
Launchd Drift Recovery
Register when critical macOS launchd agents for your AI stack drift out of the loaded state and you want guarded automatic kickstart.
Overview
Launchd Drift Recovery is an Operate-phase MCP server that kickstarts drifted critical macOS launchd plists with trust-graph gating and one recovery per plist per hour.
What is this MCP server?
- Watcher detects drifted critical launchd plists
- Runs launchctl kickstart -k to reload drifted jobs
- Trust-graph-gated recovery before acting
- Throttled to one kickstart per plist per hour
- Pairs with other DailyAI drift watchers on the same host
- Throttle: 1 kickstart per plist per hour
- Recovery action: launchctl kickstart -k
- Server version 0.1.0, MIT license
What problem does it solve?
Critical launchd jobs for your agent stack quietly unload after sleep, updates, or config edits and stay broken until you manually kickstart them.
Who is it for?
Mac-based solo builders running launchd-managed DailyAI or OpenClaw services who want MCP-gated auto-recovery.
Skip if: Linux-only deploys, Docker-only production, or teams forbidding automated launchctl from agent tooling.
What do I get? / Deliverables
Drifted plists get a controlled kickstart -k so collectors and sidecars return without unbounded automated restart loops.
- Detection of drifted critical launchd plists
- Trust-graph-approved kickstart -k actions
- Rate-limited recovery at most once per plist per hour
Recommended MCP Servers
Journey fit
Operate infra is the canonical home for macOS launchd plist drift and automated kickstart recovery on a dev machine. Infra reflects kickstart -k on drifted critical plists with trust-graph gating and hourly per-plist throttling.
How it compares
MCP macOS launchd recovery watcher, not a cross-platform Kubernetes reconciler or generic process manager UI.
Common Questions / FAQ
Who is launchd-drift-recovery for?
Indie developers on macOS who run agent automation through launchd and need drift detection plus throttled kickstart recovery.
When should I use launchd-drift-recovery?
Use it in Operate infra when plist-loaded state diverges from expectation and manual launchctl kickstart is too easy to forget.
How do I add launchd-drift-recovery to my agent?
Register io.github.dailyaiagents-cpu/launchd-drift-recovery in your MCP client, define which critical plists the trust graph allows, and run on the Mac host that owns those jobs.