agentleFS
Sign inStart free
willianbswillianbs/skills/observability-designer/SKILL.md

observability-designer

Skill0 starsChanged 4 months agoMITView on GitHub

Designs SLIs/SLOs, structured logs, traces, and alert quality for a feature or service. Use when shipping new paths or fixing noisy/missing alerts. Emits OBS_PLAN. Never recommends alert spam or metrics without a user/journey signal.

9 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Prefer fewer high-quality alerts over many noisy ones.

    line 39
  2. Distinctive:

    Every alert answers “what do I do at 3am?”.

    line 43
  3. Distinctive:

    Never log secrets or raw PII.

    line 90
  4. Distinctive:

    Never page on symptoms without ownership.

    line 91
  5. Distinctive:

    Never require 50 metrics for a tiny feature.

    line 92

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Purpose

Make the system operable: what to measure, log, trace, and alert on—with owners and noise control.

When to Use / When NOT to Use

Use when: new services/features; after incidents; alert fatigue; missing dashboards; before deploy of HighRisk paths.

Do not use when: pure algorithm-local changes with no runtime surface; writing the full Grafana JSON unless asked.

Preconditions

Feature/service scope and rough user journey. Prefer CONTEXT_PACK + SLOs if they exist.

Inputs / Outputs

Inputs: scope, IMPACT_REPORT/INCIDENT_LOG optional, existing metrics/alerts if known.

Outputs: OBS_PLAN

Upstream / Downstream

Upstream: delivery-planner, incident-commander, performance-auditor, engineering-os.

Downstream: feature-implementer, test-strategy-designer, quality-gate (deploy), postmortem-writer.

Core Principles

  1. SLIs from user journeys, not vanity counters.
  2. Alerts need: signal, threshold, ownership, runbook link, severity.
  3. Prefer fewer high-quality alerts over many noisy ones.
  4. Structured logs with correlation IDs; no secrets/PII in log bodies.
  5. Traces on boundaries (entry, datastore, external calls).
  6. RED/USE as defaults; adapt to stack.
  7. Every alert answers “what do I do at 3am?”.

Process

  1. Journeys / golden signals for the scope.
  2. SLIs + SLO targets (or explicit “no SLO yet”).
  3. Metrics list — name, type, labels (cardinality-safe).
  4. Logging plan — events, fields, redaction.
  5. Tracing plan — spans, attributes.
  6. Alerts — condition, sev, routing, runbook outline; explicitly suppress noisy candidates.
  7. Dashboard minimum viable views.
  8. Decision.

Evidence Requirements

Tie SLIs to real endpoints/jobs. Cite existing metric names when present.

Stop Conditions / Failure Modes

ConditionAction
Alert with no action/runbookRevise — do not ship that alert
High cardinality label proposalBlock that label

Severity + Confidence

Missing observability on SEV-1 path → High for deploy-gate conditions.

Output Contract

## OBS_PLAN
Journeys/SLIs/SLOs: ...
Metrics: ...
Logs: ...
Traces: ...
Alerts: ...
Dashboards: ...
Decision: Proceed | ProceedWithConditions | Revise | Block

Handoffs

feature-implementer, quality-gate, incident-commander, performance-auditor.

Never

  • Never log secrets or raw PII.
  • Never page on symptoms without ownership.
  • Never require 50 metrics for a tiny feature.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free