Monoscope full color logo
Sign Up
Home Getting Started SDK Guides AI Dashboard API Reference
CLI & Agents
MCP Server

On this page

Claude Code Skills

Three drop-in skills that turn natural-language requests into well-formed CLI invocations. A skill is a named bundle of instructions plus shell access that Claude Code loads on demand — when your request matches the skill’s description, Claude applies its playbook.

The skills are thin wrappers around the monoscope CLI — no separate API key, no extra runtime. If the CLI works, the skills work. Prefer wiring tools in over the network instead? Use the MCP server.

Available skills

Skill Activates on What it does
investigate “investigate the 500 errors”, “look into payment-api errors”, “what happened at 10:34?” Runs the discover → search → context → triage chain. Uses facets to find services/values, logs search to drill in, events context --summary to pull the surrounding window.
triage “do an on-call sweep”, “ack the noisy issues for X”, “clear the alert queue” Reviews open issues, log patterns, and monitors. Acknowledges, archives, mutes, or resolves with the right granularity.
kql-reference Any time another skill needs to write KQL Operator/function reference for Monoscope’s KQL dialect. Rarely invoked directly — other skills pull it in for query construction.

The skills are open-source and live at github.com/monoscope-tech/skills.

Install

Claude Code

# Add the marketplace once
claude plugin marketplace add monoscope-tech/skills

# Install
claude plugin install monoscope-skills@monoscope-skills

Restart Claude Code. The skills activate automatically when relevant — no explicit invocation needed.

Cursor, Cline, Copilot, and other agents

npx skills add monoscope-tech/skills

This drops the skill files into your project’s .claude/skills/ directory (or equivalent) so any tool that reads that format picks them up.

Prerequisites

The skills assume the monoscope binary is on your PATH and you’re authenticated:

curl monoscope.tech/install.sh | sh
export MONOSCOPE_API_KEY=<your-api-key>
export MONOSCOPE_PROJECT=<your-project-uuid>
monoscope auth status   # should report authenticated

The first time a skill runs, it reads these env vars (or falls back to ~/.config/monoscope/config.yaml and ~/.config/monoscope/tokens.json).

Usage examples

Investigate

> Investigate the 500 errors in checkout-api over the last hour.

[Claude runs:]
  monoscope facets resource.service.name --top 5
  monoscope logs search 'attributes.http.response.status_code == 500' \
    --service checkout-api --since 1h --first --id-only
  monoscope events get <id>
  monoscope events context --at <timestamp> --window 5m --summary

[Then summarises:]
  - 47 errors in the last hour, all from checkout-api
  - Started at 10:34 after the deploy at 10:32 (correlated trace IDs span
    payments and inventory)
  - Likely cause: connection pool exhaustion in payments — 3 traces show
    `attributes.db.client.connections.usage` jumping from 12 → 100

Triage

> Do an on-call sweep for the past day.

[Claude runs:]
  monoscope issues list --status open
  monoscope log-patterns list --per-page 30
  monoscope monitors list

[Then bulk-acks the noisy ones, flags the surprising ones for review,
 and produces a summary the team can paste into Slack.]

Custom skill

The shipped skills are a starting point. You can write your own — they’re just markdown files with a frontmatter description Claude uses for matching. See SKILL.md examples for the format.

Pointing at a self-hosted server

Set MONOSCOPE_API_URL:

export MONOSCOPE_API_URL=http://monoscope.internal:8080

The skills inherit the env, so the same investigate/triage workflows run unchanged against your private instance.

See also