What Claude can do for you here, and when to reach for each one.
Showing all 29
Writes crisp, answer-first summaries in Claude's voice, for you to post with your own intro.
claude-summary
Use it when you ask
How to run In the Claude app, just ask, or pick claude-summary by name. In Claude Code, type /hia-skills:claude-summary.
Writes and polishes Race for Impact copy for emails, social and web pages in the RFI voice.
rfi-copywriting
Use it when you ask
How to run In the Claude app, just ask, or pick rfi-copywriting by name. In Claude Code, type /hia-skills:rfi-copywriting.
Scores an RFI email on subject line, body copy, CTA and HYROX lingo, so you know whether it works structurally.
rfi-email-audit
Use it when you ask
How to run In the Claude app, just ask, or pick rfi-email-audit by name. In Claude Code, type /hia-skills:rfi-email-audit.
Files every Canva design Claude makes into the team-shared HIA folder, so teammates can open it without asking for access.
canva
Use it when you ask
How to run In the Claude app, just ask, or pick canva by name. In Claude Code, type /hia-skills:canva.
Researches a UI pattern on Mobbin and packages screenshots, an opinionated README and a Claude Design handover prompt.
design-reference-pack
Use it when you ask
How to run In the Claude app, just ask, or pick design-reference-pack by name. In Claude Code, type /design-reference-pack:design-reference-pack.
Drafts HIA, RFI and partner decks in Figma Slides from the team template.
figma-deck-draft
Use it when you ask
How to run In the Claude app, just ask, or pick figma-deck-draft by name. In Claude Code, type /hia-skills:figma-deck-draft.
Makes on-brand charts, diagrams, stat tiles and impact graphics from the master design system.
hia-brand-visuals
Use it when you ask
How to run In the Claude app, just ask, or pick hia-brand-visuals by name. In Claude Code, type /hia-skills:hia-brand-visuals.
Checks a Customer.io campaign before it goes live: triggers, timezones, audiences, double sends, links and sender details.
cio-go-live-qa
Use it when you ask
How to run In the Claude app, just ask, or pick cio-go-live-qa by name. In Claude Code, type /hia-skills:cio-go-live-qa.
Answers fundraising number questions from the Filament data warehouse, checking the live tables first so queries don't fail.
filament-query
Use it when you ask
How to run In the Claude app, just ask, or pick filament-query by name. In Claude Code, type /hia-skills:filament-query.
Reviews open Intercom conversations and surfaces the ones that need Paddy personally.
intercom-inbox-triage
Use it when you ask
How to run In the Claude app, just ask, or pick intercom-inbox-triage by name. In Claude Code, type /hia-skills:intercom-inbox-triage.
Builds a role-specific onboarding plan in Notion and walks the new starter through it from day one.
onboard-new-starter
Use it when you ask
How to run In the Claude app, just ask, or pick onboard-new-starter by name. In Claude Code, type /hia-skills:onboard-new-starter.
Pulls action items from the weekly Scorecard call, drafts a Slack summary and adds Paddy's items to Things.
process-scorecard-todos
Use it when you ask
How to run In the Claude app, just ask, or pick process-scorecard-todos by name. In Claude Code, type /hia-skills:process-scorecard-todos.
Summarises the DMs, mentions and threads aimed at you since you last checked.
slack-summary
Use it when you ask
How to run In the Claude app, just ask, or pick slack-summary by name. In Claude Code, type /hia-skills:slack-summary.
Moves a candidate to the next stage and preps whoever runs that stage.
advance-candidate
Use it when you ask
How to run In the Claude app, just ask, or pick advance-candidate by name. In Claude Code, type /hia-recruitment:advance-candidate.
Drafts an honest, specific feedback email when a candidate asks why they didn't advance.
candidate-feedback
Use it when you ask
How to run In the Claude app, just ask, or pick candidate-feedback by name. In Claude Code, type /hia-recruitment:candidate-feedback.
Starts a hiring round from a finished job description: round config, rubric, screening questions, sourcing plan and the Notion role hub.
kickoff-round
Use it when you ask
How to run In the Claude app, just ask, or pick kickoff-round by name. In Claude Code, type /hia-recruitment:kickoff-round.
Puts candidates on hold with a holding-email draft and the right paused stage in Notion.
pause-candidate
Use it when you ask
How to run In the Claude app, just ask, or pick pause-candidate by name. In Claude Code, type /hia-recruitment:pause-candidate.
Files a work-trial candidate's references, signed letter, NDA and invoice.
process-work-trial-admin
Use it when you ask
How to run In the Claude app, just ask, or pick process-work-trial-admin by name. In Claude Code, type /hia-recruitment:process-work-trial-admin.
Drafts the rejection email that fits how far they got and closes their Notion card.
reject-candidate
Use it when you ask
How to run In the Claude app, just ask, or pick reject-candidate by name. In Claude Code, type /hia-recruitment:reject-candidate.
Reviews a recorded interview, checks the planned questions were asked and scores the candidate against the rubric.
review-call
Use it when you ask
How to run In the Claude app, just ask, or pick review-call by name. In Claude Code, type /hia-recruitment:review-call.
Writes the end-of-round retro from the funnel numbers and five reflection questions.
round-retro
Use it when you ask
How to run In the Claude app, just ask, or pick round-retro by name. In Claude Code, type /hia-recruitment:round-retro.
Finds new applications, dedupes them, scores each against the rubric and cards them in Notion with screening-call questions.
screen-applications
Use it when you ask
How to run In the Claude app, just ask, or pick screen-applications by name. In Claude Code, type /hia-recruitment:screen-applications.
Builds a signed-off spec in a fresh session: implementer agents build, a separate verifier tries to break it, at most two fix rounds.
fable-build
Try
How to run Claude Code only. Type /fable-orchestration:fable-build.
Plans a build with you and writes the spec, decisions and acceptance criteria. Stops for your sign-off before any code.
fable-plan
Try
How to run Claude Code only. Type /fable-orchestration:fable-plan.
Nothing matches that. Try another word, or press "Show all".
Helpers, a connector and a copy-paste prompt. Mostly for building things in Claude Code.
Lets Claude Code read and write your Google Sheets. Runs a pinned, reviewed copy of an open-source connector.
hia-google-sheets
Needs Claude Code, not claude.ai in a browser
How to run Claude Code only. Install it from the team marketplace; it needs your own Google login.
Writes the code for one phase of a spec and reports back. Used by /fable-build.
implementer
Dispatched by /fable-build
How to run You don't run it yourself; Claude Code calls it for you.
Read-only fact-finding: where code lives, how something works, external docs.
researcher
Dispatched by other agents
How to run You don't run it yourself; Claude Code calls it for you.
Independently re-checks a finished change and tries to break it. It can't edit code.
verifier
Dispatched by /fable-build
How to run You don't run it yourself; Claude Code calls it for you.
Paste at the end of a working session to wrap up and leave a handover the next agent can start from cold.
session-handover
Copy from the repo and paste
How to run Nothing to install. Open it on GitHub, copy it and paste it into Claude.
claude.ai, desktop and Cowork
Open Customize and look under the organisation plugins. The main set, hia-skills, is installed for everyone by default. Other plugins, such as hia-recruitment, can be switched on from the same place.
Add the team marketplace once, then install the plugin you want.
/plugin marketplace add High-Impact-Athletes/claude-plugins
/plugin install hia-skills@hia-claude-plugins
To pick up the latest versions later, run /plugin marketplace update hia-claude-plugins.
Every skill has an owner, who decides what it should do and reviews changes. Some also have a maintainer, who does the day to day fixing. The owner's first name is at the bottom of each card. If something goes wrong, press "Report a problem" on the card and it goes straight to them.
Owners marked proposed have been suggested but not yet agreed by that person; confirmed means they have agreed. To confirm, open a one-line pull request changing owner_status: proposed to owner_status: confirmed in the skill's SKILL.md (or in owners.yaml for a whole plugin). The open list is in OWNERS-TODO.md.
Claude Code runs only. Use in claude.ai, desktop chat and Cowork is not counted, so a skill with few or no runs here may still be well used.
Claude Code runs only, top 5
Claude Code runs only, all time
Claude Code runs only. Agents, the Google Sheets connector and copy-paste prompts are not tracked.
Press "Report a problem" on the card. The short form goes to the skill's owner. Saying which kind of problem it is helps, because each is fixed in a different place.
Not sure which kind it is? Pick "not sure" on the form. The owner will sort it.
Skills marked for possible retirement, plus any skill with no Claude Code runs in the last 60 days. Nothing is removed automatically; these are reviewed at the monthly retro. claude.ai use is invisible here, so ask the team before retiring anything.