Executes raw SQL in the database. LLMs should use this for regular queries that don't change the schema.
AI agents invoke execute_sql to trigger actions in PostgREST. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.
This tool executes arbitrary SQL code provided by the LLM. While the description suggests it is intended for non-schema-altering queries, the ability to execute 'raw SQL' means an AI agent could misuse it to run DELETE, UPDATE with broad WHERE clauses, or other state-modifying operations that cause significant data loss or corruption.
From the tool's definition Tool name 'execute_sql' combined with description 'Executes raw SQL in the database' indicates arbitrary SQL execution capability.
Documented attack patterns abuse exactly the kind of access execute_sql gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and PostgREST, and nothing reaches the server without passing your rules. This is the rule we recommend for execute_sql:
{
"version": "1",
"default": "deny",
"tools": {
"execute_sql": {
"limits": [
{
"counter": "execute_sql_rate",
"window": "minute",
"max": 10,
"scope": "grant"
}
]
}
}
}execute_sql stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Executes raw SQL in the database. LLMs should use this for regular queries that don't change the schema. It is categorised as a Execute tool in the PostgREST MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.
Register the PostgREST MCP server in PolicyLayer and add a rule for execute_sql: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches PostgREST. Nothing to install.
execute_sql is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.
Yes. Add a rate_limit block to the execute_sql rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for execute_sql. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
execute_sql is provided by the PostgREST MCP server (supabase/mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 32 PostgREST tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
32 PostgREST tools catalogued and risk-classified — across an index of 42,500+ MCP servers.