KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#codex #developer-tools #typescript #ai-agents #claude-code #python #mcp #react #cli #macos #ai-agent #deepseek-harness

CerebroCortex

★ 37 Open GitHub ↗

AI Brain

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h1 align="center">CerebroCortex</h1>
<h3 align="center">A brain that remembers like a brain.</h3>

<a href="#architecture">Architecture</a> &bull;
 <a href="#quick-start">Quick Start</a> &bull;
 <a href="#cognitive-bootstrap">Cognitive Bootstrap</a> &bull;
 <a href="#brain-regions">Brain Regions</a> &bull;
 <a href="#multi-agent-memory">Multi-Agent</a> &bull;
 <a href="#dream-engine">Dream Engine</a> &bull;
 <a href="#interfaces">Interfaces</a> &bull;
 <a href="#agent-framework-integration">Integration</a> &bull;
 <a href="#dashboard">Dashboard</a>

---

CerebroCortex is a **brain-analogous memory system for AI agents**. It doesn't just store and retrieve text — it *encodes*, *associates*, *consolidates*, and *dreams*, using the same cognitive architecture that makes biological memory so powerful.

Built for the [Model Context Protocol (MCP)](https://modelcontextprotocol.io), it gives [Hermes Agent](https://github.com/NousResearch/hermes-agent), Claude, or any MCP-compatible LLM persistent, associative, emotionally-weighted memory that strengthens with use, decays with neglect, and reorganizes itself while it sleeps.

> **TL;DR** — Imagine if your AI assistant had a hippocampus, an amygdala, and could dream.
> That's CerebroCortex.

---

## Why CerebroCortex?

Most AI memory systems are **glorified vector databases** — dump text in, cosine-similarity it back out. That works for search. It doesn't work for *memory*.

Real memory is:

| What real memory does | What vector DBs do | What CerebroCortex does |
|---|---|---|
| **Associates** — links ideas by meaning, emotion, causality | Returns top-k similar docs | 9 typed link types + spreading activation through an associative graph |
| **Forgets intelligently** — old unused memories fade | Stores everything forever | ACT-R power-law decay + FSRS spaced-repetition |
| **Strengthens with use** — frequently recalled memories persist | Static similarity scores | Hebbian "neurons that fire together wire together" strengthening |
| **Has emotional weight** — traumatic/joyful moments stick | All memories are equal | Amygdala engine scores valence + arousal, boosts salience |
| **Consolidates offline** — sleep reorganizes and prunes | Nothing | 6-phase Dream Engine with LLM-powered pattern extraction |
| **Layers memories** — sensory -> working -> long-term -> cortex | Flat storage | 4-layer promotion system with configurable decay |

---

## Architecture

```mermaid
graph TB
 subgraph Interfaces["🔌 Interfaces"]
 MCP["MCP Server 63 tools"]
 API["REST API 78 endpoints"]
 CLI["CLI cerebro"]
 DASH["Dashboard Alchemical UI"]
 WATCH["File Watcher Auto-ingest"]
 CCBS["CCBS Cognitive Bootstrap"]
 end

subgraph Cortex["🧠 CerebroCortex Coordinator"]
        REM["remember() / bulk_remember()"]
        REC["recall() — text + vision"]
        ASS["associate()"]
        EPI["episode_*()"]
        CRUD["restore() / purge() / versions()"]
    end

subgraph Engines["⚡ Brain Region Engines"]
 TH["Thalamus Gating"]
 AM["Amygdala Affect"]
 TL["Temporal Lobe Semantics"]
 HC["Hippocampus Episodes"]
 AC["Association Cortex Linking"]
 CB["Cerebellum Procedural"]
 PF["Prefrontal Cortex Executive"]
 NC["Neocortex Schemas"]
 DR["Default Mode Network Dream Engine"]
 end

subgraph Activation["📐 Activation System"]
 ACTR["ACT-R B(t) = ln(Σ t_k⁻ᵈ)"]
 FSRS["FSRS R(t,S) = (1+t/9S)⁻¹"]
 SA["Spreading Activation Collins & Loftus"]
 end

subgraph Storage["💾 Triple-Backend Storage"]
 SQ["SQLite Source of truth"]
 IG["igraph In-memory traversal"]
 CH["ChromaDB Vector search + FTS5 fallback"]
 VS["VisionVectorStore CLIP embeddings"]
 end

MCP --> Cortex
    API --> Cortex
    CLI --> Cortex
    DASH --> API
    WATCH --> Cortex
    Cortex --> Engines
    Engines --> Activation
    Engines --> Storage
    Activation --> Storage
    CCBS --> Cortex

style Interfaces fill:#1a1a2e,stroke:#00ff41,color:#00ff41
    style Cortex fill:#1a1a2e,stroke:#00d4ff,color:#00d4ff
    style Engines fill:#1a1a2e,stroke:#ff0055,color:#ff0055
    style Activation fill:#1a1a2e,stroke:#ffd700,color:#ffd700
    style Storage fill:#1a1a2e,stroke:#9d4edd,color:#9d4edd
```

---

## Quick Start

### Install

```bash
git clone https://github.com/buckster123/CerebroCortex.git
cd CerebroCortex
pip install -e ".[all]"
```

For vision support (image/PDF ingestion, cross-modal recall), include the `[vision]` extra:

```bash
pip install -e ".[all,vision]"
```

### Run as MCP Server (for Hermes Agent, Claude, or any MCP client)

```bash
./cerebro-mcp
```

Add to your Hermes config (`~/.hermes/config.yaml`), Claude config (`~/.claude.json`), or any MCP client:

```json
{
  "mcpServers": {
    "cerebro-cortex": {
      "command": "/path/to/CerebroCore/cerebro-mcp"
    }
  }
}
```

For Hermes Agent, add to your `~/.hermes/config.yaml`:

```yaml
mcp:
  servers:
    cerebro-cortex:
      command: cerebro-mcp  # or /path/to/CerebroCore/cerebro-mcp
```

See **[HERMES_INTEGRATE.md](HERMES_INTEGRATE.md)** for full Hermes setup including the memory provider plugin.

Optionally set an agent identity via environment variable:

```bash
CEREBRO_AGENT_ID=my-agent ./cerebro-mcp
```

### Run as REST API

```bash
./cerebro-api
# → http://localhost:8767
# → Dashboard at http://localhost:8767/ui
```

### Use the CLI

```bash
# Store a memory
cerebro remember "The mitochondria is the powerhouse of the cell" --type semantic

# Search memories
cerebro recall "cell biology" -n 5

# Check stats
cerebro stats

# Run the Dream Engine
cerebro dream --run
```

### Python API

```python
from cerebro.cortex import CerebroCortex
from cerebro.types import MemoryType, LinkType, Visibility

brain = CerebroCortex()
brain.initialize()

# Remember (with multi-agent support)
node = brain.remember(
    "Deployment failed due to missing env var DATABASE_URL",
    memory_type=MemoryType.EPISODIC,
    tags=["deployment", "bug"],
    salience=0.8,
    agent_id="ALICE",                  # Which agent owns this
    visibility=Visibility.SHARED,       # SHARED, PRIVATE, or THREAD
)

# Recall (scoped to agent's visible memories)
results = brain.recall("deployment issues", top_k=5, agent_id="BOB")
for memory, score in results:
    print(f"  [{score:.2f}] {memory.content[:80]}")

# Associate
brain.associate(node.id, other_id, LinkType.CAUSAL, weight=0.9)

brain.close()
```

---

## Cognitive Bootstrap System (CCBS)

> **Your AI's operating system for memory.**

CerebroCortex v0.5.0 introduces the **Cerebro Cognitive Bootstrap System** — dynamic, query-aware cognitive priming that teaches any connected AI agent how to use its own brain effectively.

Instead of burning context on static system prompts, CCBS loads only the cognitive modules the current session needs. It's a living, recallable, token-budgeted cognitive architecture stored inside Cerebro itself.

### How It Works

```mermaid
graph LR
    QUERY["User Query"] --> ANALYZE["Query Analysis"]
    ANALYZE --> TRIGGERS["Manual Triggers?"]
    ANALYZE --> KEYWORDS["Keyword Detection"]
    TRIGGERS --> ASSEMBLE["Assemble Block"]
    KEYWORDS --> ASSEMBLE
    ASSEMBLE --> BUDGET["Token Budget"]
    BUDGET --> PROMPT["Cognitive Prompt Block"]
    
    style QUERY fill:#1a1a2e,stroke:#00ff41,color:#00ff41
    style ANALYZE fill:#1a1a2e,stroke:#00d4ff,color:#00d4ff
    style ASSEMBLE fill:#1a1a2e,stroke:#ffd700,color:#ffd700
    style PROMPT fill:#1a1a2e,stroke:#ff0055,color:#ff0055
```

### 12 Modular Cognitive Modules

| Module | Type | Loaded | Purpose |
|--------|------|--------|---------|
| **soul** | manifest | always | Master bootstrap — loading protocol, triggers, budget tiers |
| **module-core** | identity | always | Base reasoning style, communication patterns |
| **module-cerebro-index** | navigation | always | Table of contents for all cerebro operations |
| **module-cerebro-ops** | operations | always | remember/recall/associate/search patterns |
| **module-cerebro-session** | tracking | always | Checkpoint protocol, episode tracking |
| **module-cerebro-meta** | strategy | always | 90/10 rule, auto-remember triggers |
| **module-cerebro-intentions** | prospective | auto | TODOs, reminders, deferred tasks |
| **module-cerebro-agents** | multi-agent | auto | Cross-agent messaging, shared memory |
| **module-technical** | engineering | auto | Code, systems, architecture reasoning |
| **module-analysis** | evaluation | auto | Debugging, measurement, trade-offs |
| **module-creative** | design | auto | Ideation, lateral thinking, aesthetics |
| **module-research** | synthesis | auto | Literature review, evidence evaluation |
| **module-communicate** | explanation | auto | Teaching, documenting, presenting |

### Token Budget Tiers

| Mode | Est. Tokens | Use Case |
|------|-------------|----------|
| **Minimal** | ~900 | soul + core only — quick queries |
| **Standard** | ~1,600 | Mandatory + 1-2 detected — normal sessions |
| **Full** | ~4,200 | All modules — deep work, complex tasks |

### Manual Triggers

Override auto-detection by saying:

| Trigger | Action |
|---------|--------|
| "Full load" / "Max brain" / "All in" | Load ALL modules |
| "Solo core" / "Minimal" | Load only soul + core |
| "Debug mode" | core + cerebro + technical + analysis |
| "Creative mode" | core + cerebro + creative |
| "Research mode" | core + cerebro + research + analysis |
| "Cerebro mode" | core + all cerebro sub-modules |
| "Teach me" / "Explain" | core + cerebro + communicate |

### Using CCBS

**Via REST API:**
```bash
curl -X POST http://localhost:8767/bootstrap \
  -H "Content-Type: application/json" \
  -d '{"query": "Debug why the API is failing"}'
```

**Via MCP tools:**
```
mcp_cerebro_recall(query="CCBS bootstrap soul module", top_k=3)
mcp_cerebro_recall(query="CCBS module core", top_k=2)
mcp_cerebro_recall(query="CCBS module technical", top_k=1)
```

**Via Python:**
```python
from cerebro.cortex import CerebroCortex
from cerebro.interfaces.api_server import CognitiveBootstrapAssembler

brain = CerebroCortex()
assembler = CognitiveBootstrapAssembler(brain)

block = assembler.assemble(
    query="Design a new dashboard layout",
    mode="standard",
    max_tokens=2000
)
print(block["modules_loaded"])  # ['soul', 'module-core', ...]
print(block["total_tokens"])    # 1650
```

### The 90/10 Rule

CCBS enforces a memory ops discipline: **Cerebro first, harness memory second.**

| Data Type | Destination | Why |
|-----------|-------------|-----|
| Session summaries, discoveries | Cerebro `remember()` | Grows, needs recall |
| TODOs, intentions | Cerebro `store_intention()` | Participates in recall |
| Project ecosystem map | Cerebro `remember()` | Changes over time |
| Host OS, hardware facts | Harness memory | Static, never changes |
| User communication prefs | Harness memory | Stable conventions |

Harness memory has a ~2.2K char limit. Cerebro has no such limit.

---

## Brain Regions

CerebroCortex maps 9 engines to their biological counterparts. Each engine handles a specific cognitive function, just like the real brain:

```mermaid
graph LR
    INPUT["📥 Input"] --> TH

subgraph Encoding["Encoding Pipeline"]
 TH["🫁 Thalamus Gating & Filtering"]
 TL["🧩 Temporal Lobe Concept Extraction"]
 AM["❤️ Amygdala Emotional Analysis"]
 end

subgraph Association["Association & Storage"]
 AC["🔗 Association Cortex Auto-Linking"]
 HC["🐴 Hippocampus Episode Binding"]
 CB["🎯 Cerebellum Procedural Memory"]
 end

subgraph Executive["Executive & Abstraction"]
 PF["🎓 Prefrontal Cortex Ranking & Promotion"]
 NC["🌐 Neocortex Schema Formation"]
 end

subgraph Offline["Offline Consolidation"]
 DR["💤 Default Mode Network Dream Engine"]
 end

TH --> TL --> AM --> AC
    AC --> HC
    AC --> CB
    HC --> PF
    CB --> PF
    PF --> NC
    NC --> DR
    DR -.->|"strengthens"| AC

style Encoding fill:#0d1117,stroke:#00ff41,color:#c9d1d9
    style Association fill:#0d1117,stroke:#00d4ff,color:#c9d1d9
    style Executive fill:#0d1117,stroke:#ffd700,color:#c9d1d9
    style Offline fill:#0d1117,stroke:#9d4edd,color:#c9d1d9
```

| Engine | Brain Region | What It Does |
|---|---|---|
| **GatingEngine** | Thalamus | First line of defense. Filters noise, detects duplicates, assigns memory layer, estimates initial salience. If it's not worth remembering, it dies here. |
| **AffectEngine** | Amygdala | Emotional processor. Scores valence (positive/negative/mixed) and arousal (calm→intense). High-emotion memories get salience boosts — just like real trauma and joy. |
| **SemanticEngine** | Temporal Lobe | Concept extractor. Identifies key concepts, creates semantic links between related memories, detects contradictions. |
| **EpisodicEngine** | Hippocampus | Episode manager. Binds memories into temporal sequences (episodes), tracks context flow, enables "what happened during X" retrieval. |
| **LinkEngine** | Association Cortex | The connector. Auto-links new memories to related ones, runs spreading activation, applies Hebbian strengthening ("fire together, wire together"). |
| **ProceduralEngine** | Cerebellum | Skill memory. Identifies workflows, strategies, and repeatable patterns. Tracks success/failure to reinforce what works. |
| **ExecutiveEngine** | Prefrontal Cortex | The executive. Ranks recall results, decides layer promotions (working→long-term), manages prospective memory (TODOs). |
| **SchemaEngine** | Neocortex | The abstractifier. Extracts general principles from specific episodes. "3 deployments failed from missing env vars" → schema: "always validate env before deploy". |
| **DreamEngine** | Default Mode Network | Offline consolidation. Replays, clusters, abstracts, prunes, and recombines — exactly like biological sleep. Details [below](#dream-engine). |

---

## Memory Types

Six modalities, modeled after human cognitive memory research:

```mermaid
mindmap
 root((Memory Types))
 Episodic
 Temporal sequences
 "What happened"
 Context-rich
 Semantic
 Facts & concepts
 "What I know"
 Relationship-rich
 Procedural
 Strategies & workflows
 "How to do it"
 Success-tracked
 Affective
 Emotional markers
 "How it felt"
 Salience-boosted
 Prospective
 Future intentions
 "What to do next"
 Deadline-aware
 Schematic
 Abstractions
 "The pattern"
 Dream-derived
```

| Type | Analogy | Example |
|---|---|---|
| **Episodic** | Diary entry | "During the 2024-03-15 deploy, the DB migration failed at step 4" |
| **Semantic** | Encyclopedia | "PostgreSQL supports JSONB columns for semi-structured data" |
| **Procedural** | Recipe | "To deploy: run tests → build image → push → apply k8s manifests" |
| **Affective** | Gut feeling | "Last time we rushed a Friday deploy, it went badly — high anxiety" |
| **Prospective** | Sticky note | "TODO: Migrate user table to new schema before v2 launch" |
| **Schematic** | Mental model | "Schema: Services that skip staging tend to have 3x more prod incidents" |

---

## The Activation System

This is where CerebroCortex diverges from every other memory system. Instead of static similarity scores, memories have *dynamic activation levels* that change over time.

### Combined Recall Score

Every memory retrieval produces a score blending four signals:

```mermaid
pie title Recall Score Composition
    "Vector Similarity" : 35
    "ACT-R Activation" : 30
    "FSRS Retrievability" : 20
    "Salience" : 15
```

```
score = 0.35 × vector_similarity
      + 0.30 × actr_activation
      + 0.20 × fsrs_retrievability
      + 0.15 × salience
```

---

<details>
<summary>🔬 Deep Dive: ACT-R Base-Level Activation</summary>

### ACT-R — Adaptive Control of Thought (Rational)

CerebroCortex implements the **base-level activation equation** from John Anderson's ACT-R cognitive architecture (Carnegie Mellon, 1993):

```
B(t) = ln( Σ t_k^{-d} )
```

Where:
- `t_k` = time since the k-th access (in seconds)
- `d` = decay rate (default: **0.5** — square-root power law)
- The sum runs over all recorded access timestamps

**What this means in practice:**
- A memory accessed once 10 minutes ago has *higher* activation than one accessed once 10 days ago
- A memory accessed 50 times over months has *higher* activation than one accessed twice yesterday
- Activation follows a **power law**, not exponential — old memories don't vanish, they just get quieter

**Parameters:**

| Parameter | Value | Meaning |
|---|---|---|
| `ACTR_DECAY_RATE` | 0.5 | Power-law exponent (higher = faster decay) |
| `ACTR_NOISE` | 0.4 | Stochastic noise in retrieval (models human variability) |
| `ACTR_RETRIEVAL_THRESHOLD` | 0.0 | Minimum activation to be retrievable |
| `MAX_STORED_TIMESTAMPS` | 50 | Individual timestamps kept; older ones get compressed |

</details>

<details>
<summary>🔬 Deep Dive: FSRS Retrievability</summary>

### FSRS — Free Spaced Repetition Scheduler

Alongside ACT-R, each memory tracks an FSRS-based **retrievability** score using the forgetting curve from modern spaced-repetition research:

```
R(t, S) = (1 + t / (9 × S))^{-1}
```

Where:
- `t` = elapsed time since last review (in days)
- `S` = stability (how long the memory resists forgetting)

**Stability grows with successful recalls.** Every time you access a memory, its stability increases, meaning it takes longer to forget. This models the "spacing effect" — the finding that distributed practice beats massed practice.

| Parameter | Value | Meaning |
|---|---|---|
| `FSRS_INITIAL_STABILITY` | 1.0 day | New memories start fragile |
| `FSRS_INITIAL_DIFFICULTY` | 5.0 / 10 | Mid-range difficulty |
| `FSRS_MIN_STABILITY` | 0.1 days | Floor (2.4 hours) |
| `FSRS_MAX_STABILITY` | 365 days | Ceiling (1 year) |

</details>

<details>
<summary>🔬 Deep Dive: Spreading Activation</summary>

### Collins & Loftus Spreading Activation (1975)

When you recall a memory, activation doesn't stay local — it **spreads** through the associative network to related memories, exactly like priming in human cognition.

style A fill:#00ff41,stroke:#000,color:#000
    style B fill:#ffd700,stroke:#000,color:#000
    style C fill:#ffd700,stroke:#000,color:#000
    style D fill:#ff6b35,stroke:#000,color:#000
    style E fill:#ff6b35,stroke:#000,color:#000
```

**How it works:**
1. Seed memories start with activation = 1.0
2. Activation flows through links, weighted by link type and link weight
3. Each hop applies a **decay factor** (default: 0.6, so 40% loss per hop)
4. Stops at **max 2 hops** or when activation drops below 0.05
5. Maximum **50 activated nodes** per spread

**Link Type Weights** (how well each type conducts activation):

| Link Type | Weight | Rationale |
|---|---|---|
| `causal` | 0.9 | Cause→effect is the strongest association |
| `semantic` | 0.8 | Conceptual relatedness is highly relevant |
| `supports` | 0.8 | Evidence is closely tied to claims |
| `part_of` | 0.8 | Components are tightly bound to wholes |
| `contextual` | 0.7 | Shared context is moderately relevant |
| `derived_from` | 0.7 | Abstractions connect to source material |
| `temporal` | 0.6 | "Happened around the same time" is weaker |
| `affective` | 0.5 | Emotional similarity is a softer signal |
| `contradicts` | 0.3 | Contradictions are related, but oppositionally |

</details>

---

## Memory Layers

Memories flow through four durability layers, inspired by the Atkinson-Shiffrin multi-store model:

style S fill:#1a1a2e,stroke:#00ff41,color:#00ff41
    style W fill:#1a1a2e,stroke:#00d4ff,color:#00d4ff
    style LT fill:#1a1a2e,stroke:#ffd700,color:#ffd700
    style C fill:#1a1a2e,stroke:#ff0055,color:#ff0055
```

| Layer | Half-Life | Promotion Criteria | Description |
|---|---|---|---|
| **Sensory** | 6 hours | Automatic on 2nd access | Raw input buffer. Most memories die here. |
| **Working** | 3 days | 5+ accesses, 24h+ old | Active working memory. Frequent use keeps it alive. |
| **Long-Term** | 30 days | Dream Engine consolidation only | Stable storage. Only the Dream Engine can promote further. |
| **Cortex** | Permanent | — | Crystallized knowledge. No decay. The stuff you'll never forget. |

---

## Associative Network

Memories don't exist in isolation — they form a **typed, weighted graph** of associations:

style A fill:#00ff41,color:#000
    style B fill:#00d4ff,color:#000
    style C fill:#ff6b35,color:#000
    style D fill:#00ff41,color:#000
    style E fill:#ff0055,color:#000
    style F fill:#ffd700,color:#000
    style G fill:#00d4ff,color:#000
```

**9 link types:**

| Type | Meaning | Example |
|---|---|---|
| `temporal` | A happened before/after B | Deploy → Hotfix |
| `causal` | A caused B | Missing var → Deploy failure |
| `semantic` | A is conceptually related to B | PostgreSQL ↔ Database |
| `affective` | A evokes similar emotion to B | Friday deploy ↔ Anxiety |
| `contextual` | A and B share context | Same session, same project |
| `contradicts` | A conflicts with B | "Safe to deploy" vs. "Never deploy Fridays" |
| `supports` | A provides evidence for B | Incident report → Best practice |
| `derived_from` | B was abstracted from A | Episodes → Schema |
| `part_of` | A is a step within B | Step → Episode |

---

## Multi-Agent Memory

CerebroCortex supports **solo agents, multi-agent teams, and hive-mind configurations** with full visibility scoping:

| Visibility | Who can see it | Use case |
|---|---|---|
| **SHARED** (default) | All agents | Collective knowledge, shared facts |
| **PRIVATE** | Only the owner | Internal state, secrets, personal notes |
| **THREAD** | Participants in a conversation thread | Scoped collaboration |

### How It Works

Every memory carries `agent_id` and `visibility`. Enforcement flows through every layer:

- **Recall** — ChromaDB vector search uses `$or` visibility clauses; post-filter applies `_can_access()`
- **Spreading activation** — batch visibility cache prevents leaking across agent boundaries
- **All 9 engines** — SQL queries append scope filters via `_scope_sql()`
- **Dream Engine** — runs a scoped cycle per agent; shared memories consolidate in every agent's dream
- **Auto-linking** — cross-agent PRIVATE links are prevented at creation time
- **Link pruning** — when visibility changes to PRIVATE, cross-agent links are automatically deleted
- **Episodes** — `get_episode()` and `get_episode_memories()` verify agent ownership

### Quick Example

```bash
# Agent ALICE stores private and shared memories
cerebro remember "Alice's secret" --agent ALICE --visibility private
cerebro remember "Shared knowledge" --agent ALICE

# BOB can't see ALICE's private memories
cerebro recall "secret" --agent BOB    # returns nothing
cerebro recall "knowledge" --agent BOB # finds shared knowledge

# Owner can share private memories
cerebro share <mem_id> shared --agent ALICE

# Dream runs per-agent automatically
cerebro dream --run  # One cycle per registered agent
```

### Backwards Compatibility

Omitting `--agent` or `agent_id` disables filtering entirely — all memories are visible. This means single-agent setups work unchanged, and the default visibility of SHARED ensures all existing memories remain accessible.

---

## Dream Engine

The crown jewel. Inspired by sleep research, the Dream Engine runs **offline consolidation cycles** that reorganize, strengthen, prune, and recombine memories — exactly like the brain does during sleep.

```mermaid
graph TD
 subgraph Cycle["🌙 Dream Cycle"]
 direction TB
 P1["Phase 1: SWS Replay Replay episodes, strengthen temporal links"]
 P2["Phase 2: Pattern Extraction Cluster similar memories, find themes"]
 P3["Phase 3: Schema Formation Abstract episodes into general principles"]
 P4["Phase 4: Emotional Reprocessing Adjust salience based on outcomes"]
 P5["Phase 5: Pruning Decay stale low-salience memories"]
 P6["Phase 6: REM Recombination Discover unexpected connections"]
 end

P1 --> P2 --> P3 --> P4 --> P5 --> P6

P1 -.- N1["⚡ Algorithmic"]
    P2 -.- N2["🤖 LLM-Assisted"]
    P3 -.- N3["🤖 LLM-Assisted"]
    P4 -.- N4["⚡ Algorithmic"]
    P5 -.- N5["⚡ Algorithmic"]
    P6 -.- N6["🤖 LLM-Assisted"]

style Cycle fill:#0d1117,stroke:#9d4edd,color:#c9d1d9
    style N1 fill:none,stroke:none,color:#00ff41
    style N2 fill:none,stroke:none,color:#00d4ff
    style N3 fill:none,stroke:none,color:#00d4ff
    style N4 fill:none,stroke:none,color:#00ff41
    style N5 fill:none,stroke:none,color:#00ff41
    style N6 fill:none,stroke:none,color:#00d4ff
```

<details>
<summary>🧬 Deep Dive: Dream Phase Details</summary>

### Phase 1 — Slow-Wave Sleep Replay

Like the hippocampal replay observed during deep sleep, this phase **replays recent episodes** and strengthens temporal links between memories that occurred together.

- Algorithmic (no LLM cost)
- Strengthens `temporal` and `contextual` links within episodes
- Focuses on most recent episodes first

### Phase 2 — Pattern Extraction

Clusters similar memories using a similarity threshold (default: 0.80), then uses LLM to **summarize recurring patterns** from each cluster.

- Minimum cluster size: 3 memories
- Extracts procedures from repeated workflows
- Creates new `semantic` links between cluster members

### Phase 3 — Schema Formation

The most powerful phase. Takes episodes and **abstracts them into general schemas** — reusable principles and mental models.

- LLM generates schema descriptions from episode summaries
- Creates `derived_from` links: schema → source episodes
- New schemas are typed as `MemoryType.SCHEMATIC`

### Phase 4 — Emotional Reprocessing

Adjusts salience scores based on outcomes. Memories associated with negative outcomes get **boosted salience** (learn from mistakes), while repeatedly successful patterns get normalized.

- Algorithmic (no LLM cost)
- Models the finding that emotional memories are preferentially consolidated during sleep

### Phase 5 — Pruning

Removes noise. Memories that are:
- Older than **48 hours**
- Below **0.3 salience**
- In the **sensory** layer
- Have **no links** to other memories

...get pruned. This prevents unbounded growth while preserving everything meaningful.

### Phase 6 — REM Recombination

The creative phase. Samples **20 diverse memories**, then checks **10 random pairs** for unexpected connections using LLM analysis.

- Models the "creative insight during REM sleep" phenomenon
- Discovers non-obvious relationships between distant memories
- Minimum connection strength: 0.4 (only confident connections)
- Creates new `semantic` or `causal` links

</details>

### Running the Dream Engine

Dream cycles run **per-agent automatically** — each registered agent gets a scoped dream where shared memories participate in every cycle but private memories stay isolated.

```bash
# CLI (auto per-agent)
cerebro dream --run

# REST API
curl -X POST http://localhost:8767/dream/run

# MCP (Claude will call this automatically)
dream_run(max_llm_calls=20)
```

**LLM Configuration:**
- Primary: **OpenAI-compatible** (LM Studio, vLLM, or any local endpoint)
- Fallback: **Claude** (Anthropic API)
- Max LLM calls per cycle per agent: 20
- Temperature: 0.7
- Configurable at runtime via `data/settings.json`

---

## Storage Architecture

CerebroCortex uses a **triple-backend architecture** — each backend handles what it's best at:

```mermaid
graph LR
    subgraph SQLite["SQLite"]
        S1["Nodes"]
        S2["Links"]
        S3["Episodes"]
        S4["Agents"]
    end

subgraph igraph["igraph (C library)"]
        I1["In-memory graph"]
        I2["Spreading activation"]
        I3["Neighbor traversal"]
    end

subgraph ChromaDB["ChromaDB"]
        C1["cerebro_memories"]
        C2["cerebro_knowledge"]
        C3["cerebro_sessions"]
    end

SQLite -->|"rebuild on init"| igraph
    igraph -->|"persist changes"| SQLite

style SQLite fill:#1a1a2e,stroke:#ffd700,color:#ffd700
    style igraph fill:#1a1a2e,stroke:#00ff41,color:#00ff41
    style ChromaDB fill:#1a1a2e,stroke:#00d4ff,color:#00d4ff
```

| Backend | Role | Why |
|---|---|---|
| **SQLite** | Canonical persistent store | Reliable, zero-config, great for single-node (RPi5-friendly) |
| **igraph** | In-memory graph operations | C-speed traversal, spreading activation in microseconds |
| **ChromaDB** | Vector similarity search | 384-dim SBERT embeddings, fast ANN retrieval |

**Design choice:** igraph + SQLite instead of Neo4j. Neo4j is powerful but heavy — CerebroCortex runs comfortably on a **Raspberry Pi 5** with 8GB RAM.

---

## Interfaces

CerebroCortex exposes three interfaces — use whichever fits your workflow:

### MCP Server (56 Tools)

The native interface for Hermes Agent, Claude, and any MCP-compatible agent. All tool descriptions are written in plain English so agents can pick tools by description alone — no jargon.

```bash
./cerebro-mcp
```

<details>
<summary>Full MCP Tool List (63 tools)</summary>

| Tool | Description |
|---|---|
| **Core Memory** | |
| `remember` | Save information to long-term memory |
| `recall` | Search memories by meaning, ranked by relevance, importance, and recency |
| `get_memory` | Get a single memory by ID with all metadata |
| `update_memory` | Update content, tags, importance, or visibility |
| `delete_memory` | Delete a memory (soft-delete by default) |
| `restore_memory` | Restore a soft-deleted memory from trash |
| `purge_memory` | Permanently delete a memory or old trash |
| `share_memory` | Change who can see a memory (owner only) |
| `associate` | Create a link between two memories (improves search) |
| **Versions** | |
| `get_memory_versions` | List edit history for a memory |
| `restore_version` | Roll back to a previous revision |
| **Trash** | |
| `list_deleted` | List soft-deleted memories in trash |
| `purge_all_deleted` | Empty trash older than N days |
| **Bulk Operations** | |
| `bulk_delete` | Soft-delete multiple memories at once |
| `export_memories` | Export memories to JSON or Markdown |
| **Tags** | |
| `list_tags` | List all tags in use |
| `rename_tag` | Rename a tag across all memories |
| `merge_tags` | Merge two tags into one |
| `delete_tag` | Remove a tag from all memories |
| **Threads** | |
| `list_threads` | List conversation threads |
| `get_thread_memories` | Get all memories in a thread |
| `prune_thread` | Remove old memories from a thread |
| **Vision** | |
| `ingest_file` | Read a file and store as searchable memories (supports .md, .json, .txt, .pdf, .html, .csv, images, code) |
| `describe_image` | Generate a text description of an image |
| `search_vision` | Search by image similarity |
| **Episodes** | |
| `episode_start` | Start recording a sequence of related events |
| `episode_add_step` | Add a memory as the next step in an episode |
| `episode_end` | Finish recording an episode |
| `list_episodes` | List recent episodes |
| `get_episode` | Get episode details (scope-checked) |
| `get_episode_memories` | Get all memories in an episode, in order |
| **Sessions** | |
| `session_save` | Save a session summary for future continuity |
| `session_recall` | Retrieve notes from previous sessions |
| **Agents** | |
| `register_agent` | Register a new agent in the memory system |
| `list_agents` | List all registered agents |
| **Intentions** | |
| `store_intention` | Save a TODO or reminder for future action |
| `list_intentions` | List pending TODOs and reminders |
| `resolve_intention` | Mark a TODO as done |
| **Schemas & Procedures** | |
| `create_schema` | Create a general pattern or principle from multiple memories |
| `list_schemas` | List stored patterns and principles |
| `find_matching_schemas` | Find patterns matching tags or concepts |
| `get_schema_sources` | Get original memories a pattern was derived from |
| `store_procedure` | Store a workflow, strategy, or how-to guide |
| `list_procedures` | List stored workflows and how-to guides |
| `find_relevant_procedures` | Find workflows matching tags or concepts |
| `record_procedure_outcome` | Record whether a procedure worked or failed |
| **Graph Exploration** | |
| `memory_neighbors` | Get memories directly linked to a given memory |
| `find_path` | Find the shortest chain of links between two memories |
| `common_neighbors` | Find memories linked to both A and B |
| **Cognitive Bootstrap (CCBS)** | |
| `cognitive_bootstrap` | Assemble dynamic cognitive priming modules based on query intent |
| **Near-Duplicate Detection** | |
| `check_near_duplicates` | Preview potential duplicates before ingestion |
| **Activation & Decay** | |
| `activation_heatmap` | Get activation data for all memories (scatter-plot ready) |
| `activation_at_risk` | Find fading memories that need revival |
| `activation_curve` | Project ACT-R decay curve for a specific memory |
| **Audit Logging** | |
| `query_audit` | Query the audit log with filters |
| `audit_summary` | Summary of audit events by type |
| **System** | |
| `cortex_stats` | Comprehensive system statistics (raw JSON) |
| `memory_health` | Memory system health report |
| `memory_graph_stats` | Detailed graph structure metrics |
| `emotional_summary` | Breakdown of memories by emotional tone |
| `dream_run` | Run offline memory maintenance cycle |
| `dream_status` | Get status of last maintenance cycle |
| **Compatibility** | |
| `memory_store` | Alias for `remember` |
| `memory_search` | Alias for `recall` |

</details>

### REST API (78 Endpoints)

Full HTTP API with interactive docs at `/docs`.

```bash
./cerebro-api
# → http://localhost:8767
# → Swagger docs at http://localhost:8767/docs
```

<details>
<summary>Full Endpoint List</summary>

| Method | Path | Description |
|---|---|---|
| `GET` | `/` | API info |
| `GET` | `/health` | Health check |
| `GET` | `/stats` | System statistics |
| `GET` | `/ui` | Web dashboard |
| **Memory CRUD** | | |
| `POST` | `/remember` | Store memory |
| `POST` | `/recall` | Search memories |
| `GET` | `/q/{query}` | Quick search |
| `GET` | `/memory/{id}` | Get memory by ID |
| `PATCH` | `/memory/{id}` | Update memory |
| `DELETE` | `/memory/{id}` | Delete memory |
| `POST` | `/memory/{id}/share` | Change visibility (owner only) |
| `GET` | `/memory/{id}/versions` | Get version history |
| `POST` | `/memory/{id}/versions/{vid}/restore` | Restore a version |
| `POST` | `/associate` | Create link |
| **Ingestion** | | |
| `POST` | `/ingest/upload` | Upload and ingest a file (multipart) |
| **Cognitive Bootstrap** | | |
| `POST` | `/bootstrap` | Assemble cognitive prompt block from CCBS modules |
| **Trash** | | |
| `GET` | `/trash` | List soft-deleted memories |
| `POST` | `/trash/{id}/restore` | Restore from trash |
| `DELETE` | `/trash/{id}` | Permanently delete |
| `POST` | `/trash/purge-all` | Bulk purge old trash |
| **Tags** | | |
| `GET` | `/tags` | List all tags |
| `POST` | `/tags/rename` | Rename a tag |
| `POST` | `/tags/merge` | Merge tags |
| `DELETE` | `/tags/{tag}` | Delete a tag |
| **Threads** | | |
| `GET` | `/threads` | List conversation threads |
| `GET` | `/threads/{id}/memories` | Get thread memories |
| `DELETE` | `/threads/{id}` | Prune (soft-delete) a thread |
| **Bulk Operations** | | |
| `POST` | `/bulk/delete` | Soft-delete multiple memories |
| `POST` | `/bulk/visibility` | Change visibility for multiple |
| `POST` | `/export` | Export memories to JSON/Markdown |
| **Episodes** | | |
| `POST` | `/episodes/start` | Start episode |
| `POST` | `/episodes/{id}/step` | Add episode step |
| `POST` | `/episodes/{id}/end` | End episode |
| `GET` | `/episodes` | List episodes |
| `GET` | `/episodes/{id}` | Get episode (scope-checked) |
| `GET` | `/episodes/{id}/memories` | Get episode memories (scope-checked) |
| **Sessions** | | |
| `POST` | `/sessions/save` | Save session note |
| `GET` | `/sessions` | Recall sessions |
| **Agents** | | |
| `GET` | `/agents` | List agents |
| `POST` | `/agents` | Register agent |
| **Intentions** | | |
| `POST` | `/intentions` | Store intention |
| `GET` | `/intentions` | List pending |
| `POST` | `/intentions/{id}/resolve` | Resolve intention |
| **Schemas & Procedures** | | |
| `GET` | `/schemas` | List schemas |
| `POST` | `/schemas` | Create schema |
| `GET` | `/procedures` | List procedures |
| `POST` | `/procedures` | Store procedure |
| **Graph & System** | | |
| `GET` | `/memory/health` | Health report |
| `GET` | `/graph/stats` | Graph statistics |
| `GET` | `/graph/data` | Graph data (for visualization) |
| `GET` | `/graph/neighbors/{id}` | Memory neighbors |
| `GET` | `/graph/path/{a}/{b}` | Find path |
| `POST` | `/dream/run` | Run Dream Engine (per-agent) |
| `GET` | `/dream/status` | Dream status |
| `GET` | `/emotions` | Emotional summary |
| **File Watcher** | | |
| `GET` | `/watch/status` | Watcher status |
| `POST` | `/watch/toggle` | Enable/disable watcher |
| **Settings** | | |
| `GET` | `/settings` | Get current settings |
| `PUT` | `/settings` | Update settings (hot-reload) |
| `POST` | `/settings/reset` | Reset to defaults |

</details>

### CLI

```bash
# Core memory
cerebro remember "..." --type semantic # Store
cerebro recall "query" -n 10 # Search
cerebro recall "query" --agent BOB --thread t1 # Scoped search
cerebro get <id> --agent ALICE # Get by ID (scope-checked)
cerebro update <id> --salience 0.9 --tags py # Update
cerebro delete <id> --force # Delete
cerebro share <id> shared --agent ALICE # Change visibility
cerebro associate <id1> <id2> causal # Link

# Episodes
cerebro episode start --title "Debug"
cerebro episode get <ep_id> --agent ALICE # Scope-checked
cerebro episode list

# Intentions, schemas, procedures
cerebro intention add "TODO: fix auth"
cerebro intention list
cerebro schema list
cerebro procedure add "Step 1: ..." --tags ops

# Watch directories for auto-ingestion
cerebro watch ~/Dropbox/CerebroInbox --tags "auto,production"

# System
cerebro stats
cerebro health
cerebro emotions
cerebro agents list --json
cerebro dream --run                             # Per-agent dream cycles

# Migration
cerebro import neocortex data.json
cerebro import json data.json
cerebro import markdown knowledge.md
```

---

## Agent Framework Integration

CerebroCortex works as a **drop-in memory upgrade** for any AI agent framework that supports MCP. No protocol changes needed — the MCP server speaks standard JSON-RPC 2.0 over stdio.

### Hermes Agent

CerebroCortex is a community memory provider for [Hermes Agent](https://github.com/NousResearch/hermes-agent). Two integration paths:

1. **MCP Server** — Add to `config.yaml`, get 56+ tools instantly (see [Quick Start](#quick-start))
2. **Memory Provider Plugin** — Deeper integration with prefetch, background sync, and session summaries (see [PR #7913](https://github.com/NousResearch/hermes-agent/pull/7913))

See **[HERMES_INTEGRATE.md](HERMES_INTEGRATE.md)** for the complete Hermes integration guide.

### Claude Code / OpenClaw / Other Frameworks

1. Point your framework's MCP adapter at `./cerebro-mcp`
2. All 63 tools are auto-discovered with plain-English descriptions
3. Set `CEREBRO_AGENT_ID` to identify your agent

See **[INTEGRATE.md](INTEGRATE.md)** for detailed setup with Claude Code and OpenClaw:
- Quick start for Claude Code and OpenClaw
- Configuration (agent ID, API keys, runtime settings)
- Tool reference table (most useful tools for agents)
- Multi-agent setup with visibility scoping
- Copy-paste CLAUDE.md snippets for agent system prompts

### What CerebroCortex adds over file-based memory

| Capability | File-based memory | CerebroCortex |
|---|---|---|
| **Decay & promotion** | Everything lives forever | Unused memories fade, active ones get promoted |
| **Semantic search** | Keyword / filename lookup | Meaning-based search with 384-dim embeddings |
| **Associative graph** | Flat files | 9 link types with spreading activation |
| **Multi-agent** | Per-agent directories | Shared/private/thread visibility with scoped recall |
| **Session continuity** | Manual notes | Structured session save/recall with priority |
| **Offline consolidation** | None | Dream Engine extracts patterns, prunes noise, discovers connections |
| **File ingestion** | Built-in | `ingest_file` tool imports .md, .json, .txt, and code files |

### Runtime Settings

Settings can be changed without restarting the server:

```bash
# Environment variable for agent identity
export CEREBRO_AGENT_ID="my-agent"

# Runtime config persisted to data/settings.json
# API keys in data/.env
```

Priority: `config.py` defaults < `data/settings.json` < `data/.env`

---

## Dashboard

A fully redesigned **alchemical-themed web dashboard** with real-time neural graph visualization, drag-drop file ingestion, and complete CRUD for all v0.4+ features.

**Theme:** Dark void palette with gold (#D4AF37) accents, Cinzel display font, JetBrains Mono body, animated golden particle canvas background with proximity connections. Glassmorphism cards with gradient top borders.

**Panels:**
- **☼ Cortex** — Live stats, memory type distribution, layer breakdown
- **⚛ Memory Vault** — Semantic search with bulk select, type badges, tag filters
- **⚡ Ingest** — Drag-and-drop file upload with auto-format detection
- **☉ Graph** — Interactive 3D/2D force-directed neural graph with type-colored nodes
- **☠ The Ash Heap** — Trash can with restore and permanent purge
- **✵ Sigils** — Tag management: rename, merge, delete across all memories
- **♛ Threads** — Conversation thread browser with memory counts
- **☽ Oneiros** — Dream Engine control and last-run reports
- **♨ Vital Signs** — System health diagnostics
- **⚙ Configuration** — Hot-reload settings, LLM config, watcher toggles

**Real-time features:**
- WebSocket event stream for live stats refresh
- Toast notifications for all operations
- Watcher toggle in header with on/off indicator

```bash
./cerebro-api
# Open http://localhost:8767/ui
```

---

## File Watcher

Auto-ingest files dropped into watched directories. Uses `watchdog` (inotify on Linux) with debounced ingestion and fingerprint-based deduplication.

```bash
# CLI
cerebro watch ~/Dropbox/CerebroInbox ~/Downloads --tags "auto-ingested"

# Config via settings.json
{
  "watch": {
    "enabled": true,
    "dirs": ["/home/user/Dropbox/CerebroInbox"],
    "tags": ["auto-ingested"],
    "patterns": ["*.md", "*.txt", "*.pdf", "*.png"]
  }
}
```

**Features:**
- Watches recursively — subdirectories included
- Fingerprint tracking (mtime+size) prevents re-ingestion across restarts
- 1-second debounce — waits for file moves to complete
- Supports all ingestion formats: text, markdown, JSON, PDF, HTML, CSV, images, code
- State persisted in `~/.cerebro-cortex/.cerebro-watch-state.json`
- Toggle on/off from dashboard header or `POST /watch/toggle`
- Auto-starts on API server launch if `WATCH_ENABLED=true`

---

## Migration

Import from other systems or bulk-load knowledge:

### From Neo-Cortex

```bash
cerebro import neocortex /path/to/export.json
```

Maps 11 Neo-Cortex message types → 6 CerebroCortex modalities. Preserves access counts, creates contextual links from `responding_to` chains, and tags imported memories with `neo:{original_type}` for provenance.

### From JSON

```bash
cerebro import json /path/to/data.json
```

Two formats supported:

```jsonc
// Simple format
[
  {"content": "The speed of light is 299,792,458 m/s", "type": "semantic", "tags": ["physics"]}
]

// Full format (with metadata and strength)
[
  {
    "content": "...",
    "metadata": {"memory_type": "semantic", "salience": 0.9},
    "strength": {"stability": 10.0, "difficulty": 3.0}
  }
]
```

### From Images, PDFs, HTML, and CSV

```bash
# Image — auto-captioned and OCR'd, vision embeddings extracted
cerebro ingest image /path/to/photo.jpg --tags ["reference"]

# PDF — text extracted per-page, embedded images linked as attachments
cerebro ingest pdf /path/to/document.pdf --tags ["research"]

# HTML — text + image references ingested
cerebro ingest html /path/to/page.html --tags ["web"]

# CSV — row-per-memory or schema-mode ingestion
cerebro ingest csv /path/to/data.csv --tags ["dataset"]
```

Requires `[vision]` extras: `pip install "cerebro-cortex[vision]"`

### From Markdown

```bash
cerebro import markdown /path/to/knowledge.md
```

Splits by `##` headings (or paragraphs if no headings). Supports YAML frontmatter:

```markdown
---
type: semantic
tags: [python, async]
salience: 0.8
---

## Asyncio Event Loop

The event loop is the core of every asyncio application...
```

### From Text & Code Files

```bash
# Via MCP tool — auto-detects format from extension
ingest_file(file_path="/path/to/notes.txt", tags=["project"])
ingest_file(file_path="/path/to/paper.pdf", tags=["research"])
ingest_file(file_path="/path/to/screenshot.png", tags=["reference"])

# Supports: .txt, .md, .json, .pdf, .html, .csv, .png, .jpg, .jpeg, .webp
#           .py, .js, .ts, .go, .rs, .java, .rb, .sh, and more
```

Splits by blank lines into paragraphs. Paragraphs exceeding ~500 words are further split at sentence boundaries. Duplicates are automatically detected and skipped.

---

## Vision & Cross-Modal Recall

CerebroCortex v0.4.0 introduces **cross-modal memory** — images are first-class citizens alongside text.

### How It Works

1. **Image ingestion**: `ImageAdapter` generates a text caption (via Ollama llava fallback) and optional OCR (pytesseract). The image is also embedded via CLIP (sentence-transformers) into a separate vision vector space.
2. **PDF ingestion**: `PDFAdapter` extracts text per-page and pulls out embedded images, linking them as `part_of` the document.
3. **Cross-modal recall**: When `recall(include_vision=True)` is called, text results from ChromaDB are merged with vision results from `VisionVectorStore`. Existing text memories get a boost if their attached images match; new image-only candidates are inserted with a slightly lower initial score.

```python
# Search text and vision together
results = brain.recall("deployment diagram", include_vision=True, top_k=10)

# Search vision only
results = brain.recall("red server icon", include_vision=True, top_k=5)
```

**Requirements:** Install with `pip install "cerebro-cortex[vision]"` (includes pillow, pytesseract, pymupdf, beautifulsoup4).

---

## Configuration

All tunables live in `src/cerebro/config.py`. Key parameters:

<details>
<summary>ACT-R Parameters</summary>

| Parameter | Default | Description |
|---|---|---|
| `ACTR_DECAY_RATE` | 0.5 | Power-law exponent |
| `ACTR_B_CONSTANT` | 0.0 | Additive constant |
| `ACTR_NOISE` | 0.4 | Retrieval noise |
| `ACTR_RETRIEVAL_THRESHOLD` | 0.0 | Minimum activation |
| `MAX_STORED_TIMESTAMPS` | 50 | Individual timestamp limit |

</details>

<details>
<summary>FSRS Parameters</summary>

| Parameter | Default | Description |
|---|---|---|
| `FSRS_INITIAL_STABILITY` | 1.0 day | Starting stability |
| `FSRS_INITIAL_DIFFICULTY` | 5.0 | Starting difficulty (1-10) |
| `FSRS_MIN_STABILITY` | 0.1 days | Stability floor |
| `FSRS_MAX_STABILITY` | 365 days | Stability ceiling |

</details>

<details>
<summary>Scoring Weights</summary>

| Signal | Weight | Description |
|---|---|---|
| Vector similarity | 35% | Semantic match from ChromaDB |
| ACT-R activation | 30% | Recency + frequency power law |
| FSRS retrievability | 20% | Spaced-repetition curve |
| Salience | 15% | Emotional importance |

</details>

<details>
<summary>Dream Engine</summary>

| Parameter | Default | Description |
|---|---|---|
| `DREAM_MAX_LLM_CALLS` | 20 | Budget per cycle |
| `DREAM_CLUSTER_SIMILARITY_THRESHOLD` | 0.80 | Clustering threshold |
| `DREAM_CLUSTER_MIN_SIZE` | 3 | Minimum cluster |
| `DREAM_PRUNING_MIN_AGE_HOURS` | 48 | Don't prune young memories |
| `DREAM_PRUNING_MAX_SALIENCE` | 0.3 | Only prune low-salience |
| `DREAM_REM_SAMPLE_SIZE` | 20 | Diversity sample |
| `DREAM_REM_PAIR_CHECKS` | 10 | Random pairs to evaluate |

</details>

<details>
<summary>LLM Configuration</summary>

| Parameter | Default | Description |
|---|---|---|
| `LLM_PRIMARY_PROVIDER` | `"openai_compat"` | Primary LLM provider (LM Studio, vLLM, etc.) |
| `LLM_PRIMARY_MODEL` | `"qwen/qwen3-4b-2507"` | Primary model |
| `LLM_FALLBACK_PROVIDER` | `"anthropic"` | Fallback (Claude API) |
| `LLM_FALLBACK_MODEL` | `"claude-sonnet-4-5-20250929"` | Fallback model |
| `LLM_TEMPERATURE` | 0.7 | Generation temperature |
| `LLM_MAX_TOKENS` | 1024 | Max tokens per call |
| `OPENAI_COMPAT_BASE_URL` | `"http://localhost:1234"` | OpenAI-compatible API endpoint |

</details>

---

## Project Structure

```
CerebroCortex/
│── cerebro-mcp                    # MCP server launcher
│── cerebro-api                    # REST API launcher
│── INTEGRATE.md                   # Agent framework integration guide
│── pyproject.toml                 # Package config & dependencies
│
│── src/cerebro/
│   │
│   ─── bootstrap/                    # CCBS cognitive modules
│   │   ─── modules/                  #   12 markdown module sources
│   │
│   ─── types.py                   # Core enums (6 memory types, 9 link types, 4 layers, MediaType)
│   ├── config.py                  # All tuneable parameters
│   ├── settings.py                # Runtime settings (hot-reload from settings.json/.env)
│   ├── cortex.py                  # Main coordinator — the brain itself
│   │
│   ├── models/                    # Pydantic data models
│   │   ├── memory.py              #   MemoryNode, StrengthState, MemoryMetadata, Attachment
│   │   ├── link.py                #   AssociativeLink
│   │   ├── episode.py             #   Episode, EpisodeStep
│   │   ├── agent.py               #   AgentProfile
│   │   └── activation.py          #   ActivationResult, RecallResult
│   │
│   ├── storage/                   # Storage backends
│   │   ├── coordinator.py         #   StorageCoordinator — single write gateway
│   │   ├── graph_store.py         #   SQLite + igraph hybrid
│   │   ├── chroma_store.py        #   ChromaDB vector store
│   │   ├── vision_embeddings.py   #   CLIP vision embeddings + sidecar store
│   │   ├── sqlite_schema.py       #   Schema initialization (v6: FTS5, attachments, versions)
│   │   ├── embeddings.py          #   SBERT / Ollama embedding functions
│   │   └── base.py                #   Abstract VectorStore interface
│   │
│   ├── ingestion/                 # Unified ingestion pipeline
│   │   ├── pipeline.py            #   IngestionPipeline — adapter registry + orchestration
│   │   ├── base.py                #   IngestionAdapter base class
│   │   ├── text_adapter.py        #   Plain text & code files
│   │   ├── markdown_adapter.py    #   Markdown with YAML frontmatter
│   │   ├── json_adapter.py        #   Generic JSON arrays
│   │   ├── image_adapter.py       #   Images: caption + OCR + vision embed
│   │   ├── pdf_adapter.py         #   PDFs: text + image extraction
│   │   ├── html_adapter.py        #   HTML: text + image references
│   │   └── csv_adapter.py         #   CSV: row-per-memory or schema mode
│   │
│   ├── activation/                # Biologically-inspired activation
│   │   ├── strength.py            #   ACT-R base-level + FSRS retrievability
│   │   ├── decay.py               #   Layer decay + promotion logic
│   │   └── spreading.py           #   Collins & Loftus spreading activation
│   │
│   ├── engines/                   # 9 brain-region engines + tag manager
│   │   ├── thalamus.py            #   GatingEngine — sensory filter
│   │   ├── amygdala.py            #   AffectEngine — emotional processing
│   │   ├── temporal.py            #   SemanticEngine — concept extraction
│   │   ├── hippocampus.py         #   EpisodicEngine — episode management
│   │   ├── association.py         #   LinkEngine — associative network
│   │   ├── cerebellum.py          #   ProceduralEngine — skill memory
│   │   ├── prefrontal.py          #   ExecutiveEngine — ranking & promotion
│   │   ├── neocortex.py           #   SchemaEngine — abstraction
│   │   ├── dream.py               #   DreamEngine — offline consolidation
│   │   └── tag_manager.py         #   TagManager — tag CRUD
│   │
│   ├── watch/                     # File watcher for auto-ingestion
│   │   ├── __init__.py            #   FileWatcher export
│   │   └── watcher.py             #   watchdog-based ingestion handler
│   │
│   ├── interfaces/                # External interfaces
│   │   ├── mcp_server.py          #   MCP server (63 tools)
│   │   ├── api_server.py          #   FastAPI REST server (55+ endpoints)
│   │   └── cli.py                 #   Click CLI
│   │
│   ├── migration/                 # Data import tools
│   │   ├── neo_cortex_import.py   #   Neo-Cortex migration
│   │   ├── json_import.py         #   Generic JSON import
│   │   ├── markdown_import.py     #   Markdown knowledge base import
│   │   └── text_import.py         #   Plain text & code file import
│   │
│   └── utils/
│       └── llm.py                 #   LLM client (Anthropic + Ollama fallback)
│
├── web/
│   └── index.html                 #   Dashboard shell (modular JS/CSS)
│
├── tests/                         #   453 tests across 27 test files
│   ├── conftest.py                #   Shared fixtures (incl. multi_agent_cortex)
│   ├── test_models/
│   ├── test_storage/
│   ├── test_activation/
│   ├── test_engines/              #   (11 files — engines + scope enforcement)
│   ├── test_interfaces/           #   MCP, API, CLI
│   ├── test_migration/
│   ├── test_ingestion/            #   Adapter tests (7 adapters)
│   ├── test_vision/               #   Vision embeddings + cross-modal recall
│   ├── test_crud/                 #   Soft-delete, versioning, bulk ops
│   └── test_watch.py              #   File watcher lifecycle + ingestion
│
└── data/                          #   Runtime data (gitignored)
    ├── cerebro.db                 #   SQLite graph database
    ├── chroma/                    #   ChromaDB vector store
    └── exports/                   #   Data exports
```

---

## Testing

467 tests. Run in groups to avoid file-descriptor exhaustion from ChromaDB client cleanup:

```bash
# Core + engines + storage + activation
PYTHONPATH=src pytest tests/test_engines tests/test_models tests/test_activation tests/test_storage -q --tb=short

# Migration
PYTHONPATH=src pytest tests/test_migration -q --tb=short

# Interfaces (MCP, API, CLI)
PYTHONPATH=src pytest tests/test_interfaces -q --tb=short

# New features (ingestion, vision, CRUD, watch)
PYTHONPATH=src pytest tests/test_ingestion tests/test_vision tests/test_crud tests/test_watch -q --tb=short
```

```
tests/test_models/test_memory.py              ✓  (data model validation)
tests/test_storage/test_graph_store.py        ✓  (SQLite + igraph operations)
tests/test_storage/test_embedder_fingerprint.py ✓  (embedder health)
tests/test_activation/test_strength.py        ✓  (ACT-R + FSRS math)
tests/test_activation/test_spreading.py       ✓  (spreading activation)
tests/test_engines/test_thalamus.py           ✓  (gating + dedup)
tests/test_engines/test_amygdala.py           ✓  (emotional analysis)
tests/test_engines/test_temporal.py           ✓  (concept extraction)
tests/test_engines/test_hippocampus.py        ✓  (episode management)
tests/test_engines/test_association.py        ✓  (linking + Hebbian)
tests/test_engines/test_cerebellum.py         ✓  (procedural memory)
tests/test_engines/test_prefrontal.py         ✓  (executive ranking)
tests/test_engines/test_neocortex.py          ✓  (schema formation)
tests/test_engines/test_dream.py              ✓  (dream + per-agent scoping)
tests/test_engines/test_cortex.py             ✓  (integration tests)
tests/test_engines/test_scope_enforcement.py  ✓  (multi-agent scope + link pruning)
tests/test_interfaces/test_mcp_server.py      ✓  (MCP protocol, 63 tools)
tests/test_interfaces/test_api_server.py      ✓  (REST API)
tests/test_interfaces/test_cli.py             ✓  (CLI commands)
tests/test_migration/test_*.py                ✓  (all importers)
tests/test_ingestion/test_adapters.py         ✓  (7 adapters)
tests/test_vision/test_vision_store.py        ✓  (CLIP embeddings)
tests/test_vision/test_cross_modal.py         ✓  (cross-modal recall)
tests/test_crud/test_soft_delete.py           ✓  (soft-delete + versioning)
tests/test_watch.py                           ✓  (file watcher lifecycle + e2e ingestion)
tests/test_bootstrap/test_ccbs.py             ✓  (cognitive bootstrap assembler)
tests/test_utils/test_llm.py                  ✓  (LLM client)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 467 passed ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
```

---

## Requirements

- **Python 3.11+**
- **~200MB disk** for SBERT model (downloaded on first run)
- **Runs on Raspberry Pi 5** (8GB recommended) — no GPU required

### Dependencies

| Package | Purpose | Required |
|---|---|---|
| `chromadb` | Vector similarity search | ✓ |
| `sentence-transformers` | SBERT embeddings (384-dim) + CLIP vision | ✓ |
| `python-igraph` | C-speed graph operations | ✓ |
| `pydantic` | Data validation | ✓ |
| `python-dateutil` | Date parsing | ✓ |
| `watchdog` | File system events (auto-ingestion) | ✓ |
| `click` | CLI framework | ✓ |
| `fastapi` + `uvicorn` | REST API (optional) | ✗ |
| `mcp` | Model Context Protocol (optional) | ✗ |
| `anthropic` | Claude API for Dream Engine (optional) | ✗ |
| `pillow` | Image processing (vision extra) | ✗ |
| `pytesseract` | OCR (vision extra) | ✗ |
| `pymupdf` | PDF text/image extraction (vision extra) | ✗ |
| `beautifulsoup4` | HTML parsing (vision extra) | ✗ |

Install subsets:

```bash
pip install 'cerebro-cortex[mcp]'           # MCP server only
pip install 'cerebro-cortex[mcp,llm]'       # MCP + Dream Engine LLM support
pip install 'cerebro-cortex[api]'           # REST API + dashboard
pip install 'cerebro-cortex[vision]'        # Image/PDF/HTML/CSV ingestion + cross-modal recall
pip install 'cerebro-cortex[all]'           # Everything
```

---

## Roadmap

- [x] Wire ChromaDB vector search into recall pipeline
- [x] Multi-agent memory scoping (SHARED/PRIVATE/THREAD visibility)
- [x] Per-agent Dream Engine consolidation cycles
- [x] Cross-agent link pruning on visibility changes
- [x] Episode query scope enforcement
- [x] OpenAI-compatible LLM provider (LM Studio support)
- [x] WebSocket real-time updates for dashboard
- [x] Runtime settings management (hot-reload from settings.json/.env)
- [x] Agent framework integration (Hermes Agent, OpenClaw, Claude Code — plain-English tool descriptions)
- [x] Hermes Agent memory provider plugin ([PR #7913](https://github.com/NousResearch/hermes-agent/pull/7913))
- [x] Configurable agent ID via environment variable
- [x] File ingestion MCP tool (.md, .json, .txt, code files)
- [x] **v0.4.0 — Unified StorageCoordinator (SQLite as source of truth)**
- [x] **v0.4.0 — Unified ingestion pipeline with pluggable adapters**
- [x] **v0.4.0 — Vision & cross-modal memory (CLIP + image/PDF/HTML/CSV ingestion)**
- [x] **v0.4.0 — SQLite FTS5 keyword fallback**
- [x] **v0.4.0 — Soft delete, trash can, memory versioning**
- [x] **v0.4.0 — Tag management, bulk operations, thread management**
- [x] **v0.4.0 — Pydantic v2 compliance (@field_serializer)**
- [x] **v0.4.1 — File watcher bridge for auto-ingesting directories**
- [x] **v0.4.1 — Alchemical dashboard overhaul (modular, 11 panels, drag-drop ingest)**
- [x] **v0.4.1 — 20 new REST endpoints (trash, versions, tags, threads, bulk, export, watch)**
- [x] **v0.4.1 — Watcher toggle in dashboard + API background thread auto-start**
- [x] **v0.4.2 — Semantic chunking for long documents**
- [x] **v0.4.2 — Near-duplicate detection at ingestion time**
- [x] **v0.4.2 — Temporal decay visualization dashboard panel**
- [x] **v0.4.2 — Audit logging for access denials and visibility changes**
- [x] **v0.5.0 — Cerebro Cognitive Bootstrap System (CCBS) — dynamic cognitive priming**
- [x] **v0.5.0 — 12 modular cognitive modules with query-aware loading**
- [x] **v0.5.0 — POST /bootstrap endpoint with token budget enforcement**
- [x] **v0.5.0 — Harness memory migration to Cerebro (90/10 rule)**
- [ ] pgvector backend option for multi-node deployments
- [ ] Semantic similarity-based query analysis (upgrade from keyword matching)
- [ ] Dashboard panel for CCBS module management
- [ ] Auto-remember integration into Hermes session start loop

---

## License

[MIT](LICENSE) — Use it, fork it, give your AI a brain.

---

Built with 🧠 by humans and Claude &bull; Integrated with <a href="https://github.com/NousResearch/hermes-agent">Hermes Agent</a> &bull; Powered by cognitive science, not just cosine similarity

CerebroCortex

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README