Guide
A team prompt library template, and when a prompt should become a skill
A team prompt library works when every entry has the same few fields: a name that says the task, one line on when to use it, the inputs to paste, the prompt, one good example output, an owner, and the date it was last tested. Keep it in one shared place sorted by job, review it each month, and retire prompts nobody uses. When several people use a prompt every week, turn it into a skill and share it through AI Skill Pod: everyone gets it in their Claude, Cursor or ChatGPT (Pro, Business, Enterprise) from one invite link, and nobody pastes it again.
Short answer
- Same fields for every prompt.
- One owner, a monthly review, a retire rule.
- Sort by job, not by tool.
- Weekly, shared prompts become skills.
The template to copy
One entry per prompt. Copy the block into your doc, wiki or database, and fill in every field before the prompt counts as approved.
Name: [verb + object, such as "Draft a refund reply"]
When to use: [one line: the situation, such as "A customer asks for a refund within 30 days"]
Job: [sales / support / marketing / finance / HR / operations]
Inputs to paste: [what the person adds each time: the customer's message, the order number]
Prompt:
[the prompt, with [brackets] where the inputs go]
Good output: [one real example the team approved, with names removed]
Watch for: [the one mistake it makes, such as "promises a refund date"]
Owner: [name]
Last tested: [date] in [Claude / ChatGPT / Copilot]
Status: [draft / approved / retired]Why each field is there
| Field | Why |
|---|---|
| Name and when to use | People find a prompt by the task, and the same line later becomes a skill's description, which says what it does and when to use it |
| Inputs to paste | Most bad outputs come from a missing input, not a bad prompt |
| Good output | A concrete example shows the style better than a description of it |
| Owner and last tested | Someone answers for it, and a date shows when it was last checked against a real task |
| Status | An approved prompt and someone's draft look the same without it |
The naming, example and testing advice follows Anthropic's skill authoring best practices, read 5 November 2026.
Keep the library alive
- One owner for the library, and one owner per prompt.
- Sort by job, because people look for "support" before they look for "ChatGPT".
- Review once a month: retest the approved prompts on a real task and update the date.
- Retire what nobody used in two months, rather than letting the list grow.
- Promote the weekly ones to skills, and leave a line in the library pointing to the skill.
A worked example: thirty prompts, eleven kept
Illustration: a made-up business. A support team of nine kept thirty prompts in a shared doc: many near-duplicates, no owners, and several written for a refund policy that had changed. The team lead put each through the template. Eleven passed with an owner and a test date, fifteen were merged or retired, and four that every agent used daily became skills.
The four skills went into the team's pod, so each agent's Claude or ChatGPT reads them without anyone opening the doc. When the refund policy changed again, the owner edited the refund skill once and every agent's next reply used the new rule.
AI Skill Pod is this site's product. Pod pricing: Free up to 3 members and 10 skills; Team $5 per seat a month from 3 seats.
Next step: make your first skill
Pick the library prompt people use most and turn it into a skill. How to make your first Claude skill from a doc you have walks through it.
Skills for this
- AI practiceTurn a prompt into a skillTurns a prompt that works into a skill the whole team can reuse, with a valid name, a when-to-use description, steps, an output template, an example, and checks.
- AI practiceContext brief for an AI assistantWrites instructions for an AI assistant or agent: role, audience, can and can't, sources of truth, tools, format, examples, edge cases, and handoff.
- Operations and adminSOP from an explanation or recordingTurns an explanation or screen-recording transcript into an SOP with a trigger, roles, exact-label steps, decision points, checks, and an owner.
Questions
What should a team prompt library include?
For every prompt: a name that says the task, one line on when to use it, the inputs to paste, the prompt itself, one good example output, an owner, and the date it was last tested with which AI. Sort entries by job, and add a status so people can tell an approved prompt from a draft.
Where should a team keep its prompt library?
In one place everyone already opens, such as a shared doc, a wiki page or a database, with one owner for the library. What matters more than the tool is that each entry has the same fields and that someone reviews them each month.
When should a prompt become a skill?
When several people use it every week, or when its output must follow the team's rules whoever asks. A skill is read by the AI itself, so nobody has to find, copy and paste the prompt, and an edit reaches everyone at once.
AI skill
Have your AI turn a prompt that works into a skill your team can reuse
Put this skill in AI Skill Pod and everyone on your team gets it in their Claude, Cursor or ChatGPT (Pro, Business, Enterprise): one invite link, free to start. Turn a prompt into a skill has your AI turn a library prompt that works into a skill: a valid name, a description that says when to use it, numbered steps with your rules, an output template, one example and checks.
14 days free · No card needed
Sources
- Claude Docs: Skill authoring best practices — read at 19:20 JST: the description says what a skill does and when to use it; consistent naming; concrete examples; avoid time-sensitive information; test with real usage and share with teammates to see what's missing (read 2026-11-05)
- Agent Skills: Specification — a skill is a folder with a SKILL.md: a name and a description, then the instructions (read 2026-11-05)