RobustMQ Metrics
SkillMonitoring & opsLets your agent design and add metrics and Grafana dashboards for RobustMQ services.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the RobustMQ Metrics skill
About this capability
Designs and implements minimal, high-value metrics for RobustMQ services and dashboards. Use when the user asks to add metrics, improve observability, or update Grafana panels for core processing pipelines.
What this skill tells your AI
The instructions your AI receives, as published by robustmq/robustmq in .claude/skills/robustmq-metrics/SKILL.md and read by ahel’s review.
Purpose
Add observability with minimal but complete coverage of a target pipeline:
- process count and process duration
- retry and terminal outcomes (if applicable)
- key failure points in read/process/write path
- liveness/health
Do not over-instrument. Prefer compact metrics that support operations and debugging.
Metric Design Rules
-
Cover chain, not everything:
- success/failure + duration
- retry/terminal path (when strategy exists)
- read/commit/write failures
- up/down
-
Always include count and latency for processing path.
-
Label policy:
- required: stable low-cardinality dimensions only
- optional only when bounded enum (for example
result,strategy,protocol) - service/entity name labels are optional and must pass cardinality review
- forbidden high-cardinality labels:
topic,error_message, payload-derived labels
-
Naming:
- use module prefix (for example
mqtt_,raft_,handler_) - counters end with
_total - duration histogram uses
_ms - liveness gauge uses
_up
- use module prefix (for example
Minimal Metric Set Template
Adapt this template to the target module:
<module>_messages_processed_success_total{...}<module>_messages_processed_failure_total{...}<module>_process_duration_ms{...}(histogram)<module>_retry_total{...,strategy}(if retry exists)<module>_terminal_total{...,result}(discard/dlq/drop etc., if applicable)<module>_critical_step_failure_total{...}(read/write/commit/etc.)<module>_up{...}(gauge)
Implementation Workflow
-
Define metrics in
src/common/metrics- register counters/histogram/gauge
- add record helper APIs
- keep API signatures consistent with chosen low-cardinality labels
-
Insert runtime instrumentation
- add metrics at success, failure, retry, terminal, and liveness points
-
Fix call sites impacted by signature changes
- update all metric helper users
- ensure compile passes across dependent crates
-
Validation
- run targeted
cargo checkfor impacted crates - run lints for touched files
- run targeted
Grafana Decision Rule
After metrics are added, decide whether to update grafana/robustmq-broker.json:
- Update dashboard if metrics are operationally critical and stable.
- Skip dashboard only when user explicitly says not to update or metrics are temporary.
When updating dashboard:
- add/extend compact row
- prioritize low-cardinality dimensions and trend panels
- avoid panel explosion; 4-6 panels for first iteration
Output Format
Return:
- Metrics added/changed (names + labels)
- Instrumentation points (files/functions)
- Whether Grafana was updated and why
- Validation commands and results
Guardrails
- Do not add metrics without clear operational use.
- Do not introduce high-cardinality labels.
- Do not duplicate semantically equivalent metrics.
- Do not break existing metric names unless migration is requested.
Signals
- GitHub stars
- 2k
- Forks
- 256
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
robustmq-metrics- Source
- github.com/robustmq/robustmq