Healer Agent
An autonomous agent that diagnoses production issues, investigates root causes, and applies fixes without human intervention. Part of StrongDM's factory infrastructure. Represents the logical extension of the dark factory — not just autonomous creation, but autonomous maintenance.
The Healer is StrongDM’s autonomous production maintenance agent — the logical complement to Attractor (which builds software) and the external scenario harness (which validates it).
Where traditional on-call engineering means humans diagnosing alerts at 3am, the Healer:
- Diagnoses — reads production alerts, logs, and metrics
- Investigates — determines root cause using the factory’s observability infrastructure
- Applies fixes — runs through the factory pipeline to generate and validate a fix
- Ships — deploys the fix without human involvement
The Level 5 Extension
A dark factory that only automates creation but still requires humans for maintenance is only half complete. The full factory covers the entire software lifecycle:
- Specification → human writes NLSpec (the only human step)
- Implementation → Attractor implements
- Validation → external scenario harness validates
- Deployment → CI/CD ships
- Monitoring → observability layer (CXDB + production telemetry)
- Repair → Healer diagnoses and fixes
The Healer closes the loop, removing humans from the maintenance cycle as well as the development cycle.
Prerequisites
A working Healer requires:
- Observable systems — sufficient telemetry that the agent can diagnose problems
- Validated test harness — the same scenario infrastructure used in development, now used to validate fixes
- Bounded problem domains — the agent’s understanding of “correct behavior” must be formalized in the scenario suite
Current Status
The Healer is described by Justin McCarthy in StrongDM’s published materials as part of their factory infrastructure. It is not an open-source release. No commercial off-the-shelf product currently provides equivalent autonomous production repair with the same guarantees.