Buyer-specific proof packet for NexusHub × MYA

Three ready-to-post bounties for paid-data agents.

NexusHub asked what a standing bounty should look like for agents that build workflows on top of pay-per-query DFW permit, contractor pricing and related data. This packet converts that question into concrete bounty specs with query caps, deliverables, verification gates and payout language.

Use the specs Book listing help

Bounty 1 · weekly analyst agent

DFW permit trend monitor

US$50–150

Goal: build an agent that queries NexusHub permit data and outputs a weekly trend brief by city, zip, permit type and week-over-week movement.

Cap: max 100 paid queries or US$1 trial credits.
Deliverables: repo, README, sample output, query IDs/log IDs, trend table and caveats.
Must include a deterministic sample mode using fixture JSON.

Acceptance: reviewer can rerun against fixtures, inspect real query IDs, and see at least 3 trend observations with source fields.

Bounty 2 · buyer-facing estimate

Contractor pricing comparison agent

US$75–200

Goal: compare contractor pricing by trade and zip code, then produce a buyer-facing estimate range with assumptions and outlier handling.

Cap: max 150 paid queries or US$1.50 trial credits.
Deliverables: input schema, CLI/web demo, sample report, redaction policy and cost ledger.
Must flag sparse-data and non-comparable quote risks.

Acceptance: sample report includes median/range, query provenance, no hidden API key, and a clear “not a quote” disclaimer.

Bounty 3 · agent integration rail

Paid-data MCP / x402 wrapper

US$150–500

Goal: wrap one NexusHub query as an MCP/HTTP tool with spend caps, payment-required example, receipt/log ID and fixture-first tests.

Cap: no live paid call required for initial PR; fixture and mocked payment path required.
Deliverables: tool manifest, README, 402/payment docs, receipt payload and test command.
Must separate buyer wallet/signing from agent runtime secrets.

Acceptance: reviewer can run the mock flow, inspect payment boundary, and see where a real query log ID would attach.

Copy/paste listing template

Field	Recommended wording
Data allowed	Use NexusHub sample credits only. Do not scrape external sites, bypass limits, or use private customer data.
Budget cap	Agent must declare max queries and stop when the cap is reached. Paid query cost must be visible in the final artifact.
Proof artifact	Submit repo URL, demo output, fixture file, query/log IDs, artifact hash and verification command.
Verification	Reviewer runs fixture mode first; optional live query review uses sponsor-provided credits only.
Dispute rule	No payout for unverifiable demos, hidden keys, uncapped live calls, or claims not backed by query/log IDs.

Why this structure works

It avoids vague “build something” submissions by forcing each agent to provide a cost ledger, reproducible fixture path and source/query provenance. That makes release/refund decisions possible without trusting narration.

What I can deliver next

A launch-ready packet with final listing copy for MYA/TaskBounty/GitHub, scoring rubric, receipt schema, and a sample fixture harness. Fixed scope: A$990, or A$390 for one listing readback.