supabase

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSPRIVILEGE_ESCALATION
Full Analysis
  • [COMMAND_EXECUTION]: The scripts/supabase.sh wrapper script executes shell commands including curl and jq to facilitate communication with the Supabase REST API and OpenAI's embeddings API.
  • [DYNAMIC_EXECUTION]: The query command enables the execution of arbitrary SQL statements at runtime. This allows the AI agent to dynamically assemble and run database code based on user requests or its own analysis of the schema.
  • [PRIVILEGE_ESCALATION]: The skill documentation provides instructions for creating a database function (exec_sql) using the SECURITY DEFINER property. This configuration allows the function to execute with the privileges of the user who created it (typically a database administrator), effectively allowing the agent to perform actions at a higher privilege level than the connection might otherwise permit.
  • [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it retrieves and processes data from external database records.
  • Ingestion points: Data returned from SQL queries, table selections, and vector searches (handled by supabase.sh) is ingested directly into the agent's context (e.g., in SKILL.md workflows).
  • Boundary markers: The instructions do not define specific delimiters or instructions for the agent to treat data retrieved from the database as untrusted content.
  • Capability inventory: The skill has powerful capabilities including database write access (insert, update, delete), schema modification (query), and network connectivity for embeddings generation.
  • Sanitization: While jq is used for JSON formatting in the shell script, there is no logic to sanitize or escape the content of database records before they are processed by the agent's language model.
  • [EXTERNAL_DOWNLOADS]: The vector-search command performs network requests to OpenAI's official embeddings API (api.openai.com) to generate vector representations of text for similarity search purposes.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 03:48 AM