This MIT-licensed pack of three skills from GlobalMPC sets safer defaults for Claude Code, Codex and Cursor when they write backend code. The api-endpoint skill scopes every query to the calling user, turns bad input into 4xx errors and rate-limits login and costly routes. The data-integrity skill adds locks or conditional updates, unique constraints and idempotency keys on anything that charges or sends, and the db-migration skill uses expand-then-contract changes, concurrent indexes, lock timeouts, batched backfills and down migrations. Each skill has a section where you write in your team's own conventions. You get backend code from your agent that avoids common auth, double-charge and mid-deploy schema mistakes.
Turn a codebase into a graph your coding assistant can queryGraphify is a skill and command-line tool that turns a project's code, docs, SQL schemas, configs, PDFs and media into a knowledge graph that your assistant queries rather than searching files one by one. Source code goes through tree-sitter on your own machine, so that part needs no model and nothing leaves your machine, while documents and images get a semantic pass from your assistant's model or an API key you set. Every link in the graph is tagged as either read directly from the source or inferred, and the run produces an interactive HTML view, a report of key concepts and a JSON graph for later questions. It adds a /graphify command to Cursor, Claude Code, Codex, Copilot, Gemini CLI and other assistants, and a hosted Graphify Cloud version is sold separately.Coding agentCode intelligenceKnowledge graphs
Take an AI-built app live on your own accountsThis skill takes an app written by an agent from a repository to live production on accounts you own, with providers you choose. You connect the accounts and approve each change, and a local command-line script carries out the supported setup steps, such as environment variables, webhooks, auth settings, and DNS, then records what its checks prove. It tests flows like signup, confirmation email, login, and password recovery, and it covers email, payments, auth, a database, hosting, and a domain. It never prints secrets or asks you to paste them into chat, and it reports what was and was not verified. Security note: it changes hosting, database, auth, email, DNS, and payments settings on real accounts. It requires your approval and a plan before each apply, so read that plan every time.Coding agentDeployment
Check GitHub Actions workflows that run AI agents for injection risksTrail of Bits' agentic-actions-auditor skill reviews GitHub Actions workflows that call AI coding agents such as Claude Code Action, Codex, Gemini CLI or GitHub's AI Inference action. It finds the steps that start an agent, traces into shared actions and called workflows, and records how each is configured. It then reports places where text an outsider controls, like an issue title, can reach the agent, along with loose sandbox settings and allowlists that accept any user. It only reads workflow files and does not run them.Coding agentSecurityCI/CD
Mod a PC game you own, from idea to recorded gameplayUniversal Modder is a skill pack and toolkit that lets Claude Code carry a modding idea for a PC game you own through to a version that runs in the actual game, recorded on video. It scans which games are installed and their engine, mod loaders and anti-cheat, picks the cheapest route, sets up a safe test copy and builds one small working slice first. Art and sound can be generated through the fal MCP, changes are checked in the running game, and it can package the mod and record a showcase clip. The author's examples include a Terraria mod, an extra civilization for the second Age of Empires and a Minecraft-in-GTA V mashup.Coding agentGame moddingClaude Code
Review pull requests using Sentry's engineering checklistSentry's code-review skill gives an agent the checklist Sentry engineers use on pull requests. It covers runtime errors, performance traps such as N+1 queries, side effects, backward compatibility and security holes, then asks whether the design fits the existing architecture and whether tests cover the real requirements. It says which kinds of change, like schema or API contract changes, should go to a senior engineer, and how to word feedback politely and usefully.Coding agentSentryCode review
Set up and query Cloud FirestoreThis skill sets up and manages Cloud Firestore databases in Standard or Enterprise edition. It helps with data models, indexes, queries, and SDK setup for web, Python, iOS, Android, and Flutter apps. Writing security rules is handed to a separate Firebase skill.Coding agentFirebaseDatabase
Turn a codebase into a graph your coding assistant can query
Check GitHub Actions workflows that run AI agents for injection risks
Mod a PC game you own, from idea to recorded gameplay
Review pull requests using Sentry's engineering checklist
Set up and query Cloud Firestore