socratic-problem-framing
Skill2 starsChanged 2 months agoMITView on GitHub
Clarify an ambiguous or possibly misframed problem through a bounded Socratic sequence. Use when one missing user fact could change the route, terms or goals are unclear, or the user explicitly asks to be questioned before advice. Skip simple facts, fixed-scope execution, emergencies needing immediate containment, and questions answerable from available artifacts.
In this repository9 other files it gives its agents.
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Use when the request mixes facts, interpretations, values, and goals.
- Distinctive: line 19
Ask only the unknown most likely to change the route.
- Distinctive: line 20
After each answer, state in one sentence what was confirmed, weakened, or reframed.
- Distinctive: line 21
Check vague terms, hidden premises, evidence sources, credible counterexamples, and conclusion-flipping variables.
- Distinctive: line 22
Stop as soon as the frame is decision-ready.
- Very specific: line 28
Use $fact-claim-audit when external claims need verification.
- Very specific: line 29
Use $bidirectional-steelman only after there is a contestable judgment.
- Distinctive: line 30
Do not ask for information already present in supplied files or tool state.
≈ “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.
Files
Socratic Problem Framing
Own problem definition and uncertainty reduction. The active domain skill still owns facts, execution, and the eventual recommendation.
Route
Use when the request mixes facts, interpretations, values, and goals; key terms are vague; or one unavailable user fact could change the method or conclusion. Also use when the user explicitly says to ask first and not advise yet.
Skip direct lookups, translations, deterministic work, complete specifications, and questions whose missing details cannot change the direction. In an active incident, give the minimum reversible containment first, then clarify.
Procedure
- Restate the current concern and separate observed facts, interpretations, value judgments, desired outcomes, and unknowns.
- Ask only the unknown most likely to change the route. Ask one question per turn.
- After each answer, state in one sentence what was confirmed, weakened, or reframed.
- Check vague terms, hidden premises, evidence sources, credible counterexamples, and conclusion-flipping variables.
- Stop as soon as the frame is decision-ready; never ask questions to fill a quota. Six turns is a ceiling, not a target.
- Return the initial question, the real question, confirmed facts, unverified assumptions, key variables, and one precise next question or handoff.
Handoffs and Boundaries
- Hand the clarified frame to the relevant domain skill.
- Use
$fact-claim-auditwhen external claims need verification. - Use
$bidirectional-steelmanonly after there is a contestable judgment; do not manufacture opposing camps during framing. - Do not ask for information already present in supplied files or tool state.
Completion Gate
The reframed question names the object, desired outcome or decision, binding constraints, and the variable that matters. Facts and assumptions are separate, and every question asked had decision value.
Source Status
Distilled from the user-supplied “苏格拉底提问” method card. The card provides a proposed prompt structure, not empirical evidence that six questions or this method improves outcomes.
<!-- integration-watermark: 红猴子整合 -->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
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.

