sparkgen-chat

name	sparkgen-chat
description	Send a chat message to the running agent server and display the response
user_invokable	true
auto_invokable	false
arguments	<message> [--agent name] [--session id] [--max-iterations N]

SparkGen Chat

Send a chat message to the running SparkGen agent and display the full response.

Dynamic Context

Before sending:

1. Check server health: curl -sf http://localhost:8000/health
   • If not running: suggest /sparkgen-deploy local or make local
2. Get API key from .env (default: dev-local-key)

Actions

Send Chat Message

Build and send the request:

curl -s -X POST http://localhost:8000/v1/chat \
  -H "Content-Type: application/json" \
  -H "X-API-Key: ${API_KEY:-dev-local-key}" \
  -d '{
    "message": "<user message>",
    "agent_name": "<agent or empty for default>",
    "session_id": "<session or auto-generated>",
    "max_iterations": <N or 10>
  }'

Display Response

Parse and display the JSON response:

• Response text: the agent's reply
• Agent used: which agent handled the request
• Tool calls: list any tools that were invoked (name + args + result summary)
• Iterations: how many LLM calls were made
• Latency: response time
• Session ID: for follow-up messages

Multi-Turn Conversation

When --session is provided, the agent maintains conversation history. Use the same session ID for follow-up messages to test multi-turn behavior.

Examples

/sparkgen-chat "What tools do you have?"
/sparkgen-chat "Search for deployment docs" --agent researcher
/sparkgen-chat "Follow up on that" --session abc123

Error Handling

• Server not running → suggest deployment
• 401 Unauthorized → check API_KEY in .env
• 500 error → show error detail, suggest /sparkgen-debug general
• Timeout → check Ollama is responding (curl http://localhost:11434/api/tags)