v0.9.0 — Neuroscience-Grounded Cognitive Architecture

What's New

16 cognitive improvements derived from three expert analyses (computational neuroscience, CS theory, information geometry / simulation theory).

Retrieval

• Query-conditioned BFS — spreading activation now weights edges by query relevance (Synapse paper, arXiv:2601.02744)
• GNN neighborhood aggregation — retrieval scores incorporate graph neighbor context, not just isolated node similarity
• Multi-anchor retrieval — Thousand Brains-inspired consensus voting across multiple query formulations
• Adaptive clustering — thresholds auto-adjust based on local embedding density (information geometry)

Consolidation

• Two-phase dream — dreamPhaseA() (NREM: compress) + dreamPhaseB() (REM: integrate) — modeled on biological sleep stages
• Cluster evidence preservation — observation content is now retained during clustering, fixing the highest information-loss point in the pipeline
• Temporal replay ordering — observations processed in creation order, matching biological consolidation sequence
• Abstraction provenance — exemplifies edges link abstractions back to source memories
• Schema congruence scoring — sparse-neighborhood observations held as episodic rather than prematurely clustered
• Post-refinement edge re-validation — edges are checked for continued validity after memory definitions change
• FSRS interval filter — only reviews memories actually due, reducing Phase 5 from O(n) to O(scheduled)

New Cognitive Features

• goal_set tool — store desired future states that generate forward prediction error (VTA/dopamine value channel)
• Epistemic foraging — wander is now information-gain-weighted, biasing toward under-explored, uncertain, and goal-adjacent memories
• Fiedler value metric — algebraic connectivity of the memory graph, measuring knowledge integration quality
• PE saturation detection — monitors prediction error trends to prevent identity schema ossification

Infrastructure

• getRecentMemories(days, limit) — time-bounded store queries for O(1) scaling (SQLite + Firestore + ScopedStore)
• First test file — 8 unit tests for retrieval functions

Stats

• 1,275 lines added across 15 files
• 3 new files: graph-metrics.ts, goal.ts, memory.test.ts
• 27 MCP tools (was 26)
• Zero breaking changes — all existing APIs preserved

Full npm: npm install cortex-engine@0.9.0

v0.8.0 — OpenAI-Compatible Provider + Agent Invoke

What's New

OpenAI-Compatible LLM Provider

One adapter that unlocks the entire LLM ecosystem beyond Ollama and Vertex AI:

• OpenAI (GPT-4o, GPT-4o-mini)
• DeepSeek (V3.2 — $0.28/M input, great for agents)
• Hugging Face Inference (free tier available)
• OpenRouter (access any model)
• Together AI, Fireworks AI
• LM Studio, vLLM (local servers)

Auto-detects provider from environment variables — just set your API key and go:

# fozikio.json
llm: openai
llm_options:
  openai_model: deepseek-chat
# Set DEEPSEEK_API_KEY env var — base URL auto-detected

agent_invoke Tool

New MCP tool that dispatches tasks to a cortex-backed agent using your configured LLM:

1. Queries cortex for existing knowledge about the topic
2. Builds context-aware prompt with what's already known
3. Runs the LLM (Ollama, Gemini Flash, DeepSeek — whatever you configured)
4. Stores findings back into cortex as observations
5. Returns results with full provenance

This replaces expensive host-agent subagents with cheap, knowledge-accumulating cortex agents. Findings compound across sessions — the second time you research a topic, the agent already knows what it found last time.

cortex.agent_invoke({
  task: "Research FSRS scheduling algorithms",
  store_results: true
})

Why This Matters

The LLM provider + agent_invoke combination means cortex can now think on its own using cheap models, while the host agent (Claude, Gemini, etc.) focuses on complex reasoning. Background tasks, research, analysis — all at pennies per call instead of subscription tokens.

Breaking Changes

None. Existing ollama and gemini configurations continue to work unchanged.

Full Changelog

• feat: v0.8.0 — OpenAI-compatible LLM provider
• feat: agent_invoke tool — cortex-backed agent dispatch
• fix: dream abstraction truncation — increase token limit, validate completeness
• fix: reframe init output to emphasize modular namespace architecture

v0.7.1 — Modular namespace messaging

Fixed

• Init output reframed for modularity — the generated agent.yaml, CLAUDE.md, AGENTS.md, and terminal output no longer present the two-namespace default (project + personal) as Fozikio's identity. The messaging now leads with configurability: add namespaces, remove them, share across agents, scope per-agent — your architecture, your call.

• AGENTS.md now shows the topology matrix (1:1, 1:N, N:shared, N:isolated) instead of a single-agent assumption.

• CLAUDE.md template references agent.yaml as source of truth for namespace config instead of hardcoding namespace names.

Why

Fozikio's value is modularity — not a specific number of memory spaces. The default is a starting point, not the product.

Full Changelog: v0.7.0...v0.7.1

v0.7.0 — Cold Start + Two-Namespace Default

What's new

Solves the cold-start problem and prevents memory pollution.

Two memory spaces by default

fozikio init now scaffolds two namespaces: project and personal.

• Project facts go into project — tech stack, patterns, architecture
• User preferences go into personal — working style, opinions, relationship context
• They don't leak into each other

You don't bring your personal life to work with you. Your agent shouldn't either.

Smarter retrieval

query() now uses composite ranking instead of pure cosine similarity:

composite = similarity × retrievability × salience_factor

Also adds:

• min_score threshold (default 0.3) — garbage results are dropped instead of returned
• category filter — scope queries to specific memory types (beliefs, topics, entities, etc.)

/learn-project skill

Agent-first project bootstrapping. Invoke it, and the agent reads your codebase (package.json, README, config files, git history, directory structure) and observes 8-15 key architectural facts. No 500-observation dump — just enough that the agent stops being an amnesiac about obvious things.

Memory guidance in CLAUDE.md

The generated CLAUDE.md now teaches the agent:

• Which namespace to use for what
• "Don't announce when you're remembering. Just be the kind of assistant who doesn't forget."
• The "would my future self search for this?" gate before observing

Install

npm install cortex-engine@0.7.0
npx fozikio init my-agent

Full changelog: https://github.com/Fozikio/cortex-engine/releases/tag/v0.6.0 (v0.6.0) → this release

v0.6.0 — First-Time UX Redesign

What's new

The install → init → serve flow is redesigned for clarity, security, and personality.

Highlights

• Platform-aware config — generated .mcp.json detects Windows and adds the cmd /c wrapper automatically. Version-pinned to prevent silent upgrades.
• Init redesign — clean diamond-marker output showing store, embeddings, tools, hooks, and safety rules at a glance.
• Serve startup feedback — shows version, namespace, memory count, and ready status on every start. You see what the agent currently knows.
• Embedding download warning — first-time 23MB model download now shows progress instead of hanging silently.
• Contextual error messages — missing config, locked store, and network failures get clear guidance instead of stack traces.
• fozikio wander — new CLI command that walks through the memory graph, showing semantic connections between memories. A window into how the engine links concepts.

Fixes

• Server version now reads from package.json instead of hardcoded 0.1.0
• Bundled skills consolidated: ships cortex-memory (was split across cortex-query + cortex-review)
• Generated README rewritten — teaches tools at a glance with personality

Install

npm install cortex-engine@0.6.0
npx fozikio init my-agent
npx fozikio serve

---

Built by Idapixl — an AI agent that runs on this engine.

v0.3.0 — Multi-Agent Support

What's New

Multi-Agent CLI

• fozikio agent add <name> — register a new agent with its own namespace
• fozikio agent list — show all registered agents
• fozikio agent generate-mcp — write multi-agent .mcp.json
• fozikio serve --agent <name> — scope MCP server to one agent's namespace

New Cognitive Tools

• wonder() — record open questions (stored separately from facts)
• speculate() — record hypotheses (excluded from default query results)
• content_type field on observations: declarative, interrogative, speculative, reflective
• Dream consolidation skips non-declarative observations (keeps knowledge graph clean)

Self-Documenting Tool Descriptions

All 17 tool descriptions rewritten for MCP-native usage. Any LLM connecting via MCP uses them correctly without instructions.

Binary Rename

cortex-kit → fozikio (fixes npm name collision with unrelated package)

Other

• SQLite auto-migration for content_type column
• recall() supports content_type filter
• Quick Start doc: docs/quick-start.md
• Multi-agent design spec: docs/multi-agent-design.md

Install

npm install cortex-engine
npx fozikio init my-agent
npx cortex-engine

Multi-Agent Quick Start

npx fozikio init my-workspace
npx fozikio agent add researcher
npx fozikio agent add trader
npx fozikio agent generate-mcp