A2A Communicator
Skill8 starsChanged 10 months agoMITView on GitHub
Inter-group communication skill implementing the A2A (Agent-to-Agent) protocol for information exchange between Skill Cells.
In this repository22 other files it gives its agents.
- Master ArchitectSkill
- Comunicador BackendSkill
- Orquestador BackendSkill
- Comunicador FrontendSkill
- Orquestador FrontendSkill
- Comunicador DevOpsSkill
- Orquestador DevOpsSkill
- Comunicador SeguridadSkill
- Orquestador SeguridadSkill
- Comunicador Data/MLSkill
- Orquestador Data/MLSkill
- Comunicador TestingSkill
- Orquestador TestingSkill
- Comunicador DocumentaciónSkill
- Orquestador DocumentaciónSkill
- Comunicador MarketingSkill
- Orquestador MarketingSkill
- Comunicador BusinessSkill
- Orquestador BusinessSkill
- project-memorySkill
- status-reporterSkill
- Template ManagerSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 4 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 20
When the Orchestrator needs information from another group
- Distinctive: line 29
When there are no dependencies on other groups
- Very specific: line 104
If a group receives a message with a trace_id it already processed
- Distinctive: line 201
Specific Questions: Avoid vague or too general questions
- Very specific: line 202
Appropriate Priority: Use high only when critical
- Distinctive: line 204
Error Handling: Always implement fallbacks for errors
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
🔗 A2A Communicator (Agent-to-Agent)
This skill implements the Agent-to-Agent (A2A) communication protocol to allow information exchange between different Skill Cells of the Conglomerate.
Purpose
The A2A Communicator is the "voice" of each group. When an Orchestrator needs information or services from another group, it uses this skill to send and receive messages in a standardized way.
When to Use This Skill
- When the Orchestrator needs information from another group
- To request artifacts generated by another Cell
- To coordinate tasks requiring inter-group collaboration
- To verify the status of tasks in other groups
When NOT to Use This Skill
- For communication within the same group (use direct delegation)
- For tasks an internal specialist can resolve
- When there are no dependencies on other groups
Communication Protocol
A2A Message Structure
message:
trace_id: "uuid-v4" # Unique ID for tracing
hop_count: 0 # Hop counter (max 5)
source_group: "10-backend" # Source group
target_group: "20-frontend" # Target group
request_type: "query|task|artifact"
payload:
question: "What fields does the registration form require?"
context: "Designing user table"
priority: "low|medium|high"
timeout_seconds: 30Response Structure
response:
trace_id: "uuid-v4" # Same ID as request
status: "success|error|timeout"
source_group: "20-frontend"
payload:
answer: "Mandatory fields are: name, email, password"
artifacts: [] # Paths to generated artifacts
processing_time_ms: 150Communication Flow
Step 1: Prepare Request
1. Identify target group
2. Formulate clear and specific question/task
3. Include minimum necessary context (not all context)
4. Generate unique trace_idStep 2: Validate Hops
1. Verify current hop_count
2. If hop_count >= MAX_HOPS (5), abort with error
3. Increment hop_count by 1Step 3: Send Message
1. Locate target group's Orchestrator
2. Transmit structured message
3. Set timeoutStep 4: Receive Response
1. Wait for response within timeout
2. Validate response structure
3. Extract relevant information
4. Return to requesting OrchestratorInfinite Loop Prevention
[!CAUTION] Hop Limit (Max Hops): 5
Each message carries a
hop_countcounter. If it reaches 5, the request is automatically rejected to prevent infinite loops.
Cycle Detection
The system also detects cycles via trace_id:
- If a group receives a message with a
trace_idit already processed - The request is rejected with
CYCLE_DETECTEDerror
Usage Examples
Example 1: Backend asks Frontend
Scenario: Backend Orchestrator is designing the database and needs to know required registration form fields.
# Request from Backend to Frontend
message:
trace_id: "a1b2c3d4-e5f6-7890-abcd-ef1234567890"
hop_count: 0
source_group: "10-backend-group"
target_group: "20-frontend-group"
request_type: "query"
payload:
question: "What fields are mandatory in the registration form?"
context: "Designing user table in PostgreSQL"
priority: "high"# Response from Frontend
response:
trace_id: "a1b2c3d4-e5f6-7890-abcd-ef1234567890"
status: "success"
source_group: "20-frontend-group"
payload:
answer: |
Mandatory fields:
- name (string, max 100 chars)
- email (string, valid)
- password (string, min 8 chars)
Optional fields:
- phone (string, optional)
- avatar_url (string, optional)Example 2: Frontend requests endpoint from Backend
message:
trace_id: "b2c3d4e5-f6a7-8901-bcde-f23456789012"
hop_count: 1
source_group: "20-frontend-group"
target_group: "10-backend-group"
request_type: "query"
payload:
question: "What is the endpoint to create a user?"
context: "Implementing registration form"
priority: "medium"Example 3: DevOps requests requirements from Security
message:
trace_id: "c3d4e5f6-a7b8-9012-cdef-345678901234"
hop_count: 0
source_group: "30-devops-group"
target_group: "40-security-group"
request_type: "task"
payload:
task: "Review Kubernetes configuration for compliance"
artifacts:
- "skills/30-devops-group/manifests/deployment.yaml"
priority: "high"Error Handling
| Error Code | Description | Action |
MAX_HOPS_EXCEEDED | 5 hops limit exceeded | Abort and report to user |
CYCLE_DETECTED | Communication cycle detected | Abort and report |
GROUP_NOT_FOUND | Target group does not exist | Check registry.yaml |
TIMEOUT | Response did not arrive in time | Retry or report |
INVALID_MESSAGE | Malformed message | Fix and resend |
Configuration
Communication configuration is found in:
- registry.yaml
communication:
max_hops: 5
timeout_seconds: 30
retry_attempts: 3Best Practices
- Minimal Context: Send only necessary information, not all context
- Specific Questions: Avoid vague or too general questions
- Appropriate Priority: Use
highonly when critical - Reasonable Timeouts: Adjust based on request complexity
- Error Handling: Always implement fallbacks for errors
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

