Local classification in OpenClaw with Ollama and Tev1

Description and use

Integrating local Tev1 models via Ollama into OpenClaw enables agents to make structured decisions locally without external APIs.

Problem

Integrating local Tev1 models via Ollama into OpenClaw’s Decision provider for structured agent outputs.

Outcome

The adapter validates answer types, question IDs, distribution keys, labels, numerical ranges and usage counters, limits bodies to 1 MiB, preserves caller cancellation and monotonic deadlines, and accepts only HTTP loopback origins.

Workflow

• 1. Clone the openclaw-ollama-decision repo. 2. Run npm test ollama pull tev1:0.8b and npm run smoke to validate the 0.8B model. 3. For the 4B model run ollama pull tev1:4b and npm run smoke -- --model tev1:4b --timeout-ms 120000. 4. Install the plugin via the README instructions. 5. In the target agent’s Decision picker select ollama-decision/tev1:0.8b or ollama-decision/tev1:4b. 6. Keep the chat‑model unchanged. 7. Verify integration with the desired OpenClaw version (e.g., 2026.9.8).

Evidence

OpenClaw’s developers added native support for a separate Decision model role. Our plugin connects local Tev1 models running in Ollama to that existing infrastructure.

Source

A real captured OpenClaw call with tev1:0.8b returned support, an escalation probability of about 0.8888, and an impact score of about 1.9096 on a zero-based 0–2 rubric.

Source

Requirements: Node.js 22+, an OpenClaw version exposing Decision provider contract v1 and registerDecisionProvider, and local Ollama with /v1/systemone.

Source

The adapter validates answer types, question IDs, distribution keys, labels, numerical ranges and usage counters.

Source