# Factory's Droid 1.0 makes scripted tool calls the default for coding agents

> The model-agnostic harness posts 65.2% on Factory's own Terminal-Bench 4 runs with Claude Opus 5.5. Details are still thin.

- URL: https://dev.devisdead.ai/news/factory-s-droid-1-0-makes-scripted-tool-calls-the-default-for-coding-agents
- Published: 2026-10-31T15:55:52Z
- Author: DevIsDead Newsroom
- Section: Dev
- Tags: agents, factory, coding, benchmarks
- Originally reported by MarkTechPost (https://www.marktechpost.com/2026/10/31/factory-droid-1-0-frontier-harness/).

Factory has released Droid 1.0, a coding agent harness that works with any model and makes programmatic tool calling its default, according to [MarkTechPost](https://www.marktechpost.com/2026/10/31/factory-droid-1-0-frontier-harness/)'s Asif Razzaq.

The idea behind script mode, as the headline describes it: instead of the model calling one tool, waiting, and calling the next, it writes a small program that chains the calls together. That usually means fewer round trips and less clutter in the context window, which is where long agent runs tend to fall apart.

## The number

Factory says that, paired with Claude Opus 5.5, Droid passed 65.2% of tasks in its in-house Terminal-Bench 4 runs, costing $4.73 for each task solved. Reporting cost per solved task rather than per token is the right instinct, since a cheaper call that needs three retries isn't cheaper.

## Benchmarks vs. vibes

These are vendor-run results, and so far we only have the announcement summary, not the full methodology or comparisons with other harnesses on the same model. Details are still thin. Treat 65.2% as a claim to check against your own repos, not a ranking.

The broader trend is worth noticing either way: the harness is becoming as much of a product as the model, and teams can now switch the model underneath without throwing away their agent setup.

**[Scoreboard](/scoreboard) update:** AI +1. An agent harness clears 65% of Terminal-Bench 4.