Page 94 of 97 — MCP servers, Claude Code skills, prompt packs, and workflows.
Perplexity MCP connects Claude to live internet resources, enabling comprehensive financial research beyond traditional data.
Transforms Claude into an execution tool for DeFi interactions, allowing real swaps, token transfers, and balance checks.
MCP designed for traders, providing actionable insights beyond raw data.
The CoinGecko MCP provides an extensive database of cryptocurrency data, including real-time prices and historical trends.
Widely used platform providing real-time cryptocurrency market data and statistics.
Social intelligence platform for tracking crypto trends and social engagement metrics.
A powerful Model Context Protocol (MCP) server that enables AI clients to scrape web data without blocks or CAPTCHAs.
This tool analyzes Claude session data to identify inefficiencies in workflows and suggests improvements.
Firecrawl is a web scraping tool that allows users to extract data from websites efficiently.
A tool that scrapes data and generates unique SEO content pages efficiently using AI.
A tool that extracts profile information from public LinkedIn pages, such as name, title, location, and work experiences.
A skill repository that provides coding techniques and insights inspired by Andrej Karpathy, aimed at enhancing the capabilities of AI coding workflows.
This tool allows users to inspect live website UI designs by reading HTML and CSS, enabling more effective and aesthetically pleasing webpage creation.
Vugola integrates with Claude to automate the clipping, scheduling, and posting of social media content across multiple platforms.
This tool scans real websites and rebuilds their design system instantly, allowing users to repurpose any webpage as their starting point.
A tool for verifying emails and enriching contact data with company info.
An open-source workflow automation tool for connecting apps and services.
A tool that plans, builds, deploys, and self-heals workflows from plain English, enabling users to automate processes without writing code.
Paperless-ngx scans, recognizes, and organizes documents automatically, making them searchable while keeping them stored on your own server.
Opus for planning and architecture (deeper thought). Sonnet for implementation (faster + cheaper). Plan with the thinker, build with the builder.
setup.sh that onboards a new dev in one command: deps, .env from .env.example, local DB, migrations, seed data, verify by running tests.
Before adding a package: "Is it maintained? Security issues? Bundle size? Lighter alternatives for my specific use?" Prevents bloat and risk.
Claude analyzes for N+1 queries, missing indexes, unnecessary re-renders, large imports that should lazy-load, cacheable endpoints. Prioritizes by estimated impact.
Schema change prompt that generates migration + repository update + API route updates + TS types, showing every file that needs to change BEFORE modifying anything. Claude shines on multi-layer ripple effects.
Read git log since last tag; generate release notes organized by features / bug fixes / perf / breaking changes. User-friendly language, not dev jargon.
Comprehensive dev seed: 5 users across roles, 20 realistic projects, relationships, edge cases (archived project, deleted user, empty project). No "test123" data.
Switch to plan mode before any implementation. Claude analyzes the codebase and returns an architecture plan with zero code. Review → approve → switch back to build. Prevents more bugs than any other single habit.
UI looks wrong? Screenshot + Ctrl+V + "Fix the alignment and card spacing." Visual feedback beats written descriptions every time.
"This function started failing yesterday. Read git log for this file over the past week. Identify which commit likely introduced the issue and explain. Then suggest the fix."
Before a new project: "Propose 2 architectural approaches. For each: component diagram, pros, cons, complexity, failure modes. Recommend one with reasoning." Prevents expensive rewrites.
/init scans the codebase and generates a persistent CLAUDE.md (project structure, stack, patterns, architecture decisions) that Claude auto-reads every future session. Run at the start of every new project.
/memory adds instructions that apply across all future sessions ("Always use TypeScript strict mode", "Always run tests after /src/core changes"). No repeating rules.
/compact compresses 30-45 min of conversation into a focused summary of decisions and current state. Claude stays sharp instead of slowly losing the plot.
Before every major change: `git add . && git commit -m "checkpoint before [change]"`. When Claude breaks something, you revert in seconds.
After Claude changes files: "Show me a diff of every modified file. Explain each change in one sentence." Catches unintended "helpful" edits you didn't ask for.
/clear wipes context between unrelated tasks. Carrying database-refactor context into a frontend redesign produces conflicting code. One conversation per feature.
Add to CLAUDE.md: "API routes follow src/api/example-route.ts. Repositories follow src/repositories/example-repo.ts. Components follow src/components/ExampleComponent.tsx." Patterns auto-match in new files.
Before modifying unfamiliar code: "Read src/services/ and explain the data flow from API routes to database. What patterns? What should I know before modifying?" Understanding prevents architectural mistakes.
Paste the conflict error; Claude analyzes which packages require conflicting shared-dep versions, suggests minimum-change resolution with tradeoffs.
From a bug report: create a minimal reproduction (steps + expected + actual), a failing test that captures the bug, then fix code to make the test pass.
Instead of describing code style verbally, point Claude at an existing file: "Follow the same patterns as src/auth/login.ts." Dramatic consistency improvement.
800-line file too complex? "Propose a refactoring plan splitting it into focused modules. Show proposed structure, what moves where, verify no imports break. Do NOT start yet." Plan-only output.
When Claude is stuck in a broken-implementation loop: "Stop. Read the original working version from git. Restate the goal simply. Start fresh with a different approach." Faster than accumulated-mistake debugging.
Scan codebase for SQL injection, XSS, exposed secrets, missing input validation, IDOR, rate-limit gaps. Per-finding: severity, location, danger, fix.
Never say "build the entire feature." Break into steps: schema → test → endpoint → test → validation → test → frontend → test. 5 steps with tests produces dramatically better code than one mega-prompt.
! prefix runs the message as a terminal command instead of sending to Claude. Quick tests, git status, navigation — no leaving the Claude Code UI.
GitHub Actions workflow per PR: install, test, lint, build, comment results on the PR. Uses node_modules caching.
For large features, run two terminals: one Claude Code for backend, one for frontend. Clean focused context per domain. Connect at the end.