context-optimization
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill outlines strategies for 'Compaction' (summarization) and 'Observation Masking' of tool outputs and conversation history. These techniques involve processing potentially untrusted data from external tools or retrieved documents through an LLM to generate summaries, which creates an attack surface for indirect prompt injection where malicious instructions could attempt to subvert the summarization process.
- Ingestion points: Processes tool outputs, conversational turns, and retrieved documents for summarization and masking (as described in SKILL.md).
- Boundary markers: The skill contains a 'Preservation boundaries' section advising against discarding authorization or user constraints, but does not provide specific technical delimiters (e.g., XML tags) for the data ingestion process.
- Capability inventory: This skill is a conceptual and instructional guide; it does not include active tool definitions or scripts with file system, network, or subprocess capabilities.
- Sanitization: No specific sanitization or filtering logic is provided for the content being summarized.
Audit Metadata