v2.0 · Go 1.23+ · 19 LLM Providers · MIT License

GAIA

Go Autonomous Intelligence Agent — A programming-first autonomous AI agent in Go. Single binary, zero external dependencies, Cyberpunk TUI, and real-time Web Dashboard.

Spec-Driven Development (SDD) · 19 LLM Providers with Credential Pool · Web Dashboard (`templ-islands`) · 12+ Specialized Subagents · PolicyGuard Security · Knowledge Graph Memory

🚀 Quick Start View on GitHub
GAIA — Go Autonomous Intelligence Agent
terminal
# Quick install (Linux & macOS)
$ curl -fsSL https://raw.githubusercontent.com/SalvucciFacundo/gaia/main/install.sh | bash
# Launch Cyberpunk TUI or Web Dashboard
$ gaia # Interactive Terminal UI
$ gaia serve 8080 # Real-time Web UI Dashboard
scroll

Everything You Need, Nothing You Don't

GAIA combines the best of Gentle-AI, Hermes Agent, ogcode, and pi-go into a single high-performance Go binary. No Python, no Node.js, zero heavy runtime dependencies.

12+ Specialized Subagents

Each subagent learns independently in its domain. Supports static specialized roles and dynamic runtime subagent generation.

@explorer · @implementer · @verifier

19 Providers & Credential Pool

Anthropic, OpenAI, DeepSeek, Ollama, Copilot, Groq, OpenRouter, Qwen, Kimi, and more with smart auto-failover and cooldown.

19 LLMs · Multi-Key Failover

Web UI Dashboard

Run gaia serve to open a real-time dashboard powered by templ-islands, SSE token streaming, and workspaces.

templ · htmx · SSE Live Stream

Spec-Driven Development (SDD)

Native 8-phase planning pipeline: explore → propose → spec → design → tasks → apply → verify → archive.

8-Stage Structured Pipeline

PolicyGuard & AST Audit

Tiered execution permissions (read, sandbox, full) and AST scanning to prevent destructive commands and secret leaks.

Tier Security · AST Scanner

BR Code Review

4 bounded lenses (Risk, Resilience, Readability, Reliability) + SHA256 receipts. Immutable pre-commit and pre-push validation.

4 Lenses · SHA256 Receipts

Knowledge Graph (3 Scopes)

Three-level memory recall: User scope (habits), Language scope (Go, Rust, TS patterns), and Project scope (local context).

User · Language · Project

Multi-Platform Gateway

Connect GAIA to Telegram, Discord, Slack, WhatsApp MCP, and Signal MCP from a single binary with zero external dependencies.

Telegram · Discord · Slack · WhatsApp

12+ Autonomous Subagents

Each subagent operates with isolated context, dedicated SQLite memory namespace, independent learning loop, and configurable LLM tier.

🔍 Explorer
SDD

Investigates codebase, architecture, and design patterns

💡 Proposer
SDD

Creates structured change proposals with scope and risk analysis

📋 Specifier
SDD

Authors comprehensive specifications and acceptance criteria

🏗️ Designer
SDD

Designs technical architecture and component interfaces

📐 Planner
SDD

Breaks down specifications into granular executable tasks

⚡ Implementer
SDD

Writes production code adhering to specs and task boundaries

✅ Verifier
SDD

Executes test suites and validates implementation against specs

👁️ Reviewer
BR

Bounded code review across Risk, Resilience, and Readability

🧠 Learner
BG

Analyzes usage patterns, synthesizes skills, and refines memory

📚 Researcher
OD

Performs web research, documentation discovery, and API exploration

📦 Archiver
SDD

Closes completed change sets and commits terminal records

🐛 Debugger
OD

Performs root-cause analysis, regression diagnosis, and fixes

SDD — Spec-Driven Development BR — Bounded Review BG — Background OD — On-Demand

Hexagonal Architecture & SDD

Ports and adapters. 8-phase structured SDD pipeline. 33 Go packages in a single binary.

12+
Subagents
33
Go Packages
30+
CLI Commands
19
LLM Providers
3
UI Interfaces
5
Gateway Platforms
Hexagonal Architecture Flow Interactive Ports & Adapters
flowchart TD subgraph Core ["GAIA Kernel & Orchestrator"] Loop["Main Loop: Think ➔ Act ➔ Learn ➔ Persist"] PG["PolicyGuard (Tiers: read / sandbox / full)"] KG["Knowledge Graph (User, Language, Project)"] Loop <--> PG Loop <--> KG end subgraph Subagents ["Specialized Subagents (@name)"] direction TB SDD_A["SDD: Explorer · Proposer · Specifier · Designer · Planner · Implementer · Verifier · Archiver"] OD_A["On-Demand: Reviewer (4 Lenses) · Debugger · Researcher"] BG_A["Background: Learner (Skill Synthesis) · Dynamic Subagents"] end subgraph Adapters ["Adapters & Interfaces"] direction LR LLM["19 LLM Providers & Pool"] UI["Cyberpunk TUI · Web Dashboard · Desktop"] GW["Gateways: Telegram · Discord · Slack · WhatsApp"] end Core <==> Subagents Core <==> Adapters
Spec-Driven Development Pipeline 8-Phase Continuous Delivery
flowchart LR E["1. Explore\n(@explorer)"] --> P["2. Propose\n(@proposer)"] P --> S["3. Spec\n(@specifier)"] S --> D["4. Design\n(@designer)"] D --> T["5. Tasks\n(@planner)"] T --> A["6. Apply\n(@implementer)"] A --> V["7. Verify\n(@verifier)"] V -- "Pass" --> ARC["8. Archive\n(@archiver)"] V -. "Fail / Fix" .-> A

Get Started in Seconds

Install instantly with a one-line installer or compile the standalone Go binary from source.

Linux & macOS (Installer)
curl -fsSL https://raw.githubusercontent.com/SalvucciFacundo/gaia/main/install.sh | bash
Windows (PowerShell)
iwr -useb https://raw.githubusercontent.com/SalvucciFacundo/gaia/main/install.ps1 | iex
Build from Source
git clone https://github.com/SalvucciFacundo/gaia.git
cd gaia
go build -o gaia ./cmd/gaia/
./gaia
Web UI Dashboard
gaia serve 8080
Docker Backend
gaia exec "explain this project" --backend docker
SSH Remote Backend
gaia exec "list files" --backend ssh://user@server
1

First Run Setup Wizard

On first launch, GAIA automatically starts an interactive terminal setup wizard to configure your preferred LLM provider (or auto-detect local Ollama / Copilot) and install recommended skills for your project stack.

$ gaia
# Interactive setup wizard starts automatically · Config stored in ~/.gaia/config.yaml