Operability Minimum

This page defines the minimum operational loop for small teams maintaining SMonitor-integrated libraries.

Weekly checks (recommended)

1. Top repeated warning/error codes.

2. New unknown or uncatalogued codes.

3. Error trend by library and version.

4. Bundle quality in incoming support reports.

Release-candidate checks

• pytest in clean environment.

• make -C docs html without broken references.

• API contract tests pass.

• one integration smoke run in each ecosystem library.

Support loop

1. Reproduce from message + minimal script.

2. If needed, reproduce from diagnostics bundle.

3. Triage by code and trace_hash.

4. Patch with tests.

5. Publish changelog entry referencing affected codes.

What to automate first

• contract checks (strict_signals, strict_schema);

• bundle export smoke step in CI;

• docs build gate;

• issue templates that request code/reproducer/environment.

AI-assisted loop (optional)

• use agent profile and structured bundles for triage;

• keep human review mandatory for all proposed fixes;

• store triage decisions for recurring patterns.

Why this matters

For small teams, predictable operational loops reduce support cost more than ad-hoc debugging sessions, and keep diagnostics quality stable over time.