Skills & Capabilities

Skills & Capabilities lists every skill the agent currently has — the platform-supplied skills, the skills added through Apps & Services connections, the tools exposed by attached MCP Tools, and any custom skills your team has authored. The list is the agent's complete vocabulary of actions: at runtime, the planner picks from exactly this set when it decides what the agent should do next.

Manage Agent → Skill & Capabilities, listing the agent's pre-attached skills

Reach this page from Manage Agent → Skill & Capabilities in the left navigation.

The section header shows the total count — Pre-attached Skills (<N>) — where <N> is the number of skills currently attached.

The skill list

Each row shows:

• The skill's icon (varies by type)

• The skill's identifier (example display_to_user)

• A one-line description of what the skill does

• A ⋮ overflow menu on the right for managing the row

Platform-supplied (pre-attached) skills

The platform ships a set of skills attached to every agent by default. These are the basic verbs every agent needs. The ones observed during capture include:

Skill	What it does
display_to_user	Shows a message to the user in markdown format. Use for quick updates or progress messages — not for the final answer.
reason	Uses AI to think through complex topics, make decisions, or generate summaries. Useful for reasoning or language-based tasks.
web_search	Performs a web search using the language model with web search enabled. Useful for retrieving current information. Input: a query string and optional context.
query_knowledge	Searches the connected knowledge base to find relevant answers or references. Use when the information is stored in attached knowledge.
list_files	Shows all files uploaded in the current session along with their names, details, and upload times.

The full pre-attached set in an example agent is broader (the section header in the capture showed Pre-attached Skills (19)); the table above is representative rather than exhaustive.

The row's ⋮ menu

The kebab menu on each row exposes per-skill actions — typically rename, configure, remove. These actions weren't exercised during capture.

Note — Depth gap The exact items in the ⋮ menu for each skill row were not enumerated. The standard pattern across the platform's other tables is "view / edit / remove," but the specifics are queued for a future pass.

Adding skills to this list

Skills land in this list from three places:

• The platform — pre-attached skills are added automatically when the agent is created.

• The canvas + Add Skill button — see Adding a skill. Skills from the Skill Library, Custom Skills, the MCP Marketplace, and the Agents tab are added through this flow.

• Connector configuration — when you connect an integration in Apps & Services or an MCP server in MCP Tools, the connector registers one or more skills into this list.

The reverse direction — removing a skill — is presumably reachable through the row's ⋮ menu (above). Confirm with a small test before removing a skill from a production agent, since other parts of the agent's configuration (instructions, checklists, triggers) may reference the skill by name.

How skills relate to the canvas

The Layout canvas shows skills in two forms:

• The Pre-attached Skills card (with a +<N> overflow indicator) summarizes the platform-supplied set without listing each one.

• The individual connector discs — accountancy, credit-score, banking, and so on — each represent one attached MCP, whose tools are registered as skills in this section.

Use Skill & Capabilities when you need to see the full, flat list. Use the canvas when you want a quick visual confirmation that the agent has, say, banking and analytics connectors attached.

Why this matters

This list is the planner's universe. When a user sends the agent a request, the platform's planner reads the instructions, considers the input, and picks one or more skills from this list to execute. A skill that isn't in the list cannot be used, no matter what the instructions say. A skill in the list whose description doesn't match the kinds of requests users send won't be picked, no matter how capable it is. Auditing this list — does every skill the agent needs appear? does every skill that appears earn its keep? — is the single highest-value review you can do when an agent isn't behaving as expected.