Skip to the content

connector-apis

The endpoint digest of every Connector Kwern reaches through a passthrough tool rather than a named one — Search Console, Google Analytics, Google Ads, App Store Connect, Apple Ads, LinkedIn and Stripe.

Written by Kwern. Updated 2026-10-16.

The file, as Kwern runs it

Nothing below is rewritten or cut. The description a model routes on comes first, then the file itself. The whole thing, frontmatter included, is at /skills/connector-apis.md.

description

The endpoint digest of every Connector Kwern reaches through a passthrough tool rather than a named one — Search Console, Google Analytics, Google Ads, App Store Connect, Apple Ads, LinkedIn and Stripe. Read it with read_reference before calling any connector_*_read or connector_*_write tool, to learn which path answers the question, what it takes, and what the current grant actually reaches.

A named Connector tool — connector_asc_reviews, connector_posthog_metrics — is one question with one answer. A passthrough tool is the whole API: you name the method and the path, Kwern signs the call with the Product's Connection and hands you back what the service answered, verbatim.

That trade needs the API in front of you, which is what this Skill holds: one reference file per service, listing the endpoints, their parameters, and the calls the current OAuth grant actually reaches.

APIToolsReference file
Search Consoleconnector_search_console_read, connector_search_console_writereferences/search-console.md
Google Analyticsconnector_ga4_read, connector_ga4_writereferences/google-analytics.md
Google Adsconnector_google_ads_read, connector_google_ads_writereferences/google-ads.md
App Store Connectconnector_asc_read, connector_asc_writereferences/app-store-connect.md
Apple Adsconnector_apple_ads_read, connector_apple_ads_writereferences/apple-ads.md
LinkedInconnector_linkedin_read, connector_linkedin_writereferences/linkedin.md
Stripeconnector_stripe_read, connector_stripe_writereferences/stripe.md

A Chat holds the pair of every API its Product is connected to, and none of the others.

How to use a passthrough

  1. read_reference the API's file before the first call. Each one is curated by hand and names the vendor's own index for anything it leaves out.
  2. Call the read tool for anything that only looks. It answers straight away. Which calls count as reads is the API's own judgement, not the HTTP method: a GAQL query, a GA4 report, a Search Console query and an Apple Ads report are all POST, and all reads.
  3. Call the write tool for anything that changes something. The User sees the method, the path and the body, and approves, edits or refuses the call before it leaves Kwern — one approval per call, every time. A read the model asks the write tool for still waits for an approval, so send it to the read tool.
  4. Pair a write with no other tool call in the same step: everything beside it is held until the User has decided. Make the reads first, in their own step.
  5. Read the status before the body. A 403 usually means the grant, not the path: each file says which calls the grant reaches today.
  6. Write the path only. Kwern supplies the host, the version prefix, the credentials and every header the API demands — a host written into the path is dropped.

What it discloses

The Skill reads these files only when it needs them, so the model spends nothing on them the rest of the time.

  • references/app-store-connect.md

  • references/apple-ads.md

  • references/google-ads.md

  • references/google-analytics.md

  • references/linkedin.md

  • references/search-console.md

  • references/stripe.md

Someone is asking for what you built today.

On Reddit, on X, in ChatGPT's answers. Kwern finds them every morning and drafts the reply in your voice; you decide what goes out.