More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Interview Me: Extract the Real Requirement

    1

    A pre-planning interview that turns a vague ask into a confirmed statement of intent: hypothesis with a confidence number, one question at a time, an explicit yes before anything is built.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Interview Me: Extract the Real Requirement

    Interview Me: Extract the Real Requirement

    Example session with this skill installed

    I keep losing track of freelance design clients. Build me an app for my freelance business.

    • Read your context and instructions
    • Compiled the interview me

    HYPOTHESIS: You want a small client tracker — who's active, what you owe them, what they owe you — and "app" is the conventional word for it.
    CONFIDENCE: 30% — missing: what "losing track" concretely means, whether anyone else uses it, and what would make you call it done.

    Q1: When you say losing track, what's the moment that hurts — a follow-up you forgot, an invoice you never sent, or not knowing who you're mid-project with?
    GUESS: The forgotten follow-up. "Losing track of clients" usually means people, not paperwork. If it's invoices, this becomes a money tool instead, and the design center changes completely.

    Q2 (after the answer points at follow-ups): Is this just for you, or does anyone else — a partner, an assistant — need to see it?
    GUESS: Just you. Solo freelance, most likely. A second user would double the scope with accounts and permissions, so I want this nailed before anything else.

    Q3: What does "it's working" look like in a month — never missing a follow-up, or also seeing your pipeline at a glance?
    GUESS: Never missing a follow-up. The pipeline view is the kind of thing people say they want and never open.

    What I'm deliberately not asking yet: platform, framework, reminders vs. calendar sync. Every one of those depends on the answers above, and asking them now would lock in a framing I have 30% confidence in. Once Q1–Q3 land, I'll restate the intent in your words — outcome, user, why now, success, constraint, and what's explicitly out of scope — and nothing gets built until you confirm it with an actual yes.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    What people ask for and what they actually want are different things, and the cheapest moment to find the gap is before a plan, spec, or line of code exists — after that, everyone rationalizes the wrong build into a good-enough build. This skill is that discovery interview, run as a protocol. It opens with a written hypothesis and an honest confidence percentage, so the gaps are visible before the first question. Questions go one at a time, each with your best guess attached, because reacting to a concrete guess is faster and more honest than generating answers from scratch — and because the third question usually depends on the first answer. It includes a calibration table for what each confidence band actually obliges you to do, probes for spotting "should want" answers (best-practice cosplay, deference phrases, buzzwords standing in for outcomes) and the one question that cuts through them, a restatement template that forces the out-of-scope line most projects skip, and a strict table of what counts as a yes — delegation, politeness, and silence do not. It closes with a hard stop rule: on explicit confirmation you deliver the statement of intent and end the turn, because starting downstream work in the same breath reintroduces every assumption the interview just killed. A complete worked example (the dashboard that turned out to be a list) and an edge-case set for hurried users, conflicting stakeholders, and pivoting answers are included.

    • It requires an interactive counterpart. In automated or scheduled contexts it cannot run — the skill itself says to flag the underspecified ask as a blocker there instead of guessing.
    • It depends on the requester answering honestly; it has strong probes for evasive or best-practice-shaped answers, but a determined non-answer still ends in a flagged blocker, not a requirement.
    • It optimizes for correctness of intent, not speed. For genuinely unambiguous tasks it explicitly tells you to skip it.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 24 days ago

    Permissions

    No special permissions declared or detected

    Listed24 days ago
    Updated21 days ago

    What's inside

    Frequently Asked Questions

    The difference is the machinery: a written hypothesis with a confidence number before you ask anything, guesses attached so answers are reactions instead of essays, a restatement with a mandatory out-of-scope line, and a rule for what counts as a yes. Most "clarifying questions" skip all four and still build on guesses.

    Three to six. The stop rule is predictive, not numeric: you stop when you can predict the requester's reaction to the next three questions you'd ask. If confidence isn't rising after several rounds, the skill tells you to say so and step back.

    That is delegation, not agreement, and the skill treats it as a non-yes: you come back with two concrete options as a choice. Silent agreement and polite exits get the same treatment — the yes has to be explicit.

    New in Productivity