Autonomous development workflow. Generate detailed specs, plans, and tasks for autonomous agent execution with session memory tracking.
View on GitHubanexpn/claude-plugins
jun-dev-workflows
January 21, 2026
Select agents to install to:
npx add-skill https://github.com/anexpn/claude-plugins/blob/main/skills/dev-auto/SKILL.md -a claude-code --skill dev-autoInstallation paths:
.claude/skills/dev-auto/# Autonomous Development Workflow ## Overview Generate comprehensive documentation that enables autonomous agents to implement features, fixes, or changes without constant human supervision. All work is tracked in session docs (spec, plan, tasks) that serve as long-term memory across implementation sessions. ## When to Use This Skill Use this skill when: - Working on well-defined features or fixes - Work can be fully specified upfront - Implementation can proceed autonomously with minimal supervision - Long-term session memory is desired for tracking progress **Do NOT use for:** - Exploratory work requiring human decisions at each step (use dev-guided instead) - Trivial changes that don't require planning - Quick iterations where formal documentation is overhead ## Workflow Phases This workflow consists of 5 phases. Some run interactively in the main conversation (requiring user input), while others can run as autonomous subagents. ### Phase 1: Specification (Interactive) Generate a comprehensive design specification through iterative questioning. **Process:** 1. Examine the project to understand current state 2. Ask ONE question at a time (preferring multiple choice) 3. Refine understanding through Q&A until certain 4. Present specification in 200-300 word sections 5. Get approval for each section before proceeding 6. Write final spec to `docs/development/NNN-<name>/spec.md` **Reference:** See `references/phase-spec.md` for detailed guidance **User involvement:** Answer questions, approve spec sections ### Phase 2: Planning (Interactive or Subagent) Create detailed implementation plan assuming implementer has minimal context. **Approach:** 1. **Tech stack selection** (when needed) - Research and document library/framework choices 2. **File/module mapping** - Exact paths for new/modified code 3. **Dependency order** - Build sequence so later pieces have foundations 4. **Integration points** - How new code connects to existing code 5. **Risk flags** -