skills.rest
diazMelgarejo
diazMelgarejoCOMMUNITY
Explore all skills by this author →
Website𝕏@Melgarejo_yDiazGitHub

agent-coordination-heartbeat

Monitor distributed agent liveness and auto-release claims from dead agents.

Updated Mar 22, 2026
Search Tags:#monitoring#multi-agent#orchestration#heartbeat#agent liveness#dead agent detection#claim cleanup
One-click install
npx skills add https://github.com/diazMelgarejo/orama-system --skill agent-coordination-heartbeat-diazmelgarejo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-coordination-heartbeat
Source: https://github.com/diazMelgarejo/orama-system/tree/main/bin/orama-system/skills/agent-coordination-heartbeat
Command: npx skills add https://github.com/diazMelgarejo/orama-system --skill agent-coordination-heartbeat-diazmelgarejo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? In multi-agent systems, agents can hang or die silently while still holding task claims, blocking other agents from making progress. This Skill tracks agent liveness through heartbeat events and automatically releases stale claims from agents presumed dead. ## Core Features & Use Cases - Liveness State Tracking: Classifies agents as ACTIVE, IDLE, STALLED, or DEAD based on time since last heartbeat, with a 30-minute threshold for dead-agent cleanup. - Heartbeat Operations: Register agents, emit heartbeat pulses, inspect health via list/check/dashboard commands, and force-kill unresponsive agents through scripts/agent_coordination.py. - Automatic Claim Recovery: Call cleanup_stale_claims(bus) to release claims held by DEAD agents and emit release events on the GossipBus. - Use Case: Two agents are implementing coordinated work in parallel. The reviewing agent runs a 5-minute check-fix-report loop, pulling the other agent's commits, running the test suite, and posting status notes so neither agent builds on broken code. ## Quick Start Run the heartbeat dashboard command with agent_coordination.py to check the liveness state of all registered agents.

Frequently Asked Questions about agent-coordination-heartbeat

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

FAQPage Schema
How do I check if an agent is still alive in a multi-agent system?▼

Run `python3 scripts/agent_coordination.py heartbeat check <agent_id>` to inspect a single agent, or use `heartbeat dashboard` to view all agents. Agents are classified as ACTIVE, IDLE, STALLED, or DEAD based on time since their last heartbeat.

How do I release claims held by a dead agent?▼

Call `cleanup_stale_claims(bus)` from the orchestrator.heartbeat_monitor module, or rely on the dashboard path. Agents with no activity for over 30 minutes are treated as DEAD and their claims are automatically released with events emitted on the GossipBus.

What are the heartbeat liveness states and thresholds?▼

There are four states: ACTIVE (under 60 seconds since last activity), IDLE (60-300 seconds), STALLED (300-1800 seconds), and DEAD (over 1800 seconds). The 30-minute DEAD threshold triggers automatic claim cleanup.

Can I lower the 30-minute dead agent threshold in production?▼

You should ask first before lowering the 30-minute DEAD threshold in production. The skill's boundary rules require verification of liveness before treating an agent as dead and logging a reason when force-killing an agent.

Why did my coordination message lose its backticked code text?▼

Zsh and bash treat backticks as command substitution inside double-quoted shell payloads, deleting the intended text before GossipBus receives it. Use single quotes, escaped backticks, or a here-doc for literal coordination messages.

Related Skills

Explore More Software Engineering Skills→
mtsatryanmtsatryan
1

event-driven-coordinator

Coordinate multi-agent tasks through asynchronous event-driven communication.

Community
Advanced
bismuthdigitalbismuthdigital

release-tasks

Release claimed tasks back to a shared work queue by slug.

Official
Intermediate
wewetv1987-cellwewetv1987-cell

agent-autonomy-kit

Convert waiting periods into productive task execution in multi-agent work loops.

Community
Advanced
ShadowAUSShadowAUS

health-check-endpoints

Implement liveness, readiness, and startup checks with Kubernetes probe configurations.

Community
Advanced