Fozikio

Ecosystem

Everything Fozikio provides

A cognitive engine, nine plugin packages, a CLI, and a safety layer. Use what you need. Skip what you don't.

cortex-engine

MIT

The core. 27 cognitive tools over MCP. Persistent semantic memory modeled on how biological brains actually store, consolidate, and retrieve.

$npm install @fozikio/cortex-engine
Persistent semantic memory
Dream consolidation (NREM + REM)
Belief tracking with prediction error
Spreading activation retrieval
Thousand Brains voting
Goal-directed cognition
FSRS spaced repetition
Pluggable embeddings (built-in, Ollama, OpenAI, Vertex)
Pluggable LLMs, multi-provider
Read the docsGitHubnpm

Plugin Ecosystem

9 packages

Each plugin extends cortex-engine with a focused domain. Install what your agent actually needs — cortex-engine auto-discovers installed plugins.

@fozikio/tools-threads

Thought threads: create, update, resolve ongoing lines of thinking

$npm install @fozikio/tools-threads

@fozikio/tools-journal

Session journaling: structured reflections that persist

$npm install @fozikio/tools-journal

@fozikio/tools-content

Content pipeline: draft, review, publish workflow

$npm install @fozikio/tools-content

@fozikio/tools-evolution

Identity evolution: track personality changes over time

$npm install @fozikio/tools-evolution

@fozikio/tools-social

Social cognition: interaction patterns, engagement tracking

$npm install @fozikio/tools-social

@fozikio/tools-graph

Graph analysis: memory connections, clustering, visualization

$npm install @fozikio/tools-graph

@fozikio/tools-maintenance

Memory maintenance: cleanup, deduplication, health checks

$npm install @fozikio/tools-maintenance

@fozikio/tools-vitals

Vitals tracking: agent health metrics and operational signals

$npm install @fozikio/tools-vitals

@fozikio/tools-reasoning

Cognitive reasoning: abstraction, contradiction detection

$npm install @fozikio/tools-reasoning

Install any plugin: npm install @fozikio/tools-threads — cortex-engine auto-discovers installed plugins.

fozikio CLI

MIT

Multi-agent workspace management. Initialize workspaces, start MCP servers, add isolated agents, inspect memory health.

terminal
$npx fozikio init my-agent
$npx fozikio serve
$npx fozikio agent add researcher
$npx fozikio health
$npx fozikio wander
$npx fozikio maintain fix

Reflex

MIT

Portable agent safety rules. YAML-based guardrails that work across any agent runtime.

cognitive-grounding

Forces context loading before action — prevents agents from acting on stale or missing state.

observe-first

Requires reading before writing — prevents overwriting information the agent hasn't seen.

note-about-doing

Logs intent before execution — creates an audit trail of what the agent planned to do.

Rules are portable — use them with Claude Code, Cursor, Codex, or any runtime with a Reflex adapter.

View on GitHub