Repository navigation
Restructure How it works: Configure, Control, Scale, Observe, Manage - #671
andrewleesteele wants to merge 1 commit into
Conversation
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Automations to automatically generate PRs for you. |
| title: "Developing" | ||
| title: "Develop an App" | ||
| sidebarTitle: "Develop" | ||
| description: "Build an app with actions that run next to KERNEL browsers" |
There was a problem hiding this comment.
| description: "Build an app with actions that run next to KERNEL browsers" | |
| description: "Build an app with actions that run co-located with KERNEL browsers" |
|
|
||
| | Resource | Headful | Headless | | ||
| | --- | --- | --- | | ||
| | Default memory | 8 GB | 1 GB | |
There was a problem hiding this comment.
consider mentioning support for up to 16 GB in headful.
| </CodeGroup> | ||
|
|
||
| - **[Viewport](/browsers/viewport):** defaults to 1920x1080 at 25Hz. A custom viewport restarts Chromium on creation, so use a [browser pool](/browsers/pools) if you need it to be instant. | ||
| - **[Standby](/browsers/standby):** after 5 seconds with no CDP, WebDriver, live view, or computer controls activity, a browser goes into standby. It keeps its state and stops accruing usage cost until something reconnects. |
There was a problem hiding this comment.
i'd remove this here. not sure it belongs contextually.
|
|
||
| ## More browser settings | ||
|
|
||
| Most agents never need these, but they're there when you do: |
There was a problem hiding this comment.
| Most agents never need these, but they're there when you do: | |
| Often agents don't require these, but they're there when you do: |
| | [Playwright execution](/browsers/playwright-execution) | **Default.** You know what to do on the page — navigate, fill, extract, upload. | Needs a selector or DOM path that exists. | | ||
| | [Computer controls](/browsers/computer-controls) | **Recommended fallback.** A model is looking at pixels, or the page can't be driven programmatically. | Slower per step, and the model has to see the state to act. | | ||
| | [WebMCP](/browsers/webmcp) | The site exposes structured tools for the action you need. | Only works on sites that register tools. | | ||
| | [Browser REPL](/browsers/repl) | An agent writes its own helpers and reuses them across turns. | JavaScript only, and state lives until the REPL resets. | |
There was a problem hiding this comment.
I'd actually lean on Browser Repl as the default at this time for on-vm control
There was a problem hiding this comment.
"state lives until the REPL resets." i actually think this is a positive too of repl!
There was a problem hiding this comment.
On second thought, We can leave playwrighht execution as the default for now, but worth likely overhauling to default to browser repl once the dust settles on the docs
|
|
||
| | Surface | Use it when | Trade-off | | ||
| | --- | --- | --- | | ||
| | [Playwright execution](/browsers/playwright-execution) | **Default.** You know what to do on the page — navigate, fill, extract, upload. | Needs a selector or DOM path that exists. | |
There was a problem hiding this comment.
another trade-off here is that each playwright execution api call is stateless (vs. browser repl which is stateful across cells)
| **Costs:** a network round trip per action, disconnects to handle, screenshot and DOM bandwidth, and the CDP fingerprint above. It's fine for low-frequency or deterministic work, and it hurts most in a vision loop. | ||
| </Tab> | ||
| <Tab title="Playwright execution API"> | ||
| Send code, not commands. Each call runs in the browser's VM against the live session, so state carries across calls and an agent can drive the page turn by turn — one tool call per step, structured data back. |
There was a problem hiding this comment.
"so state carries across calls" this is true for browser repl but not playwright execution api
There was a problem hiding this comment.
i'd reco adding browser repl as a tab here too
|
|
||
| If you're building your own agent, [Browser Loop](/browsers/browser-loop) packages these surfaces as tools for each model provider and runs every action against a KERNEL browser, so you don't write the translation layer yourself. | ||
|
|
||
| ## Going deeper |
There was a problem hiding this comment.
i'd add browser repl in this list of additional reading options too
| @@ -0,0 +1,87 @@ | |||
| --- | |||
| title: "Browser Loop" | |||
There was a problem hiding this comment.
Consider removing from current PR and postpone inclusion till further internal discussion is done.
4647ff9 to
18ca8c3
Compare
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using default effort and found 3 potential issues.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 18ca8c3. Configure here.
|
|
||
| Often agents don't require these, but they're there when you do: | ||
|
|
||
| - **[Regions](/browsers/regions):** run browsers in `us-east`, `eu-west`, or `ap-southeast`, closer to your code and your users. |
There was a problem hiding this comment.
Configure copies region codes
Medium Severity
The new Configure overview inlines the shipped region codes us-east, eu-west, and ap-southeast instead of only linking to the canonical Regional Browsers page.
Triggered by learned rule: Regional browsers: latency not residency; canonical /browsers/regions
Reviewed by Cursor Bugbot for commit 18ca8c3. Configure here.
| | `X-RateLimit-Remaining` | Requests remaining in the current window | | ||
| | `Retry-After` | Seconds to wait before retrying (only on `429` responses) | | ||
|
|
||
| All Kernel SDKs retry a `429` up to 2 times, honoring `Retry-After`. If retries are exhausted, the SDK raises a typed `RateLimitError` carrying the response headers, so you can apply your own backoff. Queue on your side rather than tightening the retry loop: a `429` means the org is over budget for the minute, so retrying faster doesn't help. |
There was a problem hiding this comment.
Limits page duplicates pricing
Medium Severity
The new Concurrency and Limits page restates the per-plan concurrency table, pool and standby notes, and rate-limit header behavior that still live on Pricing, so the two pages can drift.
Reviewed by Cursor Bugbot for commit 18ca8c3. Configure here.
| { "source": "/profiles/credentials.md", "destination": "/auth/credentials.md" }, | ||
| { "source": "/profiles/credentials", "destination": "/auth/configuration#credentials-and-auto-reauth" }, | ||
| { "source": "/profiles/credentials.md", "destination": "/auth/configuration.md" }, | ||
| { "source": "/auth/credentials", "destination": "/auth/configuration#credentials-and-auto-reauth" }, |
There was a problem hiding this comment.
Missing credentials markdown redirect
Medium Severity
/auth/credentials.md has no redirect after the credentials page was merged, and the new secrets, stop, and hCaptcha redirects also omit their .md sources.
Additional Locations (1)
Triggered by learned rule: Add redirects in docs.json when removing or moving pages
Reviewed by Cursor Bugbot for commit 18ca8c3. Configure here.
AnnaXWang
left a comment
There was a problem hiding this comment.
I really like this - can we double check that old links will automatically redirect to new links?
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
18ca8c3 to
fcdf3c3
Compare
|
Yes, every link that works today will keep working! |


Summary
First layer of the guides IA restructure (stack: #671 → #672 → #673 → #674). Adds the How it works section — Configure, Control, Scale, Observe, Manage — each with an overview page, and reorganizes the pages under it so readers see the features that matter before the knobs.
Configure
Control
Scale
Observe — Browser telemetry pages grouped under their own section.
Manage — New overview with a multi-tenant setups section. Network Access is retitled as the firewall allowlist. Projects headings move to sentence case.
Across the section, overview pages get specific titles (for example "Proxies Overview") and headings use sentence case.
Stack notes
Merge the stack in order. Until #673 merges, pages not yet moved stay in their existing sidebar groups, so every layer builds and has no broken links on its own.
Open PRs to reconcile when merging
introduction/scale.mdxand reapply its pools/create edits with the new limits anchors.Testing
mint validateandmint broken-linkspass on this branch.🤖 Generated with Claude Code
Note
Low Risk
Documentation-only IA, content moves, and redirects; no runtime or API behavior changes.
Overview
Reorganizes the Guides tab around a new How it works flow—Configure, Control, Scale, Observe, and Manage—each with an overview page (
introduction/configure,control,scale,observe,manage) and regrouped sidebar pages so readers hit capabilities before deep settings.Configure: Stealth (hCaptcha content moved into
stealth, standalone page removed), proxies, profiles, vaults, authentication (managed auth nested;auth/credentialsdeleted and merged intoauth/configurationwith redirects), payments, and config registry. Overview pages get clearer titles/descriptions; vaults adds a credential-source comparison table.Control: Overview reframes control surface vs where the loop runs; app platform docs consolidate—secrets into
apps/deploy, stop intoapps/invoke(old pages removed + redirects). Task-oriented titles on develop/deploy/invoke/logs/status.Scale: New
browsers/concurrency-and-limitscentralizes concurrency, app invocation limits, rate limits (429/headers), and per-browser memory; links that pointed at pricing anchors now target this page.Observe / Manage: Telemetry grouped under Observe; Manage overview covers projects, API keys, spending, audit logs, and firewall allowlist (
info/network-accessretitled).docs.jsonadds SEO indexing, redirects, and trims the old top-level create/control/observe/scale intro list from Overview.Cross-doc link and changelog updates follow the moves (e.g. Foreman cookbook path, credential links → configuration anchor).
Reviewed by Cursor Bugbot for commit fcdf3c3. Bugbot is set up for automated code reviews on this repo. Configure here.