Durable AI-assisted workflows
EDO
Let complex work continue after the conversation ends.
EDO breaks work into explicit steps and checks observed results. The original error-driven approach uses informative failures, progress, and a master log so an agent can determine what needs attention and revisit earlier work when necessary.
WHERE IT HELPS
Start with the work you want to improve.
- Coordinate a multi-step engineering task
- Make progress and failures inspectable
- Explore continuation, review, and recovery workflows
HOW IT CONNECTS
A clear responsibility.
A place in the bigger picture.
Fabric supplies intelligence execution. EDO coordinates work. An operator or runner performs an action, while the workflow keeps track of what succeeded and what remains.
IMPLEMENTATION STATUS & EVIDENCEWorking prototype
What the source record shows
The original EDO README documents numbered steps, 0–100 progress, regression handling, and MasterLog. The family ledger separately recommends reconstructing a portable kernel using several donors.
Read the relevant experiment notes ↗Current scope
The original runner’s progress/regression behavior is documented. A portable durable kernel with broader cancellation, approval, and recovery guarantees is reconstruction work, not a claim about every historical runner.
Current availability and integration fit are established in a walkthrough of the implementation.
A PRACTICAL NEXT STEP