Hey everyone. Sharing something new: ZCG Copilot and many other things.
It is a skill for AI coding agents that answers questions about ZCG funding from live public data, with a source link on every number. No made-up figures, which matters for grants, since a model’s training data is always out of date. It came out of a conversation with @paulbrigner about how hard it is to get a straight answer when the data is split across a spreadsheet, GitHub, and the forum.
npx skills add Michae2xl/openzcash --skill zcg-copilot
- One-click buttons for the common agents (Claude Code, Cursor, opencode, Codex, Windsurf, Copilot, plus 40 or so more): ZCG Copilot · OpenZcash
- Open source (MIT)
What it answers today:
- Given everything funded so far, where are the obvious gaps
- Which merchant payment grants were funded, and for how much
- How much a team has actually been paid, budgeted versus paid to date
- What is under review right now, and the combined ask
BYO-LLM. No server, no API key, nothing to pay. Your agent thinks. The skill points it at the right public sources and teaches it to read them correctly, including the parts that trip people up, like requested versus approved versus paid, or re-applications that reuse an old title.
Where it fits: the ZCG Grant Dashboard (the public spreadsheet) is the source of truth. OpenZcash mirrors and reconciles it, and this skill is the public, ask-anything front door to those same numbers. When any source disagrees, the spreadsheet wins. Everything it reads is open, so you can check its work or build on the same API.
Feedback welcome, especially on what questions or data would be most useful to you.