The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute bash commands for logging and voice notifications. Specifically, it uses curl to send status messages to a local port and echo to append execution data to a hidden log file (~/.claude/LIFEOS/MEMORY/SKILLS/execution.jsonl).
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it fetches and processes untrusted data from external URLs and local files without explicit sanitization or boundary markers to prevent the agent from obeying embedded instructions. Ingestion points: Content fetched via WebFetch and WebSearch in Workflows/Check.md based on user-provided URLs or file paths. Boundary markers: None identified in the workflow instructions. Capability inventory: The skill has access to WebFetch, WebSearch, and Read tools, as well as shell execution capabilities for curl and echo commands. Sanitization: No specific sanitization or filtering of the fetched content is described before processing.
[DATA_EXFILTRATION]: The skill performs local network operations by sending execution notifications to http://localhost:31337/notify. While targeting localhost, this involves sending data about the agent's activity to an external (though local) service.

BiasCheck