trufagent:config

trufagent:config

Reconfigures the model behind a specific agent role. Only ~/litellm-config.yaml changes — agent .md files stay the same.

Step 1 — Identify target agent

If no agent specified, show current fleet and ask which to reconfigure:

Configured agents:
!`cat ~/litellm-config.yaml 2>/dev/null | grep "model_name:" | grep -v "claude-sonnet" | sed 's/.*model_name: /  /'`

Which agent do you want to reconfigure?

Step 2 — Show alternatives

[AGENT] — [role description]
Current model: [current model]

Validated alternatives:
  1. [alternative 1]  ([provider])
  2. [alternative 2]  ([provider])
  3. Custom model (any valid LiteLLM model string)

Alternatives by role:

• scout: anthropic/claude-haiku-4-5, openai/gpt-4o-mini
• runner: anthropic/claude-haiku-4-5, openai/gpt-4o-mini, google/gemini-2.0-flash
• thinker: anthropic/claude-sonnet-4-6, openai/gpt-4o, google/gemini-2.5-pro
• builder: deepseek/deepseek-chat, openai/gpt-4o, mistral/codestral-latest
• writer: anthropic/claude-haiku-4-5, openai/gpt-4o-mini
• critic: anthropic/claude-sonnet-4-6 (cheaper, less powerful), openai/gpt-4o

Step 3 — Handle API key

If the chosen provider already has a key in config → skip. If new provider → ask for key → add as os.environ/PROVIDER_API_KEY.

Step 4 — Update ~/litellm-config.yaml

Update only the matching model_name: [agent] entry. Leave all other entries untouched.

Step 5 — Confirm

[agent] updated: [old model] → [new model]
Restart LiteLLM for the change to take effect:
  litellm --config ~/litellm-config.yaml

Notes

• Reconfiguring critic to a cheaper model reduces review quality on high-stakes changes
• Reconfiguring thinker to a faster model (groq) reduces reasoning depth
• Any LiteLLM-compatible model string works — see https://docs.litellm.ai/docs/providers