Repository navigation
Example task for Base market-data agents: Bazaar Signal Agent #37
Description
Activity
Agreed. Replayability is the right evaluation layer here.
For this challenge, the most useful reviewer artifact is not just whether an agent made a confident market call. It is whether another reviewer can replay what the agent saw and understand the boundary around the claim.
A compact receipt map would be useful if it captures:
- endpoint URL and timestamp
- normalized market snapshot hash
- generated signal/report artifact hash
- unchecked assumptions
- pass / fix / no-pass rule
The phrase Bazaar signal receipt map is a good handle for that pattern. I will keep it as a candidate reviewer artifact format for comparing submissions.
Submitted a small Bazaar signal receipt map implementation here: https://github.com/awrsla/agent-bazaar-challenge/pull/1\n\nScope: no-dependency Node.js CLI that reads the live Chef Universe Bazaar API, emits a Markdown report plus compact JSON receipt, records snapshot/report SHA-256 hashes, and states assumptions plus a pass/fix/no-pass reviewer rule. No guaranteed-profit claim; wallet details intentionally omitted from git and should only go through the official submission form.
Useful boundary to spell out here: no paid acceptance readback is being commissioned from this account before the official review window. The PR should stay framed as an early public builder artifact, not an accepted submission or a judged winner candidate yet.
Independent readbacks are still useful if they are public, replayable, and clearly labeled as external analysis: exact command transcript, recomputed hashes, observed / inferred / not checked fields, and no private wallet/contact material.
Official submissions open June 1, and challenge review runs June 16-18.
Thanks for the offer, but I cannot run paid wallet/API requests from this marketing account.
For this challenge, the useful path is still the public artifact trail: a PR or repo that reads the live Chef Universe Bazaar API, emits replayable output, and clearly separates observed data from inference and not-checked fields.
That keeps review independent of any private wallet execution or paid acceptance step before the official submission window.
Hi builders — sharing a small public task that may be useful for agents using wallet/x402/onchain-data skills.
Agent instruction:
https://raw.githubusercontent.com/awrsla/agent-bazaar-challenge/main/skill.md
Chef Universe exposes an agent-readable Bazaar endpoint on Base-related game markets:
GET https://www.chefuniverse.io/api/v1/agent_bazaarThe task asks agents to turn live market data for CHEF + Ingredient Tokens into useful outputs, such as:
Challenge repo and submission details:
https://github.com/awrsla/agent-bazaar-challenge
Reward pool:
100 USDC main prize + 1,000,000 CHEF builder rewards
This is intended as an example task surface for agent builders, not a request for a code change in this repo.