> ## Documentation Index
> Fetch the complete documentation index at: https://help.memoryplugin.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Using Chat History

> Add chat history context to your AI conversations

export const ArticleInfo = ({author, lastUpdated}) => {
  const authorAvatar = author === 'Alara' ? '/images/author-alara.jpg' : author === 'asad' ? '/images/author-alara.jpg' : null;
  const formatDate = dateInput => {
    if (!dateInput) return '';
    if (typeof dateInput === 'string' && !dateInput.match(/^\d{4}-\d{2}-\d{2}/)) {
      return dateInput;
    }
    try {
      const date = new Date(dateInput);
      const now = new Date();
      const diffTime = Math.abs(now - date);
      const diffDays = Math.floor(diffTime / (1000 * 60 * 60 * 24));
      if (diffDays === 0) return 'today';
      if (diffDays === 1) return '1 day ago';
      if (diffDays < 7) return `${diffDays} days ago`;
      if (diffDays < 30) return `${Math.ceil(diffDays / 7)} week${Math.ceil(diffDays / 7) > 1 ? 's' : ''} ago`;
      if (diffDays < 365) return `${Math.ceil(diffDays / 30)} month${Math.ceil(diffDays / 30) > 1 ? 's' : ''} ago`;
      return `${Math.ceil(diffDays / 365)} year${Math.ceil(diffDays / 365) > 1 ? 's' : ''} ago`;
    } catch {
      return dateInput;
    }
  };
  return <div style={{
    display: "flex",
    alignItems: "center",
    gap: "8px",
    marginBottom: "16px",
    padding: "8px 12px",
    backgroundColor: "var(--ifm-color-emphasis-100)",
    borderRadius: "6px",
    fontSize: "14px",
    color: "var(--ifm-color-content-secondary)",
    border: "1px solid var(--ifm-color-emphasis-200)",
    opacity: "0.8"
  }}>
      <div style={{
    width: "40px",
    height: "40px",
    borderRadius: "50%",
    background: authorAvatar || "linear-gradient(45deg, #4F46E5, #7C3AED)",
    display: "flex",
    alignItems: "center",
    justifyContent: "center",
    color: "white",
    fontWeight: "bold",
    fontSize: "18px"
  }}>
        {authorAvatar ? <img src={authorAvatar} alt={author} style={{
    width: "100%",
    height: "100%",
    borderRadius: "50%"
  }} /> : author?.[0]?.toUpperCase()}
      </div>
      <div>
        <div style={{
    fontWeight: "400",
    fontSize: "14px"
  }}>Written by <span style={{
    fontWeight: "600"
  }}>{author}</span></div>
        <div style={{
    fontSize: "14px"
  }}>Last updated <span style={{
    fontWeight: "600"
  }}>{formatDate(lastUpdated)}</span></div>
      </div>
    </div>;
};

<ArticleInfo author="Alara" lastUpdated="2026-07-04" />

Importing your chat history is step one. The payoff is what comes next: your AI reaching back into months of past conversations and pulling the relevant slice into the chat you're having right now.

## See it in action

**Without MemoryPlugin:**

<img src="https://mintcdn.com/memoryplugin/puS0CXqqwTMpEYkc/images/chat-history/mcp-without-memoryplugin.png?fit=max&auto=format&n=puS0CXqqwTMpEYkc&q=85&s=0fbd99d65aafbb1716b8a46f48bd9fe2" alt="Without MemoryPlugin" width="1526" height="1718" data-path="images/chat-history/mcp-without-memoryplugin.png" />

The AI has no context about your past conversations and gives generic answers.

**With MemoryPlugin:**

<img src="https://mintcdn.com/memoryplugin/puS0CXqqwTMpEYkc/images/chat-history/mcp-with-memoryplugin.png?fit=max&auto=format&n=puS0CXqqwTMpEYkc&q=85&s=333d21bb7b4de00b53955fee798893e6" alt="With MemoryPlugin" width="1518" height="1734" data-path="images/chat-history/mcp-with-memoryplugin.png" />

The AI sees relevant context from your chat history and responds based on your previous discussions.

## How it works behind the scenes

When chat history context is requested, MemoryPlugin doesn't dump raw chunks into your prompt. It retrieves and condenses:

1. **Query understanding.** Your query is analyzed to generate several search variations and extract any date filters (like "last month").
2. **Hybrid search.** Semantic and keyword searches run in parallel across your history.
3. **Reranking.** Results are reranked by relevance to your specific query.
4. **Context expansion.** For each match, surrounding messages are pulled in for complete context.
5. **Summarization.** The expanded context is summarized to fit your token budget. Tens of thousands of tokens of raw conversation might come back as a few hundred tokens of relevant summary.

Because it searches, reranks, and summarizes on each call, a recall takes a few seconds rather than being instant. That buys you condensed, relevant context instead of a wall of raw transcript.

## Browser Extension

The [Browser Extension](/integrations/browser-extension) works across all major AI platforms.

**Supported platforms:** ChatGPT, Claude, Gemini, Google AI Studio, Grok, Poe, DeepSeek, Qwen, and more.

**How it works:**

* Runs **one round of retrieval per message**
* Injects automatically (context on every message) or manually (only when you click the MemoryPlugin button)
* Uses the token limits configured in extension settings

The extension isn't calling a tool. It inserts an instruction asking the AI to emit a recall marker line that the extension detects and fills with context before the message is sent. Turn it on with the **Chat History Recall** toggle in the extension's settings.

<Note>
  Chat History Memory is separate from Regular Memory in the browser extension.
  They use different systems and different buttons.
</Note>

## Remote MCP Server (more control)

The [Remote MCP Server](/integrations/remote-mcp-server) is the most flexible way to use Chat History Memory. Instead of a marker, MCP clients call a real tool named **`recall_chat_history`**.

**Supported platforms:** Claude Desktop, Claude Web, Claude Mobile, Mistral AI, Cursor, Continue, and other MCP-compatible clients.

**What `recall_chat_history` can do:**

* **Multiple rounds of retrieval** in a single conversation, not just one per message
* **Parallel queries.** Send up to **15** queries that all run at once, each hitting a topic from a different angle (timeline, people, decisions, outcomes)
* **Per-query token budget.** `maxTokens` defaults to **600**, capped at **2000**; 300 to 1000 is the useful range. More tokens means richer detail but a bigger chunk of the conversation window
* **Date bounds.** Constrain a query with `before` / `after` using ISO dates like `2025-01-15`. Bare dates are interpreted in UTC and inclusive on both ends
* **Speed or quality mode.** Defaults to **speed**; ask for **quality** mode for slower but more thorough recall on hard or ambiguous queries

**How you drive it:**

You drive it with plain-language instructions to the AI:

```
"At the start of every conversation, use recall_chat_history to load
context about my skills and learning style."

"Run recall_chat_history with several parallel queries about this
project from different angles: technical details, design decisions,
user feedback, implementation challenges, and future plans."
```

The AI makes the calls you ask for, and weaves the condensed context back into its response.

<Info>
  **Best experience:** Claude (Desktop, Web, or Mobile) and Mistral AI. These
  models are good at agentic tool use and at folding retrieved context into an
  answer naturally.
</Info>

### Controlling what gets added

Use [exclusions](/features/chat-history/dashboard#excluding-chats) to keep specific chats out of context: sensitive conversations, off-topic chats, or outdated discussions.

When you exclude a chat, its content is deleted and it won't be re-imported on future uploads or syncs. A placeholder remains so it isn't re-processed.

## Searching and analyzing your history

You can also search and question your history in the dashboard:

<CardGroup cols={2}>
  <Card title="Search your history" icon="magnifying-glass">
    Find specific discussions by topic, keyword, or timeframe with hybrid search.

    [Go to Dashboard → Search](/features/chat-history/dashboard#search-tab)
  </Card>

  <Card title="Ask questions" icon="message-question">
    Get answers synthesized from your history, with citations, using the unified Ask tool.

    [Go to Ask](/features/ask)
  </Card>
</CardGroup>

## Tips for best results

<AccordionGroup>
  <Accordion title="Be specific in your instructions" icon="message">
    With the MCP server, tell the AI exactly what to recall. Instead of "use MemoryPlugin," say "recall context about my React projects" or "recall my cooking preferences."
  </Accordion>

  <Accordion title="Use parallel queries for broad topics" icon="arrows-split">
    For thorough coverage, ask for several parallel queries at once (up to 15): "recall this from a few angles: technical details, design decisions, and user feedback."
  </Accordion>

  <Accordion title="Match the depth to the task" icon="sliders">
    Sometimes you want a lot of context (many queries, higher token budgets), sometimes none. Both are fine, and depth helps even for creative writing or documentation where past context keeps things consistent.
  </Accordion>

  <Accordion title="Control timing" icon="clock">
    Tell the AI when to recall: at the start of every conversation, only when you mention it, or just for certain kinds of questions.
  </Accordion>
</AccordionGroup>

## Limitations

<AccordionGroup>
  <Accordion title="Doesn't track how facts change over time" icon="timeline">
    Chat History doesn't yet reason about how a fact evolved. If you struggled with a topic in March but understood it by June, the AI may not know which is current. We're working on this.
  </Accordion>

  <Accordion title="Results vary by model" icon="gauge">
    How well the context lands depends on how good your AI model is at using retrieved context. Some models incorporate it better than others.
  </Accordion>
</AccordionGroup>

## Troubleshooting

<AccordionGroup>
  <Accordion title="Context not appearing" icon="message-slash">
    **Try**:

    * Confirm the Browser Extension or Remote MCP Server is installed and signed in
    * For the browser extension, check that **Chat History Recall** is enabled in settings
    * For MCP, explicitly ask the AI to use `recall_chat_history`
    * Confirm you've uploaded and processed chat history
  </Accordion>

  <Accordion title="Irrelevant context" icon="filter">
    **Try**:

    * Be more specific in your prompt or instructions
    * Exclude irrelevant chats from the Chats tab
    * For MCP, ask the AI to use fewer tokens or narrower queries
  </Accordion>
</AccordionGroup>

<Warning>
  If issues persist, [contact support](https://memoryplugin.com) with the
  platform you're using, your integration method (Browser Extension or Remote
  MCP Server), and a description of the issue.
</Warning>

## Next Steps

<CardGroup cols={2}>
  <Card title="View Dashboard" icon="table-columns" href="/features/chat-history/dashboard">
    Search your history, ask questions, and manage your conversations
  </Card>

  <Card title="Browser Extension Setup" icon="chrome" href="/integrations/browser-extension">
    Install and configure the Browser Extension
  </Card>

  <Card title="Remote MCP Server Setup" icon="server" href="/integrations/remote-mcp-server">
    Set up the Remote MCP Server for Claude Desktop and other clients
  </Card>

  <Card title="Back to Introduction" icon="arrow-left" href="/features/chat-history/introduction">
    Review what Chat History Memory is and why it matters
  </Card>
</CardGroup>