← ClaudeAtlas

autodevlisted

Autonomous feature development loop for low-risk, self-contained features using git worktrees/sandboxes and the autoresearch-style experiment pattern.
ominou5/agentic-workflows · ★ 1 · AI & Automation · score 72
Install: claude install-skill ominou5/agentic-workflows
# 🤖 /autodev — Universal Autonomous Feature Development Protocol > **[FIRST-RUN / STARTUP BLOCK — AGENT INSTRUCTIONS]** > ### 🛑 AGENT SELF-CUSTOMIZATION PROTOCOL > If you are the AI agent reading this block, it means this `/autodev` protocol has not been customized for this repository yet. You **MUST** stop and execute this self-customization sequence before performing any actual feature development. > > Follow these steps immediately: > > #### Step 1: Prompt the User > Ask the user to clarify their stack, risk tolerance, and preferences. You may copy and paste the following template to prompt them: > > > 🚀 **Autodev Initialization Initiated!** > > Let's customize this protocol for your codebase. Please answer the following: > > > > 1. **What is your Tech Stack?** (e.g., React + TypeScript + Vite + Node.js, Rust + Actix, Python + FastAPI, etc.) > > 2. **What are your exact local verification commands?** > > * *Compilation/Typecheck*: (e.g., `npx tsc --noEmit` or `cargo check`) > > * *Linting/Static Analysis*: (e.g., `npm run lint` or `eslint .`) > > * *Production Build*: (e.g., `npm run build` or `vite build`) > > * *Test Suite*: (e.g., `npm run test` or `cargo test`) > > 3. **What is your Risk Tolerance level?** > > * `High`: YOLO mode is default; writing/updating existing files is relaxed; database writes are okay; less strict file protection rules. > > * `Medium` (Recommended): Balanced; isolation preferred for non-trivial edits; database writes