Factory Droid 1.0 Script Mode: Hundreds of Tool Calls from One Model Request in a JS Sandbox

说明与用途

Factory Droid 1.0 允许在单个模型请求中执行数百个工具调用，降低令牌使用和成本。

问题

In traditional agent loops, each tool result is appended to the context, causing token bloat, rate‑limit hits, and noisy decision making.

结果

Script mode consolidates many tool calls into a single model request, dramatically reducing context size, token usage, and cost while maintaining deterministic execution.

工作流

• 1. Model emits a JavaScript program that calls tools concurrently in a sandbox.
• 2. The sandbox executes the script, printing only a summary.
• 3. The printed summary is returned to the model; raw results stay in a log for later search.

证据

Factory shipped Droid 1.0 and made programmatic tool calling the default. The model no longer emits one tool call, waits for the result, and repeats.

来源

Factory’s own numbers illustrate the effect. A Linear cleanup that touched 151 issues (312 tool calls) dropped from an estimated 36 model requests and 3.1 M input tokens to 2 requests and ~61 k tokens.

来源

On Claude Opus 5.5 the same mode cut cost ~15 % versus direct tool calling.

来源