Skip to content

code-mode.md

Latest commit

 

History

History
35 lines (26 loc) · 1.52 KB

code-mode.md

File metadata and controls

35 lines (26 loc) · 1.52 KB

Code Mode

The execute-typescript MCP tool lets the AI write a small TypeScript program that composes multiple Mongo or Postgres queries inside a sandboxed QuickJS WebAssembly isolate. One round trip in, one structured result out — instead of N+1 separate tool calls.

Why it matters

  • Math is correct. Sums, averages, percentages run as actual JavaScript inside the sandbox. The model decides what to compute; the sandbox computes it.
  • Token cost drops. A query that touches 500 documents lives and dies inside the isolate. Only the final result crosses the wire to the model.
  • Security is unchanged. Every external_* call inside the sandbox routes through the same executeQueryOperation that the direct find/aggregate/count/distinct tools use. Hidden fields are stripped before data crosses into the sandbox. The isolate has no fs, no process, no require, no fetch, no globals at all beyond the four bridge functions.

Example

const top = await external_find({ collection: "products", limit: 5 });
const ratings = await Promise.all(
  top.map((p) =>
    external_find({ collection: "ratings", filter: { productId: p._id } })
  )
);
return top.map((p, i) => ({
  name: p.name,
  avgRating: ratings[i].reduce((s, r) => s + r.score, 0) / ratings[i].length,
}));

Limits

Per execution:

  • 30s wall-clock timeout
  • 128MB memory
  • 50 bridge calls
  • 256KB serialized result

Disable the tool entirely by adding execute-typescript to ASKDB_MCP_DISABLED_TOOLS.