation plans. 2. Boundary markers: No explicit markers or 'ignore' instructions are used when processing these files. 3. Capability inventory: Includes shell execution (bash), sudo access, and agent dispatching via the Task tool. 4. Sanitization: There is no evidence of validation or sanitization for the commands extracted from documentation.
  • [COMMAND_EXECUTION] (MEDIUM): Use of sudo in environment setup instructions indicates potential for privilege escalation if the source documentation is malicious.
  • [COMMAND_EXECUTION] (MEDIUM): The skill utilizes 'auto' gates for task approval and fix cycles (e.g., task_gate_mode, fix_gate_mode), which can be enabled by the agent or by modifying the tracking.md file. This reduces human oversight and could be exploited to persist malicious changes.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 22, 2026, 08:26 PM