What tools are used for codebase discovery?

The skill orchestrates several tools including parallel sub-agents, repo mapping (gkg), library documentation (exa), and external pattern searching (Librarian) to build a comprehensive Discovery Report.

What are 'beads' in the planning pipeline?

Beads are granular, executable work items or tasks stored as markdown files in your repository. They include context, spike learnings, and clear acceptance criteria for developers or AI agents.

Why does the pipeline require a separate git worktree?

Using worktrees ensures that task tracking (beads) and feature development happen in isolated environments, preventing branch-switching conflicts and ensuring a clean merge process.

How does the skill manage high-risk implementation tasks?

The skill identifies high-risk items during the Synthesis phase and triggers 'Spikes'—short, focused research tasks that produce throwaway code to verify technical feasibility before the main work begins.

Agentic Feature Planning

Name: Agentic Feature Planning
Author: draphonix

bydraphonix

•

협업 도구

Generates systematic, risk-aware implementation plans by orchestrating discovery, verification spikes, and task decomposition.

The Planning skill implements a robust multi-phase pipeline designed to transform complex feature requests into executable technical roadmaps. It leverages parallel sub-agents to perform deep codebase discovery and uses Oracle-driven synthesis to identify architectural gaps and risks. By incorporating a dedicated verification phase for high-risk items through technical 'spikes' and decomposing work into granular 'beads,' it provides a structured path from ideation to parallelized execution. This skill is essential for teams looking to automate technical design and ensure that every implementation step is validated against existing codebase patterns and constraints.

주요 기능

01Automated task decomposition into git-tracked 'beads'

02Dependency graph validation and cycle detection using BV tools

03Multi-agent discovery reports for rapid codebase intelligence

04Parallel track planning with unique agent assignments and file scopes

0549 GitHub stars

06Risk-based verification using time-boxed technical spikes

사용 사례

01Architecting large-scale features with multiple interdependent components

02Breaking down monolithic tasks into parallelizable execution tracks

03Onboarding AI agents to complex codebases through structured discovery

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add draphonix/skills planning

For use in Claude.ai and ChatGPT

Download Skill