browser-agent
Skill0 starsChanged 3 months agoLicence not identifiedView on GitHub
Execute a client's workflow LIVE in a real browser, driving a website through its UI, from a web-flow Knowledge Base (KB) plus a Statement of Work / solution doc (SOW). Built for the common case where a client wants a process automated but the product exposes NO usable API — the only surface is the screen. The KB (produced by kb-builder mapping the live site) is the sole authority on HOW the site works: its screens, URL patterns, element locators, success and failure cues, waits, and the human gates it puts in the way. The SOW is the sole authority on WHAT workflow to run. The agent resolves the two into an ordered run, confirms before anything that changes state, then executes it live — navigating, clicking, filling, asserting and capturing — and writes a redacted run record. It is INTERACTIVE by design, but for a REMOTE operator who cannot see or touch its browser (which runs headless on the server): when it meets an OTP or MFA challenge, an unexpected screen, an ambiguous target, a locator the site has since changed, or ANY value the SOW never specified, it PAUSES, says exactly where it is and what it needs, and resumes on their reply rather than guessing — asking only for things a person can send as text, never for them to take over the browser. The SOW is the single input document — it carries the environment values and credentials too, so there is no separate secrets file; where the SOW lacks a credential the agent ASKS the operator for it in chat (email, password, one-time code) and types it into the page itself, then never echoes, restates, logs or records the value. The operator never touches the browser and is never asked to. Where a gate genuinely needs a human AT the browser — a CAPTCHA or bot check, an OAuth/SSO consent screen, a signup, a bank's payment step-up — it STOPS and records that journey as not automatable unattended, which is a finding worth having rather than a failure. It signs in with email/password wherever the site offers that route, so a run never authenticates into the operator's other services just to get into one product. It never invents a page, selector or cue the KB does not document, never solves a CAPTCHA, and never treats a screen it cannot verify as success. Use when an operator has a web-flow KB and a workflow they want actually carried out in the browser. Triggers: "run this workflow in the browser", "execute the SOW against the site", "browser-agent", "automate this flow on {site} using the KB", "drive the website and do {task}", "run the browser automation for {client}".
- Reads credentials
7 other projects use a version of this.
Adopted by
Most recently taken up by
- OyadotAI/oya-browsertheir own version, first seen Nov 3, 2026
- deepgram/skillstheir own version, first seen Oct 25, 2026
- azderus/browser-agenttheir own version, first seen Oct 20, 2026
- derHaken/SuperAntigravitytheir own version, first seen Oct 8, 2026
- fabio-baumgartner/unternehmer-skillstheir own version, first seen Aug 26, 2026
- shanmukhaditya/agent-skillstheir own version, first seen Aug 23, 2026
Other versions7 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 7
- derHaken/SuperAntigravity★ 439same name
- OyadotAI/oya-browser★ 348same name
- deepgram/skills★ 26same name
- shanmukhaditya/agent-skills★ 3same name
- fabio-baumgartner/unternehmer-skills★ 1same name
- azderus/browser-agent★ 0same name
- hterzia/human-browser★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository10 other files it gives its agents.
- api-agentSkill
- apply-fixesSkill
- code-agentSkill
- config-agentSkill
- diagnose-bugSkill
- document-generatorSkill
- find-bugsSkill
- kb-builderSkill
- process-comparisonSkill
- runner-agentSkill
The file
We split this skill into the rules it gives an agent
10 rules. 0 are common, 8 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
8
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
The licence could not be identified — read it at the source. We count its rules but don't quote them. Read them on GitHub.
“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.