Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    PM Sense Check: Contradictions in Product Docs

    1

    Finds the contradictions and gaps in a PRD or doc set before your reviewers do: numbers that disagree, dates that do not add up, scope drift, conflicting requirements and missing owners, with quotes and locations.

    $7

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    PM Sense Check: Contradictions in Product Docs

    Example session with this skill installed

    a PRD for receipt auto-scan and its launch plan.

    • Read your context and instructions
    • Compiled the pm sense check

    5 blockers and 6 major findings. Verdict: needs another draft. Blocker F1: build starts 4 May and takes 6 weeks plus a 2 week beta, so it ends on 29 June, but the PRD says GA on 5 June and the launch plan says 15 June. Others: 50% vs 30% goal, 7-year image storage vs 90-day deletion, an EU processing requirement that depends on an open question, web upload in the beta while the PRD puts the web app out of scope, and "25% lift" vs "10% to 25%". 11 questions for the author. A second example shows a consistent PRD that is ready for review with 3 small findings.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    The embarrassing questions in a PRD review are rarely about the idea. They are "the summary says 50%, the goals say 30%", "build ends after the launch date" and "the launch plan promises the web app you put out of scope". PM Sense Check finds them first. Give it one doc or a doc set (PRD, launch plan, one-pager). It builds a table of every number, date, scope item, owner and term, then runs 12 checks: numbers that disagree, units and denominators, date arithmetic, scope drift, conflicting requirements, metrics without baselines, ambiguity that forks the build, missing or conflicting owners, undefined terms, assumptions stated as fact, broken references and leftover template text. You get a prioritised findings table with exact quotes and locations, the arithmetic for every date and number, questions for the author, a list of what was checked and found consistent, and a verdict. It never comments on style.

    What it does

    • Reads one doc or several related docs and cites file > section > line for every finding
    • Builds a fact table before judging anything
    • Runs 12 consistency and gap checks
    • Shows the date and number arithmetic in each finding
    • Sorts findings into Blocker, Major and Minor
    • Writes questions for the author, grouped by owner, each linked to a finding
    • Lists what it verified as consistent
    • Reruns on a revised doc and reports fixed, open and new findings

    Why this beats prompting it yourself

    "Review my PRD" prompts return generic advice about adding personas and metrics. This skill extracts the facts first, compares them across sections and docs, quotes both sides of every contradiction and does the date math, so each finding is something you can fix in a minute.

    Who it's for

    • Product managers the day before a review or engineering handoff
    • Product leaders reviewing many docs a week
    • Program managers checking a launch plan against the PRD
    • Engineering leads who want to estimate from a doc that agrees with itself

    What you get

    • SKILL.md with the 6-step workflow and severity rules
    • A reference of the 12 checks with examples and usual severity
    • A findings report template
    • Two complete sample runs, including a two-doc set

    Frameworks and tools

    Works in any agent that reads SKILL.md: Claude Code, Cursor, Codex and similar. No API keys, installs, scripts or network access.

    Permissions

    • Read: the docs you share or point to
    • Write: nothing by default. It returns the report in chat
    • Network: none

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Product Management.

    It checks a doc against itself and the docs you give it, not against reality. It cannot tell which of two conflicting numbers is right, so it asks. Very large doc sets (over 8,000 words) get blockers and majors first.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by 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 scan

    Security scan passed 21 days ago

    Permissions

    Read Files
    Write Files
    Listed21 days ago

    What's inside

    Frequently Asked Questions

    No. It reports findings and questions. Pair it with PM Doc Slop Check for wording.

    Yes. Share the revised doc and it lists fixed, open and new findings.

    No. Only whether the doc agrees with itself.

    New in Product Management