More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Narrative Tracker: What AI Engines Say About You

    1

    Audit what ChatGPT, Perplexity, and other AI engines actually say about your brand — claim by claim, with quoted evidence — and turn it into a fix list.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Narrative Tracker: What AI Engines Say About You

    Narrative Tracker: What AI Engines Say About You

    Example session with this skill installed

    We run Loopdesk, a helpdesk tool for small online shops. Lately prospects keep asking if we're "the one that's shutting down" — we're not; a similarly named company closed last year. Can this skill find where that's coming from and tell us what to do?

    • Read your context and instructions
    • Compiled the narrative tracker

    Contract (excerpt). Want echoed: "helpdesk built for shops under 50 staff," "flat pricing per shop, not per agent," "live chat and email in one inbox." Never attached: "shutting down," "enterprise call-center suite," "free forever tier" (discontinued 2024). Rivals: two named helpdesk competitors. Buyer questions include "Loopdesk vs [rival]," "best helpdesk for a small Shopify store," and "is Loopdesk still operating?"

    Battery findings. The shutdown confusion traces to two engines answering the direct query with a paragraph that merges Loopdesk with the defunct similarly named company — graded False, wrong class (wrong company facts, not a typo). One engine repeats the discontinued "free forever" line, citing a 2023 review roundup — Stale. The flat-pricing claim is Missing on three of four engines: they describe per-agent pricing "like most helpdesks," because Loopdesk's own pricing page never states the flat model in plain words. The small-shop claim is Echoed everywhere — health is uneven, not bad: scores 7, 6, 4, 8 across the four engines.

    Fix list, in order. 1) Publish a short "Loopdesk is operating and growing" status-and-history page with founding date and the name distinction stated plainly — gives engines a citable primary source against the mix-up. 2) Rewrite the pricing page's first paragraph to state the flat per-shop model in one sentence an engine can quote. 3) Ask the 2023 roundup's publisher for a correction on the free tier; update the structured data on the pricing page meanwhile. 4) Re-run the battery in 30 days and grade the same claims in the ledger — corrections land on indexing time, and the ledger is how you'll know they landed.

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

    About this skill

    Buyers are reading a summary of your company that you didn't write and nobody approved — the one AI answer engines generate on demand. Visibility tools tell you whether you appear. This skill tells you what the machines say: which of your claims they echo, which they omit, which they twist, which are stale, and which are flat-out false — graded against a contract you write before you run a single query.

    The method runs in six steps: write the claim contract (five claims you want echoed, three you never want attached, the rivals contesting each, the questions buyers actually type); run a four-shape query battery (direct, comparison, category, problem) across every engine your buyers use, saving verbatim answers; grade each claim Echoed, Missing, Twisted, Stale, or False — with falsehoods classed as typo, wrong, or dangerous; keep a claim ledger that carries from run to run so drift becomes visible; report health scores per engine, a falsehood list, drift notes, and a territory map of who owns each claim; then work the correction playbook — engines don't take correction requests, they take better sources, so every gap maps to a page to publish or repair.

    It includes the contract template, a grading decision table, a health scoring rubric with a danger override, a worked example, and a pre-delivery quality gate. Monthly runs take an afternoon and replace a quarter of guessing.

    • Engine answers vary between sessions; the skill grades claim accuracy across a battery, but a single run is a snapshot, not a census.
    • Corrections work through sources and indexing, so improvements take weeks to show in answers — the method sets honest expectations and a re-run date.
    • It grades the brands and claims you put in the contract; a rumor outside the contract's never-claims can be missed until a run surfaces it.
    • Requires access to the engines your buyers use; region- or account-specific answers may differ from what your buyers see.

    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 any payment.

    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 24 days ago

    Permissions

    No special permissions declared or detected

    Listed24 days ago
    Updated21 days ago

    What's inside

    Frequently Asked Questions

    Those tools answer "do we show up?" This skill answers "what is said about us, is it true, and is it the story we chose?" Every claim in your contract gets a grade and a quoted piece of evidence, and every gap maps to a specific source to publish or repair.

    Engines don't take correction requests — they synthesize from sources. The playbook finds what the engine is citing and fixes it there: update or redirect the stale page, publish a plain citable statement, correct third-party sources where possible. Then the next run verifies whether the answer changed.

    Monthly is the sane default for an active brand; quarterly works for quieter categories. Always run it after a rebrand, a pricing change, a product discontinuation, or a news event — those are the moments stale and false claims get written into the answers.

    New in Marketing & SEO