Why agents will soon have only one tool: Code Mode cuts spend 15%, +3.5 Terminal-Bench points
dl_weekly · x · 2026-11-11
Philschmid argues agents are collapsing entire tool catalogs into a single sandboxed exec(code) tool (Code Mode / Programmatic Tool Calling). Instead of one JSON action per turn, the model writes a TypeScript/Python script chaining dependent actions in one pass. The evolution: few tools → dozens of tools/MCP → context limits → tool search → now code. Cites Factory Droid's +3.5-point Terminal-Bench 4 gain and 15% lower model spend.
More from coding & agent
- How should UIs attribute actions taken by agents through a signed-in human account? — max__drake · 2026-11-11
- Fable 5.2 audits an Opus 5.5-built codebase and finds issues Opus missed — CtrlAltDwayne · 2026-11-11
- Google Antigravity ships Swarm Mode: parallel agents coordinating via shared message board — testingcatalog · 2026-11-11
- Comet Builds Docker-Based Incident Testbed to Evaluate Remediation Agents with Opik — craigsdennis · 2026-11-11
- Claude agent drives a real Leica microscope via MCP, autofocus and all — jwt0625 · 2026-11-11
- Yacine's agent talk: context moves outside the model, harnesses self-edit — yacinelearning · 2026-11-11