Daemion docs

How do I query memory and knowledge?

Auth: Bearer token from device pairing
Base URL: http://localhost:3001
All examples tested against live gateway

Daemion’s current memory substrate is not the old Engram/Neo4j graph. It is:

  • compiled wiki pages
  • raw captured sources
  • searchable conversation history
  • agent-specific and shared knowledge spaces

If you are looking for memory in the current product, these are the endpoints to use.

The old /graph/* endpoints are retired. If you still see references to Engram or a knowledge graph elsewhere, treat them as historical until updated.

Browse the wiki index

GET /knowledge Auth required

Return the current wiki index.

Search compiled knowledge

GET /knowledge/search Auth required

Search compiled wiki pages by query string.

Parameter Type Description
q REQUIRED string Search query
limit number Maximum results
scope string Optional scope filter
bash
Verified

curl “http://localhost:3001/knowledge/search?q=trust&limit=10
-H “Authorization: Bearer $DAEMION_TOKEN”

Read a wiki article

GET /knowledge/article/:slug Auth required

Fetch a compiled wiki article by slug.

Inspect pages, stats, and health

GET /knowledge/pages Auth required

List pages in the knowledge substrate.

GET /knowledge/stats Auth required

Return basic counts for the knowledge substrate.

GET /knowledge/health Auth required

Return shared, agent, and raw-source health information.

Inspect knowledge log

GET /knowledge/log Auth required

Return recent knowledge-substrate events.

What about conversation memory?

Compiled knowledge is only one part of memory.

Use:

  • thread endpoints for raw conversation recall
  • history tools for searching older turns
  • knowledge endpoints for durable compiled understanding

That separation is intentional:

  • history = what was said
  • knowledge = what the system now knows