convez.ai
MCP ServerDevOps & InfraestruturaAvançado
initializ

Forge — Runtime de Agentes IA Portátil e Seguro

initializ/forge

Runtime seguro e portável para agentes IA em produção

Ver no GitHub
BomPluginScore · de 0 a 100

Forge é o runtime open-source para o padrão Agent Skills do Anthropic, construído para agentes que rodam em seu próprio ambiente e infraestrutura. Escreva um SKILL.md, compile para um agente hardened e portável, e deploy em qualquer lugar: Kubernetes, on-prem, air-gapped ou CI embarcado.

Instalação

$npx convez-ai initializ/forge

1 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

Estrelas
226
Forks
17
Acompanhando
1
Saúde
87%
Último commit
há 1 hora
Versão
v0.20.1
Issues abertas
132
Contribuidores
4

Casos de uso

  • Deployar agentes IA em ambientes corporativos com controle total
  • Automatizar fluxos de trabalho com integração a Slack, Telegram e MCP servers
  • Executar agentes localmente ou em Kubernetes com as mesmas definições
  • Implementar segurança por padrão com allowlists de egresso e logs auditados

Veredito

Forge é uma solução profissional e bem arquitetada para orquestração de agentes IA em ambientes corporativos, com foco em segurança, portabilidade e observabilidade. Ideal para times que precisam de controle total sobre infraestrutura e auditoria de ações de IA em produção.

README

A tradução para o português deste README ainda está na fila.

Forge — An Open, Secure, Portable AI Agent Runtime for the Enterprise

Forge is the open-source runtime for Anthropic's Agent Skills standard — built for the agent that runs next to a service, in your environment, on infrastructure you already operate. Write a SKILL.md. Compile to a portable, hardened agent. Deploy it anywhere containers run: Kubernetes, on-prem, air-gapped, embedded in CI, or as an A2A endpoint.

Why Forge?

  • 60-second setup — forge init wizard configures provider, keys, channels, and skills
  • Secure by default — outbound-only connections, egress allowlists, encrypted secrets, no public listeners
  • Portable — same agent runs locally, in Docker, Kubernetes, or inside Initializ Command
  • Observable — structured NDJSON audit logs with correlation IDs for every action
  • Extensible — add skills, tools, channels, and LLM providers without changing core code

Quick Start

# Install (pick one)
brew install initializ/tap/forge
curl -sSL https://raw.githubusercontent.com/initializ/forge/main/install.sh | bash

# Create and run an agent
forge init my-agent && cd my-agent && forge run

# Connect to Slack
forge run --with slack

See Quick Start for the full walkthrough, or Installation for all methods.

How It Works

SKILL.md --> Parse --> Discover tools/requirements --> Compile AgentSpec
                                                            |
                                                            v
                                                    Apply security policy
                                                            |
                                                            v
                                                    Run LLM agent loop
                                               (tool calling + memory + cron)

You write a SKILL.md. Forge compiles it into a secure, runnable agent with egress controls, encrypted secrets, and audit logging.

Key Features

FeatureDescription
Atomic SkillsSKILL.md-based agent definitions with YAML frontmatter
Egress SecurityRuntime + build-time domain allowlists with subprocess proxy
Channel ConnectorsSlack (Socket Mode), Telegram (polling) — outbound-only
Cron SchedulingRecurring tasks with channel delivery
MemorySession persistence + long-term vector search
LLM FallbacksMulti-provider with automatic failover
MCP ClientConnect to any HTTP MCP server (Linear, Notion, Atlassian, ...) — tools surface as <server>__<tool> with namespaced audit
Web Dashboardforge ui for browser-based agent management
Build SigningEd25519 artifact signing & verification
Air-Gap ReadyRuns with local models, no cloud required

Documentation

Getting Started

DocumentDescription
Quick StartGet an agent running in 60 seconds
InstallationHomebrew, binary, and Windows install
ArchitectureSystem design, module layout, and data flows

Core Concepts

DocumentDescription
SkillsSkill definitions, registry, and compilation
ToolsBuilt-in tools, adapters, and custom tools
RuntimeLLM providers, fallback chains, running modes
MemorySession persistence and long-term memory
ChannelsSlack and Telegram adapter setup
SchedulingCron configuration and schedule tools
TracingOpenTelemetry distributed tracing — spans, propagation, audit cross-link

Security

DocumentDescription
Security OverviewComplete security architecture
AuthenticationPluggable auth providers — OIDC, AWS Sigv4, GCP IAP, Azure AD
Egress SecurityEgress enforcement deep dive
SecretsEncrypted secret management
Build SigningEd25519 signing and verification
GuardrailsContent filtering and PII detection
Workflow CorrelationOrchestrator correlation IDs threaded through audit events
Platform PolicyWorkspace-level deploy-time bounds on egress / tools / models

Operations

DocumentDescription
A2A Agent CardSpec-conformant Agent Card at /.well-known/agent-card.json (A2A 0.3.0)
CommandsFull CLI reference
Configurationforge.yaml schema and environment variables
DashboardWeb UI features and architecture
DeploymentContainer packaging, Kubernetes, air-gap
HooksAgent loop hook system
PluginsFramework plugin system
Command IntegrationInitializ Command platform guide

For Claude / LLM-assisted contributors

FileDescription
Forge knowledge skillSingle comprehensive markdown — drop it into a Claude conversation for full Forge context (architecture, security, APIs, CLI, audit pipeline, how to build agents + skills).
Forge skill-builder skillSame prompt the forge ui Skill Builder uses. Author a valid SKILL.md (frontmatter + body + optional scripts) in any Claude chat — no forge ui needed.

/sync-docs keeps both skills current — see .claude/commands/sync-docs.md.

Philosophy

Running agents that do real work requires atomicity (explicit skills, defined tools, declared dependencies), security (restricted egress, encrypted secrets, audit trails), and portability (runs locally, in containers, in Kubernetes, in cloud — same agent, anywhere).

Real agent systems require atomicity, security, and portability. Forge provides those building blocks.

Contributing

We welcome contributions! See CONTRIBUTING.md for development setup, how to add skills/tools/channels, and the PR process.

Please read our Code of Conduct before participating.

License

See LICENSE for details.

Ver o original no GitHub

Bom para

Relacionados