spec

Turns a fuzzy goal or task into a versioned, execution-ready plan posted to an issue tracker.

1|1|Updated Jul 23, 2026
One-click install
npx skills add https://github.com/nearmap/shipyard --skill spec-nearmap
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: spec
Source: https://github.com/nearmap/shipyard/tree/main/skills/spec
Command: npx skills add https://github.com/nearmap/shipyard --skill spec-nearmap

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning a vague objective into a plan a fresh coding session can execute without re-asking questions is slow and error-prone. This Skill scopes the work, researches load-bearing behavior, resolves repository standards, removes ambiguity, and posts a versioned execution plan to Jira or GitHub Projects so implementation can start from an approved, self-contained specification. ## Core Features & Use Cases - Premise and prior-work validation: Checks whether the goal is already delivered, invalidated, or owned by sibling work before spending research effort, and shelves the task with evidence when the premise is gone. - Delegated research with verification: Uses breadth sweeps and end-to-end traces through sub-agents, resolves standards and risk lenses, and converts each risk lens into a verification obligation with named evidence. - Adversarial plan review: Pressure-tests the core design decision through a debate pass and runs a spec-gate review of the drafted plan before sign-off, dispositioning every finding. - Versioned plan publishing: Posts the approved plan as a single new tracker comment split into a human sign-off half and a mechanical /sy:ship half, then marks the task ready. - Use Case: You have a ticket saying "add idempotent webhook retries." Run the spec skill against it; it researches the codebase, debates the design, and posts an execution plan with file anchors, invariants, and acceptance criteria that /sy:ship can implement directly. ## Quick Start Ask the AI to spec the task: turn ticket PROJ-123 into an execution-ready plan with research, standards, and verification obligations.

Frequently Asked Questions about spec

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

FAQPage Schema
How do I turn a vague feature idea into an implementation plan?▼

Run the spec skill with the goal or an existing task key. It scans the code surface, researches load-bearing behavior, resolves repository standards, and posts a versioned execution plan to the tracker that a fresh /sy:ship session can implement without missing design decisions.

What issue trackers does this planning workflow support?▼

The skill writes through the tracker skill, which supports Jira and GitHub Projects. All task creation, plan comments, and status transitions go through that adapter rather than direct tracker API calls.

Does the spec skill write any code?▼

No. Code work is read-only during spec; it ends at the approved plan or at a shelve-with-evidence closure when research invalidates the premise. Implementation is delegated to a separate /sy:ship session.

What happens if the task is too big for one pull request?▼

The skill does not split oversized tasks ad hoc. It posts a SEAMS comment describing the cut line and dependencies, then stops and hands off to /sy:plan, which performs the tracker's canonical decomposition into smaller tasks.

Why does the skill stop before planning sometimes?▼

It runs a premise and prior-work check first: if the goal is already delivered, invalidated, or owned by an open sibling task, it closes the item with an evidence comment instead of producing a plan for work that should not ship.