REST API Usage Guide

The GAM REST API provides a standalone, high-performance RESTful service built on FastAPI + Uvicorn. It offers automatic request validation, interactive API documentation, and CORS support out of the box.

Starting the API Server

# Default (port 5001, no pre-configured LLM)
python examples/run_api.py

# With a default LLM model
python examples/run_api.py --model gpt-4o-mini --api-key sk-xxx --port 5001

Once running, the interactive documentation is available at:

URL	Description
http://localhost:5001/docs	Swagger UI (interactive)
http://localhost:5001/redoc	ReDoc (read-only)

API Endpoints

1. Add Content (POST /api/v1/add)

Add text or video content to a GAM.

Request Body (JSON):

Parameter	Type	Default	Description
type	string	"text"	GAM type: "text" or "video"
gam_dir	string	(auto)	Path to the GAM directory
input	string / array	null	Input file paths or video directory
content	string / array	null	[Text] Raw text content to add
model	string	(env)	LLM model name
api_base	string	(env)	API base URL
api_key	string	(env)	API key
use_chunking	boolean	true	[Text] Use intelligent chunking
max_splits	integer	120	[Text] Maximum number of chunks
force_reorganize	boolean	false	[Text] Force hierarchical reorganization

Example:

curl -X POST http://localhost:5001/api/v1/add \
  -H "Content-Type: application/json" \
  -d '{
    "type": "text",
    "content": ["New research data...", "Secondary findings..."],
    "model": "gpt-4o",
    "use_chunking": true
  }'

import requests

payload = {
    "type": "text",
    "content": ["New research data...", "Secondary findings..."],
    "model": "gpt-4o",
    "use_chunking": True,
}
resp = requests.post("http://localhost:5001/api/v1/add", json=payload)
print(resp.json())

Response:

{
  "success": true,
  "type": "text",
  "gam_dir": "/path/to/gam",
  "output_dir": "/path/to/chunks",
  "created_files": 5,
  "new_directories": 2
}

---

2. Query / QA (POST /api/v1/query)

Query an existing GAM knowledge base.

Request Body (JSON):

Parameter	Type	Default	Description
type	string	"text"	GAM type: "text" or "video"
gam_dir	string	(required)	Path to the GAM directory
question	string	(required)	User question
model	string	(env)	LLM model name
max_iter	integer	10	Maximum exploration iterations
system_prompt	string	""	Optional system prompt override

Example:

curl -X POST http://localhost:5001/api/v1/query \
  -H "Content-Type: application/json" \
  -d '{
    "type": "text",
    "gam_dir": "./my_gam",
    "question": "Summarize the key findings.",
    "model": "gpt-4o-mini"
  }'

import requests

payload = {
    "type": "text",
    "gam_dir": "./my_gam",
    "question": "Summarize the key findings.",
    "model": "gpt-4o-mini",
}
resp = requests.post("http://localhost:5001/api/v1/query", json=payload)
print(resp.json()["answer"])

Response:

{
  "success": true,
  "question": "Summarize the key findings.",
  "answer": "The key findings are...",
  "sources": ["file1.md", "file2.md"],
  "confidence": 0.85,
  "notes": "",
  "files_read": ["dir/file1.md"],
  "dirs_explored": ["dir/"]
}

---

3. Health Check (GET /)

curl http://localhost:5001/

{
  "service": "GAM REST API",
  "version": "1.0.0",
  "docs": "/docs"
}

Python Client Example

A complete example is available at examples/rest_api_client.py.