The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill automatically installs the official Supabase CLI using well-known package managers (npm, brew, scoop, winget) if it is not detected on the system. This installation targets official packages from recognized public registries.
[COMMAND_EXECUTION]: Shell commands are executed through a dedicated wrapper that enforces a strict security policy defined in supabase_cli_policy.py. This policy whitelists approved cloud operations and explicitly blocks dangerous arguments, credential exposure in command lines, and local/Docker-dependent workflows.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it reads information from remote Supabase projects, such as table data, function source code, and service configurations.
Ingestion points: Ingestion occurs in supabase_db.py (database queries and TypeScript generation) and supabase_resources.py (Edge Function source code, secrets metadata, and service configurations).
Boundary markers: The instructions guide the agent to interpret provider results without inventing fields, although the scripts do not wrap external data in explicit delimiters.
Capability inventory: The skill can execute shell commands via the CLI wrapper in supabase_common.py and write files to the local directory via write_output in supabase_common.py.
Sanitization: The API client includes logic to redact sensitive fields (such as tokens, passwords, and keys) from configuration and secret listings in supabase_resources.py before they are processed by the agent.

supabase