# Yotto3s Oneteam Agents Team Collaboration > Team Collaboration - Skill: `tomevault-io/yotto3s-oneteam-agents-team-collaboration` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/yotto3s-oneteam-agents-team-collaboration` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/yotto3s-oneteam-agents-team-collaboration/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/yotto3s-oneteam-agents-team-collaboration --- # Team Collaboration ## Overview Four principles for effective team collaboration in multi-agent contexts. ## When to Use - Spawned with `mode: team` in initialization context - Need to communicate with teammates or leader - Blocked and need to escalate - Need to discover who owns what scope ## When NOT to Use - In subagent mode (default) -- agents skip team behaviors and return output directly ## Team Mode Activation Protocol When an agent's initialization context includes `mode: team`, apply these behaviors on top of the agent's base workflow: | Behavior | Team mode action | |----------|-----------------| | Startup | Send ready message to leader via SendMessage: `" ready. Worktree: , scope: ."` | | Progress | Send periodic updates to leader via SendMessage | | Completion | Send report to leader via SendMessage instead of returning as output | | Escalation | SendMessage to leader with context instead of returning with `ESCALATION NEEDED` flag | | Blocking | SendMessage stating what is needed and from whom instead of returning partial result | | Teammate discovery | Read team config to learn names and roles | In subagent mode (default), agents skip all of the above and return their completion report as final output. If stuck, they return partial results with an `ESCALATION NEEDED` flag instead of blocking. ## Principles ### 1. Close the Loop Every message that requests action or delivers work gets confirmed. No message disappears into the void. **How to keep it:** - When you receive work (findings, feedback, instructions): acknowledge receipt and state what you will do next. - When you hand off work: state what you delivered and what you expect the receiver to do. - When you finish a task: report the outcome to whoever assigned it. - When you send a message and get no response: follow up -- do not assume it was received. ### 2. Never Block Silently If you cannot make progress, say so immediately. Silent waiting causes cascading delays. **How to keep it:** - When you are waiting for input from another teammate: send a message saying what you need and from whom. - When you hit an unexpected obstacle: notify the leader with what is blocking you and what you have tried. - When a task is taking longer than expected: send a status update before anyone has to ask. - Never assume someone knows you are stuck -- tell them. ### 3. Know Who Owns What Every team member knows who is responsible for what area. You do not need to know the details of their work -- you need to know who to ask. **How to keep it:** - On startup: read the team config to learn who your teammates are and what they are working on. - When you need information outside your scope: message the owner directly rather than guessing or investigating yourself. - When your scope changes: notify affected teammates. - Do not broadcast full status to everyone -- share details only with those who need them. ### 4. Speak Up Early Raise concerns and doubts immediately, no matter how small. A small question now prevents a big problem later. **How to keep it:** - When something feels wrong but you are not sure: ask the relevant teammate or leader now. - When you notice a potential conflict with another teammate's work: mention it immediately. - When instructions are ambiguous: ask for clarification before proceeding. Do not interpret silently. - When you disagree with an approach: state your concern with reasoning -- then defer to the decision maker. ## Quick Reference | Principle | Core Rule | Key Practice | |-----------|----------|-------------| | 1. Close the Loop | Every message gets confirmed | Acknowledge receipt, state next action | | 2. Never Block Silently | Say so immediately if stuck | Message what you need and from whom | | 3. Know Who Owns What | Know who to ask | Read team config, message owner | | 4. Speak Up Early | Raise concerns immediately | Ask now, do not interpret silently | ## Common Mistakes | Mistake | Fix | |---------|-----| | Broadcasting status to all | Share details only with those who need them (Principle 3) | | Waiting silently when blocked | Message immediately stating what you need (Principle 2) | | Not acknowledging received work | Confirm receipt and state next action (Principle 1) | | Investigating teammate's scope | Message the owner instead of guessing (Principle 3) | | Interpreting ambiguous instructions | Ask for clarification before proceeding (Principle 4) | ## Constraints - ALWAYS send a ready message on startup in team mode - ALWAYS acknowledge receipt of work and state next action - NEVER block silently -- report blockers immediately - NEVER investigate outside your scope -- message the owner - NEVER broadcast when a direct message suffices --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/yotto3s) — claim your Tome and manage your conversions.