analyse-problem

Overview of analyse-problem skill

What analyse-problem is for

The analyse-problem skill turns a messy issue into a structured A3 problem brief: background, current condition, target state, root cause analysis, countermeasures, implementation plan, and follow-up. It is best for Strategic Planning, ops, product, and engineering work when you need a decision-ready problem statement instead of a brainstorm.

Who should install it

Install analyse-problem if you regularly need to explain a problem clearly, align stakeholders, or move from symptoms to causes. It fits people who want a repeatable format for incident review, process improvement, project blockers, or planning sessions.

What makes it different

The value is not “more AI writing”; it is constraint. analyse-problem pushes the model to document the problem in one page, separate facts from assumptions, and end with actions and verification. That makes it more useful than a generic prompt for analyse-problem for Strategic Planning because it forces a tighter decision frame.

How to Use analyse-problem skill

Install and trigger it

Use the install flow from your skills manager, then invoke the analyse-problem skill with a short problem statement. The repository’s usage pattern is /analyse-problem [problem_description], and the skill supports an optional problem description plus markdown output by default. If your environment maps skills differently, pass the same intent: a concise, concrete issue statement.

Give the skill the right input

Strong inputs name the problem, scope, and evidence. For example: “Customer onboarding completion dropped 18% this month in enterprise accounts; analyze causes and propose countermeasures.” Weak inputs like “improve onboarding” leave the model guessing the target, owner, and success metric. For analyse-problem usage, include:

• the symptom
• the affected process or team
• any data points, examples, or timelines
• the desired outcome or constraint

Read these repository parts first

Start with SKILL.md, then inspect any linked instructions in README.md, AGENTS.md, metadata.json, and folders such as rules/, resources/, references/, or scripts/ if they exist. For this repo, the main source is concentrated in plugins/kaizen/skills/analyse-problem, so there is little extra scaffolding to chase.

Shape the prompt for better output

A good analyse-problem guide prompt is specific enough to populate each A3 section. For example:
“Use analyse-problem to document why release delays increased in Q2. Include baseline metrics, likely root causes, 3 countermeasures ranked by impact and effort, and a 30-day follow-up plan.”
That produces a more usable result than asking for a generic analysis, because it gives the model facts to organize and a decision to support.

analyse-problem skill FAQ

Is analyse-problem only for Strategic Planning?

No. It is useful anywhere a problem needs to be framed, investigated, and assigned action: product issues, operational bottlenecks, team process failures, or strategic planning reviews. The A3 format is broad, but the skill is most valuable when the output will be shared with other people.

How is this different from a normal prompt?

A normal prompt can ask for an analysis, but the analyse-problem skill gives a reusable structure and output discipline. That matters when you need consistent coverage of root cause, countermeasures, and follow-up rather than a freeform explanation.

Can a beginner use it well?

Yes, if the input is concrete. Beginners usually struggle when the problem is vague, so provide what you know: observed symptom, scope, timing, and a rough target. The skill can help organize incomplete thinking, but it cannot invent missing facts safely.

When should I not use it?

Do not use analyse-problem when you only need a quick summary, a status update, or a creative ideation session. It is also a poor fit if the issue has no stable scope yet; in that case, define the problem first before asking for a structured analysis.

How to Improve analyse-problem skill

Feed it evidence, not just conclusions

The biggest quality jump comes from concrete inputs: metrics, examples, incident timestamps, user quotes, or process notes. If you already suspect a root cause, label it as a hypothesis, not a fact, so the analysis can test it instead of echoing it.

Ask for decision-useful output

For analyse-problem, ask for what will help a team act: ranked causes, countermeasures tied to causes, owners, dependencies, and a verification plan. If you want Strategic Planning value, request impact and effort tradeoffs rather than a long narrative.

Watch for common failure modes

The usual failure is over-explaining symptoms and under-specifying the goal state. Another is mixing multiple problems into one page. Split unrelated issues, and keep one analysis focused on one decision, one owner, and one measurable outcome.

Iterate after the first draft

Use the first output to check for gaps: missing data, weak causal links, or actions that do not match the root cause. Then refine the prompt with new facts and constraints, such as budget, deadline, team capacity, or a required format for leadership review.