LLM server integration
VibePod added configuration for routing supported agents, including Claude and Codex, toward Ollama, vLLM, or other compatible model servers.
Patch releases
0.8.1 fixed auth token forwarding for Claude with vLLM. 0.8.2 fixed required flags for Codex IKWID mode.
Example commands
The 0.8 line added LLM server routing through config or environment variables for supported agents such as Claude and Codex.
VP_LLM_ENABLED=true \
VP_LLM_BASE_URL=http://host.docker.internal:11434 \
VP_LLM_API_KEY=ollama \
VP_LLM_MODEL=qwen3:14b \
vp run claude
VP_LLM_ENABLED=true \
VP_LLM_BASE_URL=http://host.docker.internal:11434/v1 \
VP_LLM_MODEL=qwen3:14b \
vp run codex