Specification feature · runtime

Agent Skills manual invocationAgent Skills specification

Let a user or agent explicitly select and activate a named skill.

Terminology basis: Agent Skills specification. Agent Skills specification

Agent Skills manual invocation: 2 supported, 2 partial, 0 unsupported, 27 unreviewed across 31 cataloged products.

Markdown · JSON

Explore this familyMore in Agent Skills6 capabilities

Current evidence by product

Can my agent use Agent Skills manual invocation?

Read across for the answer. 4 of 31 current product columns have reviewed evidence; unreviewed does not mean unsupported.

  • Supported2
  • Partial2
  • Unsupported0
  • Unknown27
  • Not applicable0

Web

9 products

Desktop

13 products

CLI

9 products

Unknown means no public evidence has been reviewed for that product and capability. It does not mean unsupported.

How statuses are assigned

Definition and scope

What this capability means

Let a user or agent explicitly select and activate a named skill.

Support is recorded only when exact, dated evidence covers this behavior.

Traceable compatibility

Assertion ledger

Documentation evidence only. No runtime conformance test is implied.

Claude CLIcli · current
Partial
Target
current Claude Code skills documentation · dated-documentation
Environment
preview-enabled
Observed
2026-08-28
  • previewthe catalog's Claude Code Agent Skills core row remains qualified as beta
Evidence
Cursordesktop · current
Supported
Target
current Cursor Agent Skills documentation · dated-documentation
Environment
local-default
Observed
2026-08-28

No additional condition is recorded for this statement.

Evidence
Gemini CLIcli · current
Partial
Target
current Gemini CLI Agent Skills documentation · dated-documentation
Environment
local-default
Observed
2026-08-28
  • runtimenamed enable, disable, reload, and link controls plus per-activation consent are documented; direct `/skill-name` invocation was not established by the reviewed guide
Evidence
Copilot Chatdesktop · current
Supported
Target
current VS Code Agent Skills documentation · dated-documentation
Environment
local-default
Observed
2026-08-28

No additional condition is recorded for this statement.

Evidence
  1. 1. Evidence checked 2026-08-28: Claude Code exposes discovered skills as `/skill-name` commands and supports `disable-model-invocation: true` for user-only activation.
  2. 2. Evidence checked 2026-08-28: Cursor lets users invoke a discovered skill by selecting `/skill-name` in Agent chat and can disable automatic model invocation.
  3. 3. Evidence checked 2026-08-28: Gemini CLI lets users list, enable, disable, reload, and link named skills, and requires user consent whenever a matched skill activates; the reviewed guide does not document a direct `/skill-name` invocation form.
  4. 4. Evidence checked 2026-08-28: VS Code lists skills in the `/` menu, accepts `/skill-name` with optional context, and supports manual-only skills through `disable-model-invocation`.