PRDCT-348: fix code-vs-docs discrepancies in extractors/other

[Iamfle4ka]

Jira issue(s): PRDCT-348

Changes:

Code-vs-docs audit of all 23 connectors under components/extractors/other. Compared documented fields, options, defaults, limits and behavior against source code schemas (configSchema.json, configRowSchema.json, PHP ConfigDefinition). 20 findings total across 12 connectors with source access.

Code-accuracy fixes (schema evidence cited)

Airtable — removed non-existent filter_by_formula UI field; added use_view/view_name params; added sync_options (sync_mode, date_from, date_to)

• Evidence: component-airtable/component_config/configRowSchema.json:47-69, 90-137

GitHub — rewrote entire config section: docs described v1 Generic Extractor (OAuth + templates); deployed component is keboola.ex-github-v2 (PAT + endpoint multiselect + row config with organization_name, repository_owner, repository_name, issue_state, commits_since)

• Evidence: component-github/component_config/configSchema.json, configRowSchema.json

Azure Cost Management — added type (ActualCost/AmortizedCost/Usage), aggregation (5 values), granularity (None/Daily/Monthly); enumerated all 19 grouping dimension values and 6 time frame options; documented Service Principal auth

• Evidence: ex-azure-cost-management/src/ConfigDefinition.php:16-89

Google Search Console — added search_type filter, include_fresh option, service account auth

• Evidence: component-google-search-console/component_config/configRowSchema.json:37-46, 87-98, 183-199

Geocoding Augmentation — fixed provider name openstreetmap → nominatim; fixed Yandex apiKey claim (code passes no key)

• Evidence: geocoding-augmentation/src/.../Augmentation.php:224-236, ParametersValidation.php:50-52

What3words — added undocumented lang parameter (default en)

• Evidence: app-what3words/main.php:38-42

AWS CUR Reports — "lowest report date" → "highest" (copy-paste from Minimum date)

• Evidence: component-aws-cost-and-usage-reports/component_config/configSchema.json:133

Time Doctor 2 — parameter mislabeled users → actual name increment; fixed garbled edit-–ime → edit-time; normalized dash style across the endpoints list to --

• Evidence: component-time-doctor-2/component_config/configSchema.json:106-112

Weather API — added forecast_days limit (1–14); clarified continue_on_failure only in input-table mode

• Evidence: component-weather-api/component_config/configSchema.json:84-85, 117-119

HiBob — added scope clarification: human_readable only affects employees table

• Evidence: component-hibob/component_config/configSchema.json:54-58

Doc hygiene fixes (typos, alt-text)

• yourpass: "YourPasss" → "YourPass", "output tale" → "output table"
• pingdom: "Last 3O days" → "Last 30 days" (letter O → digit 0)
• mapbox: "columns with in" → "columns in"
• dynamodb-streams: alt text "Confguration" → "Configuration"
• servicenow: alt text "DynamoDB Streams Configuration" → "ServiceNow Configuration"
• google-search-console: "search analytics report" → "sitemaps report" in Sitemaps section
• azure-cost: "art of the primary key" → "part of"
• weather-api: "dispite" → "despite", "occurr" → "occur"
• time-doctor-2: normalized mixed en-dash/-- style in endpoints list (per review nit)

---

11 connectors had no accessible source repo and could not be verified: DynamoDB Streams, Dark Sky, Currency Rates, Mapbox, Okta, ServiceNow, YourPass, Pingdom, Papertrail, Stripe, Generic Extractor.

Note: github/index.md v2 rewrite is now image-less — the old v1 screenshot no longer matches. A maintainer with UI access should capture a current keboola.ex-github-v2 screenshot.

Release Notes

Justification, description

Documentation accuracy fixes based on code-vs-docs audit of extractors/other connectors. Aligns documented fields, options, defaults, and behavior with actual component schemas.

Plans for Customer Communication

N/A — documentation-only changes.

Impact Analysis

No code changes. Documentation corrections only.

Deployment Plan

Merge to main → auto-deploy to help.keboola.com.

Rollback Plan

Revert PR.

Post-Release Support Plan

N/A

Link to Devin session: https://app.devin.ai/sessions/c360b6efa9804125940654b8cd841f3d
Requested by: @Iamfle4ka

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[linear]

PRDCT-348

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
connection-docs	Ready	Preview	Aug 17, 2026 2:21pm

[keboola-pr-reviewer-bot]

Verdict: needs_human (risk 3/5) · profile docs

Needs human review: multiple pages introduce new product-behaviour claims that can't be verified from the diff.

Concerns:

• src/content/docs/components/extractors/other/github/index.md: Entire config rewritten to v2 PAT/endpoint model; product-behaviour claim unverifiable from diff.
• src/content/docs/components/extractors/other/azure-cost/index.md: New Type/Aggregation/Granularity fields and 19 dimension enums added; needs product confirmation.
• src/content/docs/components/extractors/other/airtable/index.md: New Use View and Sync Options params/defaults added; unverifiable product-behaviour claim.

Suggested reviewers: @keboola/docs

[jordanrburger]

Review summary

Verdict: safe to merge with product sign-off.

Structurally build-safe — 15 Markdown files, no frontmatter/admonition/code-fence changes, no broken links or images introduced. Typo/hygiene fixes are unambiguously correct.

The substantive changes rewrite documented product behavior for several connectors and can't be verified from the docs repo. Please confirm before publishing:

• github: full v1→v2 rewrite (PAT scopes read:org+repo; endpoint multiselect; row fields organization_name / repository_owner / repository_name / issue_state default all / commits_since ISO-8601). Confirm the deployed component is keboola.ex-github-v2 with these fields/defaults.
• azure-cost: Type (default ActualCost), Aggregation (default Cost + 4), Granularity (Daily/Monthly/None), the 19 grouping dimensions, Time Frame (default MonthToDate), Service Principal auth — verify enums/defaults.
• airtable: Use View / View Name, Sync Mode Full/Incremental, Date From default last run, Date To default now, and the "hidden fields retrieved when using a view" claim.
• google-search-console: Service Account auth, search_type values (web/image/video/news/discover/googleNews), Include fresh data.
• geocoding-augmentation: provider rename openstreetmap→nominatim, and the reversal that Yandex needs no API key (flips a previously stated requirement — worth an explicit check).
• what3words: new lang param default en.
• time-doctor-2: param rename users→increment, default true.
• weather-api: forecast_days 1–14; "continue_on_failure available only when using an input table".
• hibob: human_readable "only affects the employees table".
• aws-cur-reports: "Maximum date = highest report date" and removal of the first-run backfill sentence — confirm Maximum date has no first-run backfill semantics.

Minor nits:

• github/index.md: the rewrite removes the page's only screenshot, leaving the v2 config section image-less (orphaned PNG not cleaned up). Consider an updated v2 screenshot.
• time-doctor-2: dash style is now mixed — worklog/edit-time normalized to -- while timeuse/projects/tasks still use en-dash –.

Automated review pass (Claude Code), flagged for a human maintainer — not an approval.

[devin-ai-integration]

Thanks @jordanrburger. Notes on the flagged items:

Nits — addressed:

• time-doctor-2 dash style: fixed in 3a2328a7 — normalized users/timeuse/projects/tasks to -- so the whole endpoints list is consistent.
• github orphaned screenshot: leaving the v2 config section image-less for now — I can't generate an accurate keboola.ex-github-v2 UI screenshot from the docs repo. Flagging for a maintainer with UI access to capture and drop in a current one (the old v1 PNG can be removed at that point).

Substantive behavior changes — all verified against component source, not guessed. Each was checked against the deployed component's configSchema.json / configRowSchema.json / PHP ConfigDefinition (component IDs resolved via each repo's .github/workflows/push.yml). Evidence per finding:

• github — keboola/component-github, component_config/configSchema.json (PAT auth, endpoint enum) + configRowSchema.json (row fields + issue_state default all). Confirmed deployed component is keboola.ex-github-v2.
• azure-cost — keboola/ex-azure-cost-management, src/ConfigDefinition.php (Type/Aggregation/Granularity constants + defaults, 19 grouping dimensions, 6 time frames, Service Principal auth).
• airtable — keboola/component-airtable, configRowSchema.json (Use View/View Name, sync_mode, date_from default last run, date_to default now).
• google-search-console — keboola/component-google-search-console, configRowSchema.json (search_type enum, include_fresh, service-account auth).
• geocoding-augmentation — source Augmentation.php switch uses nominatim (not openstreetmap); Yandex branch passes no API key.
• what3words — keboola/app-what3words main.php, lang default en.
• time-doctor-2 — configSchema.json param is increment (default true), not users.
• weather-api — configSchema.json forecast_days range 1–14; continue_on_failure gated to input-table mode.
• hibob — human_readable applied only to the employees table in source.
• aws-cur-reports — configSchema.json Maximum date = highest report date; no first-run backfill on the max bound.

Full report with exact file/line citations is attached to PRDCT-348. Agree this still warrants product sign-off before publishing — the above is the code evidence to make that check quick.