AgentX
Menu

Quick Start

Get AgentX running and make your first API calls.

Start the Stack

# Full stack: Docker services + Django API + Tauri client
task dev

This starts Neo4j, PostgreSQL, Redis, the Django API on port 12319, and the Tauri desktop app. All support hot reload.

For API-only development:

task db:up        # Start database services
task dev:api      # Start Django API (port 12319)

Health Check

Verify the API is running:

curl http://localhost:12319/api/health
{"status": "ok", "version": "0.1.0"}

Include database status:

curl http://localhost:12319/api/health?include_memory=true

Chat (Simple Completion)

Send a message and get a response:

curl -X POST http://localhost:12319/api/agent/chat \
  -H "Content-Type: application/json" \
  -d '{
    "message": "What is the capital of France?",
    "model": "llama3.2"
  }'
{
  "status": "success",
  "response": "The capital of France is Paris.",
  "session_id": "abc123",
  "model": "llama3.2"
}

Streaming Chat

Stream responses via Server-Sent Events:

curl -N -X POST http://localhost:12319/api/agent/chat/stream \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Explain quantum computing briefly",
    "model": "llama3.2"
  }'

Events arrive as SSE:

event: start
data: {"task_id": "t_abc123", "model": "llama3.2"}

event: chunk
data: {"content": "Quantum computing uses "}

event: chunk
data: {"content": "quantum mechanical phenomena..."}

event: done
data: {"task_id": "t_abc123", "total_time_ms": 1423.5, "session_id": "s_def456"}

Session Continuity

Pass session_id to maintain conversation context:

curl -X POST http://localhost:12319/api/agent/chat \
  -H "Content-Type: application/json" \
  -d '{
    "message": "What about its population?",
    "session_id": "abc123"
  }'

Translation

Detect Language

curl -X POST http://localhost:12319/api/tools/language-detect-20 \
  -H "Content-Type: application/json" \
  -d '{"text": "Bonjour le monde"}'
{
  "status": "success",
  "language": "fr",
  "confidence": 0.99,
  "language_name": "French"
}

Translate Text

curl -X POST http://localhost:12319/api/tools/translate \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Hello, world!",
    "targetLanguage": "fra_Latn"
  }'
{
  "status": "success",
  "translated_text": "Bonjour le monde!",
  "source_language": "eng_Latn",
  "target_language": "fra_Latn"
}

Target languages use NLLB-200 codes (e.g., fra_Latn, deu_Latn, spa_Latn). See Translation for the full language code reference.

MCP Tools

Connect a Server

curl -X POST http://localhost:12319/api/mcp/connect \
  -H "Content-Type: application/json" \
  -d '{"server": "filesystem"}'

Connect all configured servers:

curl -X POST http://localhost:12319/api/mcp/connect \
  -H "Content-Type: application/json" \
  -d '{"all": true}'

List Available Tools

curl http://localhost:12319/api/mcp/tools
{
  "status": "success",
  "tools": [
    {
      "name": "read_file",
      "description": "Read the contents of a file",
      "server": "filesystem"
    }
  ]
}

Once tools are connected, the agent can use them automatically during chat. See MCP for server configuration.

Prompt Profiles

List Profiles

curl http://localhost:12319/api/prompts/profiles

Use a Profile in Chat

curl -X POST http://localhost:12319/api/agent/chat \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Help me write a poem",
    "profile_id": "creative"
  }'

See Prompts for profile management.

Memory

Recall Memories

curl -X POST http://localhost:12319/api/memory/recall \
  -H "Content-Type: application/json" \
  -d '{"query": "user preferences", "top_k": 5}'

View Memory Stats

curl http://localhost:12319/api/memory/stats

Memory is automatically populated during chat when enable_memory is true (default). See Memory for the full memory system reference.

Database Access

task db:shell:postgres    # psql shell
task db:shell:redis       # redis-cli
task db:shell:neo4j       # cypher-shell

Neo4j web browser: http://localhost:7474

Next Steps