agent-handoff-orchestrator (for hermes agent / openclaw)

What it does

The Agent Handoff Orchestrator is a specialized skill designed to solve the "context collapse" problem in multi-agent workflows and long-running sessions. It generates high-density, structured handoff packets that allow work to move seamlessly between AI agents, human operators, or future sessions without losing progress or requiring rediscovery of facts.

Why use this skill

Standard LLM prompts often lead to "drift" when tasks are delegated. This skill enforces a rigorous engineering standard for state transfer by defining clear authority boundaries, separating verified facts from assumptions, and establishing strict return contracts. It ensures that the receiver—whether it's a coder, reviewer, or QA agent—can begin productive work in under one minute with zero ambiguity.

Supported Workflows

• AI-to-AI delegation (Implementation, Research, Security)
• AI-to-Human escalation (Approvals, Risky decisions)
• Session-to-Session persistence (Context compression)
• Failure Recovery (Root cause analysis and repair handoffs)

The output is a structured Markdown packet optimized for LLM comprehension, featuring explicit mission statements, scope constraints, and observable success criteria.