More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Principal Reps: Timed PM Interview Practice

    1

    You get one question, a stated time budget (90 seconds by default), and one attempt, answered out loud.

    Free

    Security scanned
    Principal Reps: Timed PM Interview Practice

    Principal Reps: Timed PM Interview Practice

    Example session with this skill installed

    Give me a rep.

    • Read your context and instructions
    • Compiled the principal reps

    The agent states the domain and the 90-second budget and asks one question, for example: "Your team is goaled on click-through rate and it is rising. Why might that be bad news?" You answer out loud. It replies in five lines or fewer: Reversibility 1 and Cheapest test 1 were weakest, and the sentence you should have said was the cheapest holdout that would show whether clicks are displacing conversions. It logs the rep with its seven scores and your time to the ledger.

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

    About this skill

    About this skill

    Most people preparing for a Principal or Staff product loop already know how to think through a problem. What fails in the room is speed of retrieval: the right frame shows up after the answer is over. Reading frameworks does not fix that, and long coaching feedback trains reading instead of speaking.

    This skill runs timed reps. You get one question, a stated time budget (90 seconds by default), and one attempt, answered out loud. The agent scores the answer 0 to 3 on seven dimensions (ambiguity, scope, technical, operations, mechanism, cheapest test, reversibility) and replies in at most five lines: the two weakest dimensions and the one sentence you should have said. The pattern it trains is five beats: what wasn't defined, what I claim, what I'd measure, whether it can be undone, and what would prove me wrong. Other modes run a five-question set with no coaching between questions, spaced repetition on your weak dimensions, a four-minute "tell me about something you built" story, and a postmortem after a real interview.

    The proof

    There is no outcome data, and the listing claims none. The rubric is the author's own synthesis of what separates a Principal answer from a Senior one. What the skill does guarantee is arithmetic, not judgment: a small Python script logs every rep to a CSV ledger and computes averages, trends, coverage by domain, and the stop condition. The skill retires itself when every dimension averages 2.0 or better across your last 15 reps, at which point it tells you to practice from a printed card and a timer instead.

    What you get

    SKILL.md with the modes and rules; the five-beat answer pattern with anti-patterns; a question bank of 74 domain-tagged questions answerable without domain expertise; schemas for an optional personal profile and for a target job; the narrative pattern for the four-minute story; and scripts/reps.py, the ledger.

    How to use

    1. Add the skill folder to your agent's skills directory (for Claude Code, ~/.claude/skills/principal-reps/).
    2. Say "give me a rep" and answer the question out loud within the time budget.
    3. After a few sessions, run report to see your weakest dimension, and weakness to drill it.

    Source and license

    Free to use and MIT licensed. The full source, install steps, and issue tracker are on GitHub at https://github.com/robertmnyborg/principal-reps.

    Scores come from the model and are judgment against the author's rubric, not a validated hiring rubric. It does not simulate interviewer follow-ups inside a rep. Without Python and a filesystem the ledger lives only in the current conversation. The ledger defaults to ~/Projects/principal-reps unless you set PRINCIPAL_REPS_DIR. Spoken answers depend on whatever dictation your setup provides.

    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

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 8 days ago

    • Free with an Agensi account

    Requires

    Any agent that loads SKILL.md files
    Python 3 for the rep ledger (optional; without it scores stay in the conversation)

    Permissions

    File Scopes

    ~/Projects/principal-reps/
    (or
    $PRINCIPAL_REPS_DIR):
    reps.csv
    profile.md
    targets/

    Writes only its own ledger and profile files. Reads a job posting URL only when you run target add.

    Listed8 days ago

    What's inside

    Frequently Asked Questions

    An agent will happily mock-interview you, but it gives long feedback, scores generously, and remembers nothing between sessions. This skill caps feedback at five lines, tells the agent never to score generously, and keeps a ledger so it can drill your weakest dimension.

    Yes. Any agent that reads SKILL.md can run it. Without Python and a filesystem, for example in a chat app, it keeps scores in the conversation and offers CSV rows you can paste into a ledger later.

    No. Every question is meant to be answerable in the budget by naming assumptions out loud. Refusing to answer for lack of domain knowledge counts as a failed rep.

    Yes. `target add <posting URL>` writes a target file with company facts, extra questions, and landmines, and later reps sample from it.

    On GitHub at https://github.com/robertmnyborg/principal-reps, under the MIT license. Every file in this package is there, plus the README and install steps.