More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Research Novelty Check: Has Anyone Done This Before?

    1

    Check whether a research idea is really new: split it into claims, search each claim, build a closest-work dossier, and get a calibrated verdict.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Research Novelty Check: Has Anyone Done This Before?

    Research Novelty Check: Has Anyone Done This Before?

    Example session with this skill installed

    I am a researcher with an idea: an app that generates bedtime stories for children and adapts the vocabulary and sentence length in real time to the child's reading level, measured from how they read the previous story aloud. Before I build a prototype, has anyone done this before?

    • Read your context and instructions
    • Compiled the research novelty check

    CLAIMS (invented idea, fictional researcher) — A: real-time adaptation of story vocabulary and sentence length to a measured reading level. B: reading level measured from the child's read-aloud performance on the previous story, not from a test. C: stories generated fresh for each session rather than selected from a fixed library. D: adaptation happens within a single story, not only between sessions.

    SEARCH — Three angles per claim over the last 12 months: technical terms (adaptive text generation, reading-level personalization), problem-first (who adapts reading material without a test), and venue-aware sweeps. Neighbors found: a fixed-library leveling system, a tutoring system that adapts between sessions, and a read-aloud assessment tool with no story generation. Each dossier row states its difference; none contains all four claims together.

    VERDICT: PROCEED WITH CAUTION. The combination — read-aloud measurement driving within-story generated-text adaptation — is statable and not contained in any single neighbor, but the between-session tutoring system is close enough that a reviewer will cite it. Strengthening plan: a small study comparing within-story adaptation against between-session adaptation on the same children, with reading-gain measures, before building the full prototype. Search dated; verdict covers literature to that date only. Gate check: claims, log, dossier, verified citations, and positioning paragraph all present.

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

    About this skill

    Before months go into a research direction, this skill answers one question: has anyone already done this, in a form a reviewer would recognize as the same result? It starts by breaking the idea into three to five checkable claims, each naming its mechanism, the problem, and a one-sentence difference from the obvious baseline — because ideas are not searchable, claims are. Each claim is then searched independently with at least three query angles: technical vocabulary, problem-first formulations, and venue-aware sweeps over the last six to twelve months, all recorded in a search log with wording and date range. The closest work becomes a dossier of three to eight papers, and every row must end with the remaining difference stated precisely, or an honest admission that none remains. The verdict follows strict admission rules: proceed when the difference is real and statable, proceed with caution when a neighbor is close enough to demand a named strengthening, and abandon only when one specific published paper already contains the result. Tie-breaks favor the more permissive verdict, since a wrongly passed idea dies cheaply at the next gate and a wrongly killed one is never seen again. Honesty rules cover combination novelty, checking the finding as well as the method, never fabricating citations, and dating every verdict, since a novelty verdict expires at its search date.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.

    • This is an early triage gate, not a full literature review; it does not read every paper in the area.
    • Search quality depends on the literature tools you can access and the queries run.
    • A verdict expires at its search date and must be re-checked before major commitments if months have passed.

    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 3 days ago

    Permissions

    No special permissions declared or detected

    Listed3 days ago

    What's inside

    Frequently Asked Questions

    Run the naive query and the oldest-terminology query from the protocol before trusting the emptiness. If they also return nothing, record it plainly as a clean pass with its date range, and note it is the verdict most likely to be overturned by a better-phrased search.

    No. Crowding is never, by itself, a reason to abandon. Abandon requires one specific published paper that already contains the result, named by title, authors, and venue or year.

    No. Novelty can live in the combination. If the arrangement reveals something unknown — an interaction, a failure mode, a capability the parts lack alone — that revelation is the difference, and the check judges the idea, not the parts list.

    New in Research & Analysis