screenpipe vs Runbear (2026)

screenpipe

by Mediar, Inc.

Runbear

by Runbear, Inc.

Pros

• ✓Local-first by architecture rather than by policy: capture, storage and search all sit on your own machine, cloud model calls send only the query, and Ollama support means the whole loop can run without a single cloud call.
• ✓A first-party MCP server rather than a claim: npx -y screenpipe-mcp exposes search-content and export-video over a local REST API on port 3030, with a one-click install into Claude Desktop from the app's own settings panel and documented stdio setups for Claude Code, OpenAI Codex, Cursor and Warp.
• ✓Nothing about the price or the setup is gated: all three tiers publish a rate with no quote-only step and the currency sits in the site's own structured data, and getting running is a desktop install on macOS, Windows or Linux against a documented five-minute quickstart.

Limitations

• ⚠The trust portal shows SOC 2 Type 2, ISO 27001, GDPR and HIPAA all as COMPLIANT with no in-progress qualifier anywhere on the page, and the site's own security page does not corroborate it, so we publish only SOC 2 Type II, GDPR and CCPA. There is little independent evidence to check the vendor against either: 21,150 GitHub stars but only two Product Hunt reviews and no G2 profile.
• ⚠Section 4 of the terms has you consent, by using the service, to the vendor creating Deidentified Data and Analytics Data from your usage and using it to improve and train its models. Customer content, output and personal information are excluded without separate written agreement, but the deidentified pathway is consent-by-use rather than opt-in.
• ⚠Public source is not an open-source licence: commercial use requires a paid licence under the Screenpipe Commercial License, and GitHub reports the repository licence as Other. Running it also costs roughly 5 to 10 GB of local disk per month, and on macOS the app captures nothing until screen recording and accessibility access are granted in System Settings.

Pros

• ✓Works in both MCP directions, acting as a client from Slack and exposing its own hosted MCP server for Claude Code, Claude Desktop, Cursor, VS Code, Windsurf and Zed.
• ✓Credit-based pricing with no per-seat charge, so cost tracks usage rather than headcount, and both published tiers list their exact credit allowance.
• ✓Per-user authorization on shared agents plus a SIEM-exportable audit log, and the vendor states plainly that customer data is not used for model training.

Limitations

• ⚠The MCP server is in early access with an expanding tool set, and the vendor states some agent settings remain dashboard-only.
• ⚠The vendor publishes two different integration counts, 2,500+ on its integrations page and 2,000+ in its llms.txt and its own submission, and we could not establish which is current.
• ⚠Independent evidence is thin at one G2 review and seven Product Hunt reviews, with no Capterra or Trustpilot presence, and the SOC 2 Type II report is gated behind a trust center access request.