OWL Memory MCP
gauravv-g/owl-memory-mcp
Memória local que pensa, com 169+ ferramentas e zero dependência de cloud
Ver no GitHubPlataforma MCP local que fornece memória persistente com detecção de alucinações, análise de segurança, orquestração multi-agente e 169+ ferramentas especializadas. Funciona 100% offline sem dependência de cloud, economizando 95% de tokens e oferecendo inteligência em código, Git, CI/CD, pesquisa e automação.
Instalação
npx convez-ai gauravv-g/owl-memory-mcp1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Construir agentes de IA com memória persistente e hallucination firewall
- Automatizar auditorias de segurança, análise de código e geração de documentação
- Orquestrar múltiplos agentes para executar pipelines CI/CD, pesquisa profunda e testes de qualidade
- Executar web scraping stealth, monitoramento de mudanças e research multi-query offline
Veredito

OWL é um sistema robusto e ambicioso que resolve problemas reais de alucinação, token waste e falta de memória em agentes. Sua cobertura de 21 servidores especializados é impressionante, mas a curva de aprendizado é alta; ideal para times de engenharia que precisam de controle total sobre agentes locais sem cloud.
README
A tradução para o português deste README ainda está na fila.
OWL MCP
Your AI forgets. And sometimes it lies. OWL fixes both.
Tools: ~169 | Servers: 21 | Cloud required: Zero
What It Does
OWL is a local MCP server platform that gives your AI agent:
- Memory that persists across sessions (not a filing cabinet — a brain with hallucination detection)
- Code intelligence — analyze, build, test, lint, review, refactor
- Security auditing — secret scanning, OWASP Top 10, CVE lookup, auth audit
- Git intelligence — smart commits, conflict prediction, PR generation
- Auto-documentation — README, API docs, architecture diagrams from code
- CI/CD pipelines — generate GitHub Actions, GitLab CI, Jenkins, Azure DevOps
- Multi-agent orchestration — spawn workers, merge results, execute plans
- Web intelligence — scraping, stealth fetch, dynamic JS, change monitoring
- Research — multi-query deep research with article extraction
- Creative writing — story bible, continuity, prose scoring, grammar check
Servers
owl-nexus → Meta-orchestration (11 tools)
owl-security → Security audit (10 tools)
owl-research → Deep research (10 tools)
owl-workflow → CI/CD pipelines (10 tools)
owl-git → Git intelligence (10 tools)
creative-studio → Creative writing (10 tools)
owl-deploy → Infrastructure (9 tools)
owl-docs → Auto-documentation (9 tools)
owl-code → Code intelligence (8 tools)
owl-agent → Multi-agent orchestration (8 tools)
owl-data → Data operations (8 tools)
owl-qa → Quality assurance (24 tools)
owl-web → Web intelligence (11 tools)
owl-qa-android → Android automation (12 tools)
owl-qa-visual → Visual QA (9 tools)
owl-unified → Unified memory+web+research (8 tools)
owl-qa-economics → Bug economics (3 tools)
owl-sentinel → Monitoring (4 tools)
Architecture
hermes-custom-mcps/
├── owl_shared_intelligence.py # Shared utilities (imported by all servers)
├── owl_nexus_mcp.py # Task graph orchestration
├── owl_security_mcp.py # Security audit engine
├── owl_research_mcp.py # Deep research engine
├── owl_workflow_mcp.py # CI/CD pipeline generator
├── owl_git_mcp.py # Git intelligence
├── creative_studio_mcp.py # Creative writing
├── owl_deploy_mcp.py # Infrastructure
├── owl_docs_mcp.py # Auto-documentation
├── owl_code_mcp.py # Code intelligence
├── owl_agent_mcp.py # Multi-agent orchestration
├── owl_data_mcp.py # Data operations
├── owl_qa_mcp.py # QA test framework
├── owl_web_mcp.py # Web scraping
├── owl_qa_visual.py # Visual QA
├── owl_qa_android.py # Android automation
├── owl_qa_economics.py # Bug economics
├── owl_unified_server.py # Unified memory+web+research
├── owl_sentinel.py # Monitoring daemon
├── owl_unified_daemon.py # File-watcher daemon
├── .venv/ # Python 3.11.15 virtual environment
└── PHASE9.md # Phase 9 documentation
Setup
git clone https://github.com/gauravv-g/owl-memory-mcp.git
cd owl-memory-mcp
uv venv .venv --python 3.11.15
.venv/Scripts/activate
pip install mcp httpx duckduckgo-search newspaper3k lxml_html_clean
Config
All servers are registered in config.yaml:
hermes mcp list # List all registered servers
hermes mcp test <name> # Test a server connection
hermes gateway restart # Pick up code changes
Shared Utilities
Every server imports from owl_shared_intelligence:
import owl_shared_intelligence as shared
shared.now() # UTC timestamp
shared.walk_code(path) # Walk codebase (skips node_modules, .git, etc.)
shared.categorize_files(files) # Group files by type
shared.detect_project(path) # Detect language, framework, package manager
shared.SKIP_DIRS # Directories to skip
shared.SKIP_EXTENSIONS # File extensions to skip
shared.OWL_DB_PATH # SQLite database path
License
MIT