agentsclimarketplace

Convert tacit to skill

Skill hostile-shorepatrol81/polanyi-stack/skills/convert-tacit-to-skill

Convert tacit expert knowledge into reusable AI agents, skills, and training workflows to scale reliable performance.

Install
npx -y skills add hostile-shorepatrol81/polanyi-stack --skill convert-tacit-to-skill

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

Convert tacit knowledge into a reusable agent or Codex skill by extracting cues, scoping the task, defining boundaries, and packaging the workflow into SKILL.md plus tests. Use when the user wants a fixed agent, reusable skill, or repeatable copilot based on expert judgment.

SKILL.md

2.9 KB, 603 tokens by cl100k_base, as published. Nobody here has run it

Convert Tacit To Skill

Core judgment: do not build a broad agent from vague expertise. Build a narrow agent from repeated expert performance.

Use When

  • The user wants to turn tacit knowledge into a reusable agent or skill
  • The user wants a fixed copilot for one recurring type of judgment or task
  • A human expert performs well, but the behavior is not yet packaged for reuse

Target Output

Prefer a single-purpose artifact:

  • one agent
  • one task family
  • one primary user request shape
  • one clear escalation boundary

If the request is broad, split it into multiple future skills instead of one bloated skill.

Workflow

1. Scope the Agent

Define:

  • role: what the agent is responsible for
  • task: what recurring task it performs
  • scene: the high-value context where it helps
  • success standard: what a good output looks like
  • non-goals: what it must not try to do

2. Extract the Tacit Layer

Use real cases, walkthroughs, or observed performance.

Pull out:

  • trigger phrases
  • decision cues
  • default sequence
  • failure signatures
  • escalation boundary

If the tacit layer is still vague, switch to extract-tacit-knowledge.

3. Choose the Right Packaging

Pick one:

  • Codex skill: for reusable guidance, procedural structure, references, scripts, and trigger metadata
  • fixed prompt or copilot spec: for a narrower internal agent behavior
  • human-agent hybrid workflow: when the agent should stop at synthesis and hand off decisions

Default to Codex skill if the task recurs and can benefit from supporting files.

4. Build the Skill Skeleton

Define:

  • skill name
  • frontmatter name and description
  • use when triggers
  • step-by-step workflow
  • required references
  • scripts if determinism is needed
  • output contract
  • do-not-do rules

If you need a format, read references/skill-blueprint.md.

5. Add Validation

Before calling it reusable, create:

  • three trigger examples
  • three non-trigger examples
  • two success cases
  • one failure or escalation case

If you need a format, read references/skill-test-template.md.

6. Decide Install Shape

Use one of:

  • repo-local only
  • symlink into ~/.codex/skills
  • copy into shared skills directory

Prefer symlink when the source repo should remain the source of truth.

Output

Return:

  • recommended skill name
  • scope statement
  • cue and boundary summary
  • SKILL.md outline
  • required references or scripts
  • test cases
  • install recommendation

Do Not

  • package a personality instead of a task
  • hide unclear boundaries inside vague language
  • claim reusability without trigger tests and escalation tests

What ships with it: 2 files

778 B alongside SKILL.md

references/

Gives 0 of the 12 instructions most context ai engineering skills give in 603 tokens

Counted across 1,193 of the 1,976 authors here whose files we hold, read 2026-08-07

  • Dispatch a fresh implementer subagent per taskin 48 of 1193, across 19 files
  • Dispatch a final code reviewer after all tasksin 33 of 1193, across 8 files
  • Provide full task text to the subagentin 30 of 1193, across 9 files
  • Review spec compliance before code qualityin 27 of 1193, across 10 files
  • Make the hook script executablein 26 of 1193, across 8 files
  • Re-snapshot after navigation or DOM changesin 25 of 1193, across 19 files
  • Read files before editing themin 22 of 1193, across 11 files
  • Answer subagent questions before proceedingin 22 of 1193, across 7 files
  • Mark task complete in TodoWrite after approvalin 22 of 1193, across 6 files
  • Merge hook into existing settingsin 21 of 1193, across 3 files
  • Ask if installation is global or projectin 20 of 1193, across 2 files
  • Copy the hook script to target locationin 20 of 1193, across 2 files

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

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.