| name | discord-reactions |
| description | Use this skill when communicating on Discord to decide whether to use emoji reactions or messages. Also use when asked about "reaction protocol", "discord efficiency", or how agents should communicate. |
Discord Reaction Protocol for Efficient Agent Communication
Goal
Reduce unnecessary token usage by using emoji reactions for simple acknowledgments and status updates instead of full text messages.
When to Use Reactions vs. Messages
Use Reactions For:
- Acknowledgments: "I see this", "got it", "understood"
- Status updates: "working on it", "done", "blocked"
- Simple approvals: "looks good", "approved", "ship it"
- Quick feedback: "agree", "disagree", "needs clarification"
- Task tracking: "claimed this", "reviewing", "testing"
Use Messages For:
- Complex responses: Explanations, questions, detailed feedback
- Decisions requiring context: Why you chose an approach
- Blocking issues: Problems that need discussion
- Clarifying questions: When requirements are unclear
- Important updates: User needs to know details
Golden Rule: If it would take more than 5 words to explain, use a message. Otherwise, use a reaction.
Standard Reaction Vocabulary
Acknowledgment & Status
| Emoji | Meaning | When to Use |
|---|
| 👀 | "Seen/Reading" | You've seen the message and are reviewing it |
| 👍 | "Acknowledged/OK" | Simple agreement or acknowledgment |
| ✅ | "Done/Complete" | Task completed successfully |
| ⏳ | "In Progress" | Working on it now |
| 🔄 | "Processing" | Thinking/investigating, will respond soon |
| 🚫 | "Can't do/Blocked" | Unable to proceed (follow up with why) |
Task Management
| Emoji | Meaning | When to Use |
|---|
| 🎯 | "Claimed/Taking this" | You'll handle this task |
| 📋 | "Ticket created" | GitHub issue created |
| 🔍 | "Reviewing" | Code review or investigation in progress |
| 🧪 | "Testing" | Running tests or QA |
| 🚀 | "Deployed/Shipped" | Code merged or deployed |
| 📦 | "Ready for review" | Work complete, needs approval |
Feedback & Decisions
| Emoji | Meaning | When to Use |
|---|
| ✔️ | "Approved" | PR/issue/approach approved |
| ❓ | "Need clarification" | Requires more information (follow up with question) |
| 💡 | "Suggestion/Idea" | Have an alternative approach (explain in message) |
| ⚠️ | "Warning/Concern" | Potential issue (explain in message) |
| 🔥 | "Urgent/Priority" | High priority item |
| 🐛 | "Bug found" | Issue discovered |
Priority Levels
| Emoji | Meaning | When to Use |
|---|
| 🔴 | "P1/Critical" | Blocking issue, needs immediate attention |
| 🟡 | "P2/Important" | Should be done soon |
| 🟢 | "P3/Normal" | Regular priority |
| ⚪ | "P4/Low" | Nice to have |
Communication Patterns
Pattern 1: Simple Acknowledgment
Scenario: User asks you to create a ticket
- Bad: "Got it! I'll create that ticket now."
- Good: React with 👍 (then create ticket)
Pattern 2: Task Completion
Scenario: You finished creating tickets
- Bad: Long message with all details
- Good: React with ✅ + short summary message with links
Pattern 3: Status Update
Scenario: User asks if you're working on something
- Bad: "Yes, I'm currently investigating this issue and will report back."
- Good: React with ⏳ or 🔍 (they know you're on it)
Pattern 4: Agreement with Other Agent
Scenario: Another agent posts analysis you agree with
- Bad: "I agree with @dev's approach here."
- Good: React with ✅ or 👍 on their message
Pattern 5: Claiming Work
Scenario: User posts request, multiple agents could handle it
- Bad: "I can take this one!"
- Good: React with 🎯 (claims it)
Pattern 6: Escalation Needed
Scenario: You're blocked and need help
- Good: React with 🚫 + message explaining blocker
Agent-Specific Patterns
@po (Product Owner)
- 📋 = Ticket created
- 🎯 = Added to backlog
- ✔️ = Requirements approved
- 📦 = Ready for dev
@dev (Developer)
- 🎯 = Claimed ticket
- ⏳ = Working on it
- 📦 = PR ready for review
- ✅ = Merged/deployed
@architect (Architect)
- 🔍 = Researching / Investigating
- 📊 = Analysis in progress
- 💡 = Have recommendation
- ⚠️ = Found concerns
@qa (QA)
- 👀 = Reviewing/Testing PR
- ✅ = QA Approved
- ❌ = QA Failed
- 🔄 = Re-testing after fixes
Example Conversation (Before vs. After)
Before (Verbose):
User: Create a ticket for dark mode
PO: Got it! I'll create a ticket for dark mode support right now.
PO: Ticket created! Here's the link: github.com/...
User: Thanks!
Dev: I see the ticket, I can work on this.
PO: Great, thanks for taking it!
Dev: Started working on it.
Dev: PR is ready for review.
Reviewer: I'm reviewing it now.
Reviewer: Looks good, approved!
Dev: Thanks, merged it.
Token usage: ~1000 tokens
After (Efficient):
User: Create a ticket for dark mode
PO: 👍 (reacts immediately)
PO: ✅ Ticket #26 created: github.com/...
Dev: 🎯 (claims ticket)
Dev: ⏳ (starts work)
Dev: 📦 PR #27 ready: github.com/...
Reviewer: 🔍 (reviewing)
Reviewer: ✔️ (approved)
Dev: ✅ Merged
Token usage: ~400 tokens (60% reduction!)
Using the discord_add_reaction Tool
Agents can now add reactions to Discord messages using the discord_add_reaction tool.
Tool Signature
discord_add_reaction({
channel: string,
messageId: string,
emoji: string
})
Example Usage
Acknowledging a user message:
discord_add_reaction({
channel: "development",
messageId: "1471958558642929757",
emoji: "👍"
})
Claiming a ticket:
discord_add_reaction({
channel: "development",
messageId: "1471958558642929757",
emoji: "🎯"
})
Showing work in progress:
discord_add_reaction({
channel: "development",
messageId: "1471958558642929757",
emoji: "⏳"
})
Approving a PR announcement:
discord_add_reaction({
channel: "development",
messageId: "1471958558642929757",
emoji: "✔️"
})
Getting Message IDs
Message IDs are available in the discord_read_channel response:
{
"messages": [
{
"id": "1471958558642929757",
"author": "username#0",
"content": "Create a ticket for dark mode"
}
]
}
Multiple Reactions
You can add multiple reactions to the same message by calling the tool multiple times:
discord_add_reaction({ channel: "development", messageId: "123", emoji: "👀" })
discord_add_reaction({ channel: "development", messageId: "123", emoji: "✅" })
Implementation Guidelines
For Agents:
- React first, respond if needed: If simple acknowledgment suffices, just react
- Use reactions for status: Don't say "I'm working on it", react with ⏳
- Combine reactions + brief messages: React ✅ + short summary with links
- React to other agents: Show agreement/acknowledgment with reactions
- Follow up complex reactions with details: 🚫 or ⚠️ always needs explanation
For Users:
- React to acknowledge agent messages: Let agents know you saw it
- Use reactions for approval: ✔️ instead of "looks good"
- Use reactions to prioritize: 🔴/🟡/🟢 to indicate urgency
- Expect reactions from agents: Silence + 👀 reaction means "I'm on it"
Benefits
- 60-80% token reduction on simple acknowledgments
- Faster communication - no waiting for full message generation
- Clearer status tracking - visual indicators at a glance
- Less cognitive load - skim reactions instead of reading paragraphs
- Cost savings - fewer API calls, lower token usage