One-click install
npx skills add https://github.com/berlinguyinca/autospec --skill autospec-split
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: autospec-split
Source: https://github.com/berlinguyinca/autospec/tree/main/skills/autospec-split
Command: npx skills add https://github.com/berlinguyinca/autospec --skill autospec-split

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires curl, gh, jq, git, and includes scripts (resource) components.

What problem does it solve?

Autospec-split turns an already-written, tracked design spec in docs/specs into a properly labeled GitHub issue epic plus an auto-implement child-issue queue, so you can start implementation without rewriting planning.

Core Features & Use Cases

  • Existing-spec shortcut: Selects a tracked docs/specs/*.md design doc (explicit path or newest candidate) and skips spec writing.
  • Linked issue decomposition: Produces one epic umbrella issue plus N self-contained child issues ready for local-LLM execution.
  • Quality scaffolding & handoff: Stops after Phase 3.5 with the /autospec-run handoff so implementation can begin immediately.

Quick Start

Ask your agent to split an existing spec into issues and stop for the /autospec-run handoff by requesting: split latest spec.

Frequently Asked Questions about autospec-split

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

FAQPage Schema
How do I split an existing markdown spec into GitHub issues for implementation?

To decompose an existing markdown spec into GitHub issues, use autospec-split. It resolves a tracked design doc in docs/specs/*.md and generates a linked epic umbrella issue plus self-contained child issues ready for execution without rewriting your planning.

Does spec decomposition require writing a new design document from scratch?

Spec decomposition does not require writing a new design document. The process selects an already-written tracked design spec from docs/specs and skips the brainstorming phase, directly converting your existing product intent into an auto-implement issue queue.

How do I generate a GitHub issue queue with deterministic labels from a design spec?

You generate a GitHub issue queue with deterministic labels by running controlled issue generation via the GH CLI. The workflow applies model-fit labeling and lint gates to produce an epic plus N linked child issues ready for local-LLM execution.

What dependencies do I need to decompose specs and create GitHub epics?

To decompose specs and create GitHub epics, you need curl, gh, jq, and git installed. These command-line dependencies handle repo validation, origin/main tracking checks, and controlled GitHub issue generation.

Can I use autospec-split to create issues for local-LLM execution and stop before coding?

You can use autospec-split to create issues for local-LLM execution and stop before coding. The workflow halts after Phase 3.5 with a handoff step, allowing implementation to begin immediately without writing any code during the decomposition phase.

What happens if there are multiple design specs in the docs/specs directory?

When multiple design specs exist in the docs/specs directory, the workflow performs candidate-spec resolution with ambiguity prompting. You can provide an explicit path to select the correct file or let the system target the newest candidate spec automatically.