Skip to content
Claudekit.ai

Sentry Skills: Engineering Workflows from Sentry

getsentry/skills packages Sentry's internal agent skills: code review, find-bugs, security review, PR writing, commit rules and skill scanning. Install, tips.

By Claudekit Team Published 1 min read

Category
Workflow
Repository
github.com
GitHub stars
1,083
License
Apache-2.0

Install

claude plugin install sentry-skills@sentry-skills

Sentry Skills (getsentry/skills) is the set of agent skills Sentry’s own engineers use, published in the open Agent Skills format. Some are Sentry-specific (brand voice, Django reviews, internal triage), but many encode engineering practices any team can borrow: structured code review, bug hunting in local changes, security review, PR writing and CI iteration.

Who it is for

Teams that want proven, opinionated review and PR workflows without writing skills from scratch, and anyone curious how a large engineering organization structures its agent skills.

What it covers

  • Review and quality: code-review, find-bugs, code-simplifier, security-review, gha-security-review, secret-serialization, django-access-review, django-perf-review.
  • Git and PRs: commit, create-branch, pr-writer, iterate-pr (loop until CI passes and feedback is handled), pr-screenshots, pr-link-issue.
  • Agent tooling: agents-md for concise AGENTS.md and CLAUDE.md files, claude-settings-audit for recommended permissions, skill-scanner for skill security issues, skill-writer, prompt-optimizer, security-model.
  • Subagents: code-simplifier and senpai, a mentor for new hires.

Install

claude plugin marketplace add getsentry/skills
claude plugin install sentry-skills@sentry-skills

Or for other agents: npx skills add getsentry/skills. Restart Claude Code afterwards.

Example prompts

Find bugs and security issues in my local changes before I push
Iterate on this PR until CI is green and review comments are addressed
Audit this repo and propose Claude Code permission settings
Scan the skills in .claude/skills for risky instructions

Tips and limits

  • Sentry-specific skills (brand, SRED, internal triage) add noise; disable or ignore them.
  • commit says to always use it when committing, which may conflict with your own commit conventions.
  • Pairs with the Sentry MCP server for production error context.

Related: Trail of Bits skills, CLAUDE.md guide, permission modes.

Frequently asked questions

Are these skills for setting up Sentry in my app?

No. They are Sentry's internal engineering skills. For Sentry setup and production debugging, Sentry points to getsentry/sentry-for-ai.

Which skills are useful outside Sentry?

find-bugs, security-review, gha-security-review, skill-scanner, iterate-pr, agents-md, claude-settings-audit and prompt-optimizer are broadly useful.

Do I need to restart Claude Code?

Yes. The README says to restart Claude Code after installation; skills then activate automatically when relevant.

Get the weekly Claude digest

New skills, MCP servers, prompts and Claude updates — one short email a week. No spam, unsubscribe anytime.

The newsletter is launching soon. Follow the RSS feed in the meantime.