pedronauck/skills

Cloudflare Agents SDK is designed for building stateful, globally distributed AI agents on Durable Objects. It provides built-in support for persistent memory, real-time WebSocket connections, scheduled tasks, async workflows, and SQL integration.

Use this SDK when your agent requires:

• Persistent state and memory.
• Real-time WebSocket connections.
• Long-running workflows (minutes or hours).
• Chat interfaces with AI models.
• Scheduled or recurring tasks that maintain state.
• Database queries that depend on agent state.

The Mermaid Diagrams Skill allows you to create professional software diagrams using text-based syntax. It is designed to transform complex technical concepts into maintainable, version-controllable diagrams that can be rendered in Markdown files (like those on GitHub/GitLab), VS Code, Notion, Obsidian, and Confluence.

Trigger Phrases

The skill activates when you use phrases such as:

• "diagram", "visualize", "model", "map out", "show the flow"
• "architecture diagram", "class diagram", "sequence diagram", "flowchart"
• "database schema", "ERD", "entity relationship"
• "system design", "data model", "domain model"

Cloudflare Network Interconnect (CNI) provides private, high-performance connectivity to Cloudflare's network. This service is Enterprise-only.

Connection Types

• Direct: Physical fiber in a shared datacenter (10/100 Gbps). Requires ordering a cross-connect.
• Partner: Virtual connectivity via partner SDN providers like Console Connect, Equinix, or Megaport.
• Cloud: Connectivity via AWS Direct Connect or GCP Cloud Interconnect (available via Magic WAN only).

Dataplane Versions

• v1 (Classic): Supports GRE tunnels, VLAN/BFD/LACP, and peering. Note that MTU is asymmetric (1500↓/1476↑).
• v2 (Beta): Does not support GRE or VLAN/BFD/LACP (uses ECMP instead). MTU is 1500 in both directions.

Cloudflare TURN (Traversal Using Relays around NAT) is a managed relay service for WebRTC applications. It provides a relay point for traffic between WebRTC clients and SFUs when direct peer-to-peer communication is blocked by NATs or firewalls. The service utilizes Cloudflare's global anycast network across 310+ cities to connect clients to the nearest location automatically.

Key Characteristics:

• Anycast Architecture: Automatic connection to the closest Cloudflare location.
• Global Network: Available across Cloudflare's network (excluding China Network).
• Zero Configuration: No manual region or server selection required.
• Protocol Support: STUN/TURN over UDP, TCP, and TLS.
• Pricing: Free when used with Cloudflare Calls SFU; otherwise, it costs $0.05/GB outbound.

R2 Data Catalog is a managed Apache Iceberg REST catalog built directly into Cloudflare R2 buckets. It allows you to manage analytical datasets using the Apache Iceberg open table format, providing ACID transactions, schema evolution, and time-travel queries without managing any catalog infrastructure.

Key Features

• Apache Iceberg Tables: Supports ACID transactions, schema evolution, and time-travel.
• Zero-egress Costs: Query data from any cloud or region without data transfer fees.
• Standard REST API: Compatible with engines like Spark, PyIceberg, Snowflake, Trino, and DuckDB.
• Fully Managed: No catalog servers to maintain; available to all R2 subscribers at no extra cost beyond storage/operations.

Cloudflare Bot Management provides multi-tier protection based on your subscription plan:

• Free (Bot Fight Mode): Automatically blocks definite bots with no configuration required.
• Pro/Business (Super Bot Bot Fight Mode): Offers configurable actions, protection for static resources, and analytics groupings.
• Enterprise (Bot Management): Provides granular bot scores (1-99), WAF integration, JA3/JA4 fingerprinting, access to the Workers API, and Advanced Analytics.

Cloudflare Pages Functions provides several core capabilities for full-stack development:

• Method handlers: Use specific functions like onRequestGet, onRequestPost, etc., to handle different HTTP methods.
• Middleware: Implement _middleware.js for cross-cutting concerns like authentication or logging.
• Bindings: Access Cloudflare resources directly via bindings, including KV, D1, R2, Durable Objects, Workers AI, and Service bindings.
• TypeScript: Full type support is available via the wrangler types command.
• Advanced mode: For custom routing logic that exceeds file-based capabilities, use _worker.js.

Depending on your security requirements, you can implement specific features within API Shield:

• Validate request/response structure: Use Schema Validation 2.0 (see configuration.md).
• Verify auth tokens: Use JWT Validation (see configuration.md).
• Client certificates: Use mTLS (see configuration.md).
• Detect BOLA attacks: Use BOLA Detection (see patterns.md).
• Track auth coverage: Use Auth Posture (see patterns.md).
• Stop volumetric abuse: Use Abuse Detection (see patterns.md).
• Discover shadow APIs: Use API Discovery (see api.md).

This skill is designed for development targeting React 19.2 and React Compiler 1.0 (babel-plugin-react-compiler), using eslint-plugin-react-hooks 7.1+.

It is intended for writing React components or hooks, deciding on memoization (useMemo, useCallback, memo), diagnosing compiler bailouts, managing state, typing props/refs in TypeScript, and using React 19 APIs like Actions or use().

Do NOT use this skill for:

• React Native
• Non-React frameworks (Vue, Svelte, Solid)
• Backend-only Node.js code

This reference is organized into several key documents to guide your development:

• configuration.md: Covers setup, deployment, environment variables, and Wrangler configuration.
• api.md: Details Sessions, tracks, endpoints, and request/response patterns.
• patterns.md: Provides architecture patterns, use cases, and integration examples (including PartyTracks, presence systems, and scaling).
• gotchas.md: Lists common issues, debugging steps, performance considerations, and security tips.

The kubernetes-specialist skill is designed for deploying and managing Kubernetes workloads. It is appropriate for tasks involving workload deployment (Deployments, StatefulSets, etc.), networking configuration (Services, Ingress, NetworkPolicies), configuration management (ConfigMaps, Secrets), persistent storage, Helm chart creation, troubleshooting, and implementing security best practices like RBAC and Pod Security Standards.

Core Workflow:

1. Analyze requirements: Understand scaling, security, and workload characteristics.
2. Design architecture: Select workload types, networking, and storage patterns.
3. Implement manifests: Create declarative YAML with resource limits and health checks.
4. Secure: Apply RBAC, NetworkPolicies, and least privilege principles.
5. Validate: Use kubectl to confirm health and roll back if necessary.