Harness · Kilo-Org/kilocode
Kilo Code
The IDE-native end of the spectrum — lineage Cline → Roo Code → Kilo Code. Included because a harness benchmark that only admits terminal agents quietly excludes where most people actually code.
AEQI builds one of the harnesses measured here. This board is in method preview and pre-registration — no performance scores are published yet, and we lead with our own losses, not our wins. Read the method & charter →
Result: not yet measured
The facts
- Maintainer
- Kilo-Org
- Source
- Kilo-Org/kilocode
- License
- MIT
- Category
- Coding harness
- Headless run
- Partial — primarily a VS Code extension; a CLI exists, but the editor loop is the main surface
- Native OpenRouter
- — can be pinned to one shared model
- Determinism notes
- The IDE-embedded surface is the fairness problem: the CLI is the only path to a controlled run, so it has to be smoke-tested before the number means anything.
Where it sits
Performance scores land here only with published traces, under the pre-registered method. Until then, this profile is the fair-run readiness check: can we start Kilo Code headless and pin it to one shared model to isolate the scaffold?