FAQ

FAQ — Common Integration Questions

This page consolidates questions raised by developers and integration teams during onboarding and initial API review.

Federation & Search

Do I need to implement a search endpoint on my platform to participate in federated search?

No. Participation in the federated search network is a content contribution model, not an endpoint implementation model.

Partner platforms do not expose a search endpoint of their own. Instead, they register their repository as a source and upload their knowledge products to the GGKP-managed index. The entire search infrastructure — semantic embedding pipeline, query engine, result ranking — is operated by GGKP.

From a partner's perspective, integration with the Search Federator involves two operations only:

A one-time POST /search-federator/upload-source to register the repository

Periodic POST /search-federator/upload-products calls to keep the index current

There is no partner-side endpoint to build, host, or maintain. The search endpoints described in this documentation are consumed by partners, not implemented by them.

Note — Write access (Upload Source, Upload Products permissions) is required to contribute content. Querying the federated index is available to any authenticated client regardless of whether they have contributed content.

Is there support for connecting existing CMS platforms like WordPress or Drupal?

Drupal is the framework underlying the GGKP API itself, which means Drupal integrations are particularly well-aligned with the platform's data model and authentication patterns. A Drupal site can connect to the Search Federator using standard HTTP client modules, mapping content types to the product upload schema.

WordPress integration is equally viable via the WordPress HTTP API or any REST client plugin. The GGKP API is a standard REST interface and imposes no CMS-specific requirements on the caller — any system capable of making authenticated HTTP POST requests can participate as a contributing source.

Dedicated native connector plugins for WordPress and Drupal are on the GGKP platform roadmap but are not yet publicly available.

Classifier

Can we use the Automatic Classifier to improve our platform's own content, without contributing to the federated index?

Yes. The POST /classify endpoint is a standalone service. Calling it returns classification results — taxonomy tags, confidence scores, and AI-generated reasoning — directly to the caller and does not automatically push results to the GGKP federated index or modify any shared data.

A partner platform can use the classifier entirely for internal purposes:

Improving tagging consistency within its own CMS

Populating metadata fields for local search and filtering

Identifying content gaps against the GGKP taxonomy

Driving internal recommendation features

The only requirement is a valid API key. No additional permissions are needed for classification.

Integration pattern — If you want to classify content and contribute it to the federated index, the recommended workflow is:
1.
Submit the PDF to POST /classify to obtain taxonomy tags
2.
Use the returned tags as classification metadata
3.
Submit POST /search-federator/upload-products with those tags included
The two steps are independent and can be orchestrated by your own system.

Knowledge Products

What is a Knowledge Product? Why are products uploaded separately from source registration? Can I upload a PDF and have its full text indexed?

What it is

A Knowledge Product is the atomic unit of the federated index. It represents a discrete, citable resource — a report, policy brief, case study, dataset, or toolkit — and is defined by structured metadata, not by the binary content of the document itself.

Minimum required fields:

Field	Required	Description
`title`	Yes	Human-readable title of the document
`url`	Yes	Canonical URL pointing to the document or landing page
`description`	No	Abstract or descriptive text; improves search relevance significantly
`tags`	No	Array of taxonomy term strings; drives classification-based filtering
`image`	No	URL to a cover or thumbnail image; empty string if unavailable

Why sources and products are separate

A source is a persistent identity for a contributing repository — registered once. Products are the individual records that flow through that source over time as content is published or updated. This design allows a single source to manage thousands of products through incremental uploads without repeating registration.

PDF full-text indexing

The POST /upload-products endpoint accepts metadata fields — not a PDF binary. Full document text is not directly indexed via product upload.

However, POST /classify does accept a PDF binary, extracts its full text, and returns taxonomy classifications based on that content. The recommended workflow when you want document content to inform indexing is:

POST /classify          →  receive taxonomy tags from PDF analysis
POST /upload-products   →  submit product record with those tags as metadata

FAQ — Common Integration Questions#

Federation & Search#

Do I need to implement a search endpoint on my platform to participate in federated search?#

Is there support for connecting existing CMS platforms like WordPress or Drupal?#

Classifier#

Can we use the Automatic Classifier to improve our platform's own content, without contributing to the federated index?#

Knowledge Products#

What is a Knowledge Product? Why are products uploaded separately from source registration? Can I upload a PDF and have its full text indexed?#