More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    NotebookLM: Interrogate Your Sources Like an Investigator

    1

    Turn a pile of sources into defensible answers with NotebookLM — curated corpus, a question ladder from facts to critique, cross-examination, and provenance on every claim.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    NotebookLM: Interrogate Your Sources Like an Investigator

    NotebookLM: Interrogate Your Sources Like an Investigator

    Example session with this skill installed

    We have 12 docs about our product spec and they contradict each other. Use NotebookLM to map every requirement, surface the contradictions, and find what the corpus never addresses.

    • Read your context and instructions
    • Compiled the notebooklm

    Curated the corpus first: 12 docs — PRD, 4 design docs, 6 meeting notes, 1 competitor teardown — tagged by date and author team, with stance noted on each. Two maybe-useful docs failed the intake checklist and stayed out; three copies of the same press release were deduped to one. The strongest counter-source was deliberately added — a corpus where everything agrees is a filter bubble. Rung 1 extraction: every stated requirement with its source doc — 34 requirements. Rung 2 comparison: 6 direct contradictions surfaced, including the PRD saying 'real-time sync' while design doc 3 says 'hourly batch acceptable for v1'. Rung 3 synthesis: 28 requirements have consensus; 6 are contested, each with the disagreeing parties named. Rung 4 critique: no source addresses migration from the legacy system — the corpus has a hole shaped exactly like the riskiest work. Cross-examination asked the same question per source and compared in a table, demanded quotes for contested claims, and probed the negative space — which sources do not mention migration, and is that surprising? Deliverable: a requirements table with per-claim sources, a contradictions section with quotes, and a gaps section. The team resolved 5 of 6 contradictions in one meeting — because the disagreement was finally visible.

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

    About this skill

    NotebookLM is only as good as the corpus you feed it and the questions you ask it — most people dump 50 PDFs and ask 'summarize', which is how confident garbage gets generated. This skill is the discipline that turns a source collection into an evidence base: curate the inputs with a stance and recency note on each, climb a question ladder from extraction through comparison and synthesis to critique before any decision support, cross-examine the same question per source, and export every claim with its provenance attached.

    Answers are only as good as the corpus: a 15-source corpus that passes the intake checklist outperforms a 60-source dump every time.
    It cannot answer what the sources do not cover — that is what the gaps section is for, and a gaps section is not a failure.
    Model inference must be labeled as inference: 'Source A says X' and 'combining A and B suggests Y' are different kinds of claims and need different scrutiny.

    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 3 days ago

    Permissions

    No special permissions declared or detected

    Listed3 days ago

    What's inside

    Frequently Asked Questions

    You can, and you will get confident garbage. Curate first — relevance, stance, dedupe, format sanity — then climb the question ladder. A summary of an uncurated corpus is a summary of its biases.

    No. Surface the disagreement with quotes and name who said what. Averaging two contradictory claims produces a third claim nobody made.

    Then it is not a finding — it is a rumor with nice formatting. Every exported claim carries its source, or it does not leave the room.

    New in LLM Ops