Skip to main content
POST
/
api
/
chat-history
/
inject
Recall chat history
curl --request POST \
  --url https://www.memoryplugin.com/api/chat-history/inject \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "query": "What decisions did I make about the database schema?",
  "maxTokens": 600,
  "platform": "claude",
  "conversationContext": "<string>",
  "conversationHistory": [
    {
      "role": "user",
      "content": "<string>"
    }
  ],
  "queries": [
    {
      "query": "<string>",
      "maxTokens": 123
    }
  ]
}
'
{
  "content": "<string>",
  "sources": [
    {
      "conversationId": "<string>",
      "conversationTitle": "<string>",
      "messageId": "<string>",
      "date": "<string>",
      "relevanceScore": 123
    }
  ],
  "tokenCount": 123,
  "processingTime": "<string>",
  "model": "<string>"
}

​
Overview

This is the primary endpoint for retrieving context from a user’s past AI conversations. It searches across all imported chat history, then uses AI to synthesize the results into a coherent summary with source citations. This is the same function that powers the recall_chat_history tool in the MCP server.

​
How it works

  1. Your query is embedded and searched against all conversation chunks (vector + keyword hybrid search)
  2. Results are filtered and ranked by relevance
  3. An AI model synthesizes the top results into a coherent narrative
  4. Source citations are attached so you can trace back to specific conversations
EndpointReturnsBest for
Recall (/inject)AI-synthesized summary + sourcesBuilding context for AI conversations, getting a coherent answer
Search (/search)Raw matched chunks with scoresBuilding custom UIs, debugging, fine-grained control

​
Parallel queries

For complex topics, use the queries array to search from multiple angles simultaneously. Each query is searched and synthesized independently. 
curl -X POST https://www.memoryplugin.com/api/chat-history/inject \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "queries": [
      { "query": "what database schema decisions were made", "maxTokens": 500 },
      { "query": "authentication architecture discussions", "maxTokens": 500 }
    ]
  }'

​
Single query

curl -X POST https://www.memoryplugin.com/api/chat-history/inject \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "what did I decide about the API rate limiting approach?",
    "maxTokens": 800
  }'

Authorizations

​
Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json
​
query
string

Natural-language description of what you're looking for in the user's chat history.

Example:

"What decisions did I make about the database schema?"

​
maxTokens
integer
default:600

Maximum tokens to allocate for the synthesized context. More tokens = richer detail but larger response.

Required range: x <= 2000
​
platform
enum<string>

Optional hint about the downstream chat platform to influence formatting.

Available options:
claude,
chatgpt,
typingmind
​
conversationContext
string

Short plaintext summary of the current conversation to ground retrieval.

​
conversationHistory
object[]

Recent dialogue turns to help ground the search.

​
queries
object[]

Array of parallel queries for complex topics (max 15). Each query is searched and synthesized independently. Use when a topic benefits from multiple search angles.

Maximum array length: 15

Response

Synthesized context from chat history

​
content
string

AI-synthesized summary of relevant chat history. Ready to inject into a conversation as context.

​
sources
object[]

Source conversations that contributed to the synthesized content.

​
tokenCount
integer

Actual token count of the synthesized content.

​
processingTime
string

Total processing time (e.g., '342ms').

​
model
string

AI model used for synthesis.