create-course

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill relies on extensive user input collected through multi-phase interviews to generate project files and components, which creates a surface for indirect prompt injection.
  • Ingestion points: Untrusted data enters the agent context via responses to AskUserQuestion across Phases 1 through 5 in SKILL.md.
  • Boundary markers: Absent; the instructions do not require the agent to wrap user input in delimiters or include 'ignore embedded instructions' warnings when writing that data to files.
  • Capability inventory: The skill uses Write, Edit, and Bash tools (listed in allowed-tools and referenced in instructions) to create files and assemble components.
  • Sanitization: Absent; there is no mention of escaping, validation, or filtering of user-provided content before it is interpolated into scripts, files, or potentially used in shell operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 04:58 PM
Security Audit — agent-trust-hub — create-course