discovery

Manages time-boxed spike issues on GitHub for exploring unscoped work.

Updated Jul 29, 2026
One-click install
npx skills add https://github.com/chris-prener/dev-kit --skill discovery-chris-prener
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: discovery
Source: https://github.com/chris-prener/dev-kit/tree/main/dev-kit/skills/discovery
Command: npx skills add https://github.com/chris-prener/dev-kit --skill discovery-chris-prener

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When requirements are unclear, teams often jump straight into building or let exploration drag on without structure. This Skill provides a disciplined, time-boxed spike workflow that turns ambiguous questions into concrete artifacts — a requirements doc, scoped issues, an ADR, or a documented no-go decision — instead of production code. ## Core Features & Use Cases - Open Spike: Files a GitHub issue with the spike label and a frozen discovery template capturing the question, time-box, and expected artifact. - Findings Tracking: Appends dated findings to the spike issue body as exploration progresses, keeping context durable. - Structured Closure: Closes spikes only with a named artifact (requirements doc, scoped issues, ADR, or not-planned summary), refusing silent closes. - Overrun Audit: Surfaces open spikes whose stated time-box has expired and prompts continue, cut-scope, or close decisions. - Use Case: Your team is asked to integrate an unfamiliar third-party API. Open a 2-day spike, log findings as you probe the API, then close with scoped implementation issues filed via the backlog workflow. ## Quick Start Ask the assistant to open a time-boxed spike to investigate whether we should build the proposed feature, with a requirements doc as the target artifact.

Frequently Asked Questions about discovery

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

FAQPage Schema
How do I run a time-boxed spike on GitHub issues?▼

Open a spike issue with the spike label using a template that records the question, time-box, and target artifact. Append dated findings as you explore, then close with a structured comment naming the artifact produced, such as scoped issues or an ADR.

When should I use a spike instead of a regular issue?▼

Use a spike when scope itself is unknown — unfamiliar data sources, competing approaches, or whether to build at all. If scope is already clear, file a regular issue through the backlog workflow instead of opening a spike.

What artifacts can a discovery spike produce?▼

A spike closes with one of four artifacts: a requirements doc, a set of scoped issues, an architecture decision record (ADR), or a documented not-planned decision summary. Spikes never produce production code.

Can I close a spike without producing an artifact?▼

The close operation refuses to proceed without a named artifact selection. However, closing the issue directly with raw gh commands or the GitHub UI bypasses this check, since there is no periodic auditor enforcing closure discipline.

How are overrun spikes detected?▼

The audit operation queries open spike-labeled issues, parses the Time-box and Opened fields from the body, and compares expiry against current UTC time. Overrun spikes are flagged with a recommendation to continue, cut scope, or close.