Configuration
Hosomaki is zero-configuration by default. The optional config file lets you override the LLM provider, model, endpoint, and output behavior.
Config file
Hosomaki searches for a config file named config.yaml in:
~/.config/hosomaki/$HOME
Example
yaml
# ~/.config/hosomaki/config.yaml
ai:
provider: ollama
endpoint: http://localhost:11434
model: llama3.2:3b
timeout: 120s # increase for slow hardware or large models
output:
color: true
language: en| Key | Default | Description |
|---|---|---|
ai.provider | ollama | AI backend (currently only ollama is supported) |
ai.endpoint | http://localhost:11434 | Ollama API endpoint |
ai.model | llama3.2:3b | Any model available via ollama list |
ai.timeout | 120s | Request timeout |
output.color | true | Colorized terminal output |
output.language | en | Language for AI responses |
Environment variables
Every config key can be overridden with an environment variable prefixed HOSOMAKI_, using the nested key path:
bash
HOSOMAKI_AI_MODEL=mistral hosomaki status
HOSOMAKI_AI_ENDPOINT=http://gpu-box:11434 hosomaki doctor