Indexing Overview

Availability and migration

Indexing feature and tools are available in the Next environment as part of the 1.7.0 release. They replace the legacy Datasources/Datasets feature. Datasources are still available in the Nexus environment for reference, but new datasets are not supported. See the Release Notes 1.7.0.

Introduction

Indexing turns your external content (repos, wikis, issues, files, designs, tests) into searchable knowledge that Agents and Pipelines, as well as LLMs from Conversations can use. Instead of manually browsing systems, you create indexes once and then search or ask questions with natural language.

Purpose of Indexing

• Centralize knowledge from multiple tools into a consistent, searchable store.
• Improve retrieval quality for Agents and LLMs with chunking and metadata.
• Keep results fresh by re-running indexing when content changes.
• Replace legacy datasets with standardized tools across many toolkits.

Quick start

If you need a fast walkthrough, see the Next environment guide: Next – Quick Start.

Indexing Tools

ELITEA provides six standardized tools available across supported toolkits:

• Index Data — create or update an index for a chosen scope.
• Search Index — run a search query against existing indexes.
• Stepback Search Index — advanced search using stepback context for improved relevance.
• Stepback Summary Index — stepback search with on-the-fly summary generation.
• Remove Index — delete an existing index.
• List Collections — list available collections (logical index groups).

Replacement for datasets

These tools replace legacy datasets. Your old datasources remain visible for reference in Nexus, but you should use the new tools in Next env to re-create indexes.

Toolkits which support Indexes

Currently supported toolkits include:

Category	Toolkits
Repos	ADO Repos, Bitbucket, GitHub, GitLab
Wikis	ADO Wiki, Confluence, SharePoint
Issues	ADO Boards, ADO Plans, Jira
Files	Artifact, SharePoint
Designs	Figma
Tests	TestRail, Xray Cloud, Zephyr Enterprise, Zephyr Essential, Zephyr Scale

For setup of a specific toolkit, see Integrations → Toolkits (e.g., GitHub Toolkit, Confluence Toolkit, Jira Toolkit).

How to configure Toolkit for Indexing

Use these steps to prepare your project and toolkit before running Index data.

Prerequisites

• A toolkit that supports Indexing (see list below) and is added to your project/agent.
• A valid Credential for that toolkit (all toolkits except Artifact require credentials).
• Project-level AI configuration for Vector Storage (PgVector) and Embedding Model.

Steps

1) Configure Credentials (required for all except Artifact)

• Create or select a Credential for the target system (e.g., GitHub, Confluence, Jira, SharePoint, ADO, Bitbucket, GitLab).
• Assign it to your toolkit when creating or editing the toolkit.

• See: Create a Credential and specific toolkit pages under Integrations → Toolkits.

  

2) Configure PgVector (required storage for indexed data)

• By default, both Private and Team projects have a shared PgVector configuration you can select.

• To configure a new PgVector configuration:

  • Click the New private pgvector credentials or New project pgvector credentials option, regarding what type of credential you need.
  • Fill the information for the PgVector (Display Name, Connection String).
  • Save it to use configuration to use for indexing.
  • Click Refresh icon to update the configuration and have it in the PgVector Configuration dropdown to select.

  

3) Configure Embedding Model (required for indexing)

• Default available models in Private and Team projects:

  • text-embedding-ada-002 — improved, performant version of the ada embedding model
  • text-embedding-3-small — improved, performant successor in the 3-series
  • text-embedding-3-large — the most capable model for English and non‑English tasks

• Practical notes:

  • For cost/speed, ada-002 and 3-small are similar; test with your data.
  • Some reports suggest ada can work better in certain cases—ada is a solid “go-to” if you just need embeddings.

  

4) Select Indexing tools on the Toolkit

• When creating a new toolkit, all Indexing tools are selected by default.
• You can later enable/disable specific tools from the toolkit’s details page.

• Recommended minimum for creating and using indexes: Index Data, Search Index, Stepback Summary Index.

  

5) Fill toolkit-required fields (vary by toolkit)

• Examples of mandatory fields by type:
  • Repositories: organization/project, repository name, branch, path filters, blacklist/allowlist
  • Confluence/Wikis: site URL, space key, labels, CQL filters
  • Project Management (Jira/ADO Boards): project key/ID, issue filters (JQL/queries), include attachments
  • SharePoint: site/drive, library/folder path, include file types

6) Save the toolkit, if you created a new one.

Helpful links:

• Create a Credential: Guide
• Create a Toolkit: Guide
• Examples: GitHub, Confluence, Jira, SharePoint

Index data and verify

This section shows how to run indexing and validate results using the Toolkit's details page -> TEST SETTINGS section. The example below uses the Artifact toolkit, but the flow is similar for other toolkits.

Index Data tool

Prerequisites:

• You’ve already configured an Artifact toolkit and have a bucket with files to index.

Steps:

1. Open Toolkits → select your Artifact toolkit.
2. See the TEST SETTINGS section on the right side.
3. In the tool dropdown, select Index data tool.
4. Provide a meaningful Collection Suffix (for example: prod, test, v1).
5. Leave other settings at defaults for a first run.
6. Click RUN TOOL to start indexing.
7. Progress and completion details appear in the main panel; scroll if needed to view messages.

List Collections tool

Use this to view the indexes (collections) created for the toolkit.

1. In TEST SETTINGS, choose List Collections.
2. Click RUN TOOL.
3. Review the output in the main panel for available collections.

Search Index tool

Query your indexed data and review matched results.

1. In TEST SETTINGS, choose Search Index.
2. In the Query field, enter what you’re looking for.
3. (Optional) In Collection Suffix, specify a particular index name; otherwise, the search runs across all indexes for the toolkit.
4. Leave other options at defaults for a first try.
5. Click RUN TOOL and review results in the main panel.

Reference

For detailed information about each indexing tool and configuration:

• See the Indexing Tools section above for capabilities and purpose.
• Review AI Configuration for Embedding Models and Vector Storage (PgVector).
• Check the Release Notes: Indexing Tools: Replacement for Datasets.

How to configure and use Indexes from Chat

You can trigger indexing and search directly from Chat using an Agent or a Toolkit that exposes indexing tools.

1. Open Chat and start a new conversation or use an existing one. See Chat.
2. Select an Agent or Toolkit that has the Index Data tool available.
3. Ask the assistant to index your target with scope details, for example:
   • "Index the GitHub repo org/repo on branch main. Use collection suffix 'prod'."
   • "Index Confluence space 'ABC' for pages with label docs."
4. Wait for confirmation in the thinking steps. If an error appears, refine your instruction or reconfigure the attached toolkit/credential.

Once indexes exist, you can use Search Index or Stepback search tools through Chat as well (e.g., "Search the index for onboarding guidelines").

How to configure and use Indexes from Agent

You can also prepare an Agent with the required toolkit(s) and run indexing via Chat or within the Agent’s context.

1. Open your Agent. See Agents.
2. In the Toolkits section, add/select a toolkit that supports Index Data and configure it with the correct Credential.
3. Save the Agent.
4. From Chat, select the Agent and instruct it to index the desired scope (repo/site/project, branch/filters, etc.).

FAQs

1. Where do I see my created indexes?
   • A dedicated menu to review created indexes is planned. For now, use List Collections and Search Index via the toolkit Test section or through Chat.
2. Can I keep using Datasources?
   • Datasources remain in Nexus for reference, but indexing in Next replaces datasets. New datasets are not supported.
3. How do I remove an index?
   • Use Remove Index from the toolkit Test section or trigger it in Chat.
4. Can I search without an Agent?
   • Yes. In a toolkit’s Test section run Search Index, or in Chat address a Toolkit that exposes the search tools.
5. Are there usage limits?
   • Next uses shared LLMs for evaluation and may have daily limits. For production, configure your own EPAM AI DIAL keys in Settings → AI Configuration and see Configure EPAM AI DIAL Key.

Useful Information

• Release details and changes: Release Notes 1.7.0
• AI Configuration (models, embeddings, vector storage): AI Configuration
• Add Teammates to conversations: Guide
• Troubleshooting Chat and Canvas usage: Chat – Advanced Usage and Canvas – Advanced Usage

Next steps

• Index one system (e.g., a single repo or space) to validate settings.
• Try Search Index and Stepback Summary Index to compare results quality.
• Expand scope and schedule regular re-indexing as content changes.