Graphs & Sensitivity Tiers

Graphs let you organize your cortex into distinct namespaces. Sensitivity tiers let you decide how much of each graph an AI is allowed to see — including the option to block AI access entirely.

Important: even for graphs with open tiers, your AI never receives everything in that graph. On every recall call, Graphnosis performs a semantic search and returns only the nodes most relevant to your current question, subject to the token cap for that graph. The rest stays encrypted and untouched. This is intentional — Graphnosis acts as your hippocampus, retrieving targeted memory traces, not dumping your files.

Multiple graphs

A single cortex can hold multiple graphs. You might create:

work — project notes, meeting summaries, technical decisions
personal — journal entries, health notes, finance snippets
research — papers, articles, reference material

Each graph is independent: its own sensitivity tier, its own token budget, its own sources. When the AI calls recall, it searches across all graphs you’ve granted it access to, respecting each graph’s tier.

Creating a graph

Open the Graphnosis window → click + New Graph → give it a name and choose a sensitivity tier.

Sensitivity tiers

Every graph is assigned one of three tiers:

Tier	What it means
`public`	Content may be surfaced to any AI client without restriction.
`personal`	Content is surfaced only when the AI explicitly asks for context (no proactive injection). Token cap applies.
`sensitive`	Excluded from AI recall by default. Surfaced only if you explicitly turn on "share with AI" for the engram — and even then, only under a tight cap.

Tiers are a hard cap, enforced by the sidecar before any content leaves the cortex. The AI model itself never sees the tier configuration — it simply doesn’t receive content above its allowed level.

Tier behavior in practice

public graphs — chunks flow freely into AI context. Proactive injection is enabled. Best for reference material, documentation, public notes.

personal graphs — chunks are only returned when recall is called explicitly. Proactive injection is disabled. The token cap limits how much context can be returned per conversation turn. Best for personal notes, journal entries, work summaries.

sensitive graphs — by default the sidecar returns zero results for recall queries targeting this graph; the AI is never told why, it just gets no results. If you deliberately turn on “share with AI” for a sensitive graph, recall returns at most a tight cap — 5 nodes / 500 tokens. You can always search and review these memories in the Graphnosis UI. Best for health information, financial records, anything you want to keep entirely local.

How much a recall returns

Recall is budgeted per call, not by a fixed per-tier number. When an AI client calls recall, it requests a token budget and a node count, and Graphnosis enforces a hard ceiling on both:

Limit	Default	Hard cap
Tokens attached per recall	2,000	8,000
Memory nodes attached per recall	20	50

That budget is shared across every engram the recall touches — Graphnosis fills it with the most relevant memories wherever they live, so no single engram can monopolize your AI’s context window.

Tiers shape this further: a public or personal engram can contribute up to the full budget, while a sensitive engram contributes nothing unless you have explicitly turned on “share with AI” for it — and even then it is held to a tight 5 nodes / 500 tokens.

Configuring policy.json

policy.json lives in the root of your cortex folder. It is not encrypted (it contains only policy rules, not content). You can edit it directly:

{
  "graphs": {
    "work": {
      "tier": "personal",
      "maxTokensPerRecall": 3000
    },
    "research": {
      "tier": "public",
      "maxTokensPerRecall": 6000
    },
    "health": {
      "tier": "sensitive"
    }
  },
  "globalMaxTokensPerTurn": 8000
}

Changes to policy.json take effect immediately — no restart required. The sidecar watches the file for changes.

`globalMaxTokensPerTurn`

This is a hard ceiling across all graphs combined. Even if individual graph caps add up to more, the sidecar will truncate the total context to this limit. Default: 8000.

Moving sources between engrams

Sources aren’t permanently bound to the engram they were ingested into. You can reassign a source at any time from the Sources pane — hover the row, click Move to…, and pick the destination.

If you need a new engram as the destination, select New Engram… from the dropdown, type a name, and Graphnosis will create it and move the source in one step.

Moving a source is instant and non-destructive. All chunks, embeddings, and cached content travel with it. The AI clients reflect the new location on the next recall call — no restart or re-ingest needed. The moved source is governed by the destination engram’s sensitivity tier immediately after the move.

Cross-engram connections are rebuilt immediately after every move. When a source moves, Graphnosis removes its old cross-engram links (they referenced the now-deleted node IDs in the origin engram), re-ingests the content into the destination with fresh node IDs, and immediately runs a cross-engram linking pass — without waiting for the next scheduled pass (every 6 hours). Related memories in other engrams are re-connected automatically. You don’t need to trigger anything manually.

Typical reasons to move a source:

You ingested a work document into personal by accident — move it to work
A research paper turned out to be closely related to a specific project — move it into that project’s engram for tighter recall
You’re splitting a large general-purpose engram into topic-specific ones over time

Archiving an engram

Archiving hides an engram from all in-app pickers and dropdowns — sources, the engram switcher, the 3D view — without touching its data. The graph file, all nodes, embeddings, and cross-engram connections remain exactly as they were. Nothing is removed, weakened, or re-indexed.

Archiving is fully reversible: click Unarchive and the engram reappears immediately with all its connections intact. Cross-engram links survive the archive/unarchive cycle unchanged.

Only Delete permanently removes an engram and purges its cross-engram connections from the store.

Which graphs does a client see?

By default, all graphs in the cortex are visible to any connected MCP client, subject to their tier. If you want to expose only specific graphs to a specific AI client, you can scope the sidecar using the GRAPHNOSIS_GRAPHS environment variable (see Environment Variables).