OpenClaw coordinates specialists in Telegram topics
Malone reports four OpenClaw agents running in separate Telegram topics with tested cross-agent messages, after fixing routing, session discovery and memory-sync faults.
dan-malone.com
Original prompt
Shared by the source author. See source notes for context.
Use sessions_send to send a message to the finance agent session key agent:finance:main. Say: Hey Fin, testing comms.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Dan Malone
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Malone reports four OpenClaw agents running in separate Telegram topics with tested cross-agent messages, after fixing routing, session discovery and memory-sync faults.
- Human involvement
- Malone designs the roles, diagnoses configuration failures and tests handoffs. Several proposed operational automations remained future work at publication.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Create separate agents, workspaces and Telegram forum topics.
- Route each bot account to its topic with appropriate mention rules.
- Set an explicit default agent and initialise specialist sessions.
Dan Malone wanted the specialised contexts he used at his desk to remain accessible from his phone and communicate without him relaying every message. He built an OpenClaw team in a Telegram forum: a general assistant, a home-automation agent, a tasks agent and a finance agent. Each had its own workspace and topic, with mention requirements controlling when it responded.
The actual setup encountered several failures. Global group settings made all bots reply in every topic. An implicit default selected the wrong agent. Cross-agent messages failed because session visibility and initialisation were incomplete. Malone corrected the account-specific routing and session settings, then tested a message from the home agent to the finance agent and received a reply.
A scheduled repository sync also exposed agent-specific memory files to a shared workspace. He corrected ignore rules and removed the tracked files. The source states that the team was running, but finance integration, morning email triage and meeting preparation remained roadmap items. The reusable lesson is to prove routing and handoffs with a small test before claiming operational workflows are complete. Separate workspaces help preserve roles, but shared credentials and automatic sync require explicit inspection rather than assumptions about isolation.
Prompt
Original prompt
Source: Blog · Dan Malone
Frequently asked questions
What was actually demonstrated?
Four agents running in separate topics and a successful test message and reply between specialist sessions.
Was bookkeeping already integrated?
No. Xero integration for the finance agent was explicitly listed as pending on the roadmap.
What went wrong with shared storage?
A sync job committed agent-specific memory into a shared repository; Malone corrected ignore rules and removed the tracked files.
More details
Remaining steps
- Configure permitted cross-agent communication and test a handoff.
- Exclude private memory and configuration before scheduled repository sync.
About this account
The configured agents keep separate working contexts and can exchange messages directly; persistent communication was demonstrated with a successful test handoff.
Firsthand account by Dan Malone; original publication date verified against Primary page JSON-LD datePublished 2026-02-18. Outcomes are the author’s reports, not independently reproduced. Original prompt excerpt transcribed from the primary account; excerpt only, not the complete prompt. Preserve the source-specific context when adapting it.
Credit: Dan Malone
Source published
Added to Agents In Use
Source checked
Tools and topics
Telegram forum · OpenClaw workspaces · Git · Notion MCP · Granola MCP
Specialist agents · Topic routing · Context boundaries
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.