# browser-use/browser-harness

Browser Harness | Self-healing harness that enables LLMs to complete any task.

Repository: https://github.com/browser-use/browser-harness
Canonical: https://ross.abutalabs.com/products/browser-harness
Homepage: https://browser-harness.com
Language: Python
License: MIT
License Family: permissive
Topics: ai-agent, browser-agent, browser-automation, browser-use, cdp, cloud-browser, llm, playwright, web-automation, browser-use-box, browser-use-cloud, persistent-browser, telegram-agent, vps-agent
Last push: 2026-08-26T00:49:55+00:00

## Health v2 (maintenance only)
Score: 81/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 9
- inputs: {"age_days": 139, "days_push": 8, "days_rel": 8, "gap_med": 6, "n_releases_24m": 10}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 17160, forks 1653 (observed 2026-08-28T04:11:17.737952+00:00)

## What it is
Browser Harness is a self-healing Python harness that connects LLM agents directly to a real browser via a single editable CDP websocket. Agents write their own reusable helper functions as they work, so the harness improves with every task and can run on local Chrome, stealth cloud browsers, or a VPS.

## Use cases
- automate browser tasks with an LLM agent
- let a coding agent control my real Chrome browser
- scrape and download content from logged-in websites
- build self-healing browser automation workflows
- run persistent browser agents on a VPS 24/7
- automate file uploads and web interactions via CDP

## When to choose
- you want an LLM agent to operate your real, logged-in browser
- you need browser automation that adapts by writing its own helpers
- you prefer a thin layer directly on CDP rather than a heavy framework
- you want to scale from local browsing to parallel cloud browsers

## When to avoid
- you need simple deterministic scripting without an LLM in the loop
- you require a mature stable API for production-critical pipelines
- you want a no-code browser automation tool
- you cannot grant an agent remote-debugging access to your browser

## Facets
- artifact type: library
- maturity: active
- function: agent-framework, web-scraping, workflow-automation, llm-inference, developer-tools
- domain: web-development, large-language-models, developer-tools
- platform: python, cross-platform, cli
- tags: browser-automation, cdp, playwright, self-healing-agents, browser-agent, chrome-devtools-protocol, web-automation, llm-agents, ai-agents, automation

## Member repositories
- browser-use/browser-harness (main) score 81

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:17.737952+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:03:29.149736+00:00, confidence not recorded.
  - readme: https://github.com/browser-use/browser-harness (fetched 2026-08-28T04:11:17.737952+00:00, sha 772210ad95b2)
  - homepage: https://browser-harness.com (fetched 2026-08-29T08:01:28.146723+00:00, sha 53f667a04550)
  - registry_pypi: https://pypi.org/pypi/browser-harness/json (fetched 2026-08-29T08:01:28.149589+00:00, sha eeeaca3a1387)
- Data as of 2026-08-30T08:39:29.467469+00:00.