plan-task

Plans development tasks into self-sufficient execution contracts stored in .plans directories.

2|Updated Apr 29, 2026
One-click install
npx skills add https://github.com/siarhei-belavus/agent-public --skill plan-task-siarhei-belavus
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: plan-task
Source: https://github.com/siarhei-belavus/agent-public/tree/main/skills/atelier-workflow/plan-task
Command: npx skills add https://github.com/siarhei-belavus/agent-public --skill plan-task-siarhei-belavus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Ambiguous engineering work often gets implemented with hidden assumptions, unclear ownership, and missing validation criteria. This Skill turns vague development tasks into concrete, self-sufficient execution contracts that a fresh implementer can follow without access to the original conversation. ## Core Features & Use Cases - Execution Contract Authoring: Produces a canonical PLAN.md with ordered, imperative, dependency-aware steps plus INDEX.md, AMENDMENTS.md, and ARTIFACT_CANDIDATES.md packet files. - Ownership and Compatibility Discipline: Resolves code ownership and reuse before planning, prohibits backward-compatibility work by default, and forbids parallel same-purpose entities. - Amendment and Handoff Workflow: Defines stop conditions, amendment flows, and validation criteria so implementation can halt safely and resume in a clean context. - Use Case: A tech lead asks for a plan to refactor a queue runner's retry logic. The Skill produces a .plans/<task-slug>/ packet defining the final state model, validation strategy, and amendment triggers, ready for a different agent to implement cold. ## Quick Start Ask the agent to plan the development task you describe into a .plans packet with a canonical PLAN.md execution contract.

Frequently Asked Questions about plan-task

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

FAQPage Schema
How do I plan a development task for handoff to another agent?▼

Describe the task scope, constraints, and acceptance criteria, and the Skill writes a .plans/<task-slug>/ packet containing PLAN.md as the canonical execution contract. The plan is self-sufficient, naming concrete files, owners, and validation steps so a fresh-context implementer can execute it without prior conversation.

What files does a task plan packet contain?▼

A packet contains INDEX.md, PLAN.md, AMENDMENTS.md, and ARTIFACT_CANDIDATES.md, plus an optional BRIEF.md for normalized discovery output. PLAN.md is the canonical execution contract with ordered, imperative, dependency-aware steps.

Does the planning workflow allow backward compatibility changes?▼

Backward compatibility is prohibited by default and only added when a human explicitly requires it for a real external boundary. Internal-only changes are planned as clean final-state designs with no compatibility scaffolding, aliases, or fallback branches.

What happens when implementation contradicts the plan?▼

Implementation must stop when the plan becomes unsafe, impossible, or contradicted by the codebase. The amendment flow requires documenting the blocker, options with pros and cons, plan impact, a human decision, and a packet update before continuing.

When should I not use a formal execution contract for a task?▼

Formal planning packets add overhead for trivial single-file edits or exploratory spikes where the change surface and acceptance criteria are already obvious. The Skill targets tasks needing cold handoff, multi-step sequencing, or lifecycle-heavy design concerns like retry and recovery.