session-search
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the
grepcommand-line utility to perform recursive searches through local session transcript files. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from past session transcripts (
~/.feynman/sessions/*.jsonl), which is considered untrusted as it contains content from previous interactions that could be manipulated. - Ingestion points: Historical session transcript files in the
~/.feynman/sessions/directory. - Boundary markers: None identified; the skill does not use specific delimiters to separate transcript data from instructions.
- Capability inventory: Shell command execution via
grepand an interactive/searchcommand. - Sanitization: No evidence of content sanitization or instruction filtering is present in the provided skill definition.
Audit Metadata