A curated list of resources, integrations, pipes, and community projects for Screenpipe: AI that knows everything you've seen, said, or heard.
Screenpipe captures screen text and audio locally while recording is enabled, making computer history searchable for people and AI assistants. Its source is available under the Screenpipe Commercial License, not an open-source license. Optional cloud AI, transcription, sync, and connected services can transmit context off-device; see the privacy data flow.
- Official Resources
- Official Libraries
- Scheduled tasks and pipes
- MCP Server
- AI Coding Tool Integrations
- Community Projects
- Hackathons and starter projects
- Service Integrations
- Tutorials & Articles
- Research and experiments
- Screenpipe: Main repository (desktop app + CLI)
- Documentation: Official docs
- API reference: Documentation in a single text file for API users and agents
- API recipes: Authenticated examples for search, meetings, frames, and memory
- CLI reference: Recording, search, agent setup, and pipe commands
- Scheduled Task Guide: Build custom pipes
- Discord: Community chat
- Resource library: Official use cases and integration guides
- screenpipe/uniOCR: Native OCR for macOS, Windows, Linux
- screenpipe/audiopipe: Fast speech-to-text in Rust (Qwen3-ASR, CoreML, DirectML, CUDA)
- screenpipe-mcp: MCP server npm package; see setup instructions
- Screenpipe Swift SDK: Swift Package Manager distribution for embedding capture in macOS apps; enterprise license and runtime requirements apply
- sck-rs: Rust screen capture library using Apple's ScreenCaptureKit
The app calls pipes Scheduled tasks. They are AI automations defined with a prompt and schedule; the CLI still uses screenpipe pipe and pipe.md. Browse them under Scheduled tasks > Discover.
- Task library: Available automations, including Obsidian sync, meeting intelligence, and time tracking
- Build scheduled tasks: Write and configure your own pipe
- Reliable reports: Define source windows, verify saved output, and check reruns before scheduling
- Task permissions: Scope the captured data a task can access
- Task troubleshooting: Diagnose failed runs and missing output
Screenpipe exposes an MCP (Model Context Protocol) server for retrieving recorded context and managing supported actions. Use Settings > Connections in the desktop app for the recommended setup: it configures the bundled runtime and SCREENPIPE_LOCAL_API_KEY. Manual setup requires a running Screenpipe instance; follow the official MCP installation guide. Retrieved context is available to the connected assistant and its configured model provider.
- Claude Desktop: Anthropic's desktop app
- Claude Code: CLI coding assistant
- Cursor: AI code editor
- VS Code + Cline: Autonomous coding agent for VS Code
- VS Code + Continue: Open-source AI code assistant
- Codex: Screenpipe setup for the Codex app and CLI
- Any MCP-compatible client
Use these Screenpipe-specific setup guides to connect your assistant to recorded context:
- Codex: Desktop and CLI integration
- Claude Code: MCP setup and work-history queries
- Cline: Screenpipe context inside VS Code
- Continue: MCP connection for the coding assistant
- OpenCode: Screenpipe MCP configuration
- Gemini CLI extension: MCP server and a skill for workflow recall and cited procedures
- GitHub Copilot CLI: Screen history and meeting context in the terminal
- OpenClaw: Connect Screenpipe memory to the personal assistant
- Hermes: MCP and skills for recall, meeting preparation, and worklogs
- Screenpipe Cloud: Read-only plugin for searching records already uploaded through Data Sync from Claude, ChatGPT, and Codex; requires Data Sync access
Descriptions below are based on each project's public documentation, checked October 7, 2026. These are independently maintained projects; current Screenpipe compatibility has not been runtime-tested. Entries marked as earlier Next.js pipes use the previous pipe architecture and may require porting. Check each project's setup, license, and model-provider requirements.
- Note Companion: Obsidian plugin with a Screenpipe integration for meeting notes and activity search
- Screenpipe Obsidian Go: Go service that turns Screenpipe activity into Obsidian work logs and daily summaries using OpenAI
- Screenpipe Distiller: Condenses daily capture into a Markdown memory document using OpenRouter and uploads it to a compatible memory backend
- Zelin's AI Assistant: Exports Screenpipe context to an Obsidian wiki and turns incoming requests into approval cards for Claude agents
- Agentic Cortex: Personal assistant framework combining structured Markdown memory, Screenpipe context, and OpenClaw or Claude Code
- WholeMem: MCP server combining Screenpipe captures, mem0 search, Obsidian daily notes, and workspace file history
- Atriveo Cortex: Ollama-powered inspector that extracts projects, commitments, and ideas from Screenpipe's SQLite data and lets users rate the results
- Personal Memory Engine: Dashboard for searching and condensing Screenpipe and OpenChronicle captures, with an OpenRouter-backed processing pipeline
- Mac Activity Tracker: Combines ActivityWatch, Screenpipe, and browser activity into daily Obsidian reports generated through Claude Code
- Screenpipe SQLite LLM Report: Configurable pipeline for turning captured OCR into reports and procedures through an LLM API; documentation in Chinese
- Ghostwork: Community GUI with a Screenpipe timeline, app-usage analytics, and macOS workflow automation
- Screenpipe Dashboard: macOS menu bar status and a timeline viewer for screenshots, OCR, and audio transcripts
- Screenpipe Manager: macOS recording controls and a local web dashboard with search and screenshot thumbnails
- Screenpipe Raycast: Raycast extension for managing pipes; built for the earlier pipe system
- Ramondino: Floating Windows assistant with local voice transcription, Open Interpreter, and optional Screenpipe history; documentation in Spanish
- blitz.screenpipe: Omarchy bar plugin that records meetings into an inbox and lets users file them into selected folders
- ScoreGoals: macOS activity tracker that uses Screenpipe context to compare recorded work with written goals and user corrections
- Rewind pipe: Earlier Next.js pipe with a searchable timeline, app-usage estimates, and OpenAI-powered chat over recorded history
- WorkToJiraEffort: Detects Jira issue keys in captured activity and logs work time to Jira, with optional Salesforce entries
- Stream Deck Workflow Advisor: Pipe that recommends Stream Deck actions, plugins, and layouts from recent desktop activity
- kordi Subscription Finder: Pipe that finds billing signals in screen text and sends subscription records to kordi
- focuspipe: Focus and context-switch tracking pipes with Goose integration, session summaries, and configurable nudges
- Screenpipe Plugins: Next.js pipe examples for meeting summaries, Notion sync, and Reddit posting, using the earlier pipe architecture
- Screenpipe Context Pipes: Connector scripts that summarize ActivityWatch, imported WHOOP data, YouTube sessions, and other metadata into Screenpipe memories; writes are opt-in
- OysterWorkflow Core: CLI pipeline for turning captured activity into candidate OpenClaw skills using an OpenAI-compatible model; its documented setup uses a pinned Screenpipe fork
- Meeting Summarizer pipe: Earlier Next.js pipe using Screenpipe transcripts and OpenAI to produce meeting notes, decisions, and action items
- Activity Summary pipe: Earlier Next.js pipe generating daily and weekly activity reports from Screenpipe context using OpenAI
- Engineer Bot pipe: Earlier Next.js pipe that drafts GitHub and Linear issue updates from recorded context, with posting actions and provider credentials
- Screenpipe Python Client: Python client for exploring and debugging Screenpipe's API
- Screenpipe SDK for Rust: Experimental community API wrapper; the author notes that most functions were not tested
- Screenpipe Electron template: Electron and React starter with a client for the local Screenpipe health endpoint; older template
- Screenpipe Tauri template: Tauri, React, and TypeScript starter with Screenpipe health and vision hooks; older template
These projects are useful examples to inspect and adapt. Their original setup may need updates for current Screenpipe versions.
- Meeting Maestro: Live meeting transcripts organized against predefined questions; additional question suggestions are listed as unfinished
- AI Interview Coach: Interview practice combining Screenpipe audio context, answer evaluation, and MediaPipe body-language tracking
- Screen Time Wrapped: Gemini-generated slides about the previous day's app and website usage
- Screen Productivity Tracker: Activity dashboard and Gemini-powered todo suggestions from captured work
- baseballwalkerchris/screenpipe-hackathon: Community hackathon repository
- Bright Path: Hackathon learning assistant using Screenpipe screen context and Gemini to help with educational tasks
- Lenz: Hackathon reading assistant bridging Screenpipe OCR and CrewAI through an MCP server
- wzrd.work: Hackathon workflow-capture prototype combining Screenpipe with computer-use and conversational tools
- CoBrain: Voice-controlled desktop prototype using Screenpipe context, Qdrant memory, Deepgram, and OpenAI agents
- continue.ai: macOS work-resumption prototype that turns Screenpipe activity into checkpoints using OpenAI vision, with ElevenLabs voice interaction
- MiniMe: Personal-memory prototype that combines Screenpipe OCR with a preference graph and Gemini-powered screen assistance
- BrokeBro: Hackathon prototype that detects food-ordering activity through Screenpipe OCR and triggers a Vapi/Groq voice reminder
These are proposals, not evidence of a shipped integration.
- PrivateGPT: Screenpipe integration: Closed proposal for local screen/audio context
- Aider: Screenpipe for screen context: Proposal for coding-session context
- Build Agents That Remember @42Paris: October 16–18, 2026, at École 42 Paris, organized by 42Entrepreneurs and Screenpipe. Tracks cover finding work to automate, teaching agents workflows, and making automations work for a team
- Screenpipe hackathon starter: Runnable examples for work-session SOPs, workflow review and replay, support escalation, handoffs, evidence search, and repeated-task discovery, with fictional sample data and tests; sample mode needs no AI key or Screenpipe installation
- Screenpipe Agentic Hackathon: Earlier Screenpipe agent-building event on Sprint.dev
Use the connections guide and connection reference for the current registry, authentication methods, and setup. Availability and account requirements vary by service.
| Category | Examples |
|---|---|
| Communication | Slack, Discord, Telegram, Gmail, Outlook, Microsoft Teams |
| Notes and documents | Notion, Obsidian, Google Docs, Google Drive, Logseq |
| Project management | Linear, Jira, Asana, Todoist |
| CRM and support | HubSpot, Salesforce, Pipedrive, Intercom, Zendesk |
| Developer tools | GitHub, Sentry, Vercel, Supabase, PostHog |
| Automation | n8n, Make, Zapier, Pushover, ntfy |
- Build a second brain: Official guide to organizing captured context into personal knowledge
- Screenpipe Blog: Official blog with guides and updates
- Screenpipe Docs: Getting Started: Quick start guide
- Feed Screenpipe into Petals: Petals' guide to sending captured activity to its transcript-ingestion API with a custom script; requires a Petals API key and has no native Screenpipe connector
- Screenpipe at South Park Commons: Technical talk and interactive deck covering capture, storage, the local API, and agent execution
These explain earlier pipe APIs and app layouts. Use the current setup guides above for installation and authentication.
- How to create an AI agent powered by your screen & mic: Louis Beaumont's January 2025 walkthrough of a Next.js activity-analysis pipe using Ollama
- How to use Llama3.2 to write daily logs in Notion based on your screen: October 2024 walkthrough and video of a Screenpipe-to-Notion workflow
- Meeting intelligence: Transcripts, speaker context, and meeting summaries
- Obsidian integration: Query Screenpipe history from an Obsidian AI plugin
- Workflow discovery: Find repeated tasks in recorded work
- Team SOP capture: Turn captured work into procedures for review
- Local or cloud AI: Choose where model processing happens
- Ollama: Configure local models with Screenpipe
- ScreenLeak: Benchmark for sensitive-information redaction in screen text, screenshots, and computer-use traces
- River AI + Screenpipe training: Experimental toolkit for curating selected chats and Screenpipe context into reviewed training data, running River fine-tuning, and evaluating results
- OBS PII Redactor: OBS Studio proof of concept using Screenpipe's redaction models to mask detected sensitive regions in video; image detection is experimental and text detection has additional limitations
- Secret Leak Guard: Experimental credential-pattern detector that reads Screenpipe OCR and raises desktop alerts; detection happens after Screenpipe has captured the text
Contributions welcome! Include a public project URL and a short description of how it uses Screenpipe. Prefer the author's repository, documentation, or demo. Label prototypes and integration proposals, and note any required cloud service or paid account. Avoid duplicate listings and unrelated tools that only mention Screenpipe.
If you've built something with Screenpipe, please open a PR to add it here.
