list_agents
List every Agent definition this credential can use. The returned qualified names are the exact
addresses accepted by get_agent, explicit bootstrap
selection, and agent_context. This tool takes no parameters.
Parameters
None.
Response
| Field | Type | Description |
|---|---|---|
agents | array | Every currently visible Agent definition. |
agents[].name | string | Exact scope-qualified MCP address. |
agents[].description | string | Description when one is present. |
agents[].scope | string | Ownership scope such as mine, team, or organization. |
For an organization caller, me/name is the represented member's private organization space,
<team>/name belongs to a bound team, and a bare name is from the organization catalogue. A
standalone personal caller receives bare names. UUIDs are deliberately not part of the MCP address.
Every listed name round-trips through Agent reads; only personal, me/, and team-owned addresses
are currently writable over MCP.
Example
Result:
{
"agents": [
{ "name": "me/drafter", "scope": "mine" },
{
"name": "engineering/reviewer",
"description": "Reviews risky changes",
"scope": "team"
},
{ "name": "release-steward", "scope": "organization" }
]
}
Errors
| Code | Cause |
|---|---|
capability_denied | The credential lacks Agents. |
internal_error | An unexpected server error occurred. |
Notes
- The tool is governed: it requires Tools and tool-policy admission. Its content read also requires Agents; it does not require create, update, or delete permission.
- Visibility includes only the credential's usable ownership/team/policy set. Hidden definitions are omitted.