planner_skill

This skill creates executable phase plans with 2-3 task waves, dependency graphs, and goal-backward verification for rapid solo development.
  • PowerShell

497

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill toonight/get-shit-done-for-antigravity --skill planner

  • SKILL.md12.5 KB

Overview

This skill creates executable phase plans for a solo developer + AI workflow, producing small, parallel-optimized plans with explicit tasks, dependencies, and goal-backward verification. It focuses on producing PLAN.md-style prompts that the AI will execute, with measurable verification steps and strict task anatomy to prevent ambiguous work.

How this skill works

Given a phase objective and context, the skill decomposes the phase into plans containing 2-3 tasks each, builds a dependency graph, assigns execution waves, and produces must-haves using a goal-backward approach. Each task lists exact file paths, precise implementation actions (including what to avoid and why), verification commands, and measurable done criteria. The skill also enforces mandatory discovery and gap-closure workflows when needed.

When to use it

  • Planning small, autonomous feature slices for a single implementer (AI) and one visionary (user).
  • Creating clear, executable prompts (PLAN.md) that the AI will run directly.
  • Producing gap-closure plans after verification reveals missing behavior.
  • Designing plans that must include dependency waves and parallel execution constraints.
  • Preparing TDD-style plans for complex business logic or stateful systems.

Best practices

  • Keep each plan to 2-3 tasks and stop before context pressure exceeds ~50%.
  • Specify exact file paths, explicit actions, AVOID notes, verify commands, and measurable done criteria.
  • Prefer vertical slices that deliver end-to-end functionality rather than horizontal layering.
  • Run mandatory discovery: Level 0-3 assessment and route to a research phase when needed.
  • Group independent plans in the same wave only if they do not modify the same files.

Example use cases

  • Create a login endpoint: plan with tasks for route file, auth util, and verification curl commands.
  • Add a new model and CRUD API as a vertical slice split into small plans across waves.
  • Close gaps reported by automated verification: generate targeted fix plans with single-issue scope.
  • Implement TDD for financial logic: red/green/refactor tasks with tests and minimal implementations.
  • Migrate a utility into a separate module to enable parallel wave execution without file conflicts.

FAQ

Plans target one user (visionary) and one implementer (the AI). No team processes are included.

What if a decision requires research?

Discovery protocol classifies the need (Level 1-3). For Level 2+, create a research-phase and produce RESEARCH.md before planning.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational