Instant, secure orchestration for AI agents—launch, mesh, and monitor in real time.

Download the installer from the releases page, open the .pkg file, and follow the installation wizard.

For advanced users: Install via Homebrew

brew tap arkavo-org/homebrew-arkavo
brew install arkavo

brew tap arkavo-org/homebrew-arkavo
brew install arkavo

Download the installer from the releases page and run the .exe file.

# Launch agent (auto-configures on first run)
arkavo

# Launch web UI
arkavo ui

• True zero-config discovery: agents auto-find each other with mDNS/DNS-SRV.
• Built for performance: Rust core pushes ≤ 2 ms A2A round-trips on commodity Macs.
• Visual flow map: instant insight into who's talking to whom

When you run arkavo for the first time in a directory, it automatically:

• Creates an AGENTS.md configuration file
• Creates a .arkavo storage directory
• Generates a unique agent ID based on your directory name (e.g., myproject-a1b2c3d)
• Configures default settings for immediate use
• Starts the agent with mDNS discovery enabled

Arkavo Edge now supports Kimi (Moonshot AI) models including the 128k context window variant:

• Configure agents with model: kimi://moonshot-v1-128k in AGENTS.md
• Add MOONSHOT_API_KEY: sk-your-api-key to the agent configuration in AGENTS.md
• API keys are securely disseminated from the UI orchestrator to agents
• Supports 8k, 32k, and 128k context models

Arkavo Edge provides first-class support for DeepSeek's API with both standard and reasoning models:

• Anthropic-compatible API: Full support for DeepSeek's Anthropic-style endpoints
• Function Calling: Support for up to 128 tools per request with automatic schema validation
• Model Selection: Use deepseek-chat (default) or deepseek-reasoner via DEEPSEEK_MODEL environment variable
• Strict Mode (Beta): Enhanced reliability with JSON Schema validation for tool arguments
• Automatic Fallback: Seamlessly switches from deepseek-reasoner to deepseek-chat when tools are present

Configure with environment variables:

• DEEPSEEK_API_KEY: Your DeepSeek API key
• DEEPSEEK_MODEL: Model to use (deepseek-chat or deepseek-reasoner)
• DEEPSEEK_BASE_URL: Optional custom API endpoint

Usage:

# Interactive chat with DeepSeek
DEEPSEEK_API_KEY=your-key arkavo chat --model deepseek

# Single prompt
DEEPSEEK_API_KEY=your-key arkavo chat --model deepseek --prompt "Explain quantum computing"

# With reasoning model
DEEPSEEK_API_KEY=your-key DEEPSEEK_MODEL=deepseek-reasoner arkavo chat --model deepseek --prompt "Solve: 15 * 23"

Arkavo Edge integrates with OpenTDF platform for entitlement-based access control:

• Fine-grained permissions: Control MCP tool execution with attribute-based policies
• JWT-based authentication: Secure token validation via Entity Resolution Service v2
• Connect protocol support: Uses OpenTDF Authorization v2 APIs with efficient HTTP/JSON
• Smart caching: Reduces latency with TTL-aware decision caching
• Fail-closed security: Denies access by default with safe diagnostic tool allowlist

Configure with environment variables:

• OPENTDF_BASE_URL: Platform endpoint (default: https://platform.opentdf.io)
• OIDC_ISSUER: Token issuer for validation
• AUD: Expected audience claim

Arkavo Edge includes a comprehensive suite of MCP tools for AI coding agents:

• codegrep_search: Fast repository-wide code search with ripgrep
• struct_find_replace: Language-aware structural search and replace with Comby
• syntax_tree: AST parsing for syntax-aware code analysis with tree-sitter

• sec_semgrep: SAST scanning with Semgrep
• deps_osv: Dependency vulnerability scanning with OSV-Scanner
• sbom_syft: SBOM generation with Syft

• browser_cdp: Chrome DevTools Protocol automation via chromiumoxide
• test_run: Multi-language test runner (pytest, jest, go test, cargo test, xcodebuild)

• gh_checks: GitHub Checks API integration with inline annotations
• gh_pr_review: PR reviews with line-level comments

• workspace_container: Container-based isolated execution with resource quotas (Docker/Podman)

• swe_bench: Objective benchmarking harness with metrics tracking

See docs/coding-agent-toolset.md for complete tool documentation.

mDNS discovery uses pure Rust implementation (mdns-sd crate) with no system dependencies

Note: iOS simulator automation and testing capabilities are only available on macOS.

cargo build --release

The default build includes mDNS discovery using a pure Rust implementation (mdns-sd crate) that doesn't require system libraries like Avahi or Bonjour. This provides true portability across all platforms.