Operative (web-eval-agent)

By Operative. Category: MCP servers for testing. Homepage

The web-eval-agent README describes Operative's tool as an MCP server that launches a browser agent to execute and debug web apps from a code editor. Tasks are given to the agent in natural language. The agent returns a UX report rather than test code. The server runs locally and requires an Operative API key, which the README describes as free. Documented clients include Cursor, Cline, Windsurf and Claude Code. The README states that the project has been discontinued. The owner archived the repository on 27 June 2026.

Description written from the vendor's documentation.

Status

Status: Shut down

The web-eval-agent README states 'PROJECT HAS BEEN SUNSET' and 'This project has been discontinued. We're building something new at withrefresh.com'; the GitHub repository was archived by the owner on 2026-06-27.

DocumentedSourceVerified

Facts

Operative (web-eval-agent): facts with evidence level, source and verification date
FactValueEvidenceSourceVerified on
StatusShut downDocumentedgithub.com
AI capabilities
  • Autonomous testing
  • Results analysis
Documentedgithub.com
Test creation
  • Natural language
Documentedgithub.com
Test outputNot documentedNot documentedNone found
Surfaces
  • Web
Documentedgithub.com
Hosting
  • Hybrid
Documentedgithub.com
Pricing model
  • Free tier
Documentedgithub.com
Free tierYesDocumentedgithub.com
CI integrationsNot documentedNot documentedNone found
Self-healingNone documentedNot documentedNone found
Agent integrationIs itself an MCP server (web-eval-agent) exposing the tools web_eval_agent and setup_browser_state to Cursor, Cline, Windsurf and Claude Code; requires an API key from operative.sh/mcpDocumentedgithub.com
Open sourceYes, licence Apache-2.0Documentedgithub.com

Published prices

Not documentedVerified

No prices are published.

Vendor-stated, verbatim

What Operative says about its product, quoted as published and not confirmed in documentation.

  • “Navigate your webapp using BrowserUse (2x faster with operative backend)”
    SourceSeen

Alternatives

Tools in the same category (MCP servers for testing), ordered by the number of shared creation and surface values, then by name. Each line lists documented facts that differ from Operative (web-eval-agent).

  • Browser Use: documents hosting cloud, self-hosted
  • BrowserStack MCP Server: documents hosting self-hosted, cloud
  • Flutter Skill: documents output Mixed; creation by autonomous from URL, recording; hosting self-hosted
  • Glance: documents output Proprietary runner; creation by autonomous from URL, recording; hosting self-hosted
  • Vibium: documents hosting self-hosted
  • a11y MCP: documents hosting self-hosted

Questions

Is this page a review of Operative (web-eval-agent)?

No. This page is a fact sheet. Each fact shows its evidence level, its source and the date it was checked. We do not rate or rank tools.

Does Operative (web-eval-agent) have a free tier?

Yes. Operative (web-eval-agent) has a free tier. This is documented, checked on 2026-09-24.

What test output does Operative (web-eval-agent) produce?

The test output format is not documented as of 2026-09-24.

Is Operative (web-eval-agent) still available?

Status: Shut down, checked on 2026-09-24. The status note above gives the date and source.

Sources

  1. refreshdotdev/web-eval-agent README (GitHub, archived 2026-06-27) Opened
  2. Operative homepage Opened
  3. Operative MCP API key page Opened
  4. Refresh homepage (successor company named in README) Opened

Spotted an outdated or wrong fact about Operative (web-eval-agent)? Send the source and we apply it at the next monthly update.

Report a change