screenpipe

Turn your screen and audio history into context any AI agent can use

Intermediate API
Screenshot of screenpipe, Turn your screen and audio history into context any AI agent can use

What is screenpipe?

screenpipe is an AI screen analyzer tool that turns your screen and audio history into context any AI agent can use. Screenpipe records screen, audio and app activity locally and exposes it through a local database, REST API and MCP server so AI agents can search your work history, reconstruct past moments and spot recurring workflows worth automating.

Screenpipe is a desktop application that continuously records screen activity, audio and app history on your own machine, then exposes that record so AI models and agents can search it and reason over it. The premise is simple: most of the context a model needs already exists in the work you did today, in the documents you opened, the meetings you sat through and the tabs you closed. Screenpipe keeps that raw history local by default and makes it queryable instead of leaving it scattered across apps. The capture layer combines accessibility data with screen frames and OCR fallback, so text is picked up even when applications do not expose it directly, and audio transcripts cover the moments where nothing was typed. A local database, a REST API and an MCP server sit on top, which means Claude, ChatGPT, Codex or a custom agent can pull from the same store. Because the history is portable and inspectable, it is not tied to a single model or vendor. The more distinctive use is workflow discovery. Screenpipe reviews recurring routines across your apps, lays out the steps and exceptions behind a repeated process, and links every step back to source evidence. A supplier invoice reconciliation, for example, surfaces as a sequence with the invoice email, matching check and tracker update shown side by side, with a suggested automation to review. You decide what gets automated, and approval steps can deliberately stay manual. Exclusions for apps, windows and URLs plus on-device redaction keep sensitive material out of storage before it lands. It suits operations staff, consultants, analysts and developers who lose time reconstructing what happened and who want their agents grounded in real evidence rather than memory. Meetings recording is one output, not the core product. Teams that need device management, identity and audit trails can push policy from a central workspace while raw history stays on each machine. Free downloading is offered; specific plan pricing is not published on the homepage, so check the site for current terms.

How do you use screenpipe?

  1. 1Download and install the desktop app
    Grab the free build for macOS, Windows or Linux from the site and run the installer. The app starts capturing screen and audio context locally once permissions are granted.
  2. 2Set your capture exclusions first
    Before normal use, exclude apps, windows and URLs you do not want recorded and mark sensitive fields for on-device redaction. This policy is applied before anything is written to storage.
  3. 3Point an agent at the local history
    Connect Claude, ChatGPT or your own agent through the MCP server or REST API so it can query the local database. The same question can then be asked with and without your work history for comparison.
  4. 4Search back through your day
    Use the recall view to scrub to a specific screen, document or conversation you saw earlier, then continue from that point rather than rebuilding the context by hand.
  5. 5Review a discovered workflow
    Open the workflow discovery view to see recurring routines across your apps, inspect example runs and check the evidence behind each step before approving any change.
  6. 6Read real user workflows for ideas
    The wall of love collects specific setups from users, from daily summaries to audit trails, which is a useful starting point for shaping your own capture and agent setup.

Pros and cons

Pros

  • Runs on macOS, Windows and Linux with source available for inspectionAI
  • Raw history stays local by default, with app, window and URL exclusions plus on-device redactionAI
  • Local database, REST API and MCP server let Claude, ChatGPT, Codex or custom agents draw on the same historyAI
  • Workflow discovery maps recurring tasks step by step and links each step to the evidence behind itAI
  • Enterprise workspace covers devices, identity, policies and audit trails without centralising raw recordingsAI

Cons

  • Continuous capture of screen and audio raises real privacy questions, and the default setup needs careful exclusion rules before it is usefulAI
  • CPU, battery and storage overhead is acknowledged on the site but not quantified, so the cost of running it daily is hard to predict up frontAI
  • Workflow suggestions must be reviewed before use, so the automation benefit depends on manual curationAI
  • Pricing details are not published on the homepage, which makes budgeting harderAI
  • Value compounds only after weeks of history accumulate, so early use yields littleAI

How much does screenpipe cost?

Support

Questions can be sent to support@screenpi.pe. The site also carries an FAQ covering capture, privacy, resource use, agents and team deployment.

Integrations

The homepage shows the local database, REST API and MCP server as the integration paths, and lists Slack, Microsoft Teams, Notion, Linear, GitHub, Obsidian and HubSpot among the tools users connect.

Features

Combined screen, audio and app capture with accessibility data, screen frames, OCR fallback and audio transcripts. Local SQLite history exposed via REST API and MCP server. Recall search to jump back to a past screen or conversation. Workflow discovery that maps recurring routines, example runs and exceptions with source evidence. Exclusions for apps, windows and URLs, on-device redaction, and enterprise device, identity, policy and audit management.

Frequently asked questions about screenpipe

  • Does screenpipe have an API?
    Yes, screenpipe offers an API.
  • How do you use screenpipe?
    The walkthrough on this page covers 6 steps: 1. Download and install the desktop app 2. Set your capture exclusions first 3. Point an agent at the local history 4. Search back through your day 5. Review a discovered workflow 6. Read real user workflows for ideas.
  • What platforms does screenpipe support?
    screenpipe is available on MacOS, Windows and Linux.
  • What does screenpipe integrate with?
    The homepage shows the local database, REST API and MCP server as the integration paths, and lists Slack, Microsoft Teams, Notion, Linear, GitHub, Obsidian and HubSpot among the tools users connect.
  • What are the limitations of screenpipe?
    Continuous capture of screen and audio raises real privacy questions, and the default setup needs careful exclusion rules before it is useful. CPU, battery and storage overhead is acknowledged on the site but not quantified, so the cost of running it daily is hard to predict up front. Workflow suggestions must be reviewed before use, so the automation benefit depends on manual curation.

Status

StatusActive
Views0
Outbound clicks0
Added10/21/2026

Platforms

MacOSWindowsLinux

Pricing

FreeFreemiumSubscription