Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Accepted format is [A-Za-z0-9][A-Za-z0-9._:-]{7,127}; invalid injected values fall back to generated id.
The resolved value is exported to process env SKILL_SESSION_ID to keep downstream calls consistent.
Execution Flow
Before each inspection, the CLI calls ROA API POST /api/v1/openapi/initial_sysom (source=skill_hub) to verify permissions and SysOM activation.
If SysOM is not activated or role readiness is missing, the CLI interactively asks whether to continue with activation + installation.
After user confirmation, it calls InitialSysom(check_only=false, source=skill_hub) for activation, then calls InstallAgentWithType.
After installation, it re-checks readiness using InitialSysom(check_only=true, source=skill_hub). Inspection continues only when re-check succeeds.
Local threshold/event-rule configuration is not used; anomaly decisions come from the server-side inspection report.
If --instance-id is not provided, the CLI calls ListAllInstances (region / instanceType=ecs / managedType / current / pageSize) and lets the user pick an instance interactively.
It always calls ROA inspection API POST /api/v1/inspection/createInstanceInspection with source=skill_hub, and supports optional metricSource (cms / sysom / auto).
If --metric-source is not explicitly provided, the CLI maps automatically from management status: managed -> sysom, unmanaged -> cms, unknown -> auto.
To inspect all items, pass items=[] (in CLI, provide an explicit empty --inspection-items).
If the standard inspection API returns InvalidAction.NotFound, the CLI marks the API as unavailable and stops follow-up flow to avoid invalid retries.
Report lookup uses ROA API GET /api/v1/inspection/getInspectionReport.
If create API is unavailable, the CLI still sends one GetInspectionReport probe call and records the result for observability.
If report contains sysom:metric:memory_usage_rate anomaly, the CLI automatically triggers InvokeDiagnosis for memgraph.
InvokeDiagnosis injects __sysom_diagnosis_source=skill_hub into params and validates business code=Success.
After diagnosis is started, the CLI polls GetDiagnosisResult until success / fail / timeout.
Auto diagnosis can be disabled via --disable-memgraph-diagnosis.
Extensibility Notes
Inspection items can be overridden via --inspection-items.
Instance filtering can be controlled by --managed-type (managed / unmanaged / all) and pagination args --current, --page-size.
Metric source can be specified via --metric-source (cms / sysom / auto); if omitted, default behavior is preserved.
If InitialSysom indicates not activated, the CLI asks for terminal confirmation before activation attempt + recheck.
Memory anomaly trigger logic is implemented in scripts/sysom_cli/inspection/command.py.
To add more post-inspection specialized diagnosis actions, reuse the InvokeDiagnosis integration pattern.
