Your AI agent squad for Claude Code.
17 specialized agents. Persistent memory. Desktop automation. Common sense engine.

Quick Demo

Full video demo coming soon — see demo/ for recording instructions.

Three scenarios that show what Cadre does in practice:

────────────────────────────────────────────────────────────────────────────
  SCENARIO 1 — Memory recall across sessions
────────────────────────────────────────────────────────────────────────────

── Session A: prime a new project ──────────────────────────────────────────

$ claude
  System Aware — Memory: 18 stored facts | Last task: [none for this project]

> /prime

  Priming project: my-fastapi-app
  Reading README.md, pyproject.toml, scanning src/ (47 files)...
  ────────────────────────────────────────────────────────
  Stack        FastAPI 0.111 · Python 3.11 · SQLAlchemy 2
  Entry point  src/main.py  (uvicorn, port 8000)
  Auth         JWT / OAuth2 password flow
  Tests        pytest · 82% coverage · make test
  ────────────────────────────────────────────────────────
  ✓  Project context stored to memory.

── Session B: brand-new terminal, zero context loaded ───────────────────────

$ claude
  System Aware — Memory: 19 stored facts | Last task: primed my-fastapi-app

> What do we know about my-fastapi-app?

  Stack        FastAPI 0.111 / Python 3.11 / SQLAlchemy 2
  Entry point  src/main.py (uvicorn, port 8000)
  Auth         JWT / OAuth2  |  Tests: make test (82% coverage)
  Last primed  2 hours ago
  Ready. What would you like to do?

────────────────────────────────────────────────────────────────────────────
  SCENARIO 2 — Common sense engine blocks a dangerous action
────────────────────────────────────────────────────────────────────────────

> Delete all .log files in /var/log recursively

  [common sense engine] BLOCKED
  ────────────────────────────────────────────────────────
  Rule:    Never delete files in system directories without explicit
           confirmation and a stated recovery plan.
  Risk:    HIGH — system log loss breaks audit trails.
  ────────────────────────────────────────────────────────
  I will not run: rm -rf /var/log/**/*.log

  Did you mean logs inside a project directory?

────────────────────────────────────────────────────────────────────────────
  SCENARIO 3 — Automatic correction capture
────────────────────────────────────────────────────────────────────────────

> Don't delete system logs — only .log files inside ~/projects/

  Executing: find ~/projects/ -name "*.log" -type f -delete
  ✓  Deleted 14 log files across 3 project directories.

  [memory] Correction captured:
  "When asked to delete logs, scope to ~/projects/ not system directories."
  This preference will apply in all future sessions.

Quick Start

git clone https://github.com/WeberG619/cadre-ai.git
cd cadre-ai
./install.sh

Restart Claude Code. That's it. You now have memory, 17 sub-agents, 22 slash commands, safety hooks, and a common sense engine.

Full Getting Started Guide — 5-minute walkthrough with examples for each feature.

What It Does

You give Claude Code a complex task. Cadre breaks it down, dispatches specialized agents, remembers context from past sessions, checks its own work against a common sense engine, and reports back.

you: Open the project spreadsheet, pull the cost data, and build a summary deck

Claude Code ─── delegating to orchestrator agent
  ├─ memory_smart_recall("project spreadsheet, cost data")
  │   └─ Found: project_costs.xlsx on D:\Projects
  │
  ├─ launching sub-agent: excel-automation
  │   ├─ open_workbook("project_costs.xlsx")
  │   ├─ read_range("Sheet1", "A1:F48")
  │   └─ ✓ Extracted 48 line items, $2.4M total
  │
  ├─ launching sub-agent: powerpoint-builder
  │   ├─ create_presentation()
  │   ├─ add_slide("Cost Summary", table_data)
  │   ├─ add_slide("Budget vs Actual", chart_data)
  │   └─ save_as("Cost_Summary.pptx")
  │
  ├─ memory_store("cost summary created")
  └─ voice: "Done. 3-slide deck saved."

✓ Task complete in 34 seconds

Features

Core Intelligence

Strong Agent Framework — 5-phase execution: Orient, Investigate, Execute, Verify, Report
Common Sense Engine — pre-action safety checks against past mistakes, blocks destructive operations
Persistent Memory — corrections, decisions, and preferences survive across sessions
17 Sub-Agents — code analysis, architecture, ML, DevOps, full-stack, C#, Python, and more

Desktop Automation

Excel — read, write, charts, formulas, pivot tables via COM automation
Word / PowerPoint — document and presentation generation
Browser — Edge CDP control: navigate, screenshot, type, scroll, click
System Bridge — real-time awareness of open apps, monitors, clipboard, recent files

Developer Workflow

22 Slash Commands — /commit, /delegate, /review-and-fix, /prime, /fix-and-commit, and more
Safety Hooks — pre-commit guards, MCP seatbelts, secret detection
8 Claude.ai Skills — idea validation, product design, marketing, competitive analysis

Integrations

Voice/TTS — Claude speaks summaries and announcements via Edge TTS
SQLite — structured data storage via MCP
Financial — stock analysis, portfolio tracking, market data
AI Render — Flux Pro photorealistic rendering from text prompts

Architecture

 You
  │
  ▼
┌──────────────────────────────────────────────────────┐
│  Claude Code                                         │
│                                                      │
│  ┌──────────────────────────────────────────┐        │
│  │  Cadre                                    │       │
│  │                                           │       │
│  │  ┌─────────────┐  ┌──────────────────┐   │       │
│  │  │ Strong Agent │  │  Common Sense    │   │       │
│  │  │  Framework   │  │    Engine        │   │       │
│  │  └──────┬──────┘  └────────┬─────────┘   │       │
│  │         │                  │              │       │
│  │  ┌──────┴──────────────────┴─────────┐   │       │
│  │  │        Memory System              │   │       │
│  │  │  corrections · facts · decisions  │   │       │
│  │  └──────────────────────────────────-┘   │       │
│  └──────────────┬────────────────────────────┘       │
│                 │                                     │
│    ┌────────────┼────────────┐                       │
│    ▼            ▼            ▼                        │
│ 17 Agents   22 Commands   Hooks                      │
└────┬────────────┬────────────┬───────────────────────┘
     │            │            │
     ▼            ▼            ▼
  MCP Servers  System Bridge  Voice/TTS
  (Excel, Word, Browser,      (Edge TTS)
   SQLite, Financial,
   AI Render)

Configuration Tiers

Tier	What's Included
Minimal	Framework + commands + agent definitions. No MCP servers.
Developer	Framework + memory + voice + git hooks. Good for most developers.
Power User	Everything. Desktop automation, system bridge, all MCP servers.

See examples/ for ready-to-use configurations for each tier.

Requirements

Claude Code (CLI or VS Code extension)
Claude Pro or Max subscription
Python 3.8+
Windows 10/11 (for desktop automation features) or macOS/Linux (core features)

Documentation

Uninstall

./uninstall.sh

Contributing

Contributions welcome. See CONTRIBUTING.md for guidelines.

License

MIT — see LICENSE for details.

Credits

Built by Weber Gouin at BIM Ops Studio.

If this helped you, star the repo — it helps others find it.

Cadre is an independent community project. Not affiliated with or endorsed by Anthropic, PBC. "Claude" and "Claude Code" are trademarks of Anthropic.

cadre-ai