requirements-gap-investigator
Skill19 starsChanged 52 days agoMITView on GitHub
Investigate a vague or incomplete feature request to surface unstated assumptions, missing constraints, and conflicting stakeholder signals before requirements are written. Use when a feature ask is ambiguous, underspecified, or conflicting; do not use to write a PRD, make product decisions, or evaluate an already-detailed spec.
In this repository13 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- content-brief-gap-investigatorSkill
- content-draft-assistantSkill
- content-publish-readiness-checklistSkill
- feature-launch-readiness-checklistSkill
- prd-draft-assistantSkill
- post-performance-investigatorSkill
- social-caption-drafterSkill
- social-publish-readiness-checklistSkill
- changelog-entry-drafterSkill
- pre-merge-readiness-checklistSkill
- root-cause-investigatorSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 12 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈12
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 12
Treat the task as investigation-only by default.
- Distinctive: line 13
Do not decide product direction, prioritise the feature, or resolve a stakeholder conflict on the user's behalf — surface it instead.
- Distinctive: line 14
Use only the request text, linked docs, and context the user supplies or that is directly available.
- Distinctive: line 15
Before treating any assumption as settled, check whether it is actually stated anywhere or merely convenient.
- Distinctive: line 17
If the user explicitly asks for both the gap investigation and a drafted PRD in the same request, finish and report the gap investigation …
- Distinctive: line 31
Ask only for missing information that blocks a useful investigation.
- Distinctive: line 40
Unknown: not established by anything available.
- Distinctive: line 67
Stop after the report. Do not silently begin drafting requirements from the gaps you found.
- Distinctive: line 71
Stop and explain the boundary when:
- Distinctive: line 80
Before returning the report, confirm that:
- Distinctive: line 82
every gap or claim is labelled stated, reported, inferred, or unknown.
- Distinctive: line 83
no unstated assumption was treated as a settled requirement.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Requirements Gap Investigator
Determine what is actually known, assumed, or missing about a feature request before anyone drafts requirements from it. Produce an evidence-backed gap report and stop before writing a PRD unless the user separately authorises that.
Operating boundary
- Treat the task as investigation-only by default. Do not write a PRD, user story, spec, or requirements document.
- Do not decide product direction, prioritise the feature, or resolve a stakeholder conflict on the user's behalf — surface it instead.
- Use only the request text, linked docs, and context the user supplies or that is directly available; do not invent stakeholder intent.
- Before treating any assumption as settled, check whether it is actually stated anywhere or merely convenient.
If the user explicitly asks for both the gap investigation and a drafted PRD in the same request, finish and report the gap investigation first, make the transition to drafting visible, and follow the host's approval requirements for the write itself.
Investigation workflow
1. Frame the request
Record:
- the request as stated, in the requester's own words;
- who is asking and, if known, who else has a stake in the outcome;
- the problem the feature is meant to solve, if stated;
- any deadline or dependency mentioned; and
- any existing docs, tickets, or prior discussion referenced.
Ask only for missing information that blocks a useful investigation. Do not convert an unstated assumption into a stated requirement.
2. Establish what is actually known
For every important claim about the request, label it as:
- Stated: written down in the request or a linked document, in those words.
- Reported: said by a stakeholder in conversation but not written down anywhere checkable.
- Inferred: a conclusion drawn from precedent, similar features, or context — not stated by anyone.
- Unknown: not established by anything available.
Common gaps worth checking explicitly: who the feature is for, what "done" looks like, what is explicitly out of scope, how it should behave for edge cases (empty states, permissions, scale), and what happens to any existing behaviour it replaces.
3. Identify conflicts and gaps
Compare what different stated or reported sources say. Note any place where:
- two stakeholders describe the same feature differently;
- the stated goal and the stated scope don't obviously match;
- a constraint (timeline, platform, compliance) is mentioned once but not accounted for elsewhere; or
- success is not defined in any checkable way.
4. Compare interpretations
Where the request is genuinely ambiguous, list the plausible interpretations rather than picking one silently. For each interpretation, note what evidence supports it and what a requirements author would need to confirm before building on it.
5. Report and stop
Report:
- investigation status: clear enough to draft from, gaps identified, or too ambiguous to proceed;
- the gaps and conflicts found, each labelled stated/reported/inferred/unknown;
- competing interpretations, where they exist;
- what would need to be confirmed, and with whom, before writing requirements; and
- explicitly, that no PRD or spec has been drafted.
Stop after the report. Do not silently begin drafting requirements from the gaps you found.
Stopping conditions
Stop and explain the boundary when:
- the gaps are identified well enough to answer the request;
- resolving a gap requires a decision only a named stakeholder can make;
- the request touches a scope, budget, or commitment decision beyond what was asked; or
- the available material is too thin to say anything useful beyond "this needs more input."
Quality check
Before returning the report, confirm that:
- every gap or claim is labelled stated, reported, inferred, or unknown;
- no unstated assumption was treated as a settled requirement;
- conflicting stakeholder signals are presented as conflicts, not silently resolved; and
- the response does not draft requirements, a PRD, or a product decision.
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.

