skill-meta-prompt
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill generates structured prompts by interpolating user-provided information such as goals, constraints, and expected outputs. This creates a surface where a user could provide instructions intended to influence the agent's behavior.
- Ingestion points: Phase 1 requirement gathering (goals, outputs, accuracy, constraints) in
SKILL.md. - Boundary markers: The skill uses defined Markdown sections (Role, Context, Instructions, Constraints) to isolate user-provided data.
- Capability inventory: The skill generates text based on user input; it does not perform file writes, network calls, or command execution based on the processed content.
- Sanitization: The instructions do not describe explicit sanitization or filtering of the user-provided goals before they are included in the prompt template.
- [SAFE]: The skill contains references to non-existent or futuristic model names (e.g., GPT-5.6, Sonnet 5, Astra) and platform-specific internal paths (e.g.,
skills/blocks/*). These are part of the skill's conceptual framework for prompt routing and do not represent functional security vulnerabilities or external dependencies.
Audit Metadata