Research applications, evidence and funding history before voting.
Retroactive grants ask you to judge work already done. This skill turns your coding agent into a diligence copilot for the review round: every applicant cross-referenced against the audited payment ledger, prior round verdicts, and their own submitted evidence. Every figure is cited, never recalled from the model's memory.
npx skills add Michae2xl/openzcash --skill coinholder-copilotBring your own model. No key, no hosted service.
Add to your agent
…and 40+ more agents the skills CLI supports. Omit -a to auto-detect your agent, or add -g to install globally.
❯
Run diligence on proposal #21 before I vote.
The ask, the claimed work, prior round verdicts, ledger history, announcement-vs-spreadsheet drift, and the open questions worth taking to the forum thread.
Who in this round was funded before, and did they deliver?
Every applicant cross-referenced against the audited ledger and prior rounds: funded-and-delivered, rejected-before, two-track orgs, first-timers.
How does the ballot actually work?
Four options, two of which are 'no' votes with different signals. Plus the keyholder veto that already zeroed out a coinholder-approved $2.67M grant.
What has the Coinholder program actually paid so far?
The completed grants and their per-payment ZEC trail, kept distinct from the $9M currently being requested.
The skill also encodes the round's real traps: repeat submissions at higher asks, same-org multiple proposals, third-party bonus nominations, announcement-vs-spreadsheet drift, and the difference between coinholder approval and actual payment. It lays out the evidence and never tells you how to vote. Methodology: how we compute this.