/route

Route task to the best AI provider based on their capabilities.

Raw task: $ARGUMENTS

Workflow

1. Get provider capabilities - See what each AI is good at
2. Analyze task - Understand what needs to be done
3. Decide provider - Choose the best match based on capabilities
4. Execute - Call runtime with your decision or execute yourself

Step 1: Get Provider Capabilities

Run:

node "${CLAUDE_PLUGIN_ROOT}/scripts/unitor-runtime.mjs" status

This shows each provider's strengths:

• Provider A: architecture, security, complex-reasoning, backend-api
• Provider B: frontend-ui, react, css, vue
• Provider C: backend-api, python, database, sql

Step 2: Analyze Task

Understand $ARGUMENTS:

• What type of work? (UI, API, database, architecture)
• What technologies? (React, Python, SQL)
• What complexity? (simple fix vs complex design)

Step 3: Decide Provider

Based on provider capabilities and task analysis, decide which provider best matches the task requirements.

If you are the best match, execute the task directly.

Step 4: Execute

If routing to another provider:

node "${CLAUDE_PLUGIN_ROOT}/scripts/unitor-runtime.mjs" route --provider=<provider-id> "$ARGUMENTS"

Replace <provider-id> with the chosen provider.

If you are the best match:

• Execute the task in $ARGUMENTS directly
• Don't call the runtime

Output Handling

If you called runtime:

• Present provider output verbatim to user
• If provider fails, execute the task yourself

Example

Task: "fix the login button styling"

1. Get status → Provider B has frontend-ui, css expertise
2. Analyze → this is UI styling work
3. Decide → Provider B is best match
4. Execute → route to Provider B