Blog

Technical notes on working with AI agents without them stepping on each other: leases between agents, memory across sessions and auditing what they do.

• How to stop two AI agents from editing the same task — What a lease with a fencing token is, what problem it solves when several agents share a board, and how it behaves in MarQ.
• How to audit what your AI agents do — What you need to record to answer who changed what, when and from which value, how each agent is identified and how that history is reviewed in MarQ.
• What an AI agent needs to resume a project between sessions — Why a chat summary is not enough and what structured state an agent can read at the start of each session: compact overview, changes since last time, notes, dependencies and its own leases.