Card 04 · AI agents, explained
What is an AI skill? Write the method once, anyone's agent can run it
An AI skill is a written method an agent loads when a task matches: the steps, the rules and the examples, saved as a file instead of retyped in a chat. It turns one person's good process into something the whole team's agents run the same way.
01 — The habit
The good method lives in one person's chat.
Priya has spent a year getting Friday's report right, and all of it sits in her own message history. Nobody else's agent can run that, and she cannot be on holiday twice.
02 — Written once
A skill is that method, saved as a file.
Four headings and a few lines each: when to use it, what to read, what to do, how to check. It is short because it is read by an agent and by whoever reviews it.
03 — The trigger
It says when it applies, and when it does not.
Every Friday, once the week's timesheets are in, and never before Monday. The bound matters more than the steps: an agent that loads the method for the wrong job still does the wrong job.
04 — What to read
It names the sources, not everything.
The timesheet folder and last week's report, and nothing else. A method that says what to leave out is what stops the report drifting into every account in the firm.
05 — What to do
Short steps, in the order the work happens.
Total the hours per client, compare with last week, write the two paragraphs that changed. Written as steps, the same run comes back in the same shape every week.
06 — How to check
It says how to know it worked.
Every client's hours match the timesheet, and no number in the report is older than Friday. That line is what turns a draft nobody trusts into one a person can sign.
07 — Who runs it
Anyone's agent, the same way.
Priya's agent and Leo's agent load the same file, so Friday looks the same whoever is out. TalentOS keeps a public library of these at talentos.so/skills, so a good method spreads without a meeting.
08 — Where it lives
It lives with the work and is reviewed like it.
The file is text in a repository, so a change to the method shows up as a diff somebody approves. Start with the one task your team does every week and get that one file right.
Run by Priya's agent and Leo's agent.
Loaded on Friday at two, by both agents, from the same file.
Tools for this
Every price and licence below is what the tool’s own page said on 18 November 2026, with two sources behind each record. There are no scores and no order of preference; where TalentOS is one of the tools it is marked ours and listed like the rest.
An AI code editor whose agent edits across files, runs commands, reviews its own changes and can run tasks in the cloud.
For a developer who wants an agent inside the editor they already use, with team rules and skills shared across the codebase.
$8 per month (seat) · Closed source
A workspace where one assistant answers the whole team in a shared thread, backed by a memory of Markdown files that every connected AI can read.
For a team of about 5 to 15 people that wants one shared assistant plus one memory, and wants that memory kept as plain files it owns.
See its pricing page · Open source, MIT
An open format for writing a skill as a folder with a SKILL.md file: instructions, examples and any scripts, loaded by an agent when a task matches.
For a team that wants one person's method reusable by everyone's agent instead of retyped into each chat.
$0 (free) · Open source, Apache-2.0
Agent Skills format and documentationAnthropic skills repository and licence
A published set of composable skills plus a small bootstrap that makes a coding agent check for the right skill before it starts a task.
For a team that wants a shared working method its agents follow, instead of each developer describing the process again in every session.
$0 (free) · Open source, MIT
Superpowers repository and licenceSuperpowers plugin listing
Anthropic's coding agent that runs in the terminal or an IDE and works on a real repository: it reads files, edits them, runs commands and commits.
For a developer or a small engineering team that wants an agent working inside its own codebase, not in a chat window.
$20 per month (seat) · Closed source