Owner 5w2h

Skill XHdW/owner-mode/platforms/claude/skills/owner-5w2h

Owner Mode for AI agents, implemented as Agent Skills: domain judgment, deliberate choice, and responsibility for work that must hold up in real use.

Install

npx -y skills add XHdW/owner-mode --skill owner-5w2h

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

16 days oldThe repository was created 16 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Planning preflight: use when the user invokes $owner-5w2h, asks to check the completeness of a requirement, design, implementation approach, research plan, or project plan, or when the current task is explicitly to form or revise such a plan and a concrete missing or contradictory context item has plausible answers that would change its scope, architecture, sequence, authority, or lifecycle cost.

SKILL.md

5.0 KB, as published. Nobody here has run it

Owner 5W2H

Run a bounded preflight on the requirement, design, implementation approach, research plan, or project plan being formed or revised. Find only the minimum missing, inconsistent, or disconnected context needed to settle a named plan decision. This may be one gap or the smallest coupled set that must be resolved together. Use Who, What, When, Where, Why, How, and How Much as overlapping internal coordinates rather than a mandatory checklist.

Keep one probe active:

What concrete missing or contradictory context could select a different plan, and can current context or reality resolve it first?

Confirm the Planning Fork

Read the request, conversation, applicable instructions, artifacts, workspace state, and available tools before declaring a context gap. Preserve facts already established by the user or environment.

A direct request to use the lens or check a plan's completeness is sufficient. In all other cases, begin the scan only when both gates hold:

The current task is explicitly producing or revising a requirement, design, implementation approach, research plan, or project plan.
A concrete missing or contradictory item is already visible in the request, plan, artifact, or evidence, and its plausible answers would change a named decision about scope, protocol, architecture, mechanism, sequence, timing, authority, location, or lifecycle cost.

When either gate fails, continue the current task using its existing context. Begin the 5W2H scan from the visible planning fork.

Scan the Planning Context

Inspect only the dimensions and relationships that could distinguish the live planning alternatives:

Who: delegated role, authority, user, audience, approver, maintainer, and affected people.
What: practical outcome, artifact, protocol, compatibility boundary, scope, and observable completion evidence.
When: current time, deadline, dependency order, active effort, waiting, uncertainty, and re-estimation condition. Read references/time-and-estimation.md only when a deadline, dependency order, wait, or estimate could select between the live plan alternatives. Read references/personal-calibration.md only when the user explicitly asks to create, use, or design a private estimate history.
Where: active workspace, repository or worktree, source of truth, environment, destination, Git state, and publication boundary. Read references/workspace-and-repository.md only when the target location or its authority boundary could select between the live plan alternatives.
Why: user value, consequence, urgency, and the causal belief connecting the deliverable to the outcome.
How: mechanism, dependencies, tools, integration path, reality signal, recovery path, and stopping condition.
How Much: money, model or compute use, time, attention, context, rework, maintenance, delay, and downstream failure cost. Read references/cost-and-rework.md only when lifecycle cost or reversibility could select between the live plan alternatives.

Treat the dimensions as overlapping coordinates rather than exclusive categories. Look for contradictions and missing relationships between them; the controlling context may be a coupled set that belongs to no single heading.

If the plan remains underdetermined, test once whether a hidden relationship, dynamic, or higher-level purpose would change a named decision. Keep this higher-level check only when it reveals a different planning fork.

Resolve the Planning Fork

Discard candidates already established by reliable context and candidates unable to change a named plan decision. Resolve each remaining gap by its available path:

Discoverable: inspect an artifact, repository, clock, system, or authoritative source.
Reversible assumption: proceed when correction is cheap and delay costs more.
User-owned: surface the smallest choice whose alternatives materially change the plan.
Externally blocked: identify the unavailable external change and the evidence that makes it necessary.

Apply the resolved context directly to the requirement or plan. Stop when the plan choice is no longer sensitive to unresolved context, or when the only remaining dependency is the smallest user-owned choice or an evidence-backed external blocker.

End with the revised plan, planning decision, user-owned choice, or blocker rather than a separate 5W2H report. Keep private reasoning and the seven-dimension map internal.