n8n Workflow: Build an Engineering Knowledge Graph (ClickUp, GitHub, Slack)

Build an Engineering Knowledge Graph from ClickUp, GitHub, Incidents & Docs—then share updates to Slack

This n8n workflow ingests engineering signals from ClickUp, GitHub, incident updates, and documentation changes, converts them into relationship data, and upserts nodes and edges into a graph database. It also posts a clear summary to Slack—so your team can track how work, code, and knowledge connect.

What this workflow does

• Receives webhook events from ClickUp tasks, GitHub pull_request events, your incident-management tool, and your documentation platform page events.
• Normalizes payloads into a shared “graph-event” structure and sends them to an external graph-event buffer API.
• Runs hourly, pulling the latest buffered event window from the buffer API for batch processing.
• Extracts explicit relationships from structured fields (e.g., assignee-to-task, PR-to-repo, incident-to-service) and flags any events that include free text for deeper extraction.
• Uses Anthropic Claude for entity linking when ambiguous free-text appears: it retrieves the known entity registry from the graph store and generates high-confidence links grounded to existing entities.
• Merges AI links with explicit links, then creates node and edge upserts via a bulk-upsert graph database API.
• Posts an update summary to a Slack channel after each batch run.

Use cases

• Turn ClickUp ownership and task context into a graph that connects people to work.
• Link GitHub pull requests to repositories and related engineering services via consistent entities.
• Correlate incident updates with impacted services and associated documentation changes.
• Keep an engineering knowledge graph current with hourly batch ingestion and AI-assisted entity linking.

Technical details

• Integrations: ClickUp webhooks, GitHub pull_request webhooks, incident tool webhooks, documentation platform page webhooks, Slack.
• Workflow logic/nodes (n8n): if, code, merge, webhook, sticky note.
• Auth: HTTP Header Auth (or equivalent) for webhook sources.
• AI step: Anthropic Claude for free-text entity linking against the graph’s existing entity registry.
• Graph writes: bulk node/edge upserts to the graph database, plus Slack summaries.