Developer Builds Tool to Stop AI Agents Writing Fragile Polling Loops

A developer noticed that AI coding agents repeatedly generated unreliable shell sleep loops when waiting for services to become ready, often declaring tasks complete based on stale or incorrect responses. The root cause was the absence of a proper readiness primitive in the agent's toolset, forcing it to improvise with fixed sleeps and basic curl checks. To address this, the developer created opencode-waitfor, a zero-dependency plugin that adds a single wait_for tool supporting HTTP, TCP, and shell command checks. The tool can validate specific response content — such as confirming a deployed version matches an expected value — rather than accepting any HTTP 200 response. It also returns the last observed output on timeout, allowing the agent to diagnose failures without firing additional probes.