The Skills page in PushMetrics, with the list of skills on the left and the markdown editor for the selected skill on the right

A skill is a short instruction sheet you write once for your agents to follow. Think of it like a recipe or a playbook. Anything your team does the same way every time (the weekly pipeline report, the monthly board update, the new-customer welcome sequence) is a good candidate for a skill.

Skills are written in plain markdown and saved in your workspace. Once one exists, you can ask any agent that has access to it to "run the weekly pipeline report" and it'll follow the steps you wrote.

Where skills live

Skills are stored as blocks in your workspace, just like SQL queries, email templates, and metrics views. You manage them at Settings → AI Capabilities → Skills.

The Skills page has the full list on the left and an editor on the right. The editor is split: the markdown source on one side, a live preview on the other. Every save creates a new version in the skill's history, so you can see what changed and who changed it (including changes made by an agent).

What does a skill look like?

A skill is just markdown. Here's a simple one:

## Weekly Pipeline Report

When someone asks for the weekly pipeline report:

1. Run the "Pipeline Summary" SQL block to pull this week's data.
2. Build a bar chart of deals by stage.
3. Post the chart and a short summary to #sales-updates on Slack.
4. Email the full table to the "Sales Leadership" group.

Always include a week-over-week comparison in the Slack message.

Without a skill, you'd have to explain those four steps every time. With a skill, you write it once and just say "run the weekly pipeline report".

Creating a skill

1
Open Settings → AI Capabilities → Skills and click New Skill. You'll get a fresh markdown file to edit.
2
Give it a clear name. The name and the first line of the body are what an agent uses to decide whether the skill applies to the current question, so make them descriptive.
3
Write the steps in markdown. Be specific. Mention the actual SQL block, table, channel, or recipient group you want the agent to use.
4
Save. The skill is now part of the workspace. Whether it shows up for a particular agent depends on that agent's settings (next section).
📥
If you already have a folder full of markdown playbooks, you can bulk-import them. The Skills page has an upload action that takes a set of .md files and creates one skill per file.

What makes a good skill

✅
A good skill is specific (real step-by-step actions, not vague guidance), names real things (the actual block, table, channel, or recipient group), and says what success looks like (so the agent knows when it's done).

Things that tend to make a skill flaky:

  • Vague verbs like "analyze the data" or "summarise nicely". Be concrete: "build a bar chart of deals by stage", "include the week-over-week change in the first paragraph".
  • Block names that don't actually exist in the workspace. The agent will go looking and come back empty-handed.
  • Two skills that overlap in scope. The agent has to guess which one to use, and may pick the wrong one.

Which agents see which skills?

Each agent decides for itself which skills it has access to. You'll find this on the agent's edit page, in the Skills section.

The Skills section on an agent's edit page: a radio toggle between using all workspace skills or picking specific ones, with checkboxes for each skill

There's a radio toggle with two modes:

Use all workspace skills
The default. The agent automatically gets every skill in the workspace, including new ones added later. Best for general-purpose agents.
Pick specific skills
A checkbox list of every skill in the workspace. Only the ones you tick are loaded. Best for specialised agents like a Sales Assistant that shouldn't try to run engineering playbooks.

The checkboxes save the moment you tick or untick them, no separate Save button needed. Each skill in the list also has a small link to its detail page so you can read what's inside before assigning it.

✨
If a "pick specific" agent invents a new skill mid-conversation, it'll keep it. When the agent creates a new skill while you're chatting, the skill is automatically added to its own assigned list. The agent will mention it in chat, and you can untick it later on the agent's edit page if you'd rather curate manually.

How an agent uses a skill

When an agent's session starts, the system prompt includes an index of every skill the agent has access to: the name, a link to its detail page, and a one-line description (the skill's first heading, basically). The full body of each skill is not loaded up front, since that would burn tokens fast.

When the agent decides a skill is relevant, it calls a tool to read the full markdown:

FIND match question to a skill READ load the full markdown RUN work through the steps ADJUST adapt to today

You can also tell the agent which skill to use: "Use the Weekly Pipeline Report skill." That's the most reliable way to make sure it picks the right playbook on the first try.

Editing skills (yours and the agent's)

Skills live in the same place no matter who wrote them. The agent uses the same set of block tools you'd use through the UI:

  • list_blocks to see what's available.
  • read_block to open one.
  • create_block to add a new skill.
  • edit_block to update an existing one.

Every change creates an entry in the skill's history. Edits made by an agent are tagged differently from edits made by a person, so you can spot them at a glance and roll back if needed.

Skill vs knowledge vs memory

It's easy to mix these up. The short version:

  • A skill is how to do something. Step-by-step instructions for a recurring task.
  • A knowledge entry is something true about your business. Definitions, conventions, what columns mean.
  • A memory note is something the agent learned for itself, kept between conversations.

See Memory & Knowledge for more on the last two.