Production-tested CLAUDE.md files and skills for PRD writing, user story generation, sprint planning, and stakeholder communication. Your product context, loaded automatically — not re-explained every session.
Claude Code was built as a developer tool, but the fastest-growing segment using it isn't engineering — it's product. The reason is structural: PM work is unusually well-suited to Claude Code's core strength, which is reasoning over dense context.
PMs operate at the intersection of user research, engineering constraints, business goals, and design decisions. Every artifact they produce — a PRD, a user story, a sprint plan, a stakeholder brief — requires synthesizing that context into a specific format for a specific audience. That is exactly what Claude does well when properly configured.
Three things make PMs particularly high-value users:
The constraint isn't skill or technical ability. It's configuration quality. A PM who spends 30 minutes building a proper CLAUDE.md gets dramatically different results than one who types instructions directly into the chat every session.
Every Claude session starts with no memory of the last one. For most users that's a minor inconvenience. For PMs, it's a significant efficiency drain.
Consider what a PM has to re-establish every session to get useful output: the product's current stage, the target users and their primary pain points, the constraints the engineering team is working under, the design system and principles in use, the roadmap priorities for this quarter, the decisions that have already been made and why. That's 400–600 words of context before you can ask a useful question.
In practice, most PMs don't re-establish all of it. They ask a truncated question, get a generic answer, and spend more time editing the output than they saved by using Claude at all. This is the context problem — and it's entirely solvable with a properly structured CLAUDE.md.
A PM-specific CLAUDE.md loads your product context automatically at the start of every session: who your users are, what you're building, the constraints that matter, and the conventions your team uses for PRDs, stories, and communication. Claude reads it before you type your first word. The session starts with shared context, not from blank.
Writing a PRD from scratch is slow. Writing one in Claude without context is faster but often wrong — the output is structurally sound but doesn't reflect the actual constraints, user insights, or design decisions that belong in this specific document for this specific team.
A PRD writing configuration solves this by loading three things before any PRD is drafted:
With this config in place, a PM can open Claude Code and type "write a PRD for real-time alerts on inventory thresholds" — and the output will be shaped by the actual product context, not a generic B2B SaaS template. The constraints are already loaded. The user persona is already defined. The acceptance criteria come out in the right format for your engineers.
The gap between a Claude-generated user story and one that actually goes into your sprint is usually format and specificity. Generic stories are structurally correct but vague. They require manual editing before engineering can act on them.
A user story generation skill closes this gap by encoding your team's exact format and acceptance criteria structure. The output of the skill isn't a template — it's a complete story your team can pull into Jira with minimal editing.
The skill enforces the split-if-complex rule automatically. If the feature scope generates more than seven acceptance criteria, Claude flags it as a story that needs to be broken down rather than generating an unwieldy list your team will struggle to estimate. That's a judgment call that takes experience to make consistently; encoding it in the skill makes it systematic.
Sprint planning is where context-loading pays off most visibly. The decisions made in sprint planning — what goes in, what gets deferred, how much the team can realistically take on — depend on three things that live outside Claude's default knowledge: your backlog priorities, your team's historical velocity, and individual capacity for the upcoming sprint.
A sprint planning configuration loads all three at session start:
With this context loaded, a sprint planning session becomes a structured reasoning exercise rather than a blank-slate conversation. You describe the sprint goal, Claude applies the actual capacity constraints to the prioritized backlog, flags items that look undersized or oversized against team velocity, and produces a draft sprint plan you can take into the planning meeting for discussion.
Technical decisions need executive summaries. User research needs engineering-friendly distillation. Design rationale needs to be accessible to people who weren't in the design review. PMs write audience-bridging content constantly — and without context, Claude produces content calibrated to a generic audience that isn't quite any of the ones that matter.
A stakeholder communication configuration defines your specific audiences and the translation rules for each:
When the config is in place, a PM can say "write an update on the delayed auth integration for the exec team" and the output will be structured, business-impact-first, jargon-free, and concise — without a separate editing pass to strip technical language. The audience translation rules are applied at generation time.
Battle cards are the canonical PM deliverable that almost nobody is happy with. Too detailed and sales can't use them in a live call. Too thin and they don't help close anything. The structure that works varies by product, market, and sales motion — and it only emerges from iteration on what your sales team actually uses.
A competitive analysis configuration encodes your validated battle card structure and the specific competitors you track, so each new analysis starts from the right frame rather than a generic SWOT matrix:
The output of the competitive analysis skill is a battle card your sales team can use immediately, not a first draft you need to reshape before sharing. The structure is already validated because it's encoded in the configuration, not invented fresh each time.
OKRs are most useful when the connection between sprint work and quarterly objectives is visible in real time, not reconstructed at the end of the quarter. A PM who can show, mid-sprint, which key results are being moved by which stories — and flag key results that have no associated work in the current sprint — has a tool that most teams don't have.
An OKR tracking configuration loads your current objectives and key results as persistent context, then makes them referenceable in sprint planning, PRD writing, and stakeholder communication:
The instructions at the bottom of the OKR section — requiring every PRD and story to reference a key result, and flagging uncovered KRs in sprint planning — encode a governance behavior that would otherwise require a recurring manual check. Claude enforces it automatically because it's in the config. The governance cost moves from your attention to the configuration file.
SkillsMP has over 800,000 auto-indexed configurations. That number is the pitch. It's also the problem.
Ask a more precise question: how many of those 800,000 configs are built for a B2B SaaS PM at a 20-person startup, selling to operations managers in manufacturing, running two-week sprints with a team of four engineers? One, maybe. More likely zero. The configs that exist are generic — PRD templates, user story formats, sprint checklists. They're structurally correct and practically useless until you've spent two hours customizing them for your actual product context.
SmarterContext doesn't have 800,000 configurations. Every configuration in the catalog has been used in actual PM workflows, with real products and real teams. The metadata tells you what they were built for: company stage, product type, team size, sales motion. You can identify the one that fits closest to your situation and know that the customization work is genuinely minimal — not a complete rebuild disguised as a starting point.
Curated also means maintained. Auto-indexed configs reflect the state of the tool when the creator uploaded them. SmarterContext configs are updated when Claude Code releases new capabilities that change how configurations should be structured. You're not pulling a snapshot from six months ago and wondering why some instructions no longer behave as documented.
See also: Claude Code workflow config guide and startup-specific Claude Code configs.
The most common reason PMs don't configure Claude Code properly is the assumption that it requires technical work. It doesn't. CLAUDE.md is a plain text file. Rules files are plain text. Skills are plain text. There is no code, no YAML, no JSON, no environment variables to configure.
The setup process for a SmarterContext PM configuration:
Step 3 takes 8–12 minutes for most PMs. The product context section has prompts for each field so you're filling in blanks, not starting from blank. You're not writing configuration — you're describing your product in plain language, which is something PMs do continuously anyway.
The rest of the configuration — the PRD structure, the story format, the audience translation rules, the OKR enforcement instructions — are pre-built and pre-tested. You benefit from the iteration that happened before you downloaded it, without having to do that iteration yourself.
Loads product vision, team constraints, and your PRD template. Generates structured PRDs that match your team's format, with "what we're NOT building" sections built in.
Jira-ready output in Given/When/Then format. Enforces story size limits and flags scope that needs splitting. Uses only your defined personas.
Loads backlog priorities, velocity data, and capacity adjustments. Produces a draft sprint plan against real constraints, not theoretical capacity.
Audience-specific translation rules for exec, engineering, and customer-facing comms. Generated content doesn't need a separate editing pass for tone.
Your battle card format and competitor profiles as persistent context. Generates sales-ready battle cards from raw competitor information.
Loads current OKRs as persistent context. Enforces story-to-KR linkage in PRDs and flags uncovered key results in sprint planning automatically.
If you're already using Claude Code without a PM-specific configuration, the fastest path is the free config audit — it scores your current setup against production-quality standards and gives you a specific list of what's missing. Most PMs find they're re-explaining context that should be in CLAUDE.md, using prompts where skills would be more reliable, and missing enforcement rules for the conventions they care most about.
If you're starting fresh with Claude Code, the PM catalog includes configurations for B2B SaaS PMs, product-led growth teams, platform PMs managing internal developer tooling, and generalist PMs at early-stage companies who need a minimal but high-quality baseline.
All configurations are plain Markdown files — no proprietary format, no lock-in. If you cancel, you keep everything you've downloaded. They work with Claude Sonnet and Claude Opus.
See also: Claude Code workflow templates and Claude Code configs for startups.
Every configuration in SmarterContext has been used in real PM workflows. PRD writing, user stories, sprint planning, stakeholder communication — curated, not auto-indexed. Start with a free audit of your current setup.
30-day money-back guarantee · Cancel anytime · Plain Markdown — no lock-in