context-optimization
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill uses strong instructional headers such as 'MANDATORY IMPORTANT MUST CRITICAL' and 'STRICT EXECUTION CONTRACT'. These are used to enforce specific project management behaviors like task tracking and progress summarization rather than bypassing safety guidelines.
- [COMMAND_EXECUTION]: Defines token-efficient patterns for codebase searching and file reading using tools like Grep and Read with offset/limit parameters.
- [REMOTE_CODE_EXECUTION]: Instructions describe a protocol for delegating tasks to sub-agents (e.g., 'Explore', 'Plan') using a Task function to isolate context usage. This is a standard architectural pattern for multi-agent systems and does not involve executing untrusted remote code.
- [DATA_EXFILTRATION]: No network-based exfiltration patterns or access to sensitive credential files were detected. The skill focuses on local file reading for context management.
Audit Metadata