The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/supabase.sh script uses the eval command to load environment variables from the environments.json configuration file. The implementation uses jq's @sh filter to ensure that all values are correctly escaped for the shell, preventing potential command injection from configuration data.
[EXTERNAL_DOWNLOADS]: The skill uses curl to interact with api.supabase.com. This network communication is necessary for the skill's primary function of executing SQL queries against Supabase projects and targets a well-known service domain.
[INDIRECT_PROMPT_INJECTION]: The skill processes JSON responses from SQL queries which could contain untrusted content. The SKILL.md file includes explicit instructions for the agent to verify destructive actions and handle data cautiously, mitigating the risk of indirect injection.
Ingestion points: SQL query results received from the Supabase API in scripts/supabase.sh.
Boundary markers: The agent is directed to use structured JSON output via jq to interpret responses.
Capability inventory: Supabase SQL execution, local configuration file management, and authenticated network requests to the Supabase API.
Sanitization: Results are parsed and formatted using jq, ensuring consistent data structure for the agent.

supabase