Page 90 of 97 — MCP servers, Claude Code skills, prompt packs, and workflows.
AI operating system blueprint for legal, accounting, and consulting firms. Architecture: practice management (Clio/PracticePanther) + document management (NetDocuments/iManage) + accounting platforms (QuickBooks/Xero) + e-discovery + Qdrant RAG over prior client work, templates, and billable-hour patterns + Claude MCP + n8n. 10 MCP tool schemas: contract reviewer + redliner, legal research synthesizer, deposition prep-er, client-letter drafter, trial-balance reconciler, audit-workpaper generator, tax-return reviewer, consulting deliverable builder, time-entry suggester, conflicts-of-interest screener. The highest-margin knowledge work in the economy, AI-accelerated.
Auditor-grade workpaper generation: pulls trial balance + prior-year workpapers + client correspondence, generates the current-year workpaper with tick-marks, variance explanations, referenced supporting docs, and review-ready memo. Saves 40% of staff audit hours in the prep phase; every page is reviewable, nothing is hidden generation.
Claude-powered ambient scribe for clinical encounters. Transcribes the patient-clinician conversation, auto-structures into SOAP note format (Subjective / Objective / Assessment / Plan), generates ICD-10 + CPT coding suggestions, and outputs ready-to-sign EHR entries. HIPAA-aware data path: PHI stays within the clinician's control, redaction boundary sits BEFORE any external model call.
Claude workflow that intakes new client data (demographics, goals, risk profile, existing assets) and generates the full onboarding packet: Investment Policy Statement draft, fee schedule, Form ADV Part 2B for the assigned advisor, advisor agreement, risk questionnaire, and initial allocation recommendation. Compliance-aware — pulls boilerplate from firm-specific RAG, flags sections needing human sign-off.
AI-native operating system blueprint for RIAs, investment advisors, and family offices. Architecture stitches CRM (Redtail / Wealthbox / Salesforce FSC) + portfolio management (Orion / Black Diamond / Addepar) + financial planning (eMoney / RightCapital) + compliance (ComplySci / SmartRIA) + Qdrant RAG for IPS language, Form ADV boilerplate, compliance memos, and historical client notes + Claude MCP bridges + n8n automation. Ships the compliance-aware data model, 10 Claude MCP tool schemas (client onboarding, portfolio review, quarterly report, IPS drafter, fee-disclosure reviewer, ADV updater, meeting prep, prospect nurture, 13F checker, household aggregator), RAG design for regulatory language, and deployment guide. Positions the advisor to serve 3-5x more clients without adding headcount while staying audit-defensible.
Pre-meeting intel brief: pulls prospect/client data (CRM history, portfolio changes since last meeting, in-the-news triggers, tax-year context, upcoming liquidity events), generates talking-points memo, flags relationship risks (outflows, competitor contact, unopened emails), and drafts discussion agenda. Turns 30-min prep into 5-min prep.
Claude as contrarian sparring partner: takes your investment thesis, constructs the strongest opposing case, identifies weakest assumptions, surfaces disconfirming evidence, simulates the bear market scenario, and returns a pre-mortem. Catches the thesis blindspots you'd otherwise only discover after the check clears.
AI-native operating system blueprint for hedge funds, private equity, and angel investors. Architecture: CRM/deal pipeline (Affinity / DealCloud / Notion) + data room (iDeals / Intralinks / SharePoint) + LP portal + portfolio-company monitoring + Qdrant RAG over prior deal memos, LP correspondence, diligence findings, and portfolio-company board materials + Claude MCP bridges + n8n automation. Ships 10 Claude MCP tool schemas (deal memo, due diligence packet, LP quarterly, portco monitoring, thesis red-team, cap table analyzer, founder reference synthesizer, competitive intel, ICP fit scorer, exit scenario modeler), RAG design for deal-flow intelligence, and deployment guide. Turns one analyst into a team; turns one partner into a fund.
End-to-end DD packet: financial audit cross-reference, customer reference synthesis (from call transcripts), competitive threat map, team background check consolidation, regulatory/litigation search summary, technology DD (for tech deals), and a risk-weighted red-flag list. Ingests the data room and prior deals via RAG to spot pattern-matched issues the human DD team might miss.
Consistent scoring rubric Claude applies to every inbound deck: founder quality (8 sub-scores), market quality (6), product quality (5), traction (5), terms (4). Generates a normalized score + narrative rationale + suggested next-step (pass / second look / term sheet). Learns from YOUR prior investments — high-scoring winners vs. missed-opportunities — to calibrate to your actual thesis, not a generic one.
Structured deal memo: executive summary, market size + growth, team assessment, competitive landscape, financials + unit economics, deal terms + capital structure, thesis + anti-thesis, recommendation. Pulls data from pitch decks + data room + prior memos on similar-stage deals via MCP. 3-hour analyst memo in 20 minutes.
Generates personalized quarterly client reports: portfolio performance vs. benchmark, contribution/withdrawal summary, allocation drift, fee transparency, notable transactions, plain-English commentary on the quarter's market. Pulls from custodian feed (Schwab / Fidelity / Pershing) + Orion/Black Diamond via MCP. Turns a 4-hour quarterly report per client into a 5-minute review-and-send.
Fund-to-LP quarterly: fund performance (IRR, MOIC, DPI, RVPI), attribution by deal, capital calls + distributions summary, portfolio company updates, market commentary, specific follow-on activity, and next-quarter outlook. Pulls from fund admin (SS&C / Juniper Square / Carta) + portco updates via MCP. Turns a week of ops-team effort into a day.
Drafts full IPS documents from client intake data: return objectives, risk tolerance, liquidity needs, time horizon, asset-class constraints, rebalancing policy, manager selection criteria, performance measurement framework. Boilerplate language pulled from firm-approved RAG source; advisor reviews + signs. Audit-defensible structure matching CFA Institute IPS best practices.
Claude-powered compliance review: scans advisor-to-client communication (emails, performance reports, marketing copy) for missing fee disclosures, prohibited performance-based claims, required risk warnings, and cherry-picked performance. Flags every SEC-rule violation before the comms go out. Cites the specific rule citation per flag (Advisers Act 206, marketing rule 206(4)-1, custody rule 206(4)-2).
End-to-end RAG setup: ingest SOPs, pricing rules, damage policy, packing standards, contract language, and training manuals into Qdrant / Supabase pgvector. Claude answers crew questions like "how do we move a piano?", "what's our claim policy?", "what's the script for a late crew?" via semantic retrieval. Ships the chunking rules, embedding model choice, system prompt, and refresh cadence. The institutional knowledge that lives in the owner's head becomes a searchable system every new hire can tap.
Claude prompt that takes inventory list + distance + stairs/elevators + packing requirements and returns crew size, truck count, estimated hours, and a defensible price range. Replaces the manual phone-quote bottleneck that kills 30-40% of inbound moving-company leads.
Claude Code workflow that ingests daily job list + crew availability + truck capacity + geography, calls Google OR-Tools via MCP to solve the vehicle routing problem with time windows, and returns an optimized crew/truck daily plan minimizing fuel, idle time, and overtime. Replaces the whiteboard + paper-schedule dispatch process that dominates the industry.
Seven-touch Claude-powered follow-up: quote sent → reminder at T-72h → T-24h packing checklist → day-of crew intro + ETA → move-complete thank-you + review request → damage claim instructions if any → post-move upsell (storage, future moves). Every touch personalizes to the move details via MCP pulls from the job record.
Claude workflow that intakes damage claims: collects photos, pulls the move record (inventory list + crew + signed waiver), cross-references against company policy via RAG, drafts the claim-response letter, and routes to legal review. Cuts the 2-3 week manual claim cycle to 48 hours and standardizes tone across operators.
Claude skill that checks the dispatch calendar for available move windows, creates the customer record, generates the quote, writes the deposit invoice, and books the calendar hold — all via MCP calls to the CRM, calendar, and payment system. Closes the "got quote, didn't book" gap most moving companies never fix.
Full architecture blueprint for an AI-native Moving Company Operating System. Stitches open-source logistics infrastructure (Fleetbase for dispatch/drivers/routes, Budibase for internal tools, n8n for automation, Traccar for GPS fleet tracking) with Google OR-Tools (vehicle routing solver), Qdrant/pgvector (RAG memory for SOPs/pricing/training), and Claude via MCP servers. Ships the complete data model mapping (jobs→moves, drivers→crews, routes→schedules), 8 Claude MCP tool schemas, 15 n8n workflow modules, 10-category prompt pack, RAG index design for ops knowledge, and deployment sketch. Positioned for the vertical where there is NO dominant open-source operator — the opportunity is to BE the platform.
Claude workflow that watches completed jobs, times + personalizes review requests per platform (Google Business, Yelp, BBB, Moving.com), detects angry customers from survey sentiment BEFORE they post publicly, auto-escalates to owner, and drafts the owner's recovery outreach. Turns the post-move review pipeline from leaking 60% of 5-stars to capturing 80%, while intercepting 1-star reviews at the source.
Framework where agents autonomously explore market inefficiencies and optimize a stock portfolio. Multi-agent exploration + portfolio optimization loop runs without manual intervention. Claude-adjacent via the agent framework layer; plug Claude in as the reasoning engine.
Self-hosted AI clipping tool that replaces $49/mo paid clippers. Clips any YouTube or local video, no watermarks, no limits, no tier gates. Pluggable AI backend — use Claude, or any other model. Runs fully open-source and local. Direct replacement for the OpusClip / Vizard class of SaaS clippers.
Step-by-step system for turning Obsidian into a second-brain + content-production engine powered by Claude Code. Ships the exact vault architecture (00-INBOX → 01-CAPTURES by NOTE TYPE not topic → 02-CONNECTIONS → 03-BRIEFS → 04-PUBLISHED → 05-CLAUDE), the CLAUDE.md identity/voice/rules template, 4 reusable skill .md files (process-inbox, weekly-connections, generate-brief, write-content), daily 20-minute ritual, weekly Sunday connection session, and the monthly performance-feedback query that makes the system compound over 6 months. Lives at 05-CLAUDE/CLAUDE.md inside your Obsidian vault; invoked via `claude` CLI from the vault root.
The app-store-style catalog for Claude Code: 1,000+ ready-to-use agents, skills, slash commands, MCP servers, and hooks, all installable with a single command. Free to browse and install. Stops the "hunt across 20 GitHub repos to find a subagent" loop — one site, one command, one install.
Claude as a small-cap growth investor — filters for market cap <$2B, annual revenue growth >20%, expanding operating margin, strong balance sheet, insider ownership, low institutional ownership. Returns business model, growth drivers, execution risks, and reasons it's overlooked.
Technical deep-dive on the math that extracted $39.7M from Polymarket in a year (arXiv:2508.03474): marginal polytope problem, Bregman projection for calculating optimal trades, Frank-Wolfe algorithm reducing projection to iterative linear programs, Central Limit Order Book execution realities, and why naive YES+NO=$1 arbitrage fails when markets have logical dependencies. Includes the documented top-15 wallet PnL ($51M combined).
646-star GitHub dataset containing every trade ever made on Polymarket — 86M rows covering every wallet, entry, exit, and timestamp. The starting point for any serious Polymarket analysis: clone, point Claude at processed/trades.csv, ask for wallets with 100+ trades and 70%+ win rate, get a ranked target list.
Step-by-step $25/month Polymarket bot build — poly_data for the target list, polymarket-cli for scanning (three-factor scoring: price gap, order book depth, hours to resolution), Polymarket/agents for the brain (base rate / news / whale / disposition agents with 3-of-4 consensus), Polymarket-Trading-Bot strategies in consensus (2 agents → full position, 1 → half, disagree → skip), three exit triggers (target-hit at 85% of gap, volume spike 3x, 24h stale thesis). Reported 214 trades, 74% win rate, Sharpe 2.31, +$11,400 in 19 days.
"Stop copying trades — copy exits." Feed Claude full market history plus live Polymarket data and ask one question: who gets paid? Top wallets exit at 86% EV capture and cut losers at -12%; average traders capture 58% and ride losses to -40%. Author reports 191 trades, 73% win rate, -2.1% max drawdown over 19 days.
Opinionated architecture for building a Claude-driven trading bot that survives deployment day: Layer 1 market data (identical shapes for backtest + live), Layer 2 strategy (opinions not orders), Layer 3 risk gate (yes / yes-but-smaller / no-correlation / no-already-down), Layer 4 execution (unique client_order_id, WebSocket idempotency), Layer 5 ledger (log everything, broker position = truth). Build ledger and risk gate FIRST, then earn the right to add strategies.
Polymarket's open-source agent framework — 8 detectors on sports contracts via one prompt, 3,000+ GitHub stars. Reported track record: 62 contracts / 49 winners / +$6,217 net in 14 days on NBA playoffs; lifetime +$31,447 from $1,600 seed in 9 weeks. The canonical starting point for Claude-driven Polymarket work.
Claude Code + ChatGPT Images 2.0 pipeline that produces 40 production-ready static ads from one brand name and one URL. Claude researches the brand's fonts / colors / packaging / photography style, builds a Brand DNA document, fills 40 proven ad templates (headline / us-vs-them / testimonial / UGC / review cards / stat callouts), fires each prompt with product photos as reference, and downloads the finished ads into organized folders with an HTML gallery. Eliminates the "find winning ad → manually recreate" loop for DTC brands and agencies.
Claude surfaces upcoming earnings reports that may create high-volatility trading opportunities, with event-window framing and IV considerations baked in.
Claude as a macro hedge-fund trader — analyzes XAUUSD with USD strength, real yields, geopolitics, and technical structure. Returns directional bias and 3 key levels for the next 48 hours.
Claude defines a 30-minute daily trading workflow: scan, plan, review. Opinionated structure so you're not reinventing the process every morning.
Polymarket's official Rust CLI — read-only scanning with no API key required. Three commands are all you need: markets list --limit 500 for active markets, clob book $TOKEN_ID for order book depth, clob midpoint $TOKEN_ID for the price. Built for agents; Claude pipes the JSON straight in.
dylanpersonguy's 53,000-line TypeScript Polymarket bot shipping 7 strategies — use it as a module source rather than a monolith: pull the arbitrage detector for related-market price gaps, the convergence engine for entries when price moves toward estimate, the whale_copy module for 60s-delay mirroring of a target wallet list. Shared-wallet, no-shared-memory agent pattern.
Claude produces 5 trade ideas with entry, stop-loss, profit target, and risk/reward ratio — the skeleton of a day-trader's morning plan.
Claude generates a 10-stock watchlist ranked by momentum, volume, catalysts, and short-term earning potential — fast shortlist before you deploy capital.
Claude as a portfolio-construction expert — for a $50K portfolio across a chosen risk preference (conservative/moderate/aggressive) and time horizon, returns diversified allocation by sector, rationale for sizing, expected returns, rebalancing rules, and downside protection.
Migration playbook for plugging Claude into a manual 5–15m trading strategy: feed Claude the full trade history, build a model around risk/R:R/expectancy, connect APIs for live price+volatility+structure, automate execution, run everything through n8n 24/7 with Telegram alerts. The point isn't a new edge — it's removing yourself as the weakest link.
Claude as a Wall Street strategist — identifies 5 stock sectors where money is flowing right now and explains why. First pass before any individual ticker work.
Independent-skills architecture for a Polymarket trading engine — arbitrage detection, whale tracking, sentiment shifts, execution timing — layered so most trades get filtered at Level 1 (liquidity/spreads) and only high-signal opportunities pass to Level 2 (EV, probability shifts). Author reports $5,162 in the first night across 370 trades at a 73% win rate.
Claude performs full multi-timeframe analysis — Daily, 4H, 1H, 15M — returns dominant bias across timeframes and the highest-probability setup right now. Stops you from entering a 15m trade that's fighting the daily.
Claude as a senior institutional trader — analyzes [ASSET] trend on [TIMEFRAME] using price action, volume, key indicators. Returns bias, key levels, continuation vs reversal probability.