AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
Enterprise-grade linting and guardrails for AI work. Turn your team's architectural decisions into executable checks that run in CI, pre-commit hooks, and inside AI coding tools.
Developer toolsTypeScript · JavaScriptGitHub @archgateOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
Enforce Architecture Decision Records as executable rules — for both humans and AI agents
Taken from the repository README
Loading the full README. The saved excerpt is shown below.
Enterprise-grade linting and guardrails for AI work.
AI agents write code fast, but they don't know your rules. Archgate turns your team's decisions into executable checks: a lint step for architecture, conventions, and AI output. Your agents read the rules before writing code, and archgate check blocks what slips through. In CI, in pre-commit hooks, and inside every major AI coding tool.
Write an ADR once. Enforce it everywhere.Read the full README ↗Enterprise-grade linting and guardrails for AI work.
AI agents write code fast, but they don't know your rules. Archgate turns your team's decisions into executable checks: a lint step for architecture, conventions, and AI output. Your agents read the rules before writing code, and archgate check blocks what slips through. In CI, in pre-commit hooks, and inside every major AI coding tool.
Write an ADR once. Enforce it everywhere.
Archgate has two layers:
.archgate/adrs/. Each ADR records a decision: what was decided, why, and what to do and not do..rules.ts file that exports automated checks. Archgate runs these checks against your codebase and reports violations..archgate/
└── adrs/
├── ARCH-001-command-structure.md # human-readable decision
├── ARCH-001-command-structure.rules.ts # machine-executable checks
├── ARCH-002-error-handling.md
└── ARCH-002-error-handling.rules.ts
When a rule is violated, archgate check reports the file, line, and which ADR was broken. Exit code 1 means violations. Wire it into CI and it blocks merges automatically.
The CLI is free and open source. Writing ADRs, enforcing rules, running checks in CI, and wiring up pre-commit hooks all work without an account or subscription.
Install via standalone script, npm, pip, dotnet, Go, gem, Maven, or proto. See the installation guide for all options and platform support.
# 1. Initialize governance in your project
cd my-project
archgate init
# 2. Edit the generated ADR to document a real decision
# .archgate/adrs/ARCH-001-*.md
# 3. Add a companion .rules.ts to enforce it automatically
# .archgate/adrs/ARCH-001-*.rules.ts
# 4. Run checks
archgate check
Each ADR can have a companion .rules.ts file that exports automated checks. See the writing rules guide for examples and the full rule API reference.
Make your AI agent architecture-aware. With the optional editor plugins, your AI coding agent reads ADRs before writing code, validates changes against your rules, and captures new architectural patterns back into ADRs, automatically.
Plugins are available for Claude Code, Cursor, VS Code, Copilot CLI, and opencode.
archgate login # one-time GitHub auth archgate init # installs the plugin automaticallyGet started with plugins. The CLI works fully without them, but plugins close the loop between decisions and code.
Full documentation is available at cli.archgate.dev, including guides for writing ADRs, writing rules, CI integration, editor plugin setup, and the complete CLI reference.
Questions, ideas, or feedback? Join the discussion at r/archgatedev.
See CONTRIBUTING.md for development setup and workflow.