Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add evoelsewhere/evoflux --skill browser-usegit clone --depth 1 https://github.com/evoelsewhere/evofluxWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/evoelsewhere/evoflux/browser-use)<a href="https://agentmods.dev/skills/evoelsewhere/evoflux/browser-use"><img src="https://agentmods.dev/badge/skills/evoelsewhere/evoflux/browser-use/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/evoelsewhere/evoflux/browser-use"><img src="https://agentmods.dev/badge/skills/evoelsewhere/evoflux/browser-use.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00222 | $0.02921 |
| Opus 5.5 | $0.00089 | $0.01168 |
| Sonnet 5.5 | $0.00044 | $0.00584 |
| Haiku 4.5 | $0.00022 | $0.00292 |
Grade A, and why
browser-use scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 17d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 202 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Browser Use
browser_use is the one browser tool for page work. Every call runs in one
of two browsers, chosen by EvoFlux at call time, and the same pipeline works
in both: look, act a little, read the page back, repeat.
Contents
- Pick the right tool
- Which browser runs the call
- Ground rules
- The pipeline
- Choosing the action for a step
- Pages that change under you
- Verifying a web app you are building
- When something went wrong
- References
- Reporting
Pick the right tool
| The task | Use |
|---|---|
| Read a public page's text, or a document by URL | web_fetch |
| Find sources or current information | web_search, then web_fetch |
| Anything that needs a live page: clicking, forms, scripts that render content, a session where the user is signed in, a screenshot, console or network state | browser_use |
| A web app you are building | preview to start its server, then browser_use (see below) |
| Crawling many URLs, rich editors (Google Docs, Sheets, Office online), request mocking, network or CPU emulation | the webbridge tool, which is offered only while WebBridge is on (references/webbridge.md) |
| A desktop application window | the computer-app-control Skill |
Which browser runs the call
EvoFlux decides on every call; the choice is not a setting of the chat and it can change between two calls when the user flips the WebBridge switch in the workbench bar.
| In-app browser | User's browser (WebBridge) | |
|---|---|---|
| Chosen when | WebBridge is off, or no extension is connected | WebBridge is on in Settings, its switch is on, and an extension is connected |
| How to tell | results end with the list of in-app tabs | the first result line starts with Browser: the user's real Chrome/Edge through WebBridge |
| Sessions | EvoFlux's own browser profile | the user's real logins, cookies and extensions |
| Tab | tabs of the Browser panel | one tab of its own for this chat (below) |
| Refs look like | ref_12 |
e12 |
| A failed step | reported, later steps still run | the rest of the call is not run |
| Not available | — | query, page_assets, download, http, popups, dialog_behavior, clear_logs, permission requests, zoom, print, save_pdf, clipboard, dispatch_event, submit, scroll_into_view, dblclick, element screenshots, index targets |
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 17d ago First seen · 202 lines · 222 tokens per session scan A e3cb059c8f0e
browser-use is a skill published in the GitHub repository evoelsewhere/evoflux (11 stars, last pushed yesterday), licensed Apache-2.0. It adds 222 tokens to every session and 2,921 once invoked, about $0.0009 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-29.
Other skills, from other repositories
h5i
Red-team a web application under authorization by driving pages, capturing the HTTP traffic that produced them, inventorying what the target exposes, and replaying mutated requests with auditable evidence. Also covers ordinary browsing and scraping, running untrusted or agent-written code inside disposable confined…
control-chrome
Control the user's Chrome browser for tasks that depend on existing Chrome state: tabs, logged-in sessions, or extensions. Prefer purpose-built connectors, APIs, or CLIs when available.
checking-a-page
Check that a web page you wrote or changed actually works, by opening it in a browser and driving every control on it. Use after creating or editing an HTML page, or the JavaScript or CSS that a page loads.
browser
Interactive browser automation via playwright-cli. Use when you need to navigate a site, click elements, fill forms, take snapshots, or triage a live app.
yutori-computer-use
Run local Mac desktop tasks with Yutori computer use. Use when the user wants to operate macOS apps, websites in a local browser, or cross-app workflows on the visible desktop, or to work across app windows in the background while they keep working.
vibesurfer
Agent-native headless browser. 26 primitives over a Unix-socket wire protocol. Real WKWebView (macOS), WebKitGTK 6 (Linux), or WebView2 (Windows) — all three engines verified per-commit by a real-browser integration suite. Optimistic concurrency via state tokens; tree-delta wire format; durable session/page/auth state…