docs: add Scalekit integration page

[saif-at-scalekit]

Summary

Adds a Scalekit integration page to the platform documentation. Scalekit is auth infrastructure for AI agents that provides a token vault and connector layer for OAuth 2.0 flows, token storage, automatic refresh, and API proxying for 100+ third-party services.

New page contents

• What is Scalekit — overview of the token vault and connector layer
• Prerequisites — Scalekit account, Apify account, Node SDK install, env vars
• Step-by-step guide:
  1. Initialize the Scalekit client
  2. Connect a user's account via getOrCreateConnectedAccount
  3. Authorize via OAuth link with polling loop
  4. Call APIs via executeTool (pre-built tools) or actions.request (direct)
• Complete working example — full Actor code (Notion search)
• LLM agent Actor section — using executeTool in an LLM tool-calling loop
• Available connectors table — 14 common services with connectionName values
• Resources — links to AgentKit docs, quickstart, connectors, example repo

Style guide alignment

• Front matter description uses the "Learn how to..." pattern (140-160 chars)
• Prerequisites formatted with italic item names per existing integration pages
• Third-party disclaimer partial imported and placed after intro section
• Follows Conventional Commits (docs: prefix)

Related

• Notion + YouTube Agent example repo
• Apify Actor per-user OAuth cookbook

[marcel-rbro]

@saif-at-scalekit Thank you for the contribution, there are some issue in the PR that need to be resolved before we can merge it.

[marcel-rbro]

Redundant heading. We're slowly removing these "What is XYZ" headings from our docs.

[marcel-rbro]

The installation should be the Step 1, initialization of the client should be the 2nd step.

[marcel-rbro]

Good info, but this should be in the intro to the entire Step 4 section. That way you decide which approach to use before learning about both.

[marcel-rbro]

No need to list as many examples with their identifiers. For an exhaustive list, users would go to the linked documentation either way. We can just list examples of available services and navigate users to your docs for the identifiers.

[marcel-rbro]

Missing newline at the end of the file

[marcel-rbro]

Suggestion from automated review:

100+ third-party services (and 100+ connectors on line 266) is the vague-qualifier pattern our style guide flags. Prefer more than 100, or use the actual count if it's stable.

[marcel-rbro]

Suggestion from automated review:

The deep import path '@scalekit-sdk/node/lib/pkg/grpc/scalekit/v1/connected_accounts/connected_accounts_pb' reaches into the SDK's internals and is brittle if the package layout changes. Is ConnectorStatus available from a top-level export? If not, consider comparing against the string 'ACTIVE' shown in the status table on lines 82-86.

[marcel-rbro]

Suggestion from automated review:

resp.connectedAccount ?? resp defensively unwraps two possible response shapes (same pattern repeats on lines 112, 191, 214). If the SDK really returns both shapes, a one-line comment explaining why would help. If it consistently returns one shape, this is dead code - simplify to either resp.connectedAccount or resp.

[marcel-rbro]

Suggestion from automated review:

const { task } = input; assumes input is non-null, but Actor.getInput() can return null when the Actor runs without input. Either add a guard (if (!input) throw new Error('Input is required.')) or note in prose that input is required.

[marcel-rbro]

Suggestion from automated review:

The orphan [Browse all connectors](...) link standing alone after the table reads like a CTA. Consider folding it into the preceding sentence: ...The rest of the code stays the same - [browse all connectors](https://docs.scalekit.com/agentkit/connectors).