Requirement Refinery
Turn vague requests into scoped work packets with clear acceptance checks and handoffs.
$9
Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+22 moreClaude CodeCursorCodex CLIMuseDots
OpenClaw
Manus
Grok
Grok Bot+18 agentsRequirement Refinery
Example session with this skill installed
req-refine refine: Our support team has four people and tracks customer requests in a shared spreadsheet. We often reply to the same request twice or forget who owns it. I want a simple way to track ownership and status. Keep the existing spreadsheet, do not purchase tools, and do not change live data. Prepare a brief we can hand to a teammate.
- Read your context and instructions
- Compiled the requirement refinery
Execution packet
packet_version: 1
mode: refine
framework: Requirement refinement. The immediate need is to settle the ownership and status rules before choosing a spreadsheet change.
status: needs-input
Goal
Give the four-person support team a clear owner and status for each customer request, reducing duplicate replies and unassigned work.
Deliverables
A brief for a teammate to prepare a draft of the existing spreadsheet with proposed ownership and status fields, operating rules, and a verification checklist. The spreadsheet destination has not been supplied.
Facts
All facts below come from the supplied request:
- Four people handle support.
- Requests are tracked in a shared spreadsheet.
- Duplicate replies and unclear ownership occur.
- The existing spreadsheet must remain in use.
- Buying tools and changing live data are excluded.
Assumptions
- Propose one accountable owner per request. If shared ownership is required, revise this rule before preparing the draft.
- Propose New, In progress, Waiting on customer, and Closed as draft statuses. These are suggestions, not established team policy.
Unknowns
| Question | Impact | Blocking | Dependent step |
|---|---|---|---|
| Who assigns the owner, and when may a teammate take over? | Determines the claim and reassignment rules. | Yes | Finalize the operating brief. |
| How does the team recognize the same customer request across rows? | Determines whether duplicate identification is possible. | Yes | Specify duplicate handling. |
| What spreadsheet platform, columns, and access are available? | Determines the exact draft changes. | Yes | Inspect a sanitized sample and prepare the draft. |
| Which reporting views would help the team? | May affect a later summary view. | No | Optional follow-up after the basic process is agreed. |
Scope
Include proposed owner and status fields, ownership and reassignment rules, duplicate handling, and a manual review routine. Exclude live edits, migration, purchases, integrations, and automatic messages.
Constraints
Keep the existing spreadsheet. Do not change live data or buy tools. Use sanitized examples in the handoff. This refine request authorizes preparing a brief only.
Acceptance and verification
| Observable outcome | How to verify |
|---|---|
| A request has one clear accountable owner or an explicit unassigned state. | Walk through a new request and a reassignment using fictional rows in a draft. |
| Teammates can distinguish active, waiting, and completed requests. | Ask a teammate to classify one fictional request in each agreed status. |
| Duplicate requests have a defined handling rule. | Walk through two fictional rows representing the same request after its identifying key is agreed. |
| The draft preserves existing data and the live sheet remains untouched. | Compare a draft to the supplied sanitized sample; perform no live write. |
Next actions
- Await the ownership and duplicate-identification answers.
- Obtain a sanitized column list or sample and confirm the platform.
- Finalize the operating brief using the agreed rules.
- Hand the brief to a teammate to prepare a separate draft. Await separate authorization before any live change.
Executor
A normal agent or teammate familiar with the team's spreadsheet. No additional skill or paid tool is required to refine this brief.
Handoff
Prepare a separate spreadsheet draft after the blocking questions are answered. Preserve existing columns and row meaning. Add only the agreed ownership and status fields, then run the fictional-row checks above. Required inputs are the sanitized current structure, ownership policy, and duplicate key. Do not edit the live spreadsheet, send messages, or purchase tools.
Questions to resolve
- Who assigns ownership, and who can reassign a request?
- What identifies one customer request when it appears more than once?
- Which spreadsheet platform and current columns does the team use?
No spreadsheet changes have been executed.
Connects securely to your tools. The creator never sees your data.
About this skill
Make the next task clear
Requirement Refinery turns a rough request into a work packet an agent or teammate can follow. It separates supplied facts from assumptions, identifies questions that affect the work, and defines observable acceptance checks.
What is included
- The req-refine SKILL.md, version 0.1.1.
- Seven focused frameworks: requirement refinement, decision analysis, evidence research, critical review, engineering spec, debug investigation, and task specification.
- A shared execution-packet contract covering goals, deliverables, scope, constraints, unknowns, acceptance checks, next actions, executor, and handoff.
How to use it
Install the complete folder in your agent's skill location so the framework and reference files remain beside SKILL.md. Start with req-refine <your request>.
The default refine mode prepares the packet and stops. Use req-refine handoff: <request> for a self-contained handoff. Explicitly request refine+execute when you want the agent to refine and then carry out authorized work. Unanswered blockers remain blockers, and refinement does not grant permission to publish, send, delete, or access accounts.
Built for practical requests
Use it for feature briefs, workflow improvements, research plans, reviews, debugging requests, or delegated deliverables. It replies in the user's language and keeps small tasks small.
Optional use with pstack / poteto-mode
Requirement Refinery works on its own. Its execution packet can recommend a normal agent, an installed workflow skill, or poteto-mode from pstack when the task warrants it. This is an instruction-based recommendation, not a built-in integration or a mandatory automatic call.
We use the Claude Code version of pstack, pstack-claude. pstack and poteto-mode are separate, optional tools and are not included in this package. Install and invoke them separately if you want to use them for execution. The default refine mode still stops after preparing the packet; refine+execute permits authorized execution without requiring pstack.
Requirements and limits
This package contains Markdown instructions only. It includes no executable code, API keys, account bindings, or external service integration. Your agent must support loading SKILL.md and its linked local files. Output quality depends on the model and the context you provide. Broad compatibility across agents has not been independently tested. Optional workflow skills mentioned in the instructions are not bundled; the normal-agent fallback is supported.
How to use
Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scanned
Verified clean 12 days ago
- Passed all security checks, Safe to install