agentsclimarketplace

Extract tacit knowledge

Skill 11Yuxuanyang/polanyi-stack/skills/extract-tacit-knowledge

The Polanyi stack for AI agents. Convert tacit knowledge, Polanyis paradox, and expert judgment into reusable agents and skills.

Install
npx -y skills add 11Yuxuanyang/polanyi-stack --skill extract-tacit-knowledge

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Extract tacit knowledge from expert performance using think-aloud, observation, and critical incident interviews. Use when the user wants to uncover implicit judgment, hidden cues, or '会做但说不清' know-how and turn it into transferable artifacts.

SKILL.md

2.5 KB, as published. Nobody here has run it

Extract Tacit Knowledge

Core judgment: do not start from theory. Start from performance under real constraints.

Use When

  • The user mentions tacit knowledge, implicit know-how, intuition, or "we know more than we can tell"
  • The goal is to turn expert judgment into a checklist, decision tree, rubric, or agent behavior
  • A team can see good performance but cannot explain what makes it good

Do First

Define the target narrowly:

  • one role
  • one task
  • one high-value situation
  • one acceptance standard

If the scope is broad, compress it before doing anything else.

Evidence Standard

Do not abstract too early. Ask for or reconstruct:

  • one real walkthrough
  • one success incident
  • one failure or near-miss incident
  • three representative artifacts, cases, or outputs

If none of this exists, say so and treat the result as a first draft, not a finished method.

Workflow

1. Capture Performance

Use one or more:

  • think-aloud narration during execution
  • direct observation or screen recording
  • critical incident interview after a success or failure

If you need prompts, read references/interview-prompts.md.

2. Extract the Hidden Layer

Pull out five things:

  • trigger signals: what changes the expert's move
  • decision cues: what they are actually noticing
  • embodied actions: what they do automatically and may forget to mention
  • failure signatures: what usually appears before a mistake
  • escalation boundaries: when normal handling stops and handoff begins

3. Repackage Only After Extraction

Convert the findings into one or more:

  • cue list
  • checklist
  • decision tree
  • failure rubric
  • escalation rule

4. Validate on a Fresh Case

Test the artifact on an unseen case with a novice, peer, or agent.

Success means:

  • they can act, not just restate
  • they notice the right cues
  • they know when not to continue

Output

Return a compact package with:

  • task definition
  • cue list
  • trigger list
  • failure signatures
  • escalation boundary
  • recommended transfer format

Do Not

  • confuse the expert's explanation with the expert's real decision process
  • turn one anecdote into a universal rule
  • ship a polished framework without at least one failure case

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.