Guide · Agent skill
How to use the Office Hours Product Validation agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The office-hours skill is a set of instructions that tells an AI agent to act as a skeptical, YC-style product advisor. Before a feature idea becomes a requirement, the agent works through six forcing questions and checks whether there is real demand, a clear wedge, and evidence behind the assumptions. This guide explains what the skill does, how to install it, how to trigger it, and how to read its output.
Office Hours Product Validationmvschwarz/openrig · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Download the office-hours ZIP from this page. It contains SKILL.md.
- 2Unzip into a skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder for one project.
- 3Or upload in the Claude apps. Alternatively, go to Settings → Capabilities → Skills in the Claude apps and upload the ZIP.
- 4Describe the feature idea. Write two or three sentences about the feature with no jargon or implementation details, then ask the agent to run office hours on it.
- 5Answer the six questions. Respond to each forcing question with specific customers, workflows, and evidence. Expect follow-up probes on vague answers.
- 6Review validation.md. Check the saved validation.md in the feature folder for the GO, REFINE, or PAUSE verdict and the listed next steps.
What the office-hours skill makes the agent do
An agent skill is a folder containing a SKILL.md file with instructions the agent loads when a task matches. This skill's SKILL.md tells the agent to interview you about a feature idea. You describe the idea in two or three sentences, with no jargon or implementation details. The agent then works through six questions in order: Demand Reality, Status Quo, Desperate Specificity, Narrowest Wedge, Observation/Surprise, and Future-Fit.
After each answer, the agent summarizes what it heard, rates the answer as Strong, Needs Work, or Red Flag, and asks follow-up questions if the answer is vague. The skill is built to find gaps rather than generate ideas. It pushes for specifics such as named customers, meetings, or support tickets.
Install the office-hours skill
You can install the skill in either of two ways. Download the ZIP from this page and unzip it into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only in that project. Alternatively, upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
The package contains a single file, SKILL.md. Make sure the unzipped folder contains that file so the agent can find the instructions.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
How to trigger the skill with a natural request
You do not need a special command. Describe what you want in plain language that matches the skill's purpose, such as asking the agent to pressure-test a feature idea before you write requirements, or to run office hours on an idea. The skill's description focuses on pressure-testing a feature idea before it becomes a requirement, so requests that mention that goal are the clearest match.
Have your evidence ready. The skill looks for real conversations, so if you have meeting notes, customer records, or support tickets, mention where they are. The agent is designed to search meeting notes and customer records for actual conversations.
- Pressure-test this feature idea before I write the requirements: [two or three sentences]
- Run office hours on this idea and tell me whether the evidence supports building it
- Check whether customers actually asked for this, using our meeting notes
Walk through the six questions
Each question builds on the previous answer. Demand Reality asks who is actively asking and what evidence exists. Status Quo asks how people solve the problem today and why that is not good enough. Desperate Specificity asks for the single most desperate user and their day. Narrowest Wedge asks for the smallest version that would solve that user's problem. Observation/Surprise asks what you learned that you did not expect. Future-Fit asks what the feature unlocks and what you learn if it fails.
Answer with specifics. The skill treats an assumption as a gap, so if you say customers want something, expect a follow-up asking which customers and when they said it.
Read the validation.md output
At the end, the agent saves an assessment to the feature folder as validation.md. The file has YAML front matter with a verdict, date, PM name, and feature folder name, followed by a section for each of the six questions with its rating and a summary of evidence. It closes with a verdict and next steps.
The verdict is GO, REFINE, or PAUSE. GO means strong evidence and a clear wedge, so proceed to requirements. REFINE means there is promise but gaps, and the file lists what research to do first. PAUSE means there is insufficient evidence of demand, and the file says what evidence would change that. Even a PAUSE should explain what would change the verdict.
- GO: strong evidence, clear wedge, proceed to requirements
- REFINE: promise but gaps; do more research first (the file lists what)
- PAUSE: insufficient evidence of demand; revisit when evidence emerges
Safety notes: scripts and network access
This skill does not run scripts and does not use the network, according to its listing. It consists of instructions in SKILL.md. It lists no required dependencies. Because the skill reads meeting notes and customer records to find evidence, check that the agent only has access to the material you intend it to use before you start.
- Runs scripts: no
- Uses network: no
- Requires: nothing listed
- Files: SKILL.md
Tips for better results
Bring evidence before you start. The skill's output depends on what it can find. Without meeting notes, customer records, or research, the output will mostly flag gaps and will be thin.
Describe the idea without implementation details. The skill asks for two or three sentences with no jargon, so keep the pitch to the problem and the user.
Expect scope to shrink. The skill is deliberately narrowing, and the narrowest wedge is treated as the most valuable question. A REFINE or PAUSE is useful output, not a failure.
What people use it for
- Deciding whether a feature idea deserves building before anyone writes requirements
- Checking whether a feature request is backed by real customer conversations or only by assumptions
- Scoping a feature down to the smallest version that would help one specific user
- Preparing a structured, skeptical review of a founder's product idea
Frequently asked questions
Is the office-hours skill free?
The source lists the licence as Apache-2.0, which governs use of the repository the skill comes from. The listing does not state any price, so check the repository's licence file for terms.
What do I need to use the office-hours skill?
The listing says it requires nothing. It runs no scripts and uses no network. You do need an agent environment that supports skills, and meeting notes or customer records if you want the evidence checks to be useful.
Where do I install the office-hours SKILL.md?
Unzip the downloaded ZIP into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
Why does the skill keep saying my evidence is weak?
The skill separates what customers actually said from what the team assumes. If you describe an assumption without a named customer, meeting, or ticket, it rates the answer as weak and asks for specifics. Bringing real notes or records gives it more to work with.
Can I use office-hours for brainstorming?
It is not suited to that. The skill is deliberately critical and narrowing rather than expansive, so use a brainstorming approach first and then run office-hours on the ideas you want to test.
What is the difference between REFINE and PAUSE?
REFINE means there is promise but gaps, and you should do more research first, with the file listing what to research. PAUSE means there is insufficient evidence of demand, and you should revisit when evidence emerges. The skill says even a PAUSE should explain what evidence would change it.
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 office-hours if you are a product manager, team, or founder who wants skeptical, evidence-based scoping before writing requirements. Skip it if you have no meeting notes, customer records, or research to draw on, or if you want open-ended brainstorming, since its output will be thin or narrowing in those cases.
Download the skill