Tony Lewis

TL;DR​

• The Directory API reads the same index as the website. Products, scripts, assistants, and crawlers all hit it, and reading it stays free.
• Free reads stay open: 30 requests a minute from an address, no account. Directory API Pro is $299 a month and raises that to 300 a minute on the workspace, using the workspace API key you already create under Settings → Security.
• The field reference stays on the Directory API guide. This note is who the cap is for. A paced crawl still fits in the free minute. A product that searches on every keystroke does not.

Picture a catalog product where every keystroke fires a search. The first thirty calls in a minute come back fine. The thirty-first comes back 429, with a pointer at the pricing page. That is the moment Pro exists for. A nightly job that waits a couple of seconds between calls never sees it, and we are not asking it to.

Most of the tools you run day to day have some way for an AI to reach them. There's an API, or someone has published an MCP server, or the vendor has a connector you can click. But a lot of the tools that actually decide how your week goes don't have any of that. A booking portal, an ads console, a commissions report, a supplier's price file. The product never shipped an API, nobody built an MCP server for it, and the only thing the vendor gives you is a button that says Export.

That exported file is the data you care about, and it's also the one thing your AI can't see. So the usual routine goes like this: you download the file, park it in a folder, rebuild the same pivot table you rebuilt last month, and then either paste a chunk of it into the chat or describe what you found. CSV Analyzer is our answer to that gap. It's a single MCP server where you upload the files you were going to download anyway, and then you ask questions about them from Claude, Cursor, or whatever client you already use.

This is one of the most requested things we've shipped in a while: attach more than one account for the same service to a single MCP server. It's live. Same URL in Cursor or Claude. Every account you attach. Your AI routes each call to the right one.

If you've ever maintained three QuickBooks companies, two Shopify stores, or a work inbox and a personal one, you know the old pattern: duplicate the MCP server, duplicate the URL, duplicate the mental overhead. That ends here.

TL;DR​

• The Bing Webmaster Tools MCP server on MCPBundles covers keyword research, site add/verify, sitemap feeds, IndexNow URL pings, URL submission, and property traffic reads — the workflows we used to onboard mcpbundles.com on Bing.
• During build, Bing’s keyword API reported on the order of ~700 US impressions in 28 days for “model context protocol” and a cluster of related “mcp server” queries — enough signal to prioritize guides and skill pages on Microsoft search, not just Google.
• IndexNow is the fast path Bing documents for changed URLs; the same bundle exposes it next to classic Webmaster sitemap and submit-url tools so agents can notify after a deploy without a separate script.

We already run organic growth through Google Search Console in chat. Bing matters for Copilot, Windows search, and the slice of users who never touch Google’s console. Rather than bolting on a one-off script, we built a Bing Webmaster Tools MCP server and used it on our own site first: verify the property, register the sitemap index, ping launch URLs with IndexNow, and pull keyword ideas before we commit to the next /guides page.

TL;DR​

• The FreeCAD MCP server lets agents create documents, add solids, edit dimensions, and export STL or STEP on your computer — FreeCAD stays open locally; geometry isn't modeled in the cloud.
• FreeCAD is open-source parametric CAD used for product design, maker projects, and mesh export — a practical fit when you want editable solids instead of one-shot mesh prompts.
• Built for makers and designers who already slice STLs or send STEP to a shop, and who'd rather describe a bracket or enclosure in chat than click through every sketch constraint alone.

Saturday morning at the bench. You need a simple enclosure base — 80×60×12 mm with four corner posts — before the printer finishes its current job. FreeCAD is already open on your Mac. You're in chat, not in the Part Design workbench, asking for the base plus an STL on your Desktop.

That's the loop: describe → model locally → export → slice. The FreeCAD integration on MCPBundles is for that loop when Desktop is running on the same machine.

TL;DR​

• Local Browser and Remote Browser now expose browser_list_webmcp_tools and browser_call_webmcp_tool — your agent navigates to a page, discovers tools registered on document.modelContext, and calls them with JSON input.
• MCPBundles dogfoods WebMCP on www.mcpbundles.com — the skills catalog registers search_skills, get_skill, and get_connect_url so a browser session can query the marketplace from inside the page, not by guessing DOM selectors.
• Works in cloud Chrome (Remote Browser) and Desktop Chrome (Local Browser, WebMCP enabled by default) — same navigate → list → call workflow, no separate bridge extension.

Modern sites are not just HTML anymore. Some already publish structured tools the page understands — search the catalog, fetch an entity, return a connect link. WebMCP is how those tools show up to in-browser agents. Until now, MCPBundles browser automation was excellent at clicks, snapshots, and screenshots — but blind to page-native APIs. That gap is closed.

TL;DR​

• The Webhook Ingest MCP server gives each connection a hosted URL — paste it into anything that can POST content (schedulers, form tools, monitoring systems, export jobs, custom scripts) and ask in chat for the latest delivery, recent history, or a specific stored payload.
• Each connection keeps a rolling history of what arrived — CSV, JSON, raw bodies, file uploads, and common document types — so agents answer from stored data, not from memory.
• Built for teams who already have push data somewhere but still copy-paste files into chat, or maintain one-off receivers nobody wants to operate.

Brightline Ops wires a dozen small integrations. Nightly exports, form submissions, alert dumps — all of them know how to POST to an HTTPS URL. What they did not have was a place those payloads could sit until someone asked an agent a question about them.

Chat only helps when the agent reads what already landed, not when someone rebuilds a listener for every new sender.

TL;DR​

• ChatGPT Ads are sponsored chat_card units below ChatGPT answers, labeled "Sponsored" and visually separate from the model's reply. They reach Free and Go users in supported regions; Plus, Pro, Business, Enterprise, and Education do not see ads (OpenAI Help Center). OpenAI expanded self-serve beyond the US test to the UK, Mexico, Brazil, Japan, and South Korea on August 11, 2026 (Testing ads in ChatGPT).
• Early public data (third-party, US-heavy, July 2026): SE Ranking sampled 50,006 commercial prompts and saw ads on ~26%; their own spend test reported ~97k impressions, ~1,263 clicks, and ~1.3% CTR with weak downstream sign-ups. Sensor Tower estimates put active advertisers at ~300 in April → 820+ in July. Treat as directional — not industry benchmarks.
• Targeting is conversational, not keyword-based: you write context hints on ad groups describing the chats where your offer fits. There is no keyword volume or pre-flight CPC forecast; you plan with budgets and post-delivery insights instead.
• Account setup runs through Ads Manager Beta — business onboarding, Persona verification, billing, manual review, then Settings → API Keys for MCP. Campaigns won't deliver until account info (name + logo) and review finish (Account Setup help).
• The OpenAI Ads MCP server on MCPBundles wraps the API for Claude, Cursor, and other agents: account overview, context-hint and copy audits, paused campaigns, geo lookup, creative upload, conversion setup, and delivery reads from chat.

You're evaluating a channel that barely existed in paid media a year ago. ChatGPT Ads started as a US test in early 2026 and is still rolling out market by market.

Northbridge Legal runs Google Search for intake keywords and wants a second line on ChatGPT without another console tab for every standup. Their marketer's first question isn't "where's the button?" — it's whether yesterday's test bought anything worth keeping, and whether the account is still stuck in review.

Worth knowing what you're buying before you touch budget. Also worth knowing what you're not buying (keyword volume, chat transcripts, pre-flight CPC quotes).

TL;DR​

• The Xero integration on MCPBundles covers the work operators actually repeat — receivables, month-end reports, bank reconciliation, payroll reads, supplier updates, document attachments, and cached financial reports — scoped to one connected organisation.
• Xero's own figures put the platform at 4.2 million+ subscribers worldwide; most teams still live inside invoice lists, report menus, and bank feeds when the question is simply who owes us money this week?
• Connect your Xero organisation once, then ask in plain language: "List overdue sales invoices by customer and flag the three largest balances."

Thursday at Harbor Ranch Events. Weddings booked through autumn, three venue deposits still sitting in "sent" status, and the bookkeeper is exporting aged receivables because the owner wants a collections call list before Friday's site walk. That's four screens — invoices, contacts, bank rec, and the P&L shortcut someone bookmarked six months ago.

Chat only helps when the opening line sounds like that Thursday, not like a feature tour.

TL;DR​

• MCP Connect Auth gives every published /skills listing a tenant authorization server — DCR, PKCE, protected-resource metadata, and federation to your existing sign-in.
• Path B (bundle URL): publish, configure federation, point Claude/Cursor/ChatGPT at https://mcp.mcpbundles.com/bundle/{slug}/ — no pip install on your MCP server.
• Path A (vendor origin): add mcpbundles-mcp-connect when clients hit your own MCP URL.
• Full reference: MCP Connect Auth integration doc.

If you've tried to publish a remote MCP server and stalled on OAuth, you're not alone. Certified MCP clients expect dynamic client registration, PKCE, protected-resource metadata, and a coherent authorization server. Rolling that stack while also shipping tools is a common reason listings sit at /skills with zero connections.

TL;DR​

• The UKHO Tidal MCP server covers 607 ADMIRALTY tidal level stations around the United Kingdom — high water and low water predictions for today plus six days ahead.
• Ask for tides by place name or station and get quoted times and heights from official data — not tide-table guesswork from a language model.
• Built for marina operators, sailing clubs, harbour teams, coastal walkers, and location scouts who need reliable schedules with Crown copyright attribution preserved.

Coastal planning questions are easy to ask and easy to answer badly.

"What time is high water at London Bridge on Saturday?" "Can we enter the harbour after lunch?" "Which tidal station covers Falmouth?"

A language model alone will invent tide tables or blur GMT with British Summer Time. The UKHO Tidal MCP server resolves the station first, pulls official ADMIRALTY predictions, and returns high water and low water events with heights — with the required Crown copyright line intact.

TL;DR​

• You can attach an A2A agent to any MCP server in your workspace—the same catalog surface you use from Claude or Cursor—not only from the Hub.
• Agent instructions are enough to run manually or from chat; a heartbeat checklist is optional and mainly for recurring scheduled runs.
• On a bundle MCP connection, agent tools use the same names as the Hub (get_agents, upsert_agent, trigger_agent_run, and the rest).

It's Tuesday standup and someone asks whether yesterday's Google Ads spend spiked for a client you do not have open in another tab. You already have the MCP server in Cursor. What you want is a named assistant on that server that you can wake on demand or on a clock.

That is the workflow we optimized this week.

TL;DR​

• The Chess.com MCP server reads public profiles, stats, monthly games (with full move lists), clubs, and the daily puzzle after you add your member name on MCPBundles — not a password, just the handle from your profile URL.
• Chess.com reported 250 million registered members (member milestone, 2026) and ~9.7 million average daily active users in Q1 2026 (Q1 board report); 2.6 billion games were played that quarter alone — most review prep still means copying PGN by hand.
• Built for coaches, club captains, and streamers: ask about your bullet form without repeating your handle every time, then name a student or guest when the lesson shifts to someone else on the same connection.

It's 6 p.m. on a Tuesday and your first online lesson starts in twenty minutes. You know the student blitzed last night, you think they lost two rapid games in a row, and you'd like one clean game on the board with every move in order — not a scavenger hunt through archive tabs and export dialogs.

That's the loop we built for: public Chess.com data is already on the web, but assembling it fast enough to teach is still manual unless your agent can reach it in one place.

When you connect Chess.com on MCPBundles, you save your public member name once. Player lookups default to that handle when you don't name someone else. Coaches still pass another member name when they're reviewing a student, a rival, or a stream guest — same server, no second setup.

TL;DR​

• Spotify Ads on MCPBundles now resolves ad accounts by brand name, remembers your pick for the thread, and exposes read paths for pulse, approvals, and exports without a separate “find the UUID” step.
• You can name campaigns, ad sets, and ads in chat, pull last week’s metrics for one campaign, and draft a paused ad set (UK geo, daily budget, delivery off) before anything goes live.
• Planning defaults stay conservative: paused campaigns, delivery off on new ad sets and ads until you mean to launch.
• I wrote this for podcast and brand teams who already buy on Spotify and hate opening Ads Manager for a five-minute standup answer.

Three Spotify brands, one login, standup in ten minutes. “Anything stuck in approval?” shouldn’t cost you four clicks and a prayer you didn’t pause the wrong row.

TL;DR​

• The Clockify MCP server connects your agent to clients, projects, tasks, tags, and time entries in your Clockify workspace after you connect on MCPBundles.
• CAKE.com — Clockify's parent company — calls it the most popular time tracking tool in the world; agencies still lose Friday afternoons to hours recaps nobody wants to rebuild in a spreadsheet.
• Built for project managers, freelancers, and ops leads who need weekly billable splits, project burn checks, and quick entry logging without opening the timer UI for every correction.

It's 4:30 on Friday. Client invoicing goes out Monday. Three projects blew their hours budget, one freelancer forgot to log yesterday's review session, and the ops lead needs billable versus non-billable split by client — not another export you'll pivot in Excel at 6 p.m.

Clockify already has the entries. The friction is assembling them fast enough to act. The Clockify integration on MCPBundles is for that assembly.