Claude Skills: the quiet feature that’s making Discover smarter every week
If you’ve used Discover for grant matching, eligibility checking, or bid drafting, you’ve almost certainly used a Claude Skill without knowing it. This post is about what that actually means, why it matters for a charity trying to win more funding with less staff time, and what we’re building next.
What is a Claude Skill?
A Skill is a folder of instructions that teaches Claude how to do a specific job, once, and then remember how to do it every time after.
Normally, an AI assistant starts from zero in every conversation. You’d have to explain, every single time, what “good” looks like for a funding application, what a readiness score means, or how to check eligibility against a specific funder’s criteria. That’s fine for a one-off question. It’s a real drag when it’s the same task every week.
Skills solve this by packaging up the procedure: the steps, the format, the rules, the examples of what “done” looks like. Claude reads a short description of each Skill first, and only pulls in the full instructions when a request actually matches what that Skill is for. So you can stack up a whole library of specialist know-how without it slowing anything down or getting in the way of unrelated tasks.
Think of it as the difference between briefing a new team member from scratch every morning, versus handing them a proper onboarding guide they can refer back to.
The important part is what goes into that onboarding guide. A Skill is only as good as the expertise behind it — and that’s the bit we care most about.
Built by fundraisers, not just engineers
Every Skill we build starts with the same question: what would a good bid writer actually do here?
We don’t sit down and prompt-engineer our way to a generic answer. Our Skills are built by working through each task the way our own experienced fundraisers and bid writers do it — the same checks, the same order of operations, the same instinct for what a funder is really asking versus what the question literally says. That process gets written down step by step, and it’s that procedure the Skill follows, not a general “write me a good answer” instruction.
So when you use the eligibility-check Skill, it’s working through criteria the way a fundraiser who’s read hundreds of funding rounds would — not guessing. When you use the grant-answer drafting Skill, it’s structuring an answer the way someone who’s actually won grants would structure it. The AI is doing the work, but the method it’s following is ours.
Why that makes it easy to adopt, even if you’re not “technical”
This is really the point of building Skills this way. Because the expertise is already baked into the Skill, you don’t need to know how to prompt an AI well, or understand what’s happening behind the scenes, to get a good result out of it.
You don’t need to know the “right” way to ask for an eligibility check — you just ask, in your own words, and the Skill runs through the same steps a bid writer would. You don’t need to learn any special syntax or set anything up. There’s nothing to configure, no separate tool to log into, no training required beyond using Discover the way you already do.
In practice, this means the person best placed to use these Skills well isn’t the most tech-savvy person on your team — it’s whoever actually understands your organisation and your funders. The Skill supplies the fundraising method; you supply the context. That’s a much lower bar to clear than “get good at AI,” and it’s the reason we’re building it this way rather than handing charities a generic AI tool and a manual.
How this supports charities using Discover
For a small charity team, the value isn’t the AI itself — it’s the domain knowledge sitting behind it. Skills are how we bake in what makes a strong funding case, how eligibility actually works, and what funders are really asking for in that “guidance text” field, so a non-technical fundraiser gets expert-level support without needing to know how any of it works under the hood.
In practice that means:
- Consistency. The same quality bar applies whether it’s your first application or your fiftieth.
- No re-explaining. You don’t need to brief the AI on what a Section 106 clause is, or what a “readiness score” means, every time you use it.
- Specialist depth without specialist cost. Each Skill can carry a lot of detailed guidance behind the scenes without cluttering the actual conversation.
What we’ve built so far
We’ve been scoping and building Skills specifically for non-technical charity staff working through Discover, each one mapped directly to how our own fundraisers and bid writers approach that task in real applications. Six are underway right now:
- Evidence-grounding / pre-draft research — pulls together the supporting evidence and context a strong application needs before a single word of the draft gets written.
- Campaign creation — helps set up a new funding campaign correctly from the outset, so matching and reporting work properly downstream.
- Grant-answer drafting — turns a funder’s question, word limit, and guidance into a properly tailored answer, not a generic template.
- Eligibility-check — works through a funder’s criteria against your organisation’s actual profile, flagging where you’re a genuine fit and where you’re not.
- Report-drafting — supports the funder reporting side of things once a grant is won, not just the winning of it.
- Readiness-score improvement playbook — turns your readiness score into a concrete, prioritised action plan rather than just a number.
Each one is scoped around a real bottleneck we kept seeing charity teams hit — the boring, repetitive, easy-to-get-wrong parts of the funding cycle that eat hours every week.
New Skills, added weekly
This isn’t a one-off build. We’re treating Skills as a living part of the product, not a launch feature we tick off and move on from. As we see patterns in how charities are actually using Discover — where people get stuck, what questions come up again and again, what “good” looks like for a particular type of funder or application — we’re turning that straight into new Skills.
So expect this list to keep growing. If there’s a specific part of your funding workflow that feels like it’s still eating too much time, that’s exactly the kind of thing we want to hear about — it’s very likely to become the next Skill we build.
