Goodhart Audit

    1

    A metric that a team can move stops measuring what you meant it to.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Goodhart Audit

    Example session with this skill installed

    goodhart-audit: CAC +team:performance marketing

    • Read your context and instructions
    • Compiled the goodhart audit

    Example on the fictional Ridgeline Outdoor shop.

    1. Metric: paid media spend ÷ new paying customers, per month.

    Inputs: spend (performance team, daily, ✅ high: shift spend into brand budget lines that CAC excludes, or delay invoices into next month). New customers (performance team via targeting, ✅ high: discount-led first orders that get returned still count). Attribution window (analytics, ⚠️ medium).
    3.

    Cliffs: the month boundary (invoice timing), the first-order flag, and returns that arrive as negative rows after the month closes.
    4. Expose / abstract / hide: spend → expose. The new-customer definition → abstract. The returns adjustment → hide.
    5.

    Counter-incentives: discount-heavy acquisition, low-basket customers, and brand spend reclassified to flatter CAC.

    ⚠️ Share the value only and hide the formula.

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

    About this skill

    A metric that a team can move stops measuring what you meant it to. Goodhart Audit runs before you hand a metric, score or dashboard to a team that can move its inputs: marketing, growth, sales or ops.

    What it does

    1. Inputs map: every input, who owns it, how often it updates, and the cheapest way a team could push it up without creating value.
    2. Cliff edges: thresholds, date windows and status flips where the metric jumps rather than moves. That's where gaming pays.
    3. Hidden-weights call: for each input, expose it, abstract it (share the rollup only) or hide it.
    4. Counter-incentive watch: the perverse behaviour the metric would reward.
    5. Verdict: ✅ share with the formula, ⚠️ share the value and hide the formula, or 🔴 share direction only.

    Where it stops: it assumes the metric is correct. Use Red-Team Formula to check the SQL first.

    Free and MIT-licensed. It's part of pm-os: https://github.com/hartejsingh99-cloud/pm-os

    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 assumes the metric is correct (use Red-Team Formula for that) and judges incentives from what you tell it about the team receiving the metric. It doesn't read your dashboards or org chart.

    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

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 5 days ago

    • Free with an Agensi account

    Requires

    Nothing to install.

    Permissions

    Read Files

    File Scopes

    .pm-os/house-rules.md
    ~/.claude/pm-os/house-rules.md

    Reads your optional house-rules file to find the metric's inputs and known data cliffs. It never writes files or calls the network.

    Listed5 days ago

    What's inside

    Frequently Asked Questions

    Yes. It's free and MIT-licensed, and the source is on GitHub: https://github.com/hartejsingh99-cloud/pm-os

    When the metric is read-only (an audit or a quarterly review), or when the people receiving it can't move its inputs, such as a peer PM.

    No. It's about incentives, not correctness. Run Red-Team Formula (also free) on the SQL first.

    New in Product Management

    Free