planning-pipeline

Reviews product plans through five lenses in one subagent pass before implementation.

Updated Sep 10, 2026
One-click install
npx skills add https://github.com/joydai2026-del/skills --skill planning-pipeline-joydai2026-del
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: planning-pipeline
Source: https://github.com/joydai2026-del/skills/tree/main/planning-pipeline
Command: npx skills add https://github.com/joydai2026-del/skills --skill planning-pipeline-joydai2026-del

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? New products and multi-week builds often fail because expensive questions about user need, scope, architecture, design, and success criteria are never asked before code exists. This Skill forces those questions upfront through a single structured review pass, preventing costly rework on decisions that are hard to reverse. ## Core Features & Use Cases - Five-lens review in one pass: A single Plan subagent evaluates idea validation, scope, engineering, design, and success criteria, returning one structured deliverable instead of five separate review documents. - Trigger discipline: Explicit rules for when the full pipeline applies (new products, multi-week builds, costly architecture decisions) versus lightweight paths for features, bug fixes, refactors, and prototypes. - Gated implementation workflow: Batches user-facing decisions into one owner question, runs a two-reviewer plan review capped at two rounds, writes the plan to docs/plans, then verifies against machine-checkable acceptance criteria after implementation. - Use Case: When starting a new product or committing to an expensive-to-change architecture, run this pipeline to get a grounded, repo-verified plan with binary success criteria before writing any code. ## Quick Start Ask the agent to run the planning pipeline on your new product idea or multi-week build plan before any implementation begins.

Frequently Asked Questions about planning-pipeline

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I validate a new product idea before writing code?▼

Run a structured planning review that asks who needs the product, what they do instead today, and what the narrowest shippable wedge is. This Skill runs that validation as one of five lenses in a single subagent pass, grounded in the actual repository.

When should I use a full planning pipeline versus just building?▼

Use the full pipeline for new products, multi-week builds with staged milestones, and architecture decisions that are costly to reverse. Skip it for single features in existing products, bug fixes, refactors, config changes, and exploratory prototypes.

How do I define success criteria for AI or LLM features?▼

Write binary, machine-checkable acceptance criteria where each criterion names its verifier. For probabilistic surfaces like LLM or agent features, treat these as the deterministic floor only and defer eval bands, failure triage, and rollback to a dedicated AI done-ness process.

Does this planning approach work for backend-only projects?▼

Yes. The design lens is explicitly skipped for backend-only deliverables or conversational interfaces like SMS and voice, with the skip stated in the status line rather than generating filler content. The design review then runs later, gated on the first visual surface.

What are the limitations of a single-pass plan review?▼

Text-based review has diminishing returns, so the process caps plan review at two rounds with one fresh-context reviewer and one adversarial reviewer. It cannot replace real user validation, and unverifiable claims are flagged as UNVERIFIED rather than resolved.