COMMUNICATION AND PROTOCOLS BETWEEN AGENTS

PRACTICE LAB CONNECTION In the Resume Review Agent lab, you will add a shared message or log bus across the supervisor and subagents, which you will build in Chapter 69. Start simply. Create one shared log that every agent appends to, with each entry holding the time, the agent name, the message type, and a short description. Then design at least one rule for stopping runaway work, such as a maximum number of handoffs and a timeout on each subagent call. Test your design against the prompts below. ARTIFICIAL INTELLIGENCE ASSISTANT PROMPTS [START OF PROMPT] Review this multi-agent communication architecture for point-to-point coupling that a shared message bus would simplify, and check for deadlock risk (two agents waiting on each other): [PASTE LAB OR PRODUCTION AGENT COMMUNICATION CODE/DIAGRAM HERE]. Propose the specific timeout value and escalation action that would break any deadlock found. Output format: coupling-risk findings, deadlock scenario (if any) with the specific timeout/escalation fix. [END OF PROMPT] [START OF PROMPT] We run [NUMBER] separate agents including [AGENT LIST, e.g. Resume Review Agent and others]. Design a minimal meta-harness giving unified budget tracking, permissions, and logging across all of them despite differing internal frameworks, and list the top three risks we currently face without it. Output format: a meta-harness design outline (what it tracks, how each agent reports in) plus a three-item risk list. [END OF PROMPT]