AI agent infrastructure · Official sources reviewed 2026-10-07
Agentic Web Check is an open-source browser evaluator for websites used by AI agents. Its local CLI examines accessibility snapshots and live controls, runs optional browser tasks with programmatic success assertions, and produces actionable reports covering perception, navigation, interaction, machine interfaces and safety.
Choose for diagnosing whether browser agents can navigate a website you maintain and for adding explicit task assertions to CI checks.
The score is based on heuristics, not a validated vendor-performance benchmark. LLM task results vary, browser challenges are reported rather than bypassed, and upstream browser behavior can change.
Install with npm or run the documented npx command using Node 20 or newer and an available Chromium build. Scan a permitted website, define goals and success assertions for behavioral tests, and inspect failures in JSON, HTML, Markdown or screenshot artifacts. Start with read-only tasks and explicitly configure any task that submits forms or changes state.
The deterministic scan and baseline agent need no model key. The optional LLM runner accepts OpenAI-compatible endpoints or Anthropic credentials, including suitable local endpoints; hosting and model-provider usage are separate costs. No hosted subscription was established.
What it can do
Unknown is different from unavailable. Each fact carries its own evidence.
| Capability | Value | Evidence | Checked |
|---|---|---|---|
| Browser evaluation | 43 documented checks inspect a live DOM and AI-mode accessibility snapshot. | Facts sourced | 2026-10-07 |
| Task verification | Programmatic assertions produce pass, fail, blocked or inconclusive verdicts with recorded browser actions. | Facts sourced | 2026-10-07 |
Understand the total cost
Free
Free
MIT-licensed local software; no product subscription is documented. Optional model-driven runs can incur provider charges.
Pre.devpre.dev is a long-horizon coding agent that plans, builds, and verifies software over extended runs in the browser or terminal. It connects to many SaaS tools, publishes previews, and sells separate RL coding tasks to AI labs through pre.dev Labs.
Explore tool
cmuxcmux is a free, open-source macOS terminal built on Ghostty for multitasking with AI coding agents. It adds vertical tabs, split panes, notification rings, an in-app browser, and optional cloud VMs on paid plans for remote agent sandboxes.
Explore tool
AgentMailAgentMail is an email inbox API built for AI agents, giving each agent its own inbox to create threads, send and receive mail, handle attachments, and integrate via SDKs or MCP. The product targets developers wiring agents into real email workflows with multi-tenant controls and optional custom domains.
Explore toolThe practical questions
No. The deterministic perception checks and baseline agent run without an LLM API key.
No. The README describes the baseline as a floor and documents heuristic limitations; test the actual assistant and tasks separately.
Agentic Web Check is an open-source browser evaluator for websites used by AI agents. Its local CLI examines accessibility snapshots and live controls, runs optional browser tasks with programmatic success assertions, and produces actionable reports covering perception, navigation, interaction, machine interfaces and safety.
MIT-licensed local software; no product subscription is documented. Optional model-driven runs can incur provider charges.. This record lists ongoing free access; check the plan limits before starting.
No paid monthly price is listed; this record treats the product as free to start. See the plan cards for entitlements, billing commitments and seat minimums.
Local CLI and CI commands; optional task runner calls configured model-provider APIs. No hosted product API is documented.. API access and subscription access may have different terms; consult the linked sources.
The score is based on heuristics, not a validated vendor-performance benchmark. LLM task results vary, browser challenges are reported rather than bypassed, and upstream browser behavior can change.
No retained pricing changes yet. A current price alone does not establish a historical trend.
Official page retrieved and editorially reviewed; no hands-on vendor application test.
Read original source ↗Facts apply to the named version and check date. Send a sourced correction if something changed.