Oci observability db

Skill adibirzu/oci-skills/skills/oci-observability-db

Operate OCI Monitoring, Logging, APM, OpenTelemetry, Notifications, Service Connector Hub, service/custom logs, metrics, alarms, dashboards, agent traces, Grafana integrations, PromQL-to-MQL conversion, and Linux or Windows host dashboards. Use for telemetry collection, queries, alerting, trace integrity, logging pipelines, exporter metrics, or observability inventory. Route Database Management, Operations Insights, Performance Hub, AWR/ADDM/ASH, and DBSNMP to oci-dbm-opsi; route Autonomous Database lifecycle to oci-autonomous-db.From its SKILL.md

Install

npx -y skills add adibirzu/oci-skills --skill oci-observability-db

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

6 stars6 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.7 KB, as published. Nobody here has run it

OCI Observability

Preflight the named context, query current telemetry configuration, and search the KB before changing an alarm, log, connector, or APM resource. Run every CLI through oci_cli, every mutation through risk-classified run_action, and every shared output through redaction.

First move

./scripts/oci_preflight.sh -c "$COMPARTMENT_OCID"

If the resolved tenancy/compartment does not match the intended target, stop before changing an alarm, log, connector, or APM resource.

Routing

Intent	Owner
Metrics, alarms, Logging, APM, OTel, RUM, Notifications, connectors	This skill
Prometheus metric queries → OCI MQL; node/windows exporter dashboards	This skill
Log Analytics OCL/LQL sources/parsers/detections	oci-log-analytics
DBM, OPSI, Performance Hub, AWR/ADDM/ASH, DBSNMP	oci-dbm-opsi
ADB provision/scale/wallet/ACL	oci-autonomous-db
Build GenAI agents/RAG/model endpoints	Official `oracle/skills` `oci/enterprise-ai`

Read observability-db.md for command shapes, grafana-dashboards.md for dashboards-as-code, and prometheus-mql-host-dashboards.md for bounded PromQL conversion and prepared Linux/Windows profiles.

Common multi-step flows

Task	Sequence
Add an alarm	list current → verify metric namespace/query → ensure destination is active → additive action → re-list/test notification
Missing traces	list APM domains → resolve correct domain/key without printing it → verify private OTLP endpoint → send test → re-query
Service logs	list groups/logs → write source config to `0600` file → additive action → verify ingestion and retention
Gate agent traces	query integrity score/state → identify missing spans/attributes → keep prompt capture off → re-score before promotion
Convert PromQL to MQL	identify metric type + labels + units → convert only a supported shape with `python3 skills/oci-observability-db/scripts/promql_to_mql.py` → verify namespace/dimensions → compare live series with PromQL
Build Linux or Windows host dashboard	identify exporter/version → list emitted metric definitions → render the matching profile → validate every MQL panel → additive provision/import → verify populated panels

Safety

Never expose datakeys, domain IDs, endpoints, OCIDs, IPs, or tenant namespaces.
Store nested configuration in 0600 file:// payloads, not argv.
Treat empty results as inconclusive until region, subtree, time window, and permissions are checked.
Re-list after a 409 rather than retrying a create.
Every mutation runs through run_action --risk <additive|in-place> --compartment <COMPARTMENT_OCID> --description "<...>" -- oci_cli ... (honors OCI_SKILLS_DRY_RUN=true for a no-op preview); get explicit user confirmation before applying an alarm, connector, or APM change.
Treat converted MQL and rendered dashboards as candidates until every query parses and returns the intended dimensions and units; reject unsupported PromQL rather than guessing.

Expected output

Report finding, redacted evidence, exact risk-classified action or read, verification, and KB reference. For database telemetry work, state the handoff owner rather than performing DBM/OPSI here.

Official documentation

Monitoring · Logging · APM · Service Connector Hub. Full list in the observability-db reference.

Open Knowledge Format grounding - every doc link here is registered and liveness-checked in the oracle-docs.md index (the pack's single source of truth). When extending this skill, cite the most specific official page through that index; the non-official MCP gateway is never a source of truth.