Skip to content

The knowledge repository

Every project grows a knowledge repository: a structured graph of the people, places, and things your conversations touch — who works where, what someone is like, when things happened, how everything connects. The agent reads it before each response and writes back what it learns after.

Entries

Entries come in three types — entities (people, organizations — conversation participants appear here), places, and things — and carry:

  • Name and a short hint (the summary shown in search results)
  • Characteristics — simple adjectives: curious, direct, old
  • Relationships to other entries — "mentor of", "employer of" — shown as labeled edges in the graph
  • Date of origin, and a freeform status ("employed", "unemployed")
  • Events — see below

Entries can also be forgotten: tell the agent to forget someone or something and the entry (with all references to it) is removed.

Events

When a conversation mentions something that happened — "Alice was laid off", "version 6.2 shipped", "the meeting on August 21" — it is recorded as an event: immutable, append-only, and hash-chained (each event's integrity hash covers the previous, so tampering is detectable). Events are never modified or deleted. Each one distinguishes when it happened (occurred at) from when Autopoetic learned it (recorded at).

How it learns

  • Read before chat — before each response, relevant knowledge is retrieved as transient context.
  • Write after chat — after each response, new and changed facts are extracted and the repository is updated. This is the "Actively learning…" status you see streaming.
  • Catch-up — conversations with Full knowledge access re-ingest past messages when needed, so nothing is lost to a failed update.

Each conversation controls its own access with the Project knowledge setting (Full / Read-only / None) — see chat features.

The knowledge graph

Knowledge Graph (project page → Project Knowledge) renders the whole repository as an interactive force-directed graph you can pan and zoom:

  • Nodes are entries (colored by type), edges are labeled relationships. Semantically similar entries appear as neighbors too.
  • Incremental loading — the graph starts with the most-connected entries and grows toward wherever you pan. Click a node to load more of its neighborhood.
  • Search — jump to any entry by name.
  • Deep links — event pages link their related entries straight into a focused graph view.
  • Time slider — scrub through the repository's history generation by generation, and return to the live state with one click.
  • Ghost nodes — hollow circles mark entries that were referenced but have since been forgotten.

Changelog and events pages

  • Knowledge Changelog — every change to the repository as a paginated history of diffs: what was added or removed from each field, word by word, with token costs and integrity hashes per change.
  • Events — the full event timeline with topics, statuses, descriptions, occurrence timestamps, related entries, and the hash-chain integrity line.

API access

The repository is fully scriptable: search, read, create, update, and delete entries, create relationships, and record events via the platform API or the MCP server. A project API key is required — see projects and teams for creating keys.