The Agent Skills Directory

[SAFE]: The skill implements a productivity workflow using local Python scripts and documentation. No indicators of prompt injection, data exfiltration, or obfuscation were found.
[DATA_EXPOSURE]: The scripts/grill_session_tracker.py utility writes session state to ~/.grill_sessions/ as JSON files. This is a standard local persistence mechanism for session management and does not involve exfiltration to external domains.
[COMMAND_EXECUTION]: The skill instructions direct the agent to use standard tools like grep and file reading to explore the codebase. These actions are scoped to the intended purpose of resolving plan dependencies by verifying existing code.
[INDIRECT_PROMPT_INJECTION]: The scripts/decision_tree_extractor.py utility reads user-provided markdown files to identify decision branches. While this constitutes an ingestion point for external data, the extraction logic is regex-based and contained within the local environment without bypassing security boundaries.

grill-me