messaging-agents_skill

This skill enables you to message other agents on the Letta server to query knowledge, coordinate tasks, and share information.

TypeScript

1.7k

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill letta-ai/letta-code --skill messaging-agents

SKILL.md4.6 KB

Overview

This skill lets you send messages to other agents on the same Letta server to communicate, query, or coordinate. Use it when you need answers from an agent with specialized knowledge or want to consult another agent's memory. It focuses on inter-agent messaging only and does not grant the target agent local access to your environment.

How this skill works

Messages are delivered through a thread-safe conversations API and the target agent receives a system reminder identifying the sender. The caller sees only the target agent's final assistant message (no tool call traces or internal reasoning). For full transcripts or tool activity, use the message-searching utilities to list conversation history for that agent.

When to use it

Ask another agent a direct question about a domain they specialize in.
Query an agent for information stored in their memory blocks.
Coordinate or discuss a task with a separate agent running on the server.
Quickly get a second opinion without deploying a subagent with local access.
Follow up in an existing conversation to continue context-specific dialogue.

Best practices

Always include the sender agent ID when issuing a message so the recipient gets the context reminder.
Keep messages focused and include necessary context in the message body because only the final response is returned.
If the recipient needs access to your codebase or files, deploy them as a subagent via the Task tool instead.
Use agent discovery (by name, tags, or message search) to find the correct target agent before messaging.

Example use cases

Ask a BackendExpert agent about authentication design decisions and token flows.
Query a documentation agent for the location of API docs or design notes stored in their memory.
Request a specialized agent to summarize recent technical discussions it participated in.
Ping a QA agent to get a quick checklist or test case suggestions for a feature.
Continue a previous conversation by referencing the conversation ID to maintain context.

FAQ

No. Target agents cannot access your local environment or run shell commands. Use the Task tool to deploy a subagent if local access is required.

Do I receive the target agent's tool calls and internal reasoning?

No. The messaging API returns only the final assistant message. To inspect tool calls or a full transcript, use the message search/listing tools for that agent.