Skills

A skill is how your agent does a kind of task: the steps, what the result looks like and the rules to follow. For example, "Rewrite for social media" spells out how to write a Xiaohongshu note, an X thread or a WeChat article, and "Earnings quick read" takes every number from the source. Connections decide which apps your agent can reach; skills decide how a task is done well. Both live in Settings, one section each (Apps and Skills), and every agent can use them.

What it's for and how people use it: Skills on waper.ai.

How your agent uses skills

  • On its own: when what you ask matches a skill, your agent uses it without you choosing first. The steps in the chat show "Used skill · skill name"; select it to open the skill.
  • By name: choose Use a skill from the + in the input box (or type / at the start of the input box) and pick one from the list (your own first, Waper's after). A skill tag appears above the input box; write your request and send, and your agent follows that skill.
  • Questions, lookups and small talk need no skill, and your agent won't force one: each skill says what is not its job, so a passing "how is Nvidia doing?" doesn't get an earnings quick read, and "what's on my calendar tomorrow?" goes straight to your calendar.

Skills carry no permissions of their own: sending to other people, deleting and running commands on your computer follow the usual approval rules. Using a skill costs nothing extra; the turn spends credits for the model as usual.

Skills from Waper

Skill What it does
Rewrite for social media Turns a brief, article or research into an X thread, Xiaohongshu note, WeChat article or LinkedIn post
Research report Sharpens the question, chooses between your material and deep research, puts the answer first and sources every fact
Competitor weekly Sums up a week of moves by a few companies from your watches and library, ranked by impact, with what it means for you
Write a newsletter issue Turns a period of your watches into an issue ready to send
Video and podcast notes Key points by timestamp, notable quotes, the people and companies mentioned
Paper quick read Problem, method, results, limitations and significance, with a link to the paper
Earnings quick read Headline numbers, year-over-year and quarter-over-quarter changes, guidance and management commentary, numbers only from the source
Scheduling Reads your calendar before proposing times and names the time zone; says who an invite will notify, and invites still wait for your OK
Inbox Separates what needs your reply from updates and other people's work; says which messages and how many before changing anything
Shopping Picks by your needs, budget and deadline; checks the exact model, stock and total price; stops at checkout so you pay
Travel Settles dates, budget and documents first, compares two or three options with their tradeoffs; asks before booking and gives you the confirmation number
Meeting notes to actions Decisions and action items from meeting notes, or a brief for a meeting on your calendar; works better with Granola and Google Calendar connected
Learn my writing voice Says what it will read and waits for your yes, then reads emails, messages and documents you wrote and sent, shows you a profile of your voice and saves it as your own skill, "My writing voice"; from then on, drafts it writes for you to send sound like you, and when you edit one it asks whether to add the change

Open a skill in Settings → Skills (‹ Skills at the top left takes you back) to see what it does, its example prompts (select one to say that sentence in your chat with the skill already picked) and its details. The instructions an official skill gives your agent (the original English text) sit in a closed "Instructions for the Agent (original)" section lower on the page; a skill of your own always shows its steps and stays editable.

A skill's page lists the apps it uses, and you can connect any that aren't connected right there. When your agent follows a skill and an app it needs isn't connected, it asks you to connect it first.

An official skill's page has Your instructions: how you want it done, such as "keep LinkedIn posts under 120 words, no hashtags". Your agent follows them every time it uses the skill; the skill itself stays the same. Say "from now on, always…" in a chat after using a skill and your agent adds it here and tells you which line it added.

To drop one you don't want, open it and select Hide: your agent stops using it and it leaves the skills menu in the input box. Show again brings it back, and so does the + next to it in the skills list.

My skills

Save your own way of doing things as a skill, and your agent follows it next time:

  • Ask an agent to write one: in a chat, say "Write a skill for me: …" and describe how you want that kind of task done (when to use it, the steps, the result, the rules). Your agent saves it right away; the change card in the chat opens it, and if you don't want it, tell the agent and it undoes it. If your agent read outside content in that turn (a web page, an email, a transcript), it first shows you the skill and waits for your OK, so instructions hidden in that content can't become something it follows every time. Create → Describe it to your agent next to the title in Settings → Skills puts that sentence in the input box of your chat.
  • Save what just worked: when something went well, say "save how we just did this as a skill" in that chat. The agent turns what you did into a general skill and saves it. Create → Save what just worked in Settings → Skills puts that sentence in your input box.
  • Your agent saves one on its own: when you walk it through a multi-step way of doing something you'll clearly do again, or correct how it did something and say "always do it this way", it saves that as a skill right away and links it in its reply. If you don't want it, just say so and it undoes it, or undo it under What it did in its details. One-off requests, or anything it isn't sure is worth keeping, it doesn't save.
  • Upload a SKILL.md: for a skill you used in another tool, choose its SKILL.md file (a name and a description, then the steps) under Create → Upload a SKILL.md.

Open one of your skills in Settings → Skills (yours are listed first) to:

  • Edit its name, "When to use it" (your agent reads this to decide when to use the skill) and the steps, in the same editor as documents;
  • Turn it off: switch off "Agent uses this skill" and your agent stops using it; it also leaves the skills menu in the input box;
  • Delete it for good.

You can have up to 50 skills of your own, each skill file (SKILL.md) up to 64 KB. Archives or folders with reference files, and importing from GitHub, come later.

Undoing a saved skill checks that it is still the version saved by that action. Later edits, or a new skill created under the same name after deletion, are left intact. Older saves without enough version evidence cannot be undone; the skill content is retained.

Skill format

Skills follow the open Agent Skills standard: a SKILL.md file that starts with a name and a description, followed by the steps in Markdown, optionally with reference files. A single SKILL.md from another tool can be uploaded today.

Using the API and MCP

Developers can list, read, create, change and delete skills with /api/v1/skills, and pass skill (the skill's identifier) when sending a message to use one. See the API reference. In MCP clients such as Claude or Cursor, list_skills and read_skill (in the skills module, run with app_run) show your skills, and passing skill to ask_agent has the agent follow one; see MCP.