speckit-elicit

User Input

$ARGUMENTS

Why this exists

spec-challenger reads a spec cold and returns ambiguities. That is one lens, applied once. BMAD's elicitation is a different move: take an artifact you have already accepted and re-run it through a named method, because the value is in the constraint — "what would break this" produces different output from "read it again carefully", and naming the method is what stops the second reading from being the first reading with more words.

How to run it

Read the artifact in full. Then pick two or three methods — not all of them; a sweep produces volume, not insight — and apply each one separately, saying which method produced which finding.

Method	The question it forces
Inversion	What would have to be true for this to be the wrong thing to build?
Failure premortem	It is six months later and this failed. Write the one-sentence cause.
Edge enumeration	Empty, one, many, huge, malformed, concurrent, interrupted. Which does the artifact not say anything about?
Reader substitution	Read it as someone who has never seen this codebase. Where do they guess?
Constraint removal	Drop the tightest constraint. Does a simpler design appear? If so, is the constraint real?
Cost of being right	Assume every requirement is correct. What does the full implementation cost, and is any of it disproportionate?
Second-order	The change ships and is used. What does it make easy that nobody asked for?
Silence audit	What does the artifact not mention that a reader would expect it to? Absence is the hardest thing to notice.

Output

For each method: the method name, then at most three findings, each one actionable. A finding says what to change, not that something is "worth considering."

Close with one line: which finding you would act on first, and which you would drop. Then stop — this proposes; /speckit-clarify encodes answers into a spec, /speckit-correct-course handles a changed intent, and neither is this command's job to do unasked.

When not to use it

• The artifact does not exist yet — write it first; there is nothing to re-examine.
• Nobody has read it once. This is a second pass, not a substitute for a first.
• The decision is already made and is not being reopened. Elicitation that cannot change anything is theatre.