plan-llm-cost-guardrails
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill acts as a template for architectural reviews, focusing on LLM cost-velocity, token-bucket limits, and fallback strategies. It emphasizes a 'plan-only' approach where no system changes are made without explicit approval.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies LLM call sites by reading user codebase files, which is an ingestion point for external data. While this presents a theoretical surface for indirect prompt injection, the risk is mitigated by the skill's limited capabilities, which are restricted to local file analysis and report generation.
- Ingestion points: Project codebase files, specifically LLM call sites and endpoint definitions.
- Boundary markers: No specific delimiters are defined in the instructions to isolate analyzed code from agent instructions.
- Capability inventory: The skill is designed to read local files and write a markdown report (plan-llm-cost-guardrails.md). It lacks network access or subprocess execution capabilities.
- Sanitization: The skill does not define specific sanitization or filtering logic for the content it processes from the codebase.
Audit Metadata