All authors
Claude Skills by reynolds-uk
github.com/reynolds-uk12 skills0 installs0 views
- GradersNo description providedVotes: 0GitHub stars: 2
- GradersNo description providedVotes: 0GitHub stars: 2
- Agent RegisterBuild or update the register: one page per AI agent or tool that acts for a business, with its permissions, limits, grade, a named supervisor, test set, cost and history, plus a list of the gaps. Use when someone needs to know what AI is acting in their name and who is accountable for each.Votes: 0GitHub stars: 2
- Curriculum GridSort a team's junior or routine tasks by what the output is worth and what doing it taught, and decide which work to keep for people. Use before automating junior work, when AI is taking the tasks people used to learn from, or before a hiring or graduate-intake decision.Votes: 0GitHub stars: 2
- Gate ReviewCheck whether an AI agent has earned its grade (Observe, Recommend, Prepare, Act) against criteria written in advance, including an override analysis. Use when deciding whether to promote or demote an agent, or when someone asks whether an agent is ready to do more on its own.Votes: 0GitHub stars: 2
- Leader CheckFive-minute leader check on how AI is going in a business: nine plain questions, then one page with where the business stands, where it sits on the autonomy map, one place to start, three questions for the team and a sentence for the board. Use when a leader or leadership team wants an honest read on their AI programme, a board-ready view of where they are with AI, or doesn't know where to start.Votes: 0GitHub stars: 2
- Self CheckThirty-point self-check on how AI is run in a business, where each point only counts with evidence. Use when the person who builds or runs AI wants to test their own programme honestly, or asks to be challenged on whether they're doing it properly.Votes: 0GitHub stars: 2
- Test SetStart a test set for one AI agent or workflow: real cases from your own work, with the right answers marked by the people qualified to judge, versioned and run on every change. Use when someone needs to know whether an AI tool's work is right, before promoting an agent, or when the only evidence of accuracy is the vendor's word.Votes: 0GitHub stars: 2
- Two ColumnsSort what AI is said to save into two columns: what was claimed, and what actually reached the accounts, with a named line for every banked saving. Use when someone reports time or cost saved by AI, when finance can't find the savings, or before AI savings go into a board pack or a plan.Votes: 0GitHub stars: 2
- Where Each Model RunsDecide, step by step, which AI model each step of a workflow uses and where it runs: rented from the lab, licensed through your cloud provider, hosted on machines you control, or run on your own hardware. Use when choosing models or vendors, when data must not leave a country or the building, when costs need cutting, or before signing an AI contract.Votes: 0GitHub stars: 2
- Where The Revenue IsFind revenue, not only savings, from AI in a services or information business: four questions about data, explanation, records of work and pricing, then a check of how each offer is charged. Use when an AI plan is all cost savings, when someone asks how AI could make them money, or when AI is about to make hourly or per-unit work cheaper.Votes: 0GitHub stars: 2
- Where The Time GoesMap one workflow end to end with the four questions (job, pain, gain, five minutes before and after), find the waiting, and sort every step into delete, automate or augment. Use when someone wants to decide where to point AI in a process, or asks why a process takes so long.Votes: 0GitHub stars: 2