clawgs
Pass
Audited by Gen Agent Trust Hub on Mar 19, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE]: Accesses local session transcripts in the
~/.claudeand~/.codexdirectories to perform log extraction and metadata analysis. - [COMMAND_EXECUTION]: Executes the
tmuxCLI to list panes and capture terminal context for session monitoring. - [EXTERNAL_DOWNLOADS]: Communicates with the OpenRouter API (
openrouter.ai) to generate summaries of session activity based on captured logs and terminal text. - [PROMPT_INJECTION]: The skill processes untrusted terminal text and log files which are interpolated into prompts for LLM summarization; while this creates an indirect prompt injection surface, it is a known risk for this use case. Ingestion points:
.jsonlsession logs andtmux capture-paneoutput. Boundary markers: Prompts use structured labels like 'Task:', 'Actions:', and 'Screen:'. Capability inventory: Executestmuxand performs network requests viareqwest. Sanitization: Implements ANSI sequence stripping and content truncation.
Audit Metadata