Product specs
867 rows from 443 repositories
PRDs, user stories and acceptance criteria.
What Product specs skills agree on
527 skill files read, by 483 of the 502 authors on this shelf whose files we hold, 2026-09-06
The middle one of the 48 measured here is ~1.6k tokens long, counted with cl100k_base
Counted by distinct author, so one author publishing three of these counts once. Where a claim sits in fewer files than authors, that is said: a claim held by forty authors across three files is one file people copied, not forty people who agreed. Near-identical wordings are grouped and the other wordings are shown, so the grouping is yours to check.
What they tell the agent to do
- Write the PRD using the template39 of 483 in 36 filesalso worded as Write the PRD from the template; Use the PRD template verbatim
- Ask one question at a time24 of 483also worded as Ask questions one at a time; Ask discovery questions one at a time
- Write acceptance criteria in Given-When-Then format24 of 483also worded as Write acceptance criteria as Given/When/Then; Write acceptance criteria in Given/When/Then format
- Write user stories with acceptance criteria22 of 483 in 21 filesalso worded as Write testable acceptance criteria for every story; write user stories with Given/When/Then acceptance criteria
- Respect ADRs in areas you touch21 of 483 in 17 filesalso worded as Respect ADRs in the area being touched; respect ADRs in the touched area
- Use the domain glossary vocabulary throughout the PRD20 of 483 in 16 filesalso worded as Use domain glossary vocabulary throughout the PRD; use the domain glossary vocabulary throughout
- Explore the repo to understand the current codebase19 of 483 in 15 filesalso worded as Explore the repo to understand the codebase; Explore the repo to understand current codebase state
- Provide testable acceptance criteria19 of 483 in 17 filesalso worded as Write testable acceptance criteria; Make every acceptance criterion testable
- Publish the PRD to the issue tracker18 of 483 in 14 filesalso worded as Publish the PRD to the project issue tracker; Publish the templated PRD to the issue tracker
- Apply the ready-for-agent triage label15 of 483 in 13 filesalso worded as Apply ready-for-agent and backlog labels
- Prefer existing seams over new ones14 of 483 in 12 filesalso worded as Prefer existing seams to new ones; Prefer existing test seams to new ones
- Save the PRD as a markdown file14 of 483 in 9 filesalso worded as Write the PRD in Markdown format; Save the PRD to a specified filename
- Include an extensive numbered list of user stories14 of 483 in 12 filesalso worded as Write an extensive numbered list of user stories; Write an extremely extensive numbered list of user stories
- Sketch the seams for testing the feature13 of 483 in 11 filesalso worded as Sketch the testing seams for the feature; sketch the test seams first
- Confirm the seams with the user12 of 483 in 10 filesalso worded as Confirm proposed seams with the user; Confirm the proposed seams with the user
What they tell it not to do
- Do not write vague acceptance criteria30 of 483 in 25 filesalso worded as accept vague requirements; Do not write vague acceptance criteria
- Do not interview the user28 of 483 in 24 filesalso worded as Interviewing the user; Do not interview the user from scratch
- Do not include file paths or code snippets25 of 483 in 21 filesalso worded as Do not include specific file paths or code snippets; Including file paths or code snippets in the PRD
- Never include implementation details22 of 483 in 20 filesalso worded as Do not test implementation details; Do not include code or implementation details
- Do not write implementation code14 of 483also worded as Do not write implementation details; No writing code
- Do not start implementing the feature12 of 483 in 9 filesalso worded as Implementing the feature; Do not start implementation
- Never use vague wireframe placeholders11 of 483 in 6 filesalso worded as Do not use vague wireframe placeholders; Never use vague placeholders in wireframes
- Never fetch pages with raw curl or fetch11 of 483 in 6 filesalso worded as Do not fetch pages with raw curl; Never fetch the page with raw curl
- Do not write untestable requirements10 of 483 in 8 filesalso worded as Do not write untestable requirements; Do not write acceptance criteria that cannot be tested
- Never write PRD content before clarification10 of 483 in 9 filesalso worded as Drafting the PRD before clarifying answers are concrete; Don't write the PRD before exploration and approval
What they expect to be installed
- AskUserQuestion50 of 483
- git36 of 483
- grep31 of 483
- Mermaid21 of 483
- gh18 of 483
- Read17 of 483
- python316 of 483 in 14 files
- project issue tracker16 of 483 in 12 files
- Glob15 of 483
- WebSearch14 of 483 in 9 files
What they ask it to produce
- PRD markdown file46 of 483 in 45 filesalso worded as PRD markdown document; Markdown PRD file (prd-[feature-name].md)
- PRD document34 of 483 in 31 filesalso worded as PRD; structured PRD document
- PRD.md24 of 483also worded as docs/PRD.md; specs/prd.md
- User stories with acceptance criteria23 of 483 in 22 filesalso worded as user story document with acceptance criteria; user stories with acceptance criteria, priority, and dependencies
- PRD published to the issue tracker20 of 483 in 16 filesalso worded as PRD published as a tracker document; PRD published to the project issue tracker
- User stories20 of 483 in 19 filesalso worded as user-stories.md; User stories document
- Product requirements document17 of 483also worded as Requirements document; A product requirements document
- Feature specification document13 of 483 in 11 filesalso worded as Feature specification files; Feature specification
When Product specs authors say to reach for one
The situations these authors wrote into their own files, counted out of the same 483 authors, with the skills that name each one
- User asks to create a PRD85 of 483 in 80 files
- To prd
- Code to prd
- Prd generator
- Go
- Pm prd
- Prd to issues
- and 23 more on this shelf
- User asks to spec out a feature27 of 483
- Prd generator
- Write feature request
- and 1 more on this shelf
- User mentions PRD or product requirements20 of 483 in 19 files
- Go
- Prd writing
- and 3 more on this shelf
- When the user wants a PRD from the current context14 of 483 in 12 files
- To prd
- Skill prd synthesis
- and 8 more on this shelf
- Writing user stories or acceptance criteria14 of 483
- Product manager
- and 1 more on this shelf
- Creating acceptance criteria12 of 483 in 10 files
- Writing user stories12 of 483
- User asks to create a design doc12 of 483
How Product specs skills are built
800 skill directories by 420 authors, read from their repositories’ own file trees 2026-08-05
The middle bundle among those shipping files is 3 files, 11.5 KB beside SKILL.md
Counted by distinct author, same as above, so one author publishing forty template copies counts once. SKILL.md itself is not counted as a file, so a single-file skill is one where that file is the whole skill.
The shape
- SKILL.md is the whole skill216 of 420 authors, 398 of 800 skills
- files ship beside it204 of 420 authors, 402 of 800 skills
- executable scripts ship inside50 of 420 authors, 74 of 800 skills
The folders they converge on
- references/127 of 420 authors, 205 of 800 skills
- scripts/41 of 420 authors, 61 of 800 skills
- agents/33 of 420 authors, 64 of 800 skills
- assets/27 of 420 authors, 42 of 800 skills
- examples/22 of 420 authors, 42 of 800 skills
- templates/19 of 420 authors, 44 of 800 skills
anombyte93/prd-taskmaster Skill
587★Zero-config goal-to-tasks engine for Claude Code (the Atlas engine). Graded PRD validation, dependency-ordered task graph, evidence-gated execution.
golbin/agent-skills/skills/prd Skill
no license209★ repoReusable agent skills for Codex and compatible tools
liqiongyu/lenny_skills_plus/samples/writing-prds-executable Skill
51★ repo86 agent-executable skill packs converted from RefoundAI’s Lenny skills (unofficial). Works with Codex + Claude Code.
skinnerlee1225/enterprise-prd-toolkit/skills/enterprise-prd-writer Skill
40★ repo把金融級 PRD 方法論工程化成四個 Claude Skills：找洞 → 寫規格 → 產測試。含交易所提現、自營交易挑戰賽的完整 PRD 範例。
echoVic/spec-flow Skill
33★A Claude Code skill for structured, spec-driven development with phase-by-phase workflow and living documentation
aakashdhar/vibe-skill/vibe-add-feature Skill
no license7★ repovibe-* is a collection of Claude Code CLI skills. Each skill is a structured instruction set that tells Claude Code exactly what to do at a specific moment in a project — how to plan it, build it, review it, test it, deploy it, and hand it off.
jcmaker/prd-maker/skills/prd-maker Skill
6★ repoInterview your idea into a PRD.md any AI coding agent can execute. Claude Code · Codex plugin. Works for non-devs. Pure markdown output.
IncommensurableHubris/fullstack-director/.agents/skills/00-discovery Skill
5★ repoSpec-first, cross-harness SDLC skills framework — one director, AI agents, a living spec spine
endorphin-ai/hasbrains-agent-kit/agent-engineering/skills/docs-project-management Skill
4★ repoThe battle-tested Claude Code kit behind HasBrainsAI — agent skills, subagents & slash commands for production multi-agent systems. Install in one command.
yihannangua/prd-review-skill/prd-review Skill
4★ repoDeep PRD and requirements review Agent Skill / 深度 PRD 与需求文档评审 Skill
lantisprime/claude-sdlc/skills/analyze Skill
3★ repoA governance layer for AI-assisted software delivery. Plan, scope, approve, build, test, release — with human sign-off at every gate.
GrayMa77er/tiny-spec/tiny-spec-breakdown Skill
3★ repoA tiny, opinionated take on spec-driven development.
saleem-daqa/qa-test-case-generation Skill
3★Reusable Agent Skill for generating professional, non-duplicated manual QA test cases from requirements, mockups, and templates.
Sixz-AI/prd-workflow/prd_decomposer Skill
2★ repoClaude Code skills for AI-assisted PRD execution — decompose requirements into atomic REQs, enforce brainstorming before coding, track status per-REQ, and maintain git discipline from first spec to final commit.
428alexander9/claude-skills-marketplace/project-planner-skill Skill
no license2★ repo🛠️ Transform Claude into a System Architect, generating detailed project documentation for AI-assisted software development with clear boundaries and traceable requirements.
unboundinnov/specdd/plugins/specdd/skills/specdd Skill
2★ repoSpec-Driven Development as a single Agent Skill
Abhillashjadhav/AI-PM-essential-skills/.claude/skills/prd-first Skill
2★ repoInstallable Claude Code plugins for AI product evaluation, model routing, guarded loops, and MCP migration decisions.
cdeust/prd-spec-generator/io.github.cdeust/prd-spec-generator MCP server
2★ repoTurns a feature description into a verified 9-file PRD — multi-judge consensus, 17 MCP tools.
facukoren/pm-agent/skills/pm-workflow Skill
1★ repoSenior Product Manager AI agent for Claude Code: discovery interview, codebase-grounded PRD with testable FR/NFR, dependency- and risk-aware roadmap, and opt-in epic-by-epic implementation with adversarial verification.
roeibh/interactive-spec Skill
1★Turn written specs into self-contained interactive HTML mockups with a closed feedback loop back to Claude Code. A Claude skill.
hugoooo1018/pm-agent-toolkit/skills/prd-paradigm Skill
1★ repoPM agent toolkit — three-layer architecture that turns one-liner requirements into industrial-grade PRDs. Agent-neutral rules for Claude Code / Cursor / any LLM.
aks-builds/openspecpm Skill
1★Spec-driven, BDD-shaped project management for AI agents. OpenSpec proposals → BDD specs → tracked work in GitHub / Azure DevOps / Jira / Linear / GitLab.
linkc-skills/HTML-Plan Skill
1★HTML is the new Markdown. An Agent Skill for human-AI collaboration on complex tasks. Works with Claude Code, Codex CLI, Gemini CLI, Cursor.
abrahamFerga/scrum-skills/skills/audit-backlog-ado Skill
1★ repoA vendor-agnostic library of AI agent skills built around the Scrum framework. Works with Claude Code, Cursor, GitHub Copilot, and any agentskills.io-compatible agent.
DustinVK/wherefore/plugins/wherefore/skills/ask Skill
1★ repoThe why behind your code, as plain markdown in your repo. Agent skill to capture decisions, dashboard to browse them.
celestialdust/achilles-skills/skills/acceptance-criteria Skill
1★ repoAn AI engineering skill suite that takes one idea from Ideate → Spec → Plan → Implement → Verify → Review → Ship — the human owns intent, the agent owns execution, ending at risk-banded draft PRs. Installable across Claude Code, Cursor, Gemini CLI, and more.
sushil930/my-claude-skills/skills/prd-maker Skill
no license0★ repo🧠 A collection of custom Claude AI skills — prompt engineering, PRD writing, UI/UX design, and more. Built with Sushil Patel.
chenpianzhou/requirement-to-prd-skill/requirement-to-prd Skill
0★ repo让 AI 写 PRD，但不许它瞎编 — turn a one-line idea into a review-ready PRD
RJTPP/agent-software-doc-skills/skills/software-design-doc Skill
0★ repoAgent skills for software documentation inspired by IEEE standard.
ValdoTR/jeo-prd Skill
0★Quiz show skill for Claude Code that tests if you understand your code before shipping
tingwei628twdev/ralph-enhanced/skills/prd Skill
0★ repoEnhanced Ralph autonomous agent loop — phase isolation, failure tracking, token budget, and needs_human escalation. Built on snarktank/ralph + Wiggum + ralphex.
dimayip/socratic-design Skill
0★From a vague idea to a decision-locked design document — a general-purpose, domain-agnostic Socratic facilitation skill (PRD / RFC / Design Doc / ADR / One-Pager).
Prd implementation ready skills
wair56/prd-implementation-ready-skills Skill
0★Agent Skill for turning rough PRDs and product requirements into implementation-ready specs.
julijakvetna/ba-project-factory Skill
0★A complete Business Analyst lifecycle skill — 24 structured prompts for enterprise software delivery
SkillMedev/project-management/skills/jira-ticket-writer Skill
0★ repoConnect Claude to your project tracker and let it draft, file, and organize the work.
SkillMedev/product-manager-stack/skills/jira-ticket-writer Skill
0★ repoRoadmaps, specs, sprints, and analytics — the PM’s full operating kit.
jovylle/build-ready/skills/build-ready Skill
0★ repoScore a spec's autonomous-build readiness before any code exists — a portable Agent Skill that red-teams your PLAN/SPEC from a zero-context builder's seat. Works in Claude Code & opencode.
megamanics/bpel-agent/skills/bpel-prd Skill
0★ repoBPEL Transformation Agent - Specialized agents for transforming Oracle BPEL processes into modern implementations. Converts legacy SOA workflows into Spring Boot applications with complete PRD documentation and test coverage—enabling 100% feature parity in modern tech stacks.
jovd83/test-analysis-skill Skill
0★Analyzes requirements and use cases for ambiguity, testability, delivery risk, and missing stakeholder decisions.
sametcn99/product-planning-copilot/.github/skills/prd-planning Skill
no license0★ repoThis repository is designed to be forked so teams can set up their own documentation-first planning system quickly.
jovd83/acceptance-criteria-designer Skill
0★Turns raw product requirements into testable acceptance criteria, coverage notes, and machine-readable contracts.
jovd83/backlog-story-generator Skill
0★Generates epics, backlog-ready user stories, and importable story packs from requirements, discovery notes, or codebases.
sanketsao/prd-for-agents Skill
no license0★When multiple agents interacting with each other, handoff is critical to pass on the appropriate context with right level of detail. This repo will take any vision from user and convert into a prd which can easily be consumed by any agent
elevationgain/agent-disco Skill
0★Agentic, structured product discovery for AI-native delivery teams.
xyjk0511/elite-prd-skill-pack/.agents/skills/elite-prd-writer Skill
0★ repoCodex skill pack for PRD discussion, requirements clarity, PRD writing, audit, engineering handoff, and QA generation
raphaelbleier/userstory-playwright-skill/skill Skill
0★ repoPoint your coding agent at a website codebase: it writes user stories into Excel, tests them with Playwright, and writes the bugs back. Works with Claude Code, Codex, Kilo Code, OpenCode and GitHub Copilot CLI.
jinzhedev/jz-tanstack-ship Skill
0★Assemble and verify a TanStack Start SaaS from a PRD.
rvdsmaile/working-backwards/.agents/skills/break-prd-into-issues Skill
no license0★ repoComposable Working Backwards skills for turning product ideas into customer narratives and build-ready plans.
shinpr/ai-coding-project-boilerplate/.claude/skills-en/documentation-criteria Skill
225★ repoAgentic coding TypeScript boilerplate for Claude Code: sub-agent workflows with built-in quality checks and context engineering.
skinnerlee1225/enterprise-prd-toolkit/skills/prd-writer Skill
40★ repo把金融級 PRD 方法論工程化成四個 Claude Skills：找洞 → 寫規格 → 產測試。含交易所提現、自營交易挑戰賽的完整 PRD 範例。
naiersaidane/claude-mastery/skills/cadre Skill
39★ repoSkills Claude Code de la formation Claude Mastery — interroge, cadre, planifie, design, investigue, illustre
dhzrx/ai-pm-claude-skills/skills/prd-generator Skill
no license22★ repoarezous/pm-pilot/.claude/skills/break-down Skill
19★ repoAI copilot for product managers. Built for Claude Code and Cursor.
gtrabanco/agentic-workflow/skills/audit-pr Skill
19★ repoStack-agnostic agentic-programming workflow skills + documentation scaffold
javier-garzas/ia-product-management/skills/crear-historias-de-usuario Skill
12★ repoSkills de IA para Product Management y Spec-Driven Development, creadas por Javier Garzas (233 Academy) y compartidas con la comunidad.
ZeroPointRepo/CRHQ-PRDs/skills/prd-builder Skill
no license11★ repoProduction-grade, 10,000+ line PRDs that let AI agents one-shot entire applications. Built on CRHQ.ai.
ninthwave-io/ninthwave/skills/decompose Skill
8★ repoOrchestrate parallel AI coding into reviewable PRs.
afrizzal/ratchet/skills/ratchet-backlog Skill
6★ reporesumable engineering loops for Claude Code — audit a codebase into an acceptance-gated backlog, then execute it one verified commit at a time.
arsudsandesh97/Revoact/skills/design-doc Skill
3★ repoAI-powered skill pack for generating professional documentation, including PRDs, architecture, design docs, roadmaps, and READMEs
samgao2000/prd-grill/skills/prd-gate Skill
3★ repoGrill → Gate → Build: interrogate requirements until accurate and complete, score them with an independent gate, then hand the human-approved PRD to autonomous development.