Sentry for AI

Your AI coding assistant already knows how to write code.
This plugin teaches it Sentry — how to set it up, how to find and fix production issues,
and how to get the most out of every feature.

Whether you’re adding Sentry to a new project, debugging a spike in errors, or
configuring alerts, just ask.
The plugin gives your assistant the context it needs to do it right.

Supports the Agent Plugins standard,
Claude Code,
Cursor,
Codex, and
Grok.

What You Can Do

Set up Sentry in any project — SDK setup wizards that detect your stack, recommend
the right features, and walk through the full installation.
No copy-pasting from docs.

Add Sentry to my Next.js app
Set up Sentry in my Rails project
Add error monitoring to my iOS app

Find and fix production issues — Query your Sentry environment, triage errors, and
fix them in place.

What are the top errors in the last 24 hours?
Which issues are affecting the most users?
Fix the recent Sentry errors

Review code with Sentry context — Automatically resolve bugs that Sentry or Seer
flag in pull request comments.

Review PR #118 and fix the Sentry comments

Configure monitoring — Set up alerts, instrument AI/LLM calls, connect OpenTelemetry
pipelines.

Create a Slack alert for new high-priority issues
Monitor my OpenAI calls with Sentry
Set up OTel Collector with Sentry exporter

Distribution

This repository is the single source of truth for all skills, but it is not itself an
installable plugin.
Each assistant needs the plugin in a slightly different shape, so the
per-agent plugins are built from it by the build scripts under
src/plugins/<agent>/build.sh. CI runs these on every push and deploys each result to
its own distribution repository, whose root is exactly that agent’s plugin:

These repositories are generated; do not edit them.
Each one’s README has the install instructions for that agent.

The skill library is also browsable at skills.sentry.dev
and available through the skills.sh
installer.

Build it yourself

Each src/plugins/<agent>/build.sh takes an output directory and writes that agent’s
plugin into it (the Codex build moves the plugin under plugins/sentry/ and swaps the
skill tree’s disable-model-invocation flags for Codex’s agents/openai.yaml policy):

git clone https://github.com/getsentry/sentry-for-ai.git
cd sentry-for-ai
src/plugins/agent-plugin/build.sh /tmp/sentry-agent-plugin

To build any target locally, run src/plugins/<agent>/build.sh <output-dir>
(agent-plugin, claude, cursor, codex, or grok).

Skills

Core

Feature Setup

Reference Library

sentry-instrument doesn’t hard-code platform steps — it pulls them from a shared
library that the build hydrates into the skill:

• src/references/sdks/<platform>/ — per-platform install and per-signal code, one
  directory per supported platform.
• src/references/concepts/ — per-signal strategy: errors, tracing, logging,
  metrics, profiling, session replay, user feedback, crons, uptime, releases, data
  scrubbing, and choosing-a-signal.

Superseded per-SDK “wizard” skills are frozen under skills-legacy/, excluded from
the plugin build.

Prerequisites

The plugin configures the Sentry MCP server automatically on
install. No extra setup needed.

Some skills require the GitHub CLI:

brew install gh    # macOS
gh auth login

Contributing

Skills follow the Agent Skills specification.
Each skill is a directory with a SKILL.md file containing YAML frontmatter and
markdown instructions.
A skill can declare the shared references it needs in a references.yml manifest; the
build hydrates the matching files from the shared library at references/ into the
skill so the shipped copy is self-contained.

License

MIT