agentleFS
agentdynarqagentdynarq/founder-skills/skills/customer-discovery/SKILL.md

customer-discovery

Skill0 starsChanged 4 months agoMITView on GitHub

Design customer discovery interviews that produce real signal instead of polite lies, and synthesize transcripts into patterns. Use when the user wants to validate an idea, write customer interview questions, plan user research, or analyze interview notes or transcripts.

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 9 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run discovery the Mom Test way: talk about the customer's life, not your idea.

    line 8
  2. Distinctive:

    Ask what problem space and who they plan to interview, then generate a 25-30 minute script:

    line 17
  3. Distinctive:

    Ask about specific past events, never hypothetical futures.

    line 26
  4. Distinctive:

    Never mention the user's solution or idea during the interview.

    line 27
  5. Distinctive:

    Every compliment is a dodge.

    line 29
  6. Distinctive:

    Prefer "talk me through it" over yes/no questions.

    line 30
  7. Distinctive:

    Read all notes or transcripts provided, then report:

    line 34
  8. Distinctive:

    Never let the user count "they said they loved it" as validation.

    line 43
  9. Distinctive:

    If every interview confirms the idea perfectly, flag it: the questions were probably leading.

    line 45

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Customer Discovery

Run discovery the Mom Test way: talk about the customer's life, not your idea. People lie to be nice; past behavior does not.

Two modes

Script mode: the user is preparing interviews. Produce a script. Synthesis mode: the user has notes or transcripts. Extract patterns.

Script mode

Ask what problem space and who they plan to interview, then generate a 25-30 minute script:

  1. Context (5 min): their role, their normal week, where the problem space fits in it
  2. Past behavior (15 min): the core. "Walk me through the last time you..." questions only
  3. Stakes (5 min): what the problem costs them, what they have already tried or paid for
  4. Close (2 min): "Who else deals with this?" for referrals

Question rules:

  • Ask about specific past events, never hypothetical futures. "When did this last happen?" not "Would you use...?"
  • Never mention the user's solution or idea during the interview.
  • "Would you pay for X?" is banned. "What have you paid to solve this?" is the replacement.
  • Every compliment is a dodge. Script a follow-up that returns to specifics.
  • Prefer "talk me through it" over yes/no questions.

Synthesis mode

Read all notes or transcripts provided, then report:

  • Patterns: pains mentioned by 3 or more people, with quote evidence
  • Workarounds: what people already do or pay for (strongest buying signal)
  • Non-signals: compliments, hypothetical enthusiasm, feature requests with no pain behind them. List these explicitly so the user does not count them as validation.
  • Verdict: does the evidence show a problem people already spend time or money on? Yes, no, or not enough interviews yet (fewer than 5 is not enough).

Rules

  • Never let the user count "they said they loved it" as validation.
  • Commitment is the only currency: time booked, money paid, reputation staked (an intro).
  • If every interview confirms the idea perfectly, flag it: the questions were probably leading.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.