skills.rest
Prismer-AI
Prismer-AIOFFICIAL
Explore all skills by this author →
WebsiteGitHub

prismer-im-collab

Coordinate multi-agent conversations and task routing in Prismer workspaces via MCP tools.

1.4k|12|Updated Mar 21, 2026
Search Tags:#delegation#task routing#Collaboration#group chat#prismer#MCP Tools#Multi-Agent
One-click install
npx skills add https://github.com/Prismer-AI/PrismerCloud --skill prismer-im-collab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: prismer-im-collab
Source: https://github.com/Prismer-AI/PrismerCloud/tree/main/sdk/prismer-cloud/built-in-skills/prismer-im-collab
Command: npx skills add https://github.com/Prismer-AI/PrismerCloud --skill prismer-im-collab

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

In multi-agent chat workspaces, agents frequently fail at basic coordination: they impersonate other participants, @-mention display names that never route, hallucinate teammates that don't exist, or poach tasks assigned to other roles. This Skill enforces the channel-specific rules that make agent replies actually reach the right recipient.

Core Features & Use Cases

  • Identity and routing rules: Hard constraints prevent agents from impersonating the human owner or other agents, and enforce @-mentioning ASCII usernames (never display names or internal ids) so group messages route correctly.
  • Grounded participant discovery: Requires calling prismer.conversation.listAgents before naming or delegating to any teammate, eliminating hallucinated org charts.
  • Task board discipline: Agents only act on Kanban cards assigned to them, using prismer.task.create/update/complete and structured human approvals for sensitive operations.
  • Workspace asset access: Search, describe, and read bounded byte ranges of uploaded files via prismer.asset.* tools instead of asking users to paste content.
  • Use Case: In a group conversation with ceo, engineer, and marketer agents, the engineer grounds the participant list, completes only its own task card, and hands off the turn with @ceo so the discussion chain continues.

Quick Start

Ask the agent to coordinate with teammates in a Prismer group conversation, for example: "List the agents in this workspace and delegate the market research outline to the researcher."

Frequently Asked Questions about prismer-im-collab

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I route a message to another agent in a Prismer group chat?▼

Mention the agent's ASCII username with @, for example @engineer, or call prismer.agent.send for verified delivery. Never @ the displayName or internal id, since the mention parser only matches ASCII usernames and silently fails otherwise.

Why does my @-mention not reach the teammate in group conversations?▼

The mention parser only matches @ followed by ASCII characters or a quoted string. Writing a localized displayName like @工程师 or an internal id routes nowhere and the chain dies silently. Call prismer.conversation.listAgents and use the returned username verbatim.

When should I use prismer.agent.send instead of an inline @-mention?▼

Use prismer.agent.send for explicit hand-offs and delegation because it returns delivery confirmation with message_id and dispatched_to. Inline @-mentions work for casual continuations but give no error feedback if the username is wrong.

Can an agent work on a Kanban task assigned to another role?▼

No. Agents may only claim, start, or complete cards assigned to themselves; the server returns 403 on unauthorized transitions. If a card should move, @ the orchestrator or task owner and let them reassign it.

How do agents read uploaded workspace files in Prismer?▼

Call prismer.asset.search to find files, prismer.asset.describe to inspect metadata, then prismer.asset.read to fetch only the needed byte range. Never load entire large files into a reply, and cite the asset id when using file evidence.

What are the limitations of multi-agent chains in Prismer conversations?▼

Multi-agent chains are capped at roughly five hops, self-mentions are filtered server-side, and messages without an @-mention in group chats are broadcast only with no agent dispatched. End the turn explicitly when the discussion is concluded.

Related Skills

Explore More Software Engineering Skills→
CannerCanner
663

wren-mcp-setup

Deploy and configure a containerized Wren Engine MCP server for AI agents.

Official
Advanced
Pact-Community-OrganizationPact-Community-Organization
1

stakeholder-reporting

Generate sprint reviews and milestone updates from project data.

Official
Intermediate
NaveedTechLabNaveedTechLab
11

a2a-messaging

Enable real-time message passing between distributed AI agents for task coordination.

Community
Advanced
BuilderIOBuilderIO
3

delegate-to-agent

Delegate AI tasks to agent chat via sendToAgentChat and agentChat APIs.

Official
Advanced