Guide · Agent skill
How to use the Design KPIs agent skill: setup and examples
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Design KPIs skill is an agent skill for the design-kpis workflow in the XiaomiMiMo/MiMo-Code repository, released under the MIT licence. It helps a product or business team decide which success metrics to track, how to define them, and what targets are reasonable for a decision. This guide explains what an agent skill is, how to install this one, how to ask for help with it, and what it needs before it can give a useful answer.
Design KPIsXiaomiMiMo/MiMo-Code · MIT Download the skill Quick start
- 1Download the ZIP. Download the design-kpis skill ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. In the Claude apps, open Settings, go to Capabilities, then Skills, and upload the skill.
- 4Connect your data sources. Make the data behind the metrics you care about available to the agent.
- 5Describe the decision. Explain the decision, the review cadence and who will act on the result.
- 6Review the recommendation. Check the candidate metrics, definitions, caveats and any provisional targets, and answer the open questions it raises.
What is an agent skill and what does Design KPIs make the agent do?
An agent skill is a folder of instructions that an AI agent loads when a task matches the skill's description. The skill's main file, SKILL.md, tells the agent how to approach that kind of task. Here, the description is: design KPI frameworks, metric definitions, targets, guardrails and measurement plans for product or business decisions.
Once the skill is active, the agent follows a structured process. It clarifies the decision the metrics must support and who will act on them. It gathers business context before recommending anything. It then proposes candidate outcome, driver and guardrail metrics, compares them, and recommends a small set with definitions, calculation methods and caveats.
- Clarifies the decision, operating cadence and measurement constraints
- Generates a wider set of candidate metrics before narrowing
- Recommends 1-3 primary KPIs, 1-2 drivers for each KPI when useful, and 1-2 guardrails when tradeoffs are likely
- Suggests provisional targets with stated assumptions, or asks for the data needed for a firm target
Install the Design KPIs skill
Download the ZIP file from this page and unzip it. You can install the skill in one of two ways. For a personal setup, unzip it into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder. Both locations are where Claude Code looks for skills.
If you use the Claude apps instead, open Settings, go to Capabilities, and upload the skill under Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload
How do you trigger the skill with a natural request?
You do not need a special command. Describe the problem in plain language and mention metrics, targets, guardrails or a measurement plan. The skill's description tells the agent to use it when success metrics, drivers, guardrails, targets or the measurement approach need to be defined or improved.
If you have existing dashboards or conflicting metric definitions to reconcile, the skill does not handle that first. It hands that work to the analyze-data-quality skill, and returns here only when you want to define the metric going forward.
- "Help me choose success metrics for our new onboarding flow launch."
- "What guardrail metrics should we watch if we push more notifications?"
- "Suggest provisional targets for weekly active use of the new feature, based on our baseline."
- "Design a measurement plan for the pricing page redesign."
What does the skill need before it can help?
The skill works best with real data. It refers to semantic layers, catalogues and live reads, so the agent needs access to the data sources behind the metrics you are discussing. If you connect or provide them, the agent searches them and verifies selected data through live reads before it draws conclusions.
The skill also relies on two companion skills, gather-business-context and analyze-data-quality, and can optionally use build-report and visualize-data. Install those alongside this one if you want the full workflow.
- Connected or provided data sources for the metrics under discussion
- Companion skills named in the source: analyze-data-quality and gather-business-context
- Optional companion skills: build-report and visualize-data
Safety notes: scripts and network access
According to the skill's listing, this skill runs no scripts and uses no network access. Its files are agents/openai.yaml and SKILL.md. Its behaviour is set by instructions in SKILL.md, not by code it executes.
The skill's own guardrail matters more in practice. If a required data source is unavailable, the agent is told to stop that path, tell you what is missing, and not substitute weaker data. Any access to your data comes from the sources you connect, not from the skill itself.
- Runs scripts: no
- Uses network: no
- Stops and asks for the source rather than using weaker substitutes
Tips for getting better KPI recommendations
Give the agent the decision first. Say what the metrics will change, who reviews them and how often. Metrics that suit a weekly launch review differ from those suited to an annual planning cycle, and the skill weighs this explicitly.
Expect targets to be provisional unless you have baseline data. The skill advises presenting only directional ranges when the evidence is thin, and it asks for the measurement needed before setting a firm number.
- State the decision, the review cadence and the audience
- Share existing definitions and past decisions so the agent does not repeat them
- Provide baseline data if you want anything beyond a provisional range
- Ask for a visual only when it will clarify tradeoffs; the skill does not build reports by default
What people use it for
- Choosing a small set of success metrics for a new feature or product launch
- Picking guardrail metrics that stop a growth push from hurting quality, trust or cost
- Redesigning a KPI framework for a business team
- Writing a documented measurement plan with definitions, calculation methods and caveats
- Drafting provisional target ranges based on baseline data and stated assumptions
Frequently asked questions
How do I install the design-kpis skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Is the design-kpis skill free?
The source states the project is licensed under MIT. The source does not give pricing for the skill, so check the repository licence and terms before you rely on it commercially.
What do I need to run the design-kpis skill?
You need connected or provided data sources for the metrics you discuss. The skill also refers to companion skills by name: analyze-data-quality and gather-business-context, with build-report and visualize-data as optional extras.
Does the design-kpis skill run scripts or use the internet?
According to the listing, it runs no scripts and uses no network access. Its files are agents/openai.yaml and SKILL.md.
Should I use this skill to reconcile conflicting dashboards?
No. The skill says to use analyze-data-quality first when the task is reconciling existing metrics, dashboards, tables, owners or sources of truth. Come back to design-kpis when you want to define metrics going forward.
Can the skill give me firm numerical targets?
Not without real baseline data. The skill advises presenting only provisional ranges in that case, and asking for the measurement needed before setting a firm target.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you are a product or analytics lead defining success metrics for a feature or launch, a business team redesigning a KPI framework, or an analyst who needs a documented measurement plan with targets and caveats. Pick something else if your main job is reconciling existing dashboards or conflicting metric definitions, which belongs to a data-quality skill, or if you have no access to the data the metrics depend on, since the skill will stop and ask rather than guess.
Download the skill