Build your AI stack
Tools, MCP servers, and skills that work together — from editor to production.
AI Coding Tools
View all →NotebookLM
NotebookLM ingests PDFs, docs, and slides you provide, answers with inline references, and can turn dense material into spoken audio overviews—aimed at students, researchers, and analysts who need traceability more than generic chat.
Bolt
Bolt targets rapid UI shells, marketing sites, and mobile prototypes with hosting and GitHub sync—use it when speed matters more than bespoke backend complexity.
Gemini
Gemini pairs conversational answers with strong Google Workspace and Search adjacency—useful when your prompts mix text, images, and quick research across languages.
GitHub Copilot
Copilot meets developers where PRs and issues already live—inline suggestions, workspace-aware chat, and agent flows that can touch multiple files inside VS Code or JetBrains.
Replit Agent
Describe a product, iterate in the design canvas, and let Replit Agent scaffold code, dependencies, and deploys—popular with students and indie hackers validating ideas in hours.
ChatGPT
ChatGPT spans everyday Q&A, long-form writing, code explanation, and image or file inputs—plus GPTs and connectors when you need repeatable workflows instead of one-off chats.
Claude
Claude shines when you need to load large PDFs, compare versions, or iterate on nuanced prose—team features and projects help keep institutional knowledge in one thread.
DeepSeek
DeepSeek is a go-to when you want chain-of-thought style answers, math-heavy prompts, or repository-scale coding help without burning premium credits.
MCP Servers
More →Neon MCP
Connects AI assistants to Neon's serverless Postgres — branches for each query, full schema inspection, and safe query preview before execution. Ideal for development workflows that need ephemeral database instances without provisioning overhead.
Puppeteer MCP
Controls a headless Chrome browser via the MCP protocol — AI assistants can navigate pages, extract content, fill forms, take screenshots, and run JavaScript in a real browser environment. Useful for web scraping, automated testing, and scraping-heavy workflows.
Airtable MCP
Connects your Airtable base schema, tables, views, and records to any MCP-compatible AI assistant — letting AI read and write structured records, trigger automations, and reason across your Airtable data.
Supabase MCP
Connects MCP clients to Supabase projects for schema inspection, real-time database events, auth management, and Edge Functions invocation so agents can reason about backend state alongside application code.
Cloudflare MCP
Bridges MCP clients to Cloudflare Workers, KV, R2 storage, and D1 databases for agents that need to inspect edge deployment state, manage workers, or check storage buckets without leaving the editor.
Grafana MCP
Exposes Grafana dashboards, alerts, and time-series data to MCP clients so agents can read application health, inspect alert status, and correlate incidents with code changes during debugging sessions.
Claude Code Skills
More →Canary rollouts
A step-by-step checklist for safely releasing features to a small percentage of users before full rollout — covers traffic splitting, metrics monitoring, rollback triggers, and automatic promotion thresholds.
Production debugging
A structured approach to diagnosing live production incidents without causing further damage — covers log triage, metric spike correlation, deploy window filtering, and safe reproduction steps.
Safe dependency upgrades
A structured checklist for upgrading npm/pip/Cargo dependencies without breaking production — covers changelog analysis, semver risk assessment, lockfile handling, and smoke testing.
RAG pipeline construction
Builds retrieval-augmented generation pipelines: embedding chunking strategies, vector store selection, hybrid search blending, and re-ranking so agents answer from your documents rather than hallucinating generic responses.
AI cost optimization
Reduces AI spending through model selection, context window minimization, batch processing, and caching strategies while maintaining output quality — balances performance per token against actual business outcomes.
Multi-agent handoff design
Designs clean handoff protocols between specialized agents so work passes between planner, coder, reviewer, and executor without losing context or creating circular dependencies.