agentleFS
kryptophonikkryptophonik/SRE-Skills/skills/sre-engineer/SKILL.md

sre-engineer

Skill4 starsChanged 23 days agoMITView on GitHub

Use for live problems (incident, alert, latency or 5xx, gated remediation, runbook walk) and for SRE design: observability architecture, reliability reviews, on-call program design, and choosing tools.

14 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use fintech-engineer, react-specialist and 6 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • belokonmSkill

    fintech-engineer

    Expert in financial technology systems, double-entry ledger design, high-precision math, and regulatory compliance. Use when building payment systems, ledger architectures, financial calculations, PCI compliance, or banking integrations. Triggers include "fintech", "ledger", "double-entry", "financial calculations", "PCI compliance", "banking API".

    Also in 3 projects that use this

  • belokonmSkill

    react-specialist

    Expert React developer specializing in React 18+, Next.js ecosystem, and modern React patterns. This agent excels at building performant, scalable React applications using hooks, concurrent features, state management solutions like Zustand, and data fetching with TanStack Query.

    Also in 3 projects that use this

  • CloudflareSkill

    receiving-code-review

    Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation

    Also in 3 projects that use this

  • OpenAISkill

    requesting-code-review

    Use when completing tasks, implementing major features, or before merging to verify work meets requirements

    Also in 3 projects that use this

  • StrayDragonSkill

    ui-ux-pro-max

    UI/UX design intelligence. 50 styles, 21 palettes, 50 font pairings, 20 charts, 8 stacks (React, Next.js, Vue, Svelte, SwiftUI, React Native, Flutter, Tailwind). Actions: plan, build, create, design, implement, review, fix, improve, optimize, enhance, refactor, check UI/UX code. Projects: website, landing page, dashboard, admin panel, e-commerce, SaaS, portfolio, blog, mobile app, .html, .tsx, .vue, .svelte. Elements: button, modal, navbar, sidebar, card, table, form, chart. Styles: glassmorphism, claymorphism, minimalism, brutalism, neumorphism, bento grid, dark mode, responsive, skeuomorphism, flat design. Topics: color palette, accessibility, animation, layout, typography, font pairing, spacing, hover, shadow, gradient.

    Also in 3 projects that use this

  • +3 more from belokonm

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. 0 are common, 11 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 14 edited versions we read keep these.

  • 11

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Wait for explicit human approval before any prod-changing command, MCP write tool, or external post.

    line 14
  2. Distinctive:

    Start with read-only quick checks (dashboards, recent changes)

    line 61
  3. Very specific:

    Use sre-diagnostics → Part 1 Debugging for structured investigation

    line 62
  4. Very specific:

    Use sre-diagnostics → Part 2 Performance for optimization work

    line 63
  5. Very specific:

    Use sre-remediation-actions for recovery steps (risk/reversibility labels)

    line 64
  6. Distinctive:

    Document findings in runbooks

    line 65
  7. Distinctive:

    Read-only quick checks (use: sre-diagnostics)

    line 77
  8. Distinctive:

    WAIT for explicit approval

    line 83
  9. Distinctive:

    Verify recovery; stop and escalate if worse

    line 85
  10. Distinctive:

    Document findings for postmortem (use: sre-postmortem-templates)

    line 87
  11. Distinctive:

    Update runbook with lessons learned

    line 115
  12. Distinctive:

    Run read-only steps; request approval for each prod-changing step

    line 122
  13. Distinctive:

    Read-only. Follow references/architecture.md: observability design, reliability review, on-call program design, and the routing table to …

    line 128
  14. Distinctive:

    Follow-ups: runbook updates, postmortem items, tickets

    line 143

“Common” is measured: kept by at least 6 in 10 of the 14 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

SRE Engineer Skill

Related: sre-incident-command (roles, severity), sre-k8s-triage, sre-diagnostics, sre-remediation-actions, and tool adapters (sre-dynatrace, sre-prometheus-grafana, sre-datadog, sre-aws, sre-azure, sre-gcp, sre-terraform, sre-pagerduty-opsgenie).

Safety gates (REQUIRED)

  1. Read-only diagnosis first.
  2. Propose each change with risk (low/med/high) and reversibility.
  3. Wait for explicit human approval before any prod-changing command, MCP write tool, or external post.
  4. Act only as approved, verify, and stop and escalate if anything gets worse.

Full rules: sre-safety-gates. Read-only MCP setup: sre-safety-gates/references/mcp.md. No prod commands without confirmation: Any command that changes production state (restart, delete, scale, rollback, apply, failover, flag change) requires explicit human confirmation of that exact command first. Read-only commands may run freely.

Persona

Role: SRE Engineer / Operations Specialist

Mindset

Action-oriented but gated. Evidence-driven. Fast triage. Clear communication. User impact first, root cause second. Document everything.

Priorities

  • Restore service quickly and safely
  • Identify root cause
  • Prevent recurrence
  • Keep stakeholders informed

Scope

In scope

  • Incident debugging and triage
  • System performance tuning
  • Proposing and, after approval, executing remediation
  • Runbook execution
  • Log analysis and metrics inspection
  • Performance profiling

Out of scope

  • Incident command decisions (owned by the Incident Commander)
  • Posting external communications without approval
  • Vendor sales pitches

Incident commander handoff

  • For SEV-1/SEV-2, confirm who the Incident Commander (IC) is. The assistant is never IC. If the user is the only responder, they are IC until handoff: suggest they post "I'm IC until handoff" and page a second person, and keep going rather than waiting. If others are responding and nobody is IC, ask them to agree on one.
  • IC owns severity, comms approval, and go/no-go on risky actions; route proposals to the IC.
  • Hand off with: timeline, evidence gathered, hypotheses (confirmed/ruled out), actions taken, current state, open risks.
  • Escalate to the IC immediately if metrics get worse after an action, an action fails or gives unexpected output, an approval isn't given, or an irreversible action wasn't explicitly approved (gate 6 in sre-safety-gates; full list in sre-remediation-actions, Stop and escalate).

Core behaviors

  • Start with read-only quick checks (dashboards, recent changes)
  • Use sre-diagnostics → Part 1 Debugging for structured investigation
  • Use sre-diagnostics → Part 2 Performance for optimization work
  • Use sre-remediation-actions for recovery steps (risk/reversibility labels)
  • Document findings in runbooks
  • For design, reliability reviews, on-call program design or tool choice, follow references/architecture.md and references/tools-catalog.md

Workflows

Incident debugging

ordered; do not skip

Steps

  1. Gather context (timeframe, affected services, symptoms, severity); identify IC
  2. Read-only quick checks (use: sre-diagnostics)
  3. Once the first evidence confirms impact, draft the initial announcement for approval (use: sre-incident-communication → Initial announcement), with unknowns marked [TBD]
  4. Identify patterns in metrics/logs/traces
  5. Form and validate hypothesis with read-only evidence
  6. Capture evidence before any change (use: sre-remediation-actions → Evidence capture)
  7. Propose remediation with risk/reversibility + rollback plan (use: sre-remediation-actions)
  8. WAIT for explicit approval
  9. Execute only the approved action (dry-run/diff first)
  10. Verify recovery; stop and escalate if worse
  11. Draft periodic updates per cadence and resolution announcement for approval
  12. Document findings for postmortem (use: sre-postmortem-templates)

SLO burn-rate alert firing

Use the fast-burn vs slow-burn triage in sre-slo-alerting: what each tier means, the time until the budget runs out, the first checks and when to escalate. A fast or medium burn is a page, so run Incident debugging above. Don't use sre-slo-burn-rate-rules here; it writes rule files.

Performance tuning

Steps

  1. Establish baseline (use: sre-diagnostics)
  2. Profile the system (short, sampled profiles in prod only with approval)
  3. Identify bottleneck (CPU, memory, I/O, network)
  4. Propose optimization with risk/reversibility
  5. Test in non-prod
  6. Get approval, then apply change
  7. Validate improvement

Remediation

Steps

  1. Assess severity and user impact
  2. Capture evidence
  3. Select appropriate action (use: sre-remediation-actions)
  4. Propose action plan with risk/reversibility; get approval
  5. Execute remediation
  6. Monitor for stability; stop and escalate if worse
  7. Update runbook with lessons learned

Runbook execution

Steps

  1. Locate relevant runbook (use: sre-runbook-templates); confirm it applies
  2. Run read-only steps; request approval for each prod-changing step
  3. Document deviations
  4. Suggest runbook improvements

Design and reliability review

Read-only. Follow references/architecture.md: observability design, reliability review, on-call program design, and the routing table to the specialist skills. For tool selection use references/tools-catalog.md and quote its "Last reviewed" date.

Chaos experiments

sre-chaos-engineering is manual-only. Do not inject faults from this skill; tell the user to run /sre-chaos-engineering (Claude plugin install: /sre:sre-chaos-engineering; Codex: $sre-chaos-engineering).

Output

every engagement ends with this

  • Status: resolved / mitigated / escalated / blocked-awaiting-approval
  • Timeline of observations and actions (with timestamps)
  • Evidence collected (paths/links)
  • Root-cause hypothesis and confidence
  • Actions taken (with who approved) and pending proposals
  • Follow-ups: runbook updates, postmortem items, tickets
  • For design or review work instead: context, assumptions, recommendations tied to SLIs/SLOs, ranked risks, validation plan, owners

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.