Python challenge ladder
103 challenges
Pick a challenge, solve it in the browser, keep the points.
Agent machinery: tool registries, the agent loop, retries, pipelines, planning, scheduling, streaming, and budgets.
Hard — 25 challenges
loading…| Challenge | Solved by | Points |
|---|---|---|
| 01A tool registry | — | 20 pts |
| 02Describe the tools for a model | — | 20 pts |
| 03Validate arguments against a schema | — | 20 pts |
| 04Dispatch a tool call | — | 20 pts |
| 05Handle an unknown tool | — | 20 pts |
| 06One step of the agent loop | — | 20 pts |
| 07The agent loop | — | 20 pts |
| 08Retry with exponential backoff | — | 20 pts |
| 09A token bucket rate limiter | — | 20 pts |
| 10A worker pool with a concurrency limit | — | 20 pts |
| 11Memoize an expensive tool | — | 20 pts |
| 12Skip work already done | — | 20 pts |
| 13Summarize old turns to fit the window | — | 20 pts |
| 14Define a workflow of steps | — | 20 pts |
| 15Run a pipeline, passing data along | — | 20 pts |
| 16Report which step failed | — | 20 pts |
| 17Order steps by their dependencies | — | 20 pts |
| 18Match a command to a workflow | — | 20 pts |
| 19Plan the steps for a request | — | 20 pts |
| 20Run the whole weekly report | — | 20 pts |
| 21Is the job due today? | — | 20 pts |
| 22Accumulate a streamed reply | — | 20 pts |
| 23Rebuild a streamed tool call | — | 20 pts |
| 24A tiny vector store | — | 20 pts |
| 25Stop when the budget runs out | — | 20 pts |