Suggest a Tool

Email us a link and we'll review it for inclusion.

Every listing on Neighbourhood Claw is hand-reviewed by an operator. We don't accept automated submissions yet — but we read every suggestion sent to:

mail hello@nbhdclaw.com

Include the URL, a one-line pitch, and (optionally) a category. We aim to respond within a week.

Neighbourhood Claw
Browse Tools Suggest a Tool
Clawd Cursor Infrastructure tool screenshot — openclaw.ai

Clawd Cursor

Clawd Cursor is an open-source desktop control skill that gives AI agents the ability to see a screen and operate a computer using the mouse and keyboard — without requiring separate API integrations

Added
1 month ago

About

Clawd Cursor is an open-source desktop control skill that gives AI agents the ability to see a screen and operate a computer using the mouse and keyboard — without requiring separate API integrations for each application. It uses a three-layer pipeline that starts with fast pattern matching, escalates to UI tree analysis, and only falls back to full screenshot vision when needed, which keeps costs low on routine tasks. It runs on Windows, macOS, and Linux and is MIT-licensed.

person_check
Best For

Developers building AI agents that need to operate desktop software without managing a sprawl of individual API keys, especially teams that want a local-first or cost-conscious automation setup.

Pros & Cons

check_circle

Pros

  • check Works with any desktop application through the UI rather than needing per-service API keys, dramatically simplifying multi-app agent workflows
  • check "Zero-LLM fast path" for common actions executes in milliseconds and incurs no API cost
  • check Three safety tiers (Auto, Preview, Confirm) give meaningful control over which actions agents can take without user approval
  • check Fully free and open source; running costs depend only on your chosen AI backend, and local Ollama support makes it free end-to-end
  • check Cross-platform support covers Windows, macOS, and Linux without separate implementations
cancel

Cons

  • close Vision-based control is inherently slower and less reliable than purpose-built API integrations when those are available
  • close Complex multi-app tasks that escalate to the screenshot layer will incur vision model API costs that can add up at scale
  • close Self-hosted setup requires developer familiarity with the codebase to configure AI backends and safety tiers
  • close Agent reliability depends on UI stability — application updates that rearrange interfaces can break existing workflows

More Infrastructure

Other tools in the same category.

View All arrow_forward