engflow-clarify
Skill0 starsChanged 17 days agoMITView on GitHub
Explore goals, challenge assumptions and resolve consequential requirement ambiguity through focused discussion. Use on explicit clarification or deep interview request, from an early idea or an existing specification. Return decisions and proposed changes in conversation only; read existing sources but do not create or update files.
In this repository13 other files it gives its agents.
- AGENTS.mdAGENTS.md
- diagram-aestheticsSkill
- engflow-code-reviewSkill
- engflow-designSkill
- engflow-handoffSkill
- engflow-implementSkill
- engflow-planSkill
- engflow-pocSkill
- engflow-refactor-assessSkill
- engflow-researchSkill
- engflow-specifySkill
- engflow-tasksSkill
- semver-bumpSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 6 are distinctive, and 0 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
Start from the available discussion, specification and relevant code.
- Distinctive: line 19
When discussing an existing feature, use its supplied path/ID or clearly matching Spec and references.
- Distinctive: line 33
Ask the smallest useful set of questions, grounded in what is already known.
- Distinctive: line 34
Prefer one consequential question at a time when answers determine the next.
- Distinctive: line 41
Keep decision certainty separate from delivery scope: a confirmed requirement may be deferred and excluded from current acceptance.
- Distinctive: line 82
No other Skill is required for discussion.
≈ “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
Engflow Clarify
Help the user reach decisions that materially change what should be built. Start from the available discussion, specification and relevant code. No Spec, Design or other Skill is required. Continue follow-up discussion within the active task until the user pauses, ends or redirects it.
When discussing an existing feature, use its supplied path/ID or clearly matching Spec and references. If several feature directories fit, ask which one; do not assume the newest directory or current branch identifies the target. Keep the summary tied to that feature's source paths; no directory or artifact is created.
Explore or clarify
For an early idea, establish the problem, intended users, useful outcomes, first-delivery scope, constraints and consequential tradeoffs. Challenge the assumptions behind the request with concrete scenarios and alternatives. For an existing specification, focus on material gaps, contradictions, edge behavior and acceptance ambiguity; do not restart discovery without cause. Adapt as the discussion develops rather than making the user select a mode.
Ask the smallest useful set of questions, grounded in what is already known. Prefer one consequential question at a time when answers determine the next. Offer a recommendation and its tradeoff when useful; do not turn supported, ordinary choices into an interview. Accept explicit user decisions as premises. Revisit settled decisions only for new evidence or a concrete conflict, explaining why. Never invent answers to complete a questionnaire.
Distinguish requirements, preferences, recommendations and unverified assumptions. Keep decision certainty separate from delivery scope: a confirmed requirement may be deferred and excluded from current acceptance. Preserve why it was deferred and any supplied revisit condition; do not silently turn it into current work. Separate a requested deadline from feature semantics and execution permission. Resolve relevant relative dates to calendar dates in the summary; do not schedule or promise future execution from a planning statement. See judgment examples when requirement certainty, delivery scope and execution authority diverge.
Technical feasibility needs source evidence or separately authorized experiments, not user agreement. Investigate relevant facts read-only when accessible; otherwise state the uncertainty and what would resolve it. Leave solution selection to technical design unless the user has already fixed a technical constraint.
Outcome and continuity
Finish when important intent is clear enough for the next useful activity, not when every possible detail has been discussed. At meaningful discussion boundaries, return a proportional decision/change summary in conversation: confirmed decisions, current/deferred scope, unresolved questions and proposed changes. Reference existing requirement IDs and affected acceptance, constraints and downstream paths where known; do not invent IDs or force the discussion into a complete Spec template.
Preserve supplied approval evidence and pending gates in the summary. Distinguish agreed changes from proposals and identify sources still awaiting synchronization; user agreement does not mean those files were updated. Do not restart settled questions or hand off after every answer merely to follow a stage sequence.
Specification handoff and boundaries
engflow-specify owns specification synthesis, consistency checks, previews and
authorized writes. Supply the discussion or decision summary as input, not a second
interview ledger. It can resolve ordinary missing details without requiring Clarify.
This Skill is read-only even when asked to save its output or edit a Spec. Return the content and identify Specify as the next capability; do not write files or silently invoke it. If the user explicitly requests clarification followed by Spec creation/update, finish the clarification summary first. The current agent may then perform the separately scoped Specify work through permitted host invocation or composition, retaining valid write authority. If the host needs a separate explicit invocation, explain that next step rather than bypassing it or claiming a save.
No other Skill is required for discussion. Do not create reports or scratch files, run experiments, modify code or documents, manage branches, commit or publish. Any helpers remain read-only and return findings to the conversation.
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.

