Page 96 of 97 — MCP servers, Claude Code skills, prompt packs, and workflows.
Claude scans the week ahead for earnings releases, macro events, and news that could move stocks — so your watchlist aligns with calendar-driven volatility instead of you getting blindsided by a Fed meeting or CPI print mid-trade.
A prompt that asks Claude to design a capital management framework aimed at aggressive gains while capping downside — position sizing, stop-loss discipline, and max-loss-per-day rules — so the setups above don't blow up the account on a bad day.
Turns short-term trading profits into a long-term capital growth plan — a compounding framework with rules for reinvestment percentage, withdrawal schedule, and when to scale position size — so daily wins don't stay stuck as short-term cash.
Converts your watchlist into 5 concrete day-trading setups — each with an entry zone, stop loss, and profit target. Gives you a structured plan rather than staring at charts hoping for inspiration.
Prompt Claude for 10 stocks screened on three filters active traders actually care about — volatility, volume, and catalyst potential — turning the previous sector scan into a concrete watchlist you can run setups against.
Break down any ticker's business model as if you're deciding to own it for 5 years. Every revenue stream with its % of total, growth trajectory per segment, the competitive advantage holding it ahead (and whether it's widening or narrowing), what would have to go wrong for irrelevance in 5 years. Forces Claude to NOT give balanced overviews — the output is the version where Claude is actively trying to find reasons NOT to buy.
Paste any earnings call transcript — Claude separates real commitments from vague deflection, identifies dodged analyst questions, distinguishes new forward guidance from repackaged old targets, flags what management ISN'T talking about that they should be, and distills the 3 things that actually matter for the stock over the next 12 months.
Stop buying tickers — start understanding landscapes. Claude maps any sector/theme: major players and each one's specific angle, which are pure-play bets vs. diversified-with-exposure, whether the TAM is actually growing or recycled from a 2021 pitch deck, risks to the whole theme (not just individual stocks), and the 3 specific tickers to own if you could only hold 3 to capture the trend over 3-5 years.
Don't overpay because you like the company. Claude identifies the most relevant valuation metrics for THIS specific business type (not a generic DCF), compares current valuation to 3/5-year historical range and closest peers, tells you what growth rate the current price is implying and whether that's realistic, and gives you the obvious-bargain price AND the obvious-expensive price as reference anchors.
The selloff question most people ignore until it's too late. Paste your holdings with allocation %; Claude identifies which positions would all drop together (you're making the same bet multiple times), sector concentration risk, single-factor vulnerability per position, and exactly which 2 positions to cut to reduce risk without losing upside. Instructed to assume you're emotionally attached to all of them and tell you the truth anyway.
Write this BEFORE you need it. A tight investment thesis you can reference when the stock drops and you're tempted to panic sell. 2-sentence "why I own this," 3 things that must stay true for the thesis to hold, 3 things that would break it and trigger an exit, 12-month expectations, and a pre-committed response to a 20% drop tomorrow (buy more, hold, or sell — and why). Written in your voice, direct, no hedging.
The edge is in the footnotes nobody reads. Paste any 10-K section; Claude pulls what actually matters — risk factors buried past the boilerplate, red flags in revenue recognition or accounting changes or related party transactions, what the cash flow statement shows that the income statement hides, and specifically what CHANGED from the previous filing (added, removed, reworded).
Find companies that dominate. Avoid the ones pretending. Claude rates each moat factor 1-10 with evidence — switching costs, network effects, cost advantages, brand power, regulatory barriers, data advantages — tells you whether the moat is widening or narrowing and why, and compares against 2 named competitors on the same criteria. Instructed to be skeptical, not generous.
Never look at a stock in isolation. Claude compares 3 tickers in the same sector across 3-year revenue growth, margin trends, balance sheet strength, competitive positioning, and valuation-relative-to-growth. Identifies which management team is actually delivering vs. just promising, picks a winner, explains why — then deliberately tells you why you might be wrong.
Kill your bias before it kills your portfolio. Paste your 2-3 sentence thesis; Claude returns the STRONGEST bear case (not the obvious risks — the ones that would actually make you sell, from a short seller who's done 6 months of research) plus a DIFFERENT bull case you're not seeing, plus the single data point that would definitively prove your thesis right or wrong.
The underrated "Claude Chart" feature — turned into a repeatable playbook. Core pattern: prompt with "Create an interactive chart about [...]" + the override "Don't code. Interactive chart only." to force Chart mode every time. Includes three battle-tested example prompts: (1) S&P 500 financial-independence chart with inflation/returns scenarios, (2) AI exponential-curves chart, (3) Chess 80/20 rookie mastery chart. Pro tips: deep-search first, then chart; ask for more tactics recursively; force Chart mode when Claude slips back into code output. Plus two sharing workflows — in-workspace "Share" link for teams, and "I need to send it to a client. Code it." prompt that produces a downloadable HTML artifact you drop into Google Drive. Works on the regular claude.ai chat — no Claude Code required. 99% of users don't need more Claude Code tokens, they need a more visual Claude.
Translates ideas into functional prototypes quickly, emphasizing speed over perfection.
Loads everything Claude needs to know about your tone, style, voice, vocabulary, banned phrases, formatting rules, and content philosophy — so every piece of content sounds exactly like you, not a generic AI. Includes example-based calibration (paste your best work), output rules, and a self-check that rewrites anything that feels generic. Works in Claude, Manus, Perplexity, or any AI platform that supports Skills.
Turn one piece of content into a full distribution engine: X/Twitter thread with hook, 10 short-form scroll-stopping hooks, newsletter summary, key takeaways, and actionable bullet points. Adapts tone per platform, native formatting only, consistent messaging, strong openings prioritized. The most-used skill in the author's personal Claude library — maximize reach + impact from a single source.
Helps build optimal AI prompts for various AI models to improve output quality.
Train Claude on your exact standard operating procedures so every repeatable business task executes the same way every time. Per-SOP structure: trigger → steps (exactly ordered) → output format → rules. Execution rules prevent improvisation, demand confirmation on missing data, and flag tool errors immediately. Integrates with Notion/CRM so Claude reads SOP docs in real-time. Includes memory instruction — flags undocumented repeated tasks and suggests new SOPs over time.
Engages Claude as a strategic partner to enhance decision-making by exposing blind spots.
Turn Claude into your personal morning analyst: every trigger produces a structured daily brief covering top industry developments (24h), day-at-a-glance from connected tools, one high-attention item you might have missed, and urgent items needing response. Under 400 words, no filler, executive-briefing tone. Best run as a scheduled task inside Cowork so it auto-delivers every morning from Gmail + Slack + Calendar + custom connectors.
Provides tailored strategic advice based on your role and goals, adapting over time for increased relevance.
Configures Claude to structure and deliver research outputs according to your preferences.
A battle-tested Claude prompt that simulates 4 specialized expert roles at the same table — Frontier Tech VC, Deep Value Short Seller, Industry Insider, and Macro/Special Situations Trader — to produce an independent, anti-sycophantic investment thesis on any company you're researching. Four-step protocol: (1) structure the company state (tech maturity, revenue, runway, competition, key milestones), (2) generate separate theses per expert with 3 risks each, (3) run a dialogue where the experts critique each other on timeline reliability, cash burn vs. execution, and moat defensibility, (4) crystallize 3 bull arguments, 3 bear arguments, 2 critical unknowns, and a risk/reward story-quality assessment. Ends with the anti-sycophancy lock: "What is the weakest point of this analysis?" Proven on $AUR (Aurora Innovation). Works on any public or private company with enough data. Tone: technical, honest, no hype. Inspired by Karpathy-style adversarial prompt tactics.
The capstone prompt: integrates prompt improvement + workflow optimization + output quality + decision intelligence + agent expansion into ONE unified infrastructure where gains in any component raise the ceiling of every other. Unified capture (≤5 min/day overhead), cross-component learning, monthly improvement-priority framework, and compounding-velocity tracker that measures RATE of improvement, not absolute level.
Maps your agent task inventory (current + manual candidates), captures performance + failure-point per task, expands automation only into tasks similar to ones already reliable, and tracks human-intervention rate monthly so the percentage of recurring tasks running without you trends downward quarter over quarter.
A structured Claude prompt that builds a complete prompt-performance capture + pattern-identification + refinement + versioning system. Runs on your most-used prompts, tracks quality per use, identifies which prompt elements drive high-quality outputs, and updates your prompt library automatically every 30 days. Every improvement is traceable, reversible, and tied to captured data — never vibes.
Defines quality dimensions per output type, captures quality signals immediately after generation, extracts which prompt and context patterns correlate with exceptional outputs, and tracks monthly baseline so your average AI output quality trends measurably upward over 90 days — not random vibes-based adjustments.
Pre-decision evaluation framework (≤10 min per decision) + decision capture before acting + outcome tracking against original expectations + monthly pattern analysis. Stops you from applying AI to execute decisions already made, and starts applying it where the highest leverage lives — making every decision better before it's acted on.
A structured Claude prompt that maps your AI workflow end-to-end, captures execution quality + friction at every step, identifies the single highest-impact bottleneck across 5+ runs, and refines one step at a time on a fixed schedule so every execution is measurably smoother than the previous one.
Turn Claude Code into a full video production pipeline using Hyperframes (HeyGen's "write HTML, render video" engine built for agents). One-command install: `npx skills add heygen-com/hyperframes`. Claude scaffolds the full project, opens a timeline editor with compositions, scenes, captions, and narration tracks, auto-generates voiceover, and produces the full transcription. Proven on a full Artemis II mission video (lunar trajectory animations + Earth return + narration) from a single prompt. Designed to layer cleanly with HeyGen's avatar / voice tech. A different category of video production tooling compared to generic AI video apps — the quality comes from investing in the prompt, which this skill structures for you.
Open-source plugin that turns Claude Code into a senior engineer across 66 specialized skills spanning 25+ frameworks and 13 languages. Uses a Progressive Disclosure pattern: the skill loads lean (80 lines), then surgically loads only the reference files your current task needs — zero token bloat, full expert depth only when required. Detects your framework from context, activates the right specialist (React / Next.js / Django / FastAPI / NestJS / Kubernetes / Terraform / and dozens more), and combines skills for complex tasks (Security Reviewer + NestJS Expert + Test Master at once). Auto-activates a Debugging Wizard when something breaks. 284 battle-tested reference files pulled in on demand. MIT license, one-command install, works across every project immediately. The difference between a junior AI assistant and a senior engineering partner.
A full AI operating system for Claude Code: 28 AI subagents (planner → reviewer → security), 100+ skills that teach your agent how to think, auto-memory with continuous learning, built-in testing + security checks, and a reported 60% token-cost reduction via harness-level optimization. Cross-platform: Claude Code, Codex, Opencode, Cursor. Installs via marketplace. Replaces one-off prompts with a persistent agent harness that composes skills, instincts, memory, and research-first development patterns.
An automation loop that runs any Claude Code skill 10 times with varied inputs, scores outputs against 3-5 binary eval criteria, rewrites the skill prompt to fix common failures, and repeats until the score plateaus. Converts inconsistent skills (good 70% of the time, unusable 30%) into production-grade reliable outputs. Inspired by Karpathy's auto-research loop — the same method AI labs use to improve their own models, applied to creative and engineering workflows. One loop → 10 test runs, scored against an eval, prompt rewritten, retested, winner kept. A hook writer skill went from 32/50 → 47/50 overnight. Works on any Claude Code skill: hooks, briefs, ad copy, scripts, reports, agents. Outputs: hardened skill prompt, scored improvement log (exactly what changed + why), reusable eval criteria, and a method that compounds over time.
Reviews conversations to identify logical errors, incorrect assumptions, and overlooked risks.
Retains and references project context across interactions, summarizing key information without needing repeat input from users.
Analyzes a project's entire folder for architectural decisions, technical debt, security vulnerabilities, and performance bottlenecks, providing a prioritized action plan.
Tracks long-term projects by retaining decisions, context, and shifting requirements over time.
Facilitates integration of multiple projects by analyzing codebases and suggesting integration strategies and shared structures.
Assesses the depth of thinking required for complex problems and solves them accordingly.
Enables simultaneous completion of tasks by creating agent teams, coordinating between them, and merging results.
This tool allows Claude to scan real websites and rebuild their design system instantly, transforming any page into a customizable starting point for users.
This tool integrates Vugola into Claude, enabling automatic clipping, scheduling, and posting of YouTube content.
This tool allows users to copy and recreate any UI from the internet, helping them to quickly generate design prototypes.
This skill enhances Claude's coding capabilities by utilizing a playbook developed by AI expert Andrej Karpathy, improving speed and accuracy.
This MCP enables Claude to inspect live websites, extracting their UI design elements to generate custom layouts more effectively.