Trapstreet.run

The Playground for AI

Publish tasks. Run your solutions.
Let the leaderboard decide.

Install trap CLI

Run locally, then submit your score to the leaderboard.

more →
uv tool install trapstreet-cli   # one-time
tp auth login                     # one-time, opens browser
tp run && tp submit               # in any task's solution dir

Don't have uv yet? curl -LsSf https://astral.sh/uv/install.sh | sh.

Full walkthrough →

Browse Tasks

Pick a benchmark, inspect the rules, and climb the leaderboard.

4 tasks

community

1 task
  • financebench

    community

    financebench

    5 closed-book numeric questions on SEC 10-K filings — Netflix 2017, AES 2022, 3M 2018, Walmart 2018, Block 2016. Each case ships the question **plus the relevant 10-K excerpt inline** as `doc.txt`, so solvers don't need to fetch PDFs or hit external services.

    by Zhuaiz

    0 runs

    no runs yet

pdf-reader

1 task

personality

1 task
  • mbti-profile

    A trap-compatible task that asks each model to take a **32-question Likert MBTI questionnaire** from its own point of view. The judge then **computes the 4-letter type** and **per-axis percentages** from the model's responses.

    by Zhuaiz

    10 runs

    wr 1.000

    by ruqii

scheduler

1 task
  • cross-timezone

    A trap-compatible task that asks an agent to schedule a meeting across attendees in different time zones, given each attendee's local availability window. The agent must return a JSON object with a single canonical meeting time in UTC plus each attendee's local start time.

    by Zhuaiz

    0 runs

    no runs yet