steel-dev
- 12 skills
- 0 followers
- 6 hours ago last updated
- ▌ Agent Browser · steel-dev bundleBrowser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
- ▌ Awesome Web Agents Pr Review · steel-dev bundleReview pull requests for the awesome-web-agents repository. Use when asked to validate a PR, assess whether a submission fits the repo, inspect README or ARCHIVE additions, check merge readiness, or apply the repository's web-agent-first contribution policy to a GitHub pull request.
- ▌ Benchmark Discovery · steel-devSurface new benchmark result candidates for the Steel leaderboard at /Users/nikola/dev/steel/leaderboard by running the arxiv discovery pipeline, filtering out off-topic and unscorable papers (e.g. ESA Gaia space-telescope hits vs the GAIA agent benchmark), and producing a standardised review-ready markdown summary with sections for proposed additions, variant candidates, and methodology notes — optionally posted to the rolling discovery issue on GitHub. Use this skill whenever the user mentions "discovery", "discovery sweep", "check arxiv", "survey new benchmark results", "find new leaderboard candidates", "what's new on the leaderboards", "any new agents on BrowseComp/GAIA/SWE-bench/WebVoyager", "see what's been published this week", "anyone reporting new scores", or any variation that involves scanning recent papers for fresh agent or model scores to potentially add to the leaderboard. Always trigger this skill before running an ad-hoc arxiv search by hand — the skill already knows the project's schema, de
- ▌
- ▌
- ▌ Atlas Research · steel-devRun the local atlas CLI to produce a cited deep-research markdown report. Use when the user explicitly invokes atlas — e.g., `/atlas-research <question>`, "run atlas on X", "use atlas to research Y". Returns the markdown to the conversation along with the run summary. Do NOT use for casual questions Claude can answer from training/web — this spends the user's model-provider + Steel credits and takes 1–4 minutes.
- ▌ Atlas Build Researcher · steel-devScaffold a domain-specific deep-research agent on the Atlas SDK. Use when the user wants to build or customize a research agent for a domain — medical, legal, financial, an internal API, a vector store — on top of @steel-dev/atlas, e.g. "build a research agent for X", "add a PubMed source to atlas", "make atlas search our internal docs". Generates a small module (instructions + domain tools + optional output schema) and smoke-tests it. Do NOT use to run a one-off query — that's atlas-research.
- ▌ Steel Skill Creator · steel-dev bundleUse this skill when the user wants to turn a recurring browser workflow into a reusable, parameterized agent skill, especially when the task has concrete inputs and a clear output such as scheduled scrapes, form submissions, data extraction, monitoring flows, price probes, or login-gated reports. Do not use for one-off web tasks; use steel-browser.
- ▌ Steel Session Debugging · steel-dev bundleUse this skill when the user provides a failed Steel session ID, failed automation run, timeout, browser error, blocked page, or unexpected result and wants evidence-backed diagnosis from session metadata, browser logs, raw agent logs, semantic traces, replay links, screenshots, network failures, or errors. Do not use for first-pass bot mitigation; collect evidence here, then hand off to steel-reliability for proxy, CAPTCHA, identity, or anti-bot fixes.
- ▌ Steel Developer · steel-dev bundleUse this skill when the user wants reusable code, scripts, examples, or docs that build on Steel cloud browsers with SDKs, REST APIs, Playwright, Puppeteer, Stagehand, Browser Use, credentials, profiles, files, extensions, embeds, proxies, or CAPTCHA APIs. Do not use for live web browsing performed by the agent; use steel-browser. Route failed-session diagnosis to steel-session-debugging and reliability mitigation to steel-reliability.
- ▌ Steel Reliability · steel-dev bundleUse this skill when a Steel automation appears unreliable because of bot detection, CAPTCHA loops, 403/access-denied or verify-you-are-human pages, proxy failures, profile or identity problems, credential or login persistence issues, suspicious pacing, redirect loops, or retry strategy, or when the user asks about stealth, proxies, CAPTCHA solving, profiles, credentials, or reliable logins. Do not use as a first step for unknown failed sessions; collect evidence with steel-session-debugging first when needed.
- ▌ Steel Browser · steel-dev bundleUse this skill for live web tasks where WebFetch or curl would fail or be insufficient: JavaScript-rendered pages, forms, screenshots, PDFs, login flows, CAPTCHA/bot-protection flows, and multi-step navigation. It runs a real Steel cloud browser through the Steel CLI. Use when the agent should visit, click, type, scrape, screenshot, or extract now. Do not use for writing reusable Steel SDK/API code; use steel-developer for code. If a live task fails, hand off to steel-session-debugging; if evidence points to bot detection, proxies, CAPTCHA, or identity, hand off to steel-reliability.