Guide · Agent skill
How to use the SLO Implementation agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The slo-implementation skill is a guide and template library that helps an AI agent define Service Level Indicators (SLIs), Service Level Objectives (SLOs) and error budgets for a service, then write the Prometheus rules and alerts to track them. It is aimed at SRE and platform engineers, and at developers who run services on Prometheus. After this guide you will know what the skill does, how to install it, how to ask for it in plain language, and which settings to check before you copy anything into production.
SLO Implementationwshobson/agents · MIT Download the skill Quick start
- 1Download the ZIP. Download the slo-implementation skill ZIP from this page.
- 2Unzip it into a skills folder. Unzip the folder into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder to share it with that project.
- 3Or upload it in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP instead.
- 4Confirm your metrics. Check that your service exports http_requests_total and a latency histogram in a PromQL-based system.
- 5Ask for an SLO for one service. Describe the service, the user-facing behaviour and the target you are considering, such as 99.9% availability.
- 6Review the generated rules. Check the targets, windows and burn thresholds against real traffic and user expectations before deploying.
What is an agent skill?
An agent skill is a folder containing a SKILL.md file with instructions, plus optional reference files, that an AI agent can load when a task matches the skill's description. This skill's SKILL.md describes its purpose and when to use it, and points to references/details.md for a larger pattern library.
The skill does not monitor anything by itself. It gives the agent a structured method and ready-made templates, and the agent applies them to your service and metrics.
- SKILL.md: the main instructions and templates
- references/details.md: extra burn-rate alert patterns and a review process for weekly, monthly and quarterly SLO reviews
What the skill makes the agent do
Once the skill is loaded, the agent can help a team pick reliability indicators and targets, calculate the error budget that a target allows, and write Prometheus recording rules and alerting rules for that budget.
The source covers three common SLI types: availability (successful requests divided by total requests), latency (requests under a threshold divided by total requests) and durability (successful writes divided by total writes). It includes a table of downtime per month and per year for targets from 99% to 99.99%, so a target such as 99.9% can be translated into 43.2 minutes of downtime per month.
- Availability SLI built from http_requests_total with a 5xx status filter
- Latency SLI built from http_request_duration_seconds histogram buckets, using a 0.5 second threshold in the examples
- Error budget formula: 1 minus the SLO target
- An error budget policy that moves from normal velocity at 100% remaining to a feature freeze at 0%
- Recording rules and multi-window burn-rate alerts, with a Grafana dashboard layout
Install the skill
Download the skill ZIP from this page and unzip it. For a personal install, place the unzipped slo-implementation folder in ~/.claude/skills/. To share it with one project, place it in that project's .claude/skills/ folder instead.
If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/slo-implementation/
- Project install: .claude/skills/slo-implementation/ inside your repository
Trigger the skill with natural requests
You do not need a special command. The skill's description covers establishing reliability targets, implementing SRE practices and measuring service performance, so ask for those things directly. Name your service and your metrics where you can, because the templates depend on metric names such as http_requests_total.
Good requests include the service name, the user-facing behaviour you care about and the metrics you already export.
- "Define availability and latency SLOs for our checkout API using our Prometheus metrics."
- "How much downtime does a 99.9% target allow per month, and what is the error budget?"
- "Write Prometheus recording rules and burn-rate alerts for our payments service."
- "Draft a Grafana SLO dashboard layout for the search service."
- "Explain the difference between SLI, SLO and SLA to our product managers."
Check the prerequisites and the example targets
The examples are PromQL-specific. You need Prometheus or a compatible PromQL-based metrics system, with http_requests_total and latency histogram metrics. Grafana is optional and is only needed for the dashboard layout.
The example numbers are starting points, not recommendations. The 99.9% target, the 28-day window and the 14.4x fast-burn threshold should be checked against real user expectations and your traffic before you rely on them, or the alerts may be misleading. Legal or contractual SLAs need review by business and legal teams and should not be taken from a generated template.
- Requires Prometheus or a compatible PromQL system with http_requests_total and latency histogram metrics
- Grafana is optional, for the dashboard layout
- The skill runs no scripts and uses no network access, according to its listing
Example: a recording rule for availability
The skill's templates use a recording rule to store the availability ratio under a stable name, then build compliance and error budget series on top of it. The following is the availability SLI rule from the source, shown for reference. Ask the agent to adapt the metric names and window to your own setup.
groups:
- name: sli_rules
interval: 30s
rules:
- record: sli:http_availability:ratio
expr: |
sum(rate(http_requests_total{status!~"5.."}[28d]))
/
sum(rate(http_requests_total[28d]))What people use it for
- Calculating the monthly downtime allowed by a 99.9% or 99.95% availability target
- Writing recording rules that store availability and latency SLI ratios in Prometheus
- Creating multi-window burn-rate alerts that page on fast budget burn and warn on slow burn
- Laying out a Grafana SLO dashboard with compliance, remaining error budget and burn-rate panels
- Explaining SLI, SLO and SLA to non-specialist stakeholders
Frequently asked questions
How do I install the slo-implementation skill?
Download the skill ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
Is the slo-implementation skill free?
The skill is published under the MIT licence, as listed on this page. The source does not mention pricing, so check the licence file in the repository for the exact terms.
What do I need to use it?
You need Prometheus or a compatible PromQL-based metrics system with http_requests_total and latency histogram metrics. Grafana is optional and only needed for the dashboard layout.
Does the skill monitor my services?
No. It is a guide and template library. It helps the agent write SLI, SLO and alerting definitions, but it does not collect metrics or send alerts on its own.
Can I use it with a metrics tool other than Prometheus?
The examples are PromQL-specific. A system that uses PromQL-compatible queries may work with some adjustment, but other systems would need the rules rewritten, as the listing notes.
Why do the example targets need checking?
The 99.9% target, 28-day window and 14.4x burn threshold are example values. Copying them without checking against real user expectations and traffic can produce misleading alerts.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you are introducing error budgets and SLO-based alerting, or you run services on Prometheus and want starter PromQL and alert rules to adapt. Pick something else if you do not use Prometheus or a compatible PromQL metrics system, or if you need a legal or contractual SLA, which should be written with your business and legal teams.
Download the skill