supabase-automation

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the user to add an external MCP server at https://rube.app/mcp. This endpoint provides the logic for all Supabase tools described in the skill.
  • [COMMAND_EXECUTION]: The skill exposes the SUPABASE_BETA_RUN_SQL_QUERY tool, allowing the agent to execute arbitrary PostgreSQL statements. This provides full control over the database, though the skill explicitly recommends using read_only: true for non-mutating queries.
  • [CREDENTIALS_UNSAFE]: The tool SUPABASE_GET_PROJECT_API_KEYS allows the agent to retrieve live service-role and anon keys. The skill correctly identifies these as sensitive and instructs the agent to mask or truncate these values in its output.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection if it processes data from a database that contains malicious instructions.
  • Ingestion points: Data is ingested through SUPABASE_SELECT_FROM_TABLE and SUPABASE_BETA_RUN_SQL_QUERY results.
  • Boundary markers: No specific delimiters (like XML tags) are mandated for the processing of database content, though the skill provides guidance on using read-only transactions.
  • Capability inventory: The skill has high-privilege write access via SQL execution and the ability to retrieve project secrets.
  • Sanitization: There are no explicit sanitization or filtering instructions for data retrieved from the database before it is used in subsequent prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 07:47 PM