agentleFS
robertbradley-ossrobertbradley-oss/skills/knowledge-interview/SKILL.md

knowledge-interview

Skill2 starsChanged 2 months agoMITView on GitHub

Interview someone to turn their practical expertise into a usable procedure, guide, or onboarding document. Use when knowledge must be elicited; ordinary rewriting of supplied notes does not need an interview.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    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:

    Ask the expert to walk through a recent concrete example.

    line 14
  2. Distinctive:

    Ask one focused question at a time by default, or use the user's preferred interview format.

    line 18
  3. Distinctive:

    Stop when the reader can follow the central task and important exceptions are covered, when the user pauses, or when remaining gaps require …

    line 38

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Knowledge interview

Help another person perform the work using knowledge that currently lives in the expert's head. Adapt the interview to the reader and the task; do not require a repository or documentation framework.

Start with a real task

Use the conversation and supplied materials to establish what the intended reader should be able to do, their starting knowledge, and the situation in which they will use the guide. Ask only for consequential missing context. If no destination is specified, maintain the working draft in chat; save it when requested or when file creation is already part of the task.

Ask the expert to walk through a recent concrete example. Follow their account before imposing a structure. Look for inputs, triggers, choices, checks, exceptions, and the evidence that tells them the work is finished.

Elicit the judgment behind the steps

Ask one focused question at a time by default, or use the user's preferred interview format. Choose the next question from the largest gap in the reader's ability to act, not a fixed questionnaire.

Useful probes include:

  • What did you notice that made you choose that step?
  • What would make you do something different?
  • What does a good result look like, and how do you check it?
  • Where do beginners get stuck or make a plausible mistake?
  • What happens when the usual input, tool, or person is unavailable?

Distinguish what the expert actually does from what they suggest, assume, or wish the process did. Preserve consequential uncertainty and disagreements between contributors. Attribute conflicting accounts instead of silently combining them into a new policy. Treat the expert's account as source testimony; verify external or current factual claims when needed for the requested guide.

Build and check the guide

As answers arrive, incorporate their stable content into a coherent draft. Preserve existing document organization and unrelated material. Keep confirmed steps separate from unresolved questions; do not invent thresholds, owners, deadlines, or approvals.

Choose the form that serves the reader: a procedure for repeatable work, a decision guide for branching judgment, or an onboarding explanation for unfamiliar context. Include prerequisites, examples, exception handling, and completion checks only where they help someone perform the task.

Walk a representative example through the draft using only information the reader would have. Check a meaningful exception when the process has one. Whenever the walkthrough requires unstated expertise, ask a targeted follow-up or label the missing instruction. Do not call this a successful user trial unless another person actually performed it.

Stop when the reader can follow the central task and important exceptions are covered, when the user pauses, or when remaining gaps require someone else's knowledge. Deliver the usable draft plus material open questions. Do not continue interviewing merely to fill sections.

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.