Quick Start

Give your AI agent persistent, intelligent memory in under 5 minutes. This guide walks you through the essential steps to get started.

1. Get Your API Key

Your API key looks like: mem_3e245c3069f3e096...

2. Store Your First Memory

The simplest way to store a memory is the /v1/simple/remember endpoint. It handles classification, embedding, and consolidation automatically.

curl -X POST "https://fathippo.ai/api/v1/simple/remember" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"text": "User prefers dark mode and concise responses"}'

Response:

{
  "id": "mem_abc123",
  "stored": true
}

Tip

The simple API auto-detects memory type (preference, fact, identity, etc.) and importance tier (critical, high, normal). No need to specify these manually.

3. Search Your Memories

Use /v1/simple/recall to search and retrieve memories. It returns just the text content—perfect for injection into prompts.

curl -X POST "https://fathippo.ai/api/v1/simple/recall" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"query": "user preferences", "limit": 5}'

Response:

{
  "results": [
    "User prefers dark mode and concise responses",
    "User likes TypeScript over JavaScript",
    "User works best in the mornings"
  ],
  "count": 3
}

4. Get Context for a Session

For a complete session workflow, use /v1/simple/context to get an injectable context string for your AI's system prompt.

curl -X POST "https://fathippo.ai/api/v1/simple/context" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"message": "Help me design an API"}'

Response:

{
  "context": "## User Context\n- User prefers REST over GraphQL\n- Timezone: GMT+8 (Singapore)\n- Prefers concise responses",
  "tokensUsed": 42,
  "memoriesUsed": 3
}

Note

The context is formatted as markdown that you can inject directly into your AI's system prompt. This gives your agent instant awareness of the user's preferences and history.

5. Full Session Flow (Optional)

For production agents, use the full session API for better tracking and analytics.

Start a Session

curl -X POST "https://fathippo.ai/api/v1/sessions/start" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"firstMessage": "Help me design an API"}'

Record Turns (Optional)

curl -X POST "https://fathippo.ai/api/v1/sessions/{sessionId}/turn" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "turnNumber": 1,
    "messages": [
      {"role": "user", "content": "Help me design an API"},
      {"role": "assistant", "content": "I can help with that..."}
    ],
    "memoriesUsed": ["mem_abc123"]
  }'

End Session

curl -X POST "https://fathippo.ai/api/v1/sessions/{sessionId}/end" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"outcome": "success"}'

Python Example

Here's a complete Python integration:

import requests

API_KEY = "mem_your_api_key"
BASE_URL = "https://fathippo.ai/api/v1"
headers = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json"
}

# Start a session and get context
def start_session(first_message: str) -> dict:
    response = requests.post(
        f"{BASE_URL}/sessions/start",
        headers=headers,
        json={"firstMessage": first_message}
    )
    return response.json()

# Store a memory
def remember(text: str) -> dict:
    response = requests.post(
        f"{BASE_URL}/simple/remember",
        headers=headers,
        json={"text": text}
    )
    return response.json()

# Search memories
def recall(query: str, limit: int = 5) -> list[str]:
    response = requests.post(
        f"{BASE_URL}/simple/recall",
        headers=headers,
        json={"query": query, "limit": limit}
    )
    return response.json()["results"]

# Example usage
session = start_session("Help me with my project")
print(f"Session ID: {session['sessionId']}")
print(f"Context: {session['context']}")

# Store what you learned
remember("User is building a REST API with FastAPI")

# Later, recall relevant context
memories = recall("What framework is the user using?")
print(memories)  # ["User is building a REST API with FastAPI"]

Next Steps

Core Concepts — Learn about memory tiers, decay, and reinforcement
API Reference — Full documentation of all endpoints
MCP Tools Reference — All 12 tools for Codex, Claude, Cursor, and OpenClaw
OpenClaw Integration — Deep integration via plugin and hooks
Git Hooks — Auto-capture commits to memory
OpenAI Integration — Add memory to GPT-based agents
Anthropic Integration — Add memory to Claude-based agents

Quick Start

Give your AI agent persistent, intelligent memory in under 5 minutes. This guide walks you through the essential steps to get started.

1. Get Your API Key

Your API key looks like: mem_3e245c3069f3e096...

2. Store Your First Memory

The simplest way to store a memory is the /v1/simple/remember endpoint. It handles classification, embedding, and consolidation automatically.

curl -X POST "https://fathippo.ai/api/v1/simple/remember" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"text": "User prefers dark mode and concise responses"}'

Response:

{
  "id": "mem_abc123",
  "stored": true
}

Tip

The simple API auto-detects memory type (preference, fact, identity, etc.) and importance tier (critical, high, normal). No need to specify these manually.

3. Search Your Memories

Use /v1/simple/recall to search and retrieve memories. It returns just the text content—perfect for injection into prompts.

curl -X POST "https://fathippo.ai/api/v1/simple/recall" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"query": "user preferences", "limit": 5}'

Response:

{
  "results": [
    "User prefers dark mode and concise responses",
    "User likes TypeScript over JavaScript",
    "User works best in the mornings"
  ],
  "count": 3
}

4. Get Context for a Session

For a complete session workflow, use /v1/simple/context to get an injectable context string for your AI's system prompt.

curl -X POST "https://fathippo.ai/api/v1/simple/context" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"message": "Help me design an API"}'

Response:

{
  "context": "## User Context\n- User prefers REST over GraphQL\n- Timezone: GMT+8 (Singapore)\n- Prefers concise responses",
  "tokensUsed": 42,
  "memoriesUsed": 3
}

Note

The context is formatted as markdown that you can inject directly into your AI's system prompt. This gives your agent instant awareness of the user's preferences and history.

5. Full Session Flow (Optional)

For production agents, use the full session API for better tracking and analytics.

Start a Session

curl -X POST "https://fathippo.ai/api/v1/sessions/start" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"firstMessage": "Help me design an API"}'

Record Turns (Optional)

curl -X POST "https://fathippo.ai/api/v1/sessions/{sessionId}/turn" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "turnNumber": 1,
    "messages": [
      {"role": "user", "content": "Help me design an API"},
      {"role": "assistant", "content": "I can help with that..."}
    ],
    "memoriesUsed": ["mem_abc123"]
  }'

End Session

curl -X POST "https://fathippo.ai/api/v1/sessions/{sessionId}/end" \
  -H "Authorization: Bearer mem_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"outcome": "success"}'

Python Example

Here's a complete Python integration:

import requests

API_KEY = "mem_your_api_key"
BASE_URL = "https://fathippo.ai/api/v1"
headers = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json"
}

# Start a session and get context
def start_session(first_message: str) -> dict:
    response = requests.post(
        f"{BASE_URL}/sessions/start",
        headers=headers,
        json={"firstMessage": first_message}
    )
    return response.json()

# Store a memory
def remember(text: str) -> dict:
    response = requests.post(
        f"{BASE_URL}/simple/remember",
        headers=headers,
        json={"text": text}
    )
    return response.json()

# Search memories
def recall(query: str, limit: int = 5) -> list[str]:
    response = requests.post(
        f"{BASE_URL}/simple/recall",
        headers=headers,
        json={"query": query, "limit": limit}
    )
    return response.json()["results"]

# Example usage
session = start_session("Help me with my project")
print(f"Session ID: {session['sessionId']}")
print(f"Context: {session['context']}")

# Store what you learned
remember("User is building a REST API with FastAPI")

# Later, recall relevant context
memories = recall("What framework is the user using?")
print(memories)  # ["User is building a REST API with FastAPI"]

Next Steps

Core Concepts — Learn about memory tiers, decay, and reinforcement
API Reference — Full documentation of all endpoints
MCP Tools Reference — All 12 tools for Codex, Claude, Cursor, and OpenClaw
OpenClaw Integration — Deep integration via plugin and hooks
Git Hooks — Auto-capture commits to memory
OpenAI Integration — Add memory to GPT-based agents
Anthropic Integration — Add memory to Claude-based agents