Sentry

Writes or edits a technical blog post in a consistent voice with clear examples, proper depth, and a hook that lands with engineers. Handles drafting, review cycles, and final polish.

Takes a function or module that works but is tangled and produces a cleaner version with the same behavior, shorter variable names, and comments where logic isn't obvious.

Adds the GitHub issue URL and its linked Linear ticket to your PR description in one step, so reviewers see the context and the ticket auto-resolves when merged.

Audits a SKILL.md for prompt injection, malicious code, overpermissioned integrations, and unsafe patterns before you add it to your agent.

ALWAYS use this skill when committing code changes — never commit directly without it. Creates commits following Sentry conventions with proper conventional commit format and issue references. Trigger on any commit, git commit, save changes, or commit message task.

Generates UI text, error messages, docs, and marketing copy in a consistent voice and tone. Enforces your brand's word choices, personality, and grammar rules across every surface.

Pulls session replays from Sentry, groups them by intent and outcome, and surfaces the moments where users pause, backtrack, or rage-click — with counts so you can prioritize.

Reads Django models and views, finds wasteful querysets, spots N+1 loops, and flags missing indexes or select_related calls with code fixes.

Analyzes your current branch for logic errors, security vulnerabilities, and code-quality regressions. Flags only the bugs worth fixing, with severity and a specific remediation.

Scaffolds a slide deck app with live charts, animations, and speaker controls. Ships as a standalone React app you can customize, share, or embed.

Takes your project list and scattered docs, reorganizes them into the standard SRED filing format with all required sections filled and cross-referenced.

Scans code for OWASP-class vulnerabilities—injection, XSS, auth, crypto—and flags the specific line, severity, and how to fix it without guessing.

Connects to Sentry, reads your JavaScript errors, and archives the ones that are browser junk or expected—leaving only the bugs that need engineering attention.

Reads a PR diff and produces line-by-line feedback on security, performance, test coverage, and design—flagging the risky stuff first, then style nitpicks.

Scans your work history, groups related links and efforts into coherent projects, and exports them to a Notion doc ready for portfolio or review use.

Connects to GitHub and lists open PRs where your team is requested to review, sorted by age. Stops you from missing a blocking review buried in notifications.

Watches a pull request, catches CI failures and code-review comments, then commits fixes automatically until checks pass and feedback is resolved.

Takes your branch and diff, writes a structured PR with a summary, why-this-matters section, and testing steps that match your team's conventions.

Takes a vague or inconsistent prompt and sharpens the wording, adds guardrails, and tests edge cases until Claude stops making the same mistakes on that task.

Drafts or updates AGENTS.md and CLAUDE.md with clear conventions, decision trees, and constraints that keep Claude focused—without bloat or contradictions.

Creates a branch with the right prefix (fix/, feat/, etc.) based on the work type, keeping your repo history clean and searchable.

Writes a SKILL.md from scratch or iterates on an existing one, building triggers, instructions, and examples that match the Agent Skills spec and your intent.

Removes modules from your mypy exclusion list in small batches, fixes the type errors that surface, validates the fixes, and reports what's been cleared and what remains.

Scans your codebase to detect tech stack and build tools, then generates a settings.json with least-privilege bash commands and file-access rules Claude can safely execute.

Guides you through a step-by-step workflow: dump context, let Claude draft sections, you review and refine, iterate until done. Keeps the work moving without losing your voice or intent.

Scans Django views, ORM queries, and permission checks for IDOR and access-control gaps. Flags which users can see what data they shouldn't.

Scans .github/workflows/ for expression injection, credential exposure, and privilege escalation traps that let a PR attacker pwn your CI.