agentleFS
wakqasahmedwakqasahmed/ai-engineering-workflow-skills/skills/engineering/clarify-work/SKILL.md

clarify-work

Skill2 starsChanged 2 months agoMITView on GitHub

Clarify non-trivial engineering work before implementation by resolving ambiguity, terminology, constraints, and the smallest viable path. Use when a request is broad, fuzzy, architectural, cross-cutting, or has hidden product decisions.

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

8 rules. About 0 are common, 7 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this before implementation when the request is not already issue-sized.

    line 8
  2. Distinctive:

    Keep only unknowns that block a safe, issue-shaped execution boundary.

    line 14
  3. Distinctive:

    If the estimate is zero, skip the interview and continue to the execution boundary.

    line 15
  4. Very specific:

    Ask exactly one blocking question per user turn, then wait for the answer.

    line 16
  5. Distinctive:

    When no blocking questions remain, identify the planning track: quick, standard, or deep.

    line 18
  6. Distinctive:

    Do not plan around hypothetical future requirements.

    line 32
  7. Distinctive:

    Do not start implementation while key terms or ownership boundaries are unclear.

    line 34
  8. Distinctive:

    When project terminology or documented decisions change during clarification, record the outcome in the project's domain docs (glossary, …

    line 35

≈ “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

Clarify Work

Use this before implementation when the request is not already issue-sized.

Workflow

  1. Restate the outcome in one sentence.
  2. Name assumptions and unknowns that can change the implementation.
  3. Keep only unknowns that block a safe, issue-shaped execution boundary. Estimate the initial number of blocking questions from all available context before asking any. Prefer 1-3 by omitting non-blocking questions, not by combining unrelated decisions.
  4. If the estimate is zero, skip the interview and continue to the execution boundary.
  5. Ask exactly one blocking question per user turn, then wait for the answer. Prefix each planned question with Question n/N · [bar], where N is the initial estimate. Render the bar with exactly N cells: n filled ■ cells followed by N - n empty □ cells.
  6. After each answer, reassess the current blocker and the remaining blockers. Treat an incomplete or ambiguous answer, or a newly discovered blocker, as a follow-up labeled Follow-up k (after Question n/N) · [bar]. Keep the original N; do not advance n or consume the next planned question until the current blocker is resolved. Render the follow-up bar at position n, without changing N.
  7. When no blocking questions remain, identify the planning track: quick, standard, or deep.
  8. State the smallest viable approach and non-goals.
  9. Convert broad work into issue-shaped execution units when needed.

Output

  • Outcome
  • Assumptions
  • Open questions
  • Suggested execution boundary
  • Verification signal

Guardrails

  • Do not plan around hypothetical future requirements.
  • During the interview, include only the current question and context needed to answer it.
  • Do not start implementation while key terms or ownership boundaries are unclear.
  • When project terminology or documented decisions change during clarification, record the outcome in the project's domain docs (glossary, ADRs) before implementation starts.

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.