kernel

Use Kernel managed auth whenever an agent needs to do something on a website as a logged-in user, including browsing, operating, automating, testing, debugging, or inspecting authenticated content. Reuse or create auth connections, complete hosted login…

Best practices for using agent-browser with Kernel cloud browsers. Use when automating websites with agent-browser -p kernel, tuning stealth or proxy behavior, persisting profiles, handling iframes, discovering Kernel session IDs or live views, or cleaning up…

Build and debug Python browser automation with the Kernel SDK, including browser lifecycle, server-side Playwright, CDP, profiles, proxies, and reliable cleanup.

Profile a website for bot detection vendors using stealth vs non-stealth Kernel browsers. Use when analyzing bot detection on a website, comparing stealth effectiveness, identifying anti-bot vendors and products, or detecting challenge types.

Use the Kernel CLI to manage cloud browsers, apps, profiles, proxies, managed auth, API keys, projects, and organization limits. Use when installing or authenticating the CLI, looking up current command syntax, or choosing the right command group.

Systematically debug a Kernel cloud browser session — VM issues, network errors, Chrome crashes, page-load failures, and live-view problems. Use when a browser session misbehaves (e.g. ERR_HTTP2_PROTOCOL_ERROR, "browser not responding", blank/error pages,…

Compare two Kernel profile snapshots to investigate behavioral differences. Use when an issue (e.g. login failure, captcha, broken automation, vendor mismatch) reproduces on one profile but not another, or when a "good" vs "bad" profile needs to be diffed.…

Best practices for using browser-use's open-source browser-harness with Kernel cloud browsers over CDP. Use when driving a Kernel browser from browser-harness, extracting a CDP URL from `kernel browsers create`, or running multi-step or parallel harness…

Playbook for running a manual, end-to-end test of the pi-skillopt plugin in a real pi session — drive the skillopt-create / taskpack / finalize skills against an example, watch the loop progress, and analyze the run. Use when asked to "manually test…

Set up and run a SkillOpt-style optimization loop for a single agent SKILL.md, scored by Harbor rollouts. Gathers the skill, the rollout environment (backend, agent, model, preloads), generates a task pack if none exists, then starts the loop. Use when asked…

Close out a skillopt run — measure the optimized skill on the held-out test split, produce best_skill.md, and emit a clean diff + report against the original skill. Use when a skillopt loop has plateaued or the user asks to "finalize", "wrap up the skillopt…

Generate a Harbor task pack (tasks + verifiers + split manifest) from just a SKILL.md, so the skill can be scored by rollouts. Derives representative tasks from what the skill teaches, defines objective verifiers (outcome + skill-adherence), and emits…

Query and transform JSON with the jq CLI so the result is both correct and consumable by the next step. Covers raw vs encoded output, missing-key and error semantics, safe value injection with --arg/--argjson, multi-input handling, and the core transform…

Teaches an agent to author SQL schema migrations as reversible up/down pairs against a populated SQLite database, applied with the sqlite3 CLI. Every change ships as NNNN_name.up.sql plus NNNN_name.down.sql, where the down returns the database to its exact…

Toy skill used by the hello-harbor example. Teaches an agent to answer a question by writing the value, wrapped in <answer></answer> tags, to the file the task names.

browse websites, open a browser, automate a website, navigate to a URL, click buttons, fill forms, and scrape pages using kernel cloud browsers with playwright and CDP.

manage cloud browsers, take screenshots, run playwright scripts, and manage browser profiles using kernel's MCP tools. use when the kernel MCP server is connected.

build browser automation scripts using the kernel python sdk with playwright and remote browser management. use when writing python code that creates browsers, runs playwright, or automates websites with kernel.

build browser automation scripts using the kernel typescript sdk with playwright, CDP, and remote browser management. use when writing typescript code that creates browsers, runs playwright, or automates websites with kernel.

manage authentication for AI agents — log in and stay logged in across websites. use when setting up login credentials, authentication flows, or persistent authenticated sessions with kernel.

scrape websites, extract data, crawl pages, and browse the web at scale using kernel cloud browsers with stealth mode and proxies.

Improve repository test quality by eliminating flakes first, then reducing slow-test redundancy, and finally optimizing longest tests while preserving coverage. Use when asked to repeatedly run CI-equivalent no-cache test cycles, diagnose flaky tests, and…

Run Hypeman tests on a remote Linux host that supports the Linux hypervisors. Use when validating cloud-hypervisor, Firecracker, QEMU, embedded guest artifacts, or Linux-only integration behavior on a server over SSH.

Use when optimizing VM Initializing-to-Running latency while preserving functionality and low implementation complexity.

Add or adjust monitoring for a Hypeman feature using repository standards for logs, traces, and metrics. Use when a user asks for instrumentation, observability reviews, telemetry consistency changes, metric design, or production-signal improvements.

Prepare and publish @onkernel/cua-ai, @onkernel/cua-agent, and @onkernel/cua-cli npm releases from kernel/cua. Use when checking release readiness, choosing package versions, writing package changelogs, committing release metadata to main, pushing…

Keep repository documentation accurate by auditing docs against source code, configs, READMEs, and external references. Use when updating docs, checking architecture docs, reviewing README drift, or maintaining files under docs/.

Drive a Kernel cloud browser from the shell using the `cua` CLI. Use this skill when you need to open URLs, click elements, type into fields, inspect pages, fill forms, take screenshots, or chain multi-step browser tasks across shell calls. Supports named…

Discover latest OpenAI, Anthropic, Google/Gemini, Meta, xAI, Moonshot, Tzafon, and Yutori models and verify computer-use support. Use when updating CUA model defaults, checking new model releases, auditing provider-native computer tool actions, or comparing…

Publish, edit, and collaborate on HTML documents at justhtml.sh. Covers agent self-onboarding (the auth.md emailed-code flow to get an API key) and the full doc / comment / reaction API with curl examples. Use when asked to publish or host an HTML document,…

Use the Stripe Projects CLI in this repository to manage deploying and access to third party services.

Automate LinkedIn tasks using agent-browser connected to the local Chrome browser via CDP.

Drive a Kernel cloud browser to solve a web task — open a browser, read the page, act, observe, repeat — working autonomously but keeping a human in the loop, handing off when a step needs a person (a sign-in, a decision, or a sensitive action). Use whenever…