Skip to main content

Linking Entries

"How do I connect related entries so they surface together?"

Knowledge graph entries are more useful when connected. A lesson about a bug is more findable if it links to the ADR that caused the design decision behind the bug.

In any KG entry (lesson, ADR, pattern), add a ## Related section at the bottom:

## Related

- [ADR-012: Chose SQLite for local storage](../decisions/adr-012.md) — context for why this constraint exists
- [Lesson: Hook variable regression](../lessons-learned/hook-variable-regression.md) — similar failure mode

Use relative paths within the knowledge graph directory so links work across machines.

Link entries when:

  • A lesson is a downstream consequence of an architectural decision
  • Two bugs share a root cause (link both to the meta-issue)
  • A pattern came from generalizing a specific lesson

Don't over-link. One or two meaningful connections beats five weak ones.

Cross-KG references​

If you have multiple knowledge graphs (project + personal), cross-KG references use absolute paths or a convention you define in me.md. Keep cross-KG links in comments or ## See Also sections — they won't be clickable across graphs but serve as recall prompts.

When you run /kmgraph:recall, results include the entry content. If you see a related link, follow it manually — the recall command doesn't traverse links automatically. The value is in knowing the connections exist.

  • Search the Graph — full-text search with FTS5 indexing for active recall
  • Session Memory — how KMGraph automatically surfaces relevant knowledge at session start