Skip to content

Extraction Pipeline

FluidRAG can extract Jira issues into your knowledge graph on a schedule, making ticket history available for RAG queries alongside your other ingested documents.

The pipeline runs in three stages:

  1. Issue Extraction — FluidRAG queries Jira using JQL for issues updated since the last sync. Issues are fetched with pagination (50 per page) and saved as individual JSON files with all fields (summary, description, status, assignee, priority, issuetype, reporter, dates).

  2. ADF-to-Markdown Conversion — Each Jira issue is converted from Atlassian Document Format (ADF) to structured Markdown. The output includes metadata (type, status, assignee, dates), the description body, and all comments with authors and timestamps. This conversion is rule-based (no LLM required).

  3. Knowledge Graph Ingestion — Processed Markdown documents are ingested into FalkorDB with entity extraction, embedding generation, and relationship creation, making them searchable via graph and vector search.

Extraction runs automatically every 8 hours by default. You can trigger it manually at any time via the Admin UI.

  1. Open the Admin UI at http://localhost/config-ui
  2. Navigate to the Sources page
  3. Find Jira in the source list
  4. Toggle Enabled to on

Specify which Jira projects to extract from:

  1. On the Sources page, click Jira to expand its configuration
  2. Enter project keys in the Project Keys field (e.g., PROJ, ENG, SUPPORT)
  3. You can enter multiple keys separated by commas
  4. Save your configuration
SettingDefaultDescription
jira_project_key(empty)Single project key
jira_project_keys(empty)Comma-separated list of project keys
jira_sync_schedule0 */8 * * *Cron expression for extraction schedule
jira_timeout3600sMaximum extraction runtime
jira_max_retries3Retry attempts on transient errors
ActionHow
Trigger extraction nowSources → Jira → Trigger Extraction
Full resyncSources → Jira → Trigger Extraction with “Full Resync” enabled
Reset sync stateSources → Jira → Reset Sync — clears last sync timestamp so the next run extracts everything

All endpoints require admin authentication.

EndpointMethodDescription
/api/agents/jiraGETGet Jira agent status and configuration
/api/agents/jiraPATCHEnable/disable the Jira agent (hot-reload)
/api/sources/jiraGETGet Jira source configuration
/api/sources/jiraPATCHUpdate Jira settings (project keys, schedule, enabled)
/api/sources/jira/enablePOSTEnable Jira extraction
/api/sources/jira/disablePOSTDisable Jira extraction
/api/extraction/trigger/jiraPOSTTrigger extraction (supports force, full_resync, start_date_override)
/api/extraction/reset/jiraPOSTReset sync state for full resync
/api/extraction/statusGETGet extraction status for all sources