Open Source • React Native & Expo • v0.12.2
13 specialized agents. 35 knowledge rules. Video debugging. Autonomous worker. All in one npx command.
Adaptive agents, rules, commands, and hooks — matched to your exact React Native stack.
One command deep-scans 15 stack dimensions, selects variant templates, and configures everything.
That's it. ERNE detects your stack, generates adaptive config, and launches.
Real-time visibility into agents, tasks, issues, and project health. All from your browser.
New in v0.12.2
Send a screen recording. ERNE extracts key frames at critical moments, analyzes the visual timeline, and finds bugs that static screenshots miss — animation glitches, race conditions, gesture failures.
Purpose-built agents that understand React Native deeply. Not generic chatbots with framework docs bolted on.
The pipeline orchestrator coordinates agents through a 5-phase workflow — from architecture to validation — with retry logic and structured handoffs between agents.
Comprehensive knowledge base covering the entire React Native ecosystem. Auto-updates weekly via GitHub Action.
New in v0.12.2
ERNE now includes Apple Human Interface Guidelines intelligence. Spring physics, semantic colors, 8pt grid, haptic pairing, and a 24-rule compliance scanner — all enforced automatically.
Agents know what to do. Skills define how — step-by-step workflows, checklists, and guardrails that enforce best practices automatically.
New in v0.12.2
Connect your project manager. ERNE picks up tickets, validates them, routes to the right agents, and creates pull requests — with zero human intervention.
Each ticket goes through confidence scoring (threshold: 70+), isolated git worktree execution, automated self-review against ERNE coding standards, audit health comparison, and finally PR creation with summary and test results.
New in v0.12.2
Click "Fix" on any issue in the dashboard. ERNE auto-detects whether Claude Code is running and picks the best strategy.
Three profiles that control how aggressively hooks check your code. Switch anytime.
Fast iteration with no friction. Perfect for prototyping and vibe coding sessions.
Balanced quality and speed. Catches real issues without slowing you down.
Production-grade enforcement for apps shipping to millions of users.
Architecture-level savings control what enters context. Runtime optimization compresses tool outputs by 97–100% — verified across 21 benchmark scenarios.
537 KB of tool outputs compressed to 2.6 KB across 21 real-world scenarios. Full debugging sessions use only 0.4% of context window — leaving 99.6% free for problem solving.
Full experience in Claude Code. Rules and config export for every major AI editor.