Published 2026-08-20 · Tested 2026-08-19

Airbyte

D+

Airbyte received 10 PASS votes and passed 3 of five readiness checks. The clearest finding came from the find the exact limits task.

Panel: GPT 5.6 Sol, Opus 5, DeepSeek v4F Battery: v1 Verified: 2026-08-19 Read as markdown (opens in a new tab)

Three AI models, GPT 5.6 Sol, Claude Opus 5, and DeepSeek v4 Flash, each read Airbyte’s public documentation independently and attempted five first-hour developer jobs: run the first data sync, find the exact limits, recover a failed sync, verify a notification webhook, use PyAirbyte.

No accounts, API calls, or code execution were used. Every verdict came from public pages and every published quotation passed a live verification check.

Airbyte Docusaurus 3.10.1 · published 2026-08-20
D+

67.5% · 54/80 · AI Agent Readiness Score · reading 30 pts · surface 50 pts

llms.txt PASS
llms-full.txt FAIL
markdown mirror PASS
MCP server FAIL
docs AI PASS
Task GPT 5.6 SolOpus 5DeepSeek v4F Consensus
Run the first data sync PASSPARTIALPASS PASS
Find the exact limits PARTIALPARTIALFAIL PARTIAL
Recover a failed sync PASSPASSPARTIAL PASS
Verify a notification webhook PASSPASSPASS PASS
Use pyairbyte PASSPASSPASS PASS

docs platform: Docusaurus 3.10.1 (unscored) · verified 2026-08-19

What the docs get right

  • Verify a notification webhook: 3 PASS votes. The procedure covers Slack app creation, event toggles, webhook entry, test delivery to the channel, and saving.
  • Use pyairbyte: 3 PASS votes. One page installs airbyte, reads source-faker, and iterates result.streams.items() using current stream terminology.
  • 3 of 5 readiness checks. Present: llms.txt, markdown mirrors, docs AI. Missing: llms-full.txt, an MCP server.

Run the first data sync

PASS

PASS consensus from 2 PASS, 1 PARTIAL.

Linked pages cover both connectors and ELT setup; scheduled connections sync immediately, and Succeeded confirms complete loading. The source, destination, and ELT connection steps form an explicitly linked chain, but the final step stops at "Airbyte takes you to the page for that connection, where you can manage it and initiate syncs" and never names the sync trigger or the success check, so triggering and confirming the first sync requires cross-referencing the sync-schedules page for the "Sync Now" button and the connection-status page for the result. The quickstart and connector setup pages provide a complete, unambiguous path from zero to first sync for both OSS and Cloud.

Find the exact limits

PARTIAL

PARTIAL consensus from 2 PARTIAL, 1 FAIL.

The limits page lists all numbers, but /platform/next/ returned 404 and the workspace page excludes Plus. One page carries every requested number with plan attribution and no internal disagreement (Standard: 1 workspace per user and 60-minute minimum schedule; Plus: 15-minute minimum schedule; all plans: 1,000 discoverable streams, 1,000 synced streams, 20,000 fields, 20MB records), but the minimum schedule interval for the Cloud Pro and Enterprise Flex plans is unresolvable because the limits page says the restriction is removed while the schedule options list offers nothing finer than 15 minutes and labels that option Plus-only. Schedule limits are documented, but no numeric workspace, stream, field, or record-size limits are published per plan. The pricing page matrix shows Core/Standard/Pro/Enterprise Flex but omits Plus, creating ambiguity about which plan is current. Record-size limits are destination-specific only.

Recover a failed sync

PASS

PASS consensus from 2 PASS, 1 PARTIAL.

Rate-limited streams wait and retry automatically; configuration errors require re-testing and saving the affected connector. Rate limiting and configuration failure get separate named sections with different recovery instructions and different UI colors, and the notification payload reference reinforces the split by documenting the errorType values config_error and transient_error. The notification payload docs distinguish config_error from transient_error, and the monitoring blog covers rate-limit auto-recovery with countdown UI. However, there is no single troubleshooting page for sync failures; the information is scattered across blog posts and the notification docs page.

Verify a notification webhook

PASS

PASS consensus from 3 PASS.

The procedure covers Slack app creation, event toggles, webhook entry, test delivery to the channel, and saving. A single page covers the whole path in order: create the Slack app, turn on Incoming Webhooks, copy the webhook URL, toggle the events you want in the Webhook column, paste the URL, click Test, then Save changes, with a nine-row table of notifiable events and example JSON payloads. A complete two-part walkthrough covering Slack app creation, webhook URL generation, notification event selection, and a test button that verifies receipt before saving.

Use pyairbyte

PASS

PASS consensus from 3 PASS.

One page installs airbyte, reads source-faker, and iterates result.streams.items() using current stream terminology. The page gives pip install airbyte plus a runnable example that calls ab.get_source("source-faker", ...), select_all_streams(), and read(), and the connector page confirms the example is current by listing PyAirbyte in its availability row, documenting the count parameter used in the example, and naming the three streams the loop will print. Install, source config, check, stream selection, read, and iteration are all demonstrated with a single clear example using current API methods (get_source, install_if_missing, select_all_streams, result.streams).

The receipt

These limitations only apply to those using the Standard plan. If you upgrade to Plus, Pro, or Enterprise Flex, Airbyte removes these limitations.

The limits page lists all numbers, but /platform/next/ returned 404 and the workspace page excludes Plus.

Agent surface notes

Initialize returned HTTP 404, and no candidate endpoint completed initialization.

The live docs expose a public Ask AI control.

Method note

This is a reading test of public documentation, not an execution test. No accounts were created and no API calls were run. The AI Agent Readiness Score counts fifteen reading votes at PASS 2, PARTIAL 1, and FAIL 0, for 30 possible points. Five readiness checks add 10 points each. The total is 80. Consensus chips show each row majority and do not affect scoring. The panel split on 3 of five tasks. Quotes shown here were re-fetched and confirmed verbatim on 2026-08-19.

Read the full methodology

Put another docs site through the battery.

Nominate a docs site