prep-req
Skill0 starsChanged 15 days agoMITView on GitHub
Turn rough product ideas, business requests, and task discussions into concise, reviewable requirements covering the problem, desired outcome, expected behavior, scope, and acceptance criteria. Use before technical planning or when refining business-facing feature requirements. Not for architecture or implementation plans.
In this repository13 other files it gives its agents.
- AGENTS.mdAGENTS.md
- big-pictureSkill
- challengerSkill
- implementation-planSkill
- interviewSkill
- junior-delegateSkill
- checkpointSkill
- codeSkill
- commitSkill
- verify-code-reviewSkill
- verifySkill
- agents-add-taskSkill
- prompt-builderSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
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
- Distinctive: line 14
Read the request, relevant discussion, existing product documentation, business rules, and provided designs.
- Distinctive: line 21
Look past a suggested feature or UI element to understand the underlying need, without disregarding an explicitly agreed product decision.
- Distinctive: line 24
Keep business constraints, including relevant compliance, timing, or user-experience requirements.
- Distinctive: line 39
Write in Polish by default unless another language is requested.
- Distinctive: line 43
When asked for analysis only, provide the analysis without editing files.
- Very specific: line 45
Stop at the product boundary.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
Prepare requirements
Turn the available context into a business/product requirement that stakeholders can validate and engineers can use for technical planning.
Describe why and what, not how to implement it. Write only as much as needed to remove meaningful ambiguity.
Understand the request
- Read the request, relevant discussion, existing product documentation, business rules, and provided designs. Inspect the repository only when necessary to verify current behavior or material constraints.
- Identify who is affected, what problem exists today, what should change, and why it matters. Do not invent evidence, benefits, or metrics.
- Preserve agreed decisions and the project's domain terminology. Distinguish confirmed facts, explicit requirements, proposals, assumptions, and open decisions; never silently turn one into another.
- Explain relevant current behavior from the user's or business process's perspective. Consider other affected roles and downstream outcomes when material.
Shape the requirement
- Look past a suggested feature or UI element to understand the underlying need, without disregarding an explicitly agreed product decision.
- Clarify expected behavior, important scenarios, business rules, permissions, exceptions, and boundaries. Include only details that materially affect the outcome.
- State non-goals where they prevent scope creep. Do not expand the request with speculative enhancements or future-proofing.
- Keep business constraints, including relevant compliance, timing, or user-experience requirements. Leave architecture, API contracts, data schemas, module design, test organization, and task breakdowns to technical planning.
- Draft from the available evidence first. Ask only focused questions whose answers would materially change the requirement. Do not decide unresolved product questions on the user's behalf; record them with their impact.
Write the requirement
Adapt the structure to the size of the change. A useful default is:
- Problem and outcome — affected users, current problem, desired result, and reason for the change.
- Expected behavior — main scenarios, business rules, intended experience, and design references when available.
- Scope and constraints — what is included, what is excluded when relevant, and material restrictions or dependencies.
- Acceptance criteria — outcomes that can be checked from a user or business perspective, without prescribing technical implementation.
- Open decisions — only material questions still requiring a product or business decision, and what each decision changes.
Merge or omit sections that add no value. Use a small example, decision table, or user flow only when it clarifies complicated behavior. User stories and success metrics are optional; do not manufacture them to fill a template.
Write in Polish by default unless another language is requested. Use plain, concrete language and preserve exact domain terms and identifiers. Do not repeat the same rule across sections.
Deliver and check
- When asked for analysis only, provide the analysis without editing files. When asked to create or revise a document, use the requested location or existing repository convention; update the existing document rather than creating a competing version.
- The result should let someone unfamiliar with the conversation answer: Why is this needed? Who is affected? What exactly must happen? What is outside scope? How will we recognize an acceptable result? What remains undecided?
- Stop at the product boundary. The
implementation-planskill handles the technical approach, responsibilities, flows, and architectural consequences.
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.

