context-compression
Pass
Audited by Gen Agent Trust Hub on May 18, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No attempts to override agent behavior, bypass safety constraints, or extract system prompts were found. The instructions focus on improving session efficiency through summarization.
- [DATA_EXFILTRATION]: The skill does not perform network operations or access sensitive system paths (e.g., .ssh, .aws). It uses the Write tool only to manage its own conversation summaries within the session context.
- [REMOTE_CODE_EXECUTION]: No remote script downloads, package installations, or dynamic code execution patterns are present.
- [INDIRECT_PROMPT_INJECTION]: The skill processes conversation context and previous tool outputs. While it summarizes potentially untrusted data from the user or files, it does so through internal summarization logic rather than executing content from that data. The use of structured summary templates helps maintain boundaries.
- [OBFUSCATION]: No hidden characters, encoded URLs, or homoglyph substitutions were detected in the instructions or example blocks.
Audit Metadata