The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation and the wrapper script recommend installing the uv tool via a shell script from https://astral.sh/uv/install.sh. This is a well-known installation method for the Astral uv utility and is considered a safe reference to an established technology provider.\n- [COMMAND_EXECUTION]: The scripts/run.py script utilizes subprocess.run() to pre-warm the environment and os.execvp() to execute the uv command that runs the skill's engine. These calls are essential for the skill's primary function of executing its project maintenance logic.\n- [DYNAMIC_EXECUTION]: The skill dynamically constructs its execution environment by determining necessary package extras for exmergo-dex-core based on project configuration and CLI flags. It then uses uv run to install and run these dependencies hermetically.\n- [INDIRECT_PROMPT_INJECTION]:\n
Ingestion points: The skill reads dbt project files, warehouse schemas, and a local snapshot file (.dex/snapshot.json).\n
Boundary markers: Output from the underlying engine is returned in a structured JSON envelope to separate findings from operational data.\n
Capability inventory: The skill is capable of executing shell commands via uv through the run.py wrapper script.\n
Sanitization: The engine logic (within exmergo-dex-core) is designed to produce sanitized JSON output according to a defined command contract.

maintain