explore

Transforms a raw feature idea into a product brief and domain context map through structured dialogue.

Updated Jun 12, 2026
One-click install
npx skills add https://github.com/lucolucus/mismagent --skill explore-lucolucus
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: explore
Source: https://github.com/lucolucus/mismagent/tree/main/plugins/mismagent/skills/explore
Command: npx skills add https://github.com/lucolucus/mismagent --skill explore-lucolucus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Starting a feature by jumping straight into code leads to misunderstood problems, inconsistent naming, and premature design decisions. This Skill structures the discovery phase of the mismAgent methodology: it turns a raw idea into an understood problem with a validated product brief, a domain context map, and a fixed ubiquitous language before any contract, task, or code exists. ## Core Features & Use Cases - Idea stress-testing: Dispatches a fresh-context challenger subagent that tries to demolish the idea before modeling, returning KILL, RESHAPE, or PROCEED verdicts. - Domain modeling: Invokes an analyst subagent that produces a context-map.md with bounded contexts, relationships, and canonical ubiquitous language that all downstream artifacts inherit. - Gated handoff: Produces product-brief.md, tactical-model.md seeds, spikes, and optional research notes, and blocks the model phase until the brief and context map exist. - Use Case: You have a one-sentence idea for a new feature. Run the explore command, dialogue through goals and constraints, let the challenger attack the idea, and receive a product brief plus an amended project context map ready for the modeling phase. ## Quick Start Run /mismagent:explore followed by your feature idea in one sentence to begin the guided discovery dialogue.

Frequently Asked Questions about explore

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

FAQPage Schema
How do I start a new feature with the mismAgent explore skill?

Run /mismagent:explore followed by your idea in one sentence. The skill opens a dialogue about goals, users, and constraints, then dispatches a challenger subagent to attack the idea and an analyst to model the domain before any code is written.

What artifacts does the explore phase produce?

Explore produces a product-brief.md with problem, user, value, and scope, plus a context-map.md containing bounded contexts and ubiquitous language. It can also emit a tactical-model.md seed file, spikes for open unknowns, infra notes, and research documents when needed.

Can I skip explore and go straight to modeling or coding?

No. The model phase only starts when the feature's product-brief.md and the project's context-map.md exist. Explore enforces a no-premature-coding rule so that contracts, tasks, and code always inherit validated problem definitions and canonical names.

What happens if the challenger subagent rejects my idea?

A KILL verdict stops the process and reports back to you, while RESHAPE triggers a redesign dialogue. Only a PROCEED verdict, with all must-answer items closed, allows the analyst to model the domain.

Does explore work in read-only or plan mode?

Yes. The dialogue and the read-only challenger proceed normally, but file-writing subagents like the analyst are deferred. Pending files are listed in the plan and materialized first once writes are re-enabled.