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

# MCP prompts for guided note capture

> Spote's three MCP prompts — capture, daily_summary, and find_related — guide your AI agent through note capture, daily review, and past-context retrieval.

In addition to tools, Spote exposes MCP prompts — pre-written instructions that tell your agent how to handle common note-taking workflows. Prompts are activated from your AI client: Claude Desktop shows them in the slash menu when you type `/prompts`, and other clients may surface them differently depending on their interface.

<Tip>
  Prompts work alongside tools — your agent can call `capture` to structure content and then automatically invoke `create_note` to save it.
</Tip>

## Available prompts

<AccordionGroup>
  <Accordion title="capture">
    **What it does**

    The `capture` prompt takes raw text and instructs your agent to extract a title, suggest an appropriate bucket, propose relevant tags, and then call `create_note` to save the structured note to Spote. It turns unstructured input — a paste, a thought, a paragraph — into a well-organized note without any manual work.

    **Parameters**

    * `text` — string, required. The raw text you want to capture.

    **How to trigger it**

    In Claude Desktop, type `/prompts` and select **capture**, then paste the text you want to save. In other MCP clients, look for a prompts panel or slash command that lists available prompts.

    **Example interaction**

    > **You:** `/prompts` → capture → "Talked to Sara today about the onboarding redesign. Key insight: users drop off because the first screen asks for too much information. We should reduce it to email only."
    >
    > **Agent:** I'll save this as a note in Spote. I'm extracting a title ("Onboarding redesign insight — Sara"), suggesting the bucket "Product", and proposing the tags `#onboarding` and `#ux`. Saving now\...
    >
    > **Agent:** Note saved. [View it in Spote →](https://your-spote-app.com/app/note/abc123)
  </Accordion>

  <Accordion title="daily_summary">
    **What it does**

    The `daily_summary` prompt instructs your agent to fetch all notes created or updated today and produce a structured summary grouped by bucket and tag. It gives you a quick end-of-day overview of everything you captured, organized by topic.

    **Parameters**

    None.

    **How to trigger it**

    In Claude Desktop, type `/prompts` and select **daily\_summary**. No additional input is needed — the agent fetches your notes automatically.

    **Example interaction**

    > **You:** `/prompts` → daily\_summary
    >
    > **Agent:** Here's what you captured today:
    >
    > **Work (3 notes)**
    >
    > * "Meeting recap: Q3 planning" — #mcp, #sprint
    > * "API design decision: pagination" — #api, #backend
    > * "PR review notes for auth refactor" — #auth
    >
    > **Ideas (1 note)**
    >
    > * "Blog post idea: AI memory patterns" — #writing, #ai
  </Accordion>

  <Accordion title="find_related">
    **What it does**

    The `find_related` prompt instructs your agent to search for notes related to a topic and present them as a list with brief summaries. It uses Spote's semantic search to surface notes by meaning, not exact wording — so you find relevant past thinking even when the phrasing doesn't match.

    **Parameters**

    * `query` — string, required. The topic or question you want to find related notes for.

    **How to trigger it**

    In Claude Desktop, type `/prompts` and select **find\_related**, then enter your topic or question. Use this before starting a task to pull in relevant past notes and decisions as context.

    **Example interaction**

    > **You:** `/prompts` → find\_related → "authentication patterns for API tokens"
    >
    > **Agent:** Here are the most relevant notes I found in Spote:
    >
    > 1. **"PAT design: token hashing approach"** (bucket: Engineering) — Documents the decision to use secure hashing with a server-side secret. Tokens are shown only once at creation.
    > 2. **"Auth middleware refactor notes"** (bucket: Engineering) — Covers the middleware that handles both session-based and token-based authentication.
    > 3. **"Meeting: API security review"** (bucket: Work) — Summary of security review, including token expiry discussion.
  </Accordion>
</AccordionGroup>