supabase

Pass

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest and process data from external Supabase documentation and changelog files to inform database schema changes and code generation (Indirect Prompt Injection).
  • Ingestion points: The skill instructs the agent to fetch documentation and product updates from official sources at https://supabase.com/changelog.md and https://supabase.com/docs/.
  • Boundary markers: The instructions do not provide explicit delimiters or warnings to isolate external data from the primary agent context.
  • Capability inventory: The agent has powerful write capabilities, including the execute_sql tool and the Supabase CLI (db query, migration new).
  • Sanitization: There is no requirement for sanitization or validation of documentation content before it is interpolated into prompts for generating database commands.
  • [EXTERNAL_DOWNLOADS]: The skill fetches configuration and updates from official vendor resources.
  • Evidence: Retrieves the latest product changelog and documentation pages via direct markdown fetching from Supabase domains.
  • [COMMAND_EXECUTION]: The skill instructs the agent to perform administrative database and migration tasks using CLI tools.
  • Evidence: Usage of supabase db query, supabase db advisors, and supabase migration commands are integrated into the workflow.
  • [DATA_EXFILTRATION]: The skill includes a connectivity check to verify the status of the Supabase MCP server.
  • Evidence: Executes a curl request to https://mcp.supabase.com/mcp to check for server availability.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 31, 2026, 04:22 PM
Security Audit — agent-trust-hub — supabase