agentsclimarketplace

Product requirements desk

Skill MadewellRD/skills-lab/dist/vendor/anthropic/sdlc-command-desk/product-requirements-desk

Vendor-agnostic agent skill suites for the software lifecycle, web, AI engineering, product, sales, and mobile. Capability assumptions live in one versioned profile, so each new frontier LLM ships as a rebuild instead of a manual pass over every skill.

Install
npx -y skills add MadewellRD/skills-lab --skill product-requirements-desk

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

create connector-grounded product requirements documents for software work. use when the user needs a prd, requirement ids, acceptance criteria, non-goals, risks, open questions, source facts, stakeholder decision synthesis, or downstream handoff notes before technical discovery, architecture, issue planning, implementation, testing, verification, or release work.

SKILL.md

7.4 KB, as published. Nobody here has run it

Product Requirements Desk

Suite workflow mode

This desk is part of the SDLC Command Desk workflow suite. When invoked from an end-to-end workflow, do not stop with only a bare next-desk instruction. Complete this desk's artifact, emit a workflow packet, and continue to the next stage when enough facts are available.

If the next stage cannot be completed because required facts, connector access, approval, or source evidence are missing, return Workflow Halt with specific resume requirements. Use references/suite-workflow-contract.md for the packet, continuation, and halt format.

Role

Turn product ideas, customer notes, GitHub issues, roadmap context, stakeholder decisions, uploaded docs, and repo/product context into connector-grounded PRDs and requirement artifacts.

This desk is an upstream SDLC skill. Do not write implementation prompts, branch plans, or pull request instructions unless the user explicitly asks for a downstream handoff summary. When execution-ready implementation work appears after requirements are settled, continue into the implementation handoff stage or emit a workflow packet for that stage.

Workflow

Outcome. A connector-grounded requirements artifact, a new PRD from a raw idea, a PRD refined from existing docs or issues, generated acceptance criteria, a requirement normalization or review, or a connector diagnostic when required context cannot be reached.

Grounding. Use GitHub issues and repo context for product-facing bugs, feature requests, milestones, related code boundaries, and existing issue labels. Use docs connectors or uploaded files for roadmap, customer, policy, architecture, or stakeholder decision context. Use communication connectors only for decision-bearing messages, and record speaker, date, and context when available.

Source facts. Separate verified facts from assumptions, preserve the source hierarchy in references/source-hierarchy.md, and mark conflicting facts explicitly rather than resolving them silently.

Templates. Use references/prd-template.md for PRDs, references/output-contract.md for output modes and file wrappers, and references/downstream-handoff.md when preparing the next SDLC desk handoff.

Parallel surface. Evidence retrieval across independent sources, GitHub issues, docs, uploaded files, decision messages, carries no ordering dependency, and acceptance criteria for distinct requirement IDs are independent of each other. Fan out over both rather than iterating serially.

Acceptance bar. The PRD is done when every requirement carries a stable ID; each requirement has acceptance criteria that are testable without further product input; non-goals, risks, and open questions are explicit rather than implied; every load-bearing fact is attributed to its source; and assumptions are labeled inline as assumptions. Do not invent users, acceptance criteria, issue IDs, roadmap commitments, release dates, dependencies, owners, or compliance requirements.

Halt policy

Proceed by default. A requirement gap is normally recorded as an open question with a labeled working assumption, not a stop. Reserve hard halts for the consequence classes in references/halt-taxonomy.md:

  • Approval: a scope, commitment, or policy decision requires a human owner to authorize.
  • Production or destructive: the request would write to a live tracker, roadmap, or customer-facing commitment.
  • Security or privacy: the requirements would encode handling of secrets or personal data that is not sourced.
  • Source conflict: stakeholder decisions, docs, or issues genuinely disagree on a load-bearing requirement.
  • Release integrity: acceptance criteria would be stated as agreed when no source establishes agreement.
  • Connector unreachable: a required source exists but cannot be read. Context that is merely absent is a soft gap: continue with a labeled assumption or emit connector-diagnostic.md.

See references/halt-conditions.md for the artifact format a halt must take.

Default output modes

A requirements run produces these three together:

  • product-requirements-document.md
  • acceptance-criteria.md
  • requirements-source-facts.md

They are one deliverable in three files: the PRD states the requirements, the acceptance criteria make each one testable, and the source-facts file records what each rests on. A PRD handed over without its acceptance criteria pushes the same work downstream at a higher price.

Two modes are genuine alternatives rather than members of that set:

  • requirements-review.md when the request is to assess requirements that already exist rather than author new ones.
  • connector-diagnostic.md when the sources needed to ground requirements cannot be reached, produced instead of the set, not beside it.

Every downloadable Markdown artifact must start with a short "How to use this file" section.

Depth is the point of the set. Every requirement carries a stable ID and criteria a tester could execute without asking what was meant. Non-goals are stated as exclusions, not left as omissions. Open questions name who can answer them. A requirement reading "the system should handle errors appropriately" has an ID and no content.

Acceptance criteria for distinct requirement IDs are independent, so the artifacts in the set are drafted on the parallel surface the workflow declares.

Producing all three never means completing one from imagination. A requirement with no stakeholder, document, or issue behind it is recorded as a proposal labeled as such, or raised as an open question. Users, dates, dependencies, owners, and compliance obligations are sourced or absent; a PRD is where invented facts do the most damage, because every later stage treats it as settled.

Downstream handoff density

Keep downstream handoffs compact. The goal is to reduce ambiguity before coding agents spend tokens. Use requirement IDs, concise acceptance criteria, explicit non-goals, source facts, and exact open questions rather than long narrative restatement.

Bundled resources

  • references/prd-template.md: canonical PRD structure.
  • references/connector-routing.md: which connectors to use and what facts to retrieve.
  • references/source-hierarchy.md: source priority and conflict behavior.
  • references/output-contract.md: artifact names and wrappers.
  • references/halt-conditions.md: missing-context and conflict halts.
  • references/downstream-handoff.md: handoff format for later SDLC desks.
  • scripts/write_prd_markdown.py: deterministic Markdown wrapper for PRD artifacts.

Continuity Kernel Adoption

Use references/continuity-kernel.md, references/capability-baseline.md, references/readiness-gates.md, references/halt-taxonomy.md, references/preflight-cache.md, and references/handoff-density-policy.md when participating in an SDLC Command Desk workflow. Preserve and update the continuity_packet instead of reasking for facts already present. Classify missing inputs as hard halts, soft gaps, or auto-routable upstream/downstream work. Use HANDOFF_BLOCKER when implementation handoff facts are insufficient for a coding agent.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.