Sentry Skills: how Sentry engineers review code and write PRs with AI
An agent knows nothing about a team's branch names, commit style or review bar until someone writes them down. Sentry publishes the skills its own engineers use, around 30 in the open Agent Skills format, from security review and bug hunting to PR descriptions. Some are tied to Sentry's internal processes.
find-bugs and security-review look for bugs and vulnerabilities in the current branch's changes, including injection, XSS, authorization and crypto mistakes.
gha-security-review reads .github/workflows for pwn requests, expression injection and credential theft, with concrete attack scenarios.
django-access-review checks views and DRF viewsets for IDOR and broken object permissions; django-perf-review spots N+1 queries.
claude-settings-audit detects the repository's stack and drafts settings.json permissions, including which read-only shell commands to allow.
secret-serialization finds tokens and passwords leaking through dataclass repr, model_dump, JSON.stringify, structured logs and tracing spans.
When to use it
You want the agent to read the branch for bugs and security issues before each commit, not just run the tests.
Your repository has GitHub Actions workflows triggered by outside pull requests and you need to know if they can be abused.
It is time to write AGENTS.md or CLAUDE.md, and you want a short file that points to references instead of a wall of rules.
Ask your agent
Find bugs and security issues in this branch's changes against main
Check .github/workflows for pwn requests and injection through PR titles
Draft recommended settings.json permissions for this Django repository
Questions
How do I install Sentry Skills?
In Claude Code run claude plugin marketplace add getsentry/skills and claude plugin install sentry-skills@sentry-skills, then restart Claude Code. Other agents can use npx skills add getsentry/skills.
Will these skills help me set up Sentry in my app?
No, this repository is about development practice. For Sentry setup and debugging production issues the README points to getsentry/sentry-for-ai.
Does Sentry Skills include tools for writing my own skills?
Yes. skill-writer creates and iterates on skills following the Agent Skills specification, and skill-scanner checks agent skills for security issues.
