Context Extension API

Enable unlimited conversation length with automatic context management.

Overview

Context Extension automatically manages long conversations by:

• Tracking all messages in a session
• Compressing old context when approaching token limits
• Retrieving relevant historical context for new messages

This allows conversations of any length while staying within LLM context limits.

Create Session

POST /v1/chat/sessions

Request Body

Field	Type	Default	Description
name	string	null	Session name
system_prompt	string	null	System prompt
context_budget	int	8000	Max tokens to maintain
space_id	string	null	Link to memory space

Example

curl -X POST https://api.memory.tensorheart.com/v1/chat/sessions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Project Planning",
    "context_budget": 16000
  }'

Response

{
  "success": true,
  "data": {
    "id": "session_abc123",
    "name": "Project Planning",
    "context_budget": 16000,
    "message_count": 0,
    "created_at": "2024-01-15T10:30:00Z"
  }
}

---

Send Message

Add a message to a session.

POST /v1/chat/sessions/{session_id}/messages

Request Body

Field	Type	Required	Description
role	string	Yes	"user" or "assistant"
content	string	Yes	Message content

Example

curl -X POST https://api.memory.tensorheart.com/v1/chat/sessions/session_abc123/messages \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "role": "user",
    "content": "Let me tell you about the project requirements..."
  }'

The system automatically compresses old context when approaching the token budget.

---

Get Session Context

Get optimized context for LLM calls.

GET /v1/chat/sessions/{session_id}/context

Returns recent messages plus relevant historical chunks, optimized for your context budget.

---

List Sessions

GET /v1/chat/sessions

---

Get Session

GET /v1/chat/sessions/{session_id}

---

Delete Session

DELETE /v1/chat/sessions/{session_id}

How Context Management Works

Messages added → Token count tracked → Budget exceeded?
                                              ↓
                                    Compress old messages
                                              ↓
                                    Store as searchable chunks
                                              ↓
                        Retrieve relevant chunks for new queries

This ensures you always have the most relevant context, regardless of conversation length.