sessions
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python helper script (
scripts/list_sessions.py) to traverse project directories and aggregate session data. - [DATA_EXPOSURE]: The skill accesses sensitive conversational history stored in the user's home directory at
~/.claude/projects/. It reads and displays session timestamps, message counts, and the content of the initial message from each log file. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from past session logs (ingested via
scripts/list_sessions.py). While this presents a theoretical injection surface if logs contain malicious prompts, the risk is mitigated by instructions to present output verbatim, basic sanitization that removes HTML-like tags (re.sub(r"<[^>]+>", "", msg)), and strict truncation of the displayed message text.
Audit Metadata