OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Evaluation & Observability

Watch Skill

Watch Skill is a Python developer tool that indexes video, audio and screen activity into searchable, timestamped evidence and checks results against deterministic contracts. It connects to agents through MCP, CLI or REST. DeepWatch is a browser-based workspace built on DeepSeek Harness with Watch Skill integrated.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Watch Skill — OpenViST project profile in Agent Evaluation & Observability.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Watch Skill when an agent workflow needs inspectable, timestamped evidence from media or a result check that is independent of the model's judgment. Use DeepWatch when you want those capabilities composed into a browser-based workspace rather than wiring the engine into an existing agent.

Who it is for

  • Developers adding media evidence retrieval or result verification to an existing AI agent.
  • Teams that want a browser-based agent workspace with tool-call receipts and retained results.

Documented uses

  • Index a recording and ask questions whose answers point to timestamps.
  • Retrieve frames, transcripts or OCR text from media and screen activity.
  • Check files, JSON values, SQL results, HTTP responses or DOM state against a frozen contract.
  • Review tool-call receipts and compare verdicts from two runs of the same contract in DeepWatch.
Before you choose

Requirements & limits

The project is open source under the MIT license. The documented package paths include Python on PyPI and Node packages on npm.

  • The base pip install cannot extract video frames; the README directs users to the standard extra for frames, retrieval and MCP.
  • OCR and local transcription are separate optional extras.
  • Watch Skill's engine and DeepWatch workspace are distinct choices: one integrates with an existing agent, while the other includes Watch Skill in a browser-based DeepSeek Harness workspace.
  • Verification distinguishes VERIFIED, FAILED, UNVERIFIED and INCONCLUSIVE rather than reducing every non-pass result to a failure.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install the standard Watch Skill package

Prepare the documented Python package setup for media frames, retrieval and MCP.

Before you start

  • Python package installation using pip.

Steps

  1. Install the standard extra with the command below.

    Documented command
    pip install 'watch-skill[standard]'

  2. Run watch-skill doctor to check the installation and repair what it can.

Expected outcome

The README describes the standard extra as providing frames, retrieval and the MCP server.

Guide limitations

  • The base install does not extract video frames.
  • The documented setup steps do not establish that a video has been successfully indexed or queried.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
oxbshw/watch-skill · Official homepage title · retained in its source language
How we review sourcesSuggest a correction