Skip to content
ƒtsforgev0.52.0
19

Model adapter

1 min read

tsforge does not ship a model. It calls your model through any OpenAI-compatible HTTP API (local vLLM, Ollama-compatible proxy, OpenRouter, etc.).

Configure models in ~/.tsforge/models.json. Override per run with environment variables listed below.

SettingDefault
Base URLhttp://localhost:8000/v1
Modeldeepseek-v4-flash
Max tokens16384

These defaults match tsforge’s dev setup. Replace baseUrl and model with yours. The gate and guardrails stay the same regardless of provider.

VariablePurpose
TSFORGE_BASE_URLAPI endpoint
TSFORGE_MODELmodel name
TSFORGE_API_KEYAPI key when required
TSFORGE_MAX_TOKENSoutput token cap
TSFORGE_CONTEXT_WINDOWcontext size for compaction
TSFORGE_COMPACT_ATauto-compact threshold (default 0.8)

Full list: Environment variables.

If model prose repeats degenerately (not tool JSON), tsforge aborts the run to avoid infinite token burn.

Quickstart · Big picture