I solve this exact problem running OpenClaw (24/7 agent orchestrator). Here's what actually works after months of iteration:

The core insight: Multiple agents with zero shared state = chaos. Solution: State files are your control plane.

Architecture: 1. HEARTBEAT.md — Every agent reads this before acting. Routes to current priorities, cascades work through a defined sequence (not random agent picking a task) 2. TODO.md — Single source of truth with sections: "DO NOW" (unblocked), "BLOCKED" (waiting), "DONE" (completed). Agents read this, not chat history or logs 3. active-tasks.md — Breadcrumbs from interrupted runs. If an agent crashes, the next session reads what was interrupted and resumes 4. Sub-agent output files (qa-status.md, conversion-tracker.md, etc) — Each specialized agent writes structured results. Prevents re-running the same analysis.

The key: State files are deterministic. No "what was I thinking?" after a crash. The next agent reads the same input, same structure, same decision rules.

Terminal log management: - Each agent logs its own session ID (e.g., [agent-001-mar05-0430]) - Output files are prefixed by agent type (qa-, scout-, coo-) - Logs are sent to daily files (memory/2026-03-05.md) for searchability

Deployment: I run ~8 agents on cron (QA watchdog, scout, content factory, conversion tracker, etc). They wake up, read their input files, write results, die. No persistent terminal. No state collision.

The problem you're hitting is probably: agents don't know what other agents already did. Solution is just: everything runs through a shared TODO.md and writes results to named files.

Happy to share templates if useful. Full patterns at: https://osolobo.com/first-ai-agent-guide/

I'm building something that I called control plane for agents, but it's only solving half of the problem you're mentioning. Basically it allows me to connect to my agents remotely from anywhere, so I have one running on my server and several running on my home PC. I generally just use something like notion or beads for issue tracking.

Another thing I do is I have a custom Claude skill that runs every night and goes through all my repositories and Claude conversations and then updates my dashboard and tasks in notion with progress.

https://github.com/amurg-ai/amurg

Works fine for a one-man pet project, but the second you hit a real codebase and dozens of commits a day, that nightly job is just gonna choke on the context window or slam into Anthropic's rate limits. You really need to be streaming agent state into a DB on the fly via WebSockets, not piecing it together once every 24 hours

There's no off-the-shelf "Jira for AI" product out there right now tbh. Most sane teams are either hacking together thin SQLite + WebSockets dashboards, or just sucking it up and staring at raw stdout across 5 tmux panes. If you need a self-hosted setup with tracing, Langfuse should do the trick and cover like 90% of your pain points. Honestly, if you actually polish this control plane concept and open-source it, the demand will be massive

It's not for your use case per se, but at Tacnode we're building what we call a "context lake" that's essentially a living, queryable context layer. It's got way more horsepower than you're describing. Built instead for enterprise-scale needs.

So basically agents always have the most up-to-the-millisecond data freshness they need to make good decisions.

I’m building this right now, in a little beyond the POC, but would love your feedback. I have the framework down.