Skip to main content

Message, Nodes, and Agents List Tools

message — tool.message​

Sends a message to an operator-configured channel (Telegram, Slack, webhook, etc.).

Action class: send-message | Side effects: external_side_effect | Approval: may require

Input​

FieldTypeDescription
channelstringOperator-registered channel ID
contentstringMessage content
formatenumtext, markdown, or json
recipientstringOptional recipient within the channel
threadIdstringOptional thread to reply into

Safety notes​

  • Only operator-registered channels are reachable
  • All messages are logged in the audit trail
  • Sensitive content must not be sent without explicit operator policy

Default Set​

Included in manasvi.toolset.controlled-write.

nodes — tool.nodes​

Inspects the distributed node manager and remote execution surfaces.

Action class: inspect-node | Read-only | Approval: none (for read operations)

Operations​

OperationDescription
listList all available nodes with status
inspectDetailed info for a specific node
capabilitiesList capabilities of a node
dispatchDispatch a tool invocation to a specific node (approval required)

Safety notes​

  • List/inspect/capabilities are read-only and safe
  • dispatch is approval-sensitive and requires additional policy
  • Node capabilities define what tools can run on that node

Default Set​

Included in manasvi.toolset.starter-read.

agents-list — tool.agents-list​

Lists available agent definitions visible to the calling principal.

Action class: list-agents | Read-only | Approval: none

Returns agent metadata: capabilities, tool IDs, version, status, and owner. Used for agent discovery before spawning subagents or routing tasks.

Output​

{
  "agents": [{
    "agentDefinitionId": "agent-def:summariser",
    "name": "Summariser",
    "capabilities": ["web.search", "memory.write"],
    "toolIds": ["tool.web-search", "tool.memory-note-write"],
    "status": "active"
  }],
  "total": 1
}

Default Set​

Included in manasvi.toolset.starter-read.

See also​