Explore Claude connectors.

AccessEXL Agent Capabilities lets Claude use the calculations your organisation has published from its Excel models, through one connection, across every approved workbook. Each capability is a model your team has described and approved, for example "the project finance model that calculates cash flows, NPV and IRR" or "the quote estimator for residential construction". Claude can: - search the capabilities shared with you and pick the one that fits the question - read its inputs and outputs, with descriptions, limits and allowed values - run it with your inputs and return the results, calculated in Excel by the real model Claude never sees the workbooks themselves, only each capability's approved inputs and outputs. Running a capability never changes the workbook: its inputs are restored after every run. If a workbook's configuration changes after approval, the capability is held for review until its owner re-approves it, and results can carry a note that a person should check them before they are relied on. You need an AccessEXL Business account with at least one published capability. Connecting is a single sign-in with your AccessEXL account; no keys to copy.

AccessEXL connects Claude to one Excel workbook stored in your SharePoint, through a controlled API that the workbook owner configures. The owner chooses which named input cells, output cells and tables are exposed; Claude never sees the rest of the file or the underlying formulas. With this connector Claude can: read the workbook's configured inputs and outputs, with their display names, descriptions, types, limits and allowed values run a scenario: write a set of inputs, recalculate the workbook in Excel and return the resulting outputs in one step update inputs, or read the current outputs Because the answers come from your own Excel model, the numbers follow your business rules exactly: pricing models, financial forecasts, quoting tools, budgets and other calculators built by your team. Inputs are validated against the limits set in the workbook, and every call is recorded (metadata only) in the organisation's activity log. You need an AccessEXL account and a workbook connected in AccessEXL. When you connect, AccessEXL asks for that workbook's API key; one connector is one workbook. For access to all approved models through a single connection, see the AccessEXL Agent Capabilities connector.

Accessibility Tracker's MCP server gives AI agents controlled access to your accessibility workspace: projects, issue backlogs, automated scans, reports, and VPAT® records. Understand the workspace — list projects and issues, inspect scans and pages, read reports and VPAT records, search documentation, and look up WCAG criteria or account limits. Plan with evidence — use guided prompts to triage a scan, prepare a remediation plan, summarize regressions, or draft VPAT remarks for human review. Act when permitted — with scoped, opt-in capabilities, an agent can create or update issues, add comments, import scan findings, and manage scan jobs. Read access is always included; write and scan-job capabilities are explicit choices you make when generating a key, and permanent deletion is never exposed. Authentication is via scoped API keys generated in your account settings, so an agent only ever gets the access you deliberately grant it.

AccessLint monitors and scans websites for accessibility problems. Most accessibility problems live behind an interaction. AccessLint saves user journeys that cross multiple pages and interactions, that a real browser replays and checks on a schedule. AccessLint can also scan a public page ad hoc against WCAG. In either case, it returns rich findings that you can dig into with Claude and file issues or spin up remediations. What you can do: - Scan a public page against WCAG and get the violations back. - Turn plain sentences about user journeys, for example "search for a product and add it to the basket", into flows, with the steps compiled and auto-healed from the sentence. - Look at the screenshot of the page as the browser saw it, at any step. - Review the history and open accessibility violations a flow is currently finding Requires a free AccessLint account. Single page scans work out of the box; you must verify ownership of a domain to run flows against it.

AccountAI is an AI-first accounting and ERP platform for small and medium businesses, with compliance frameworks for the UAE (FTA), Bahrain, Qatar, Oman and OHADA, plus a standard framework for any other country. This connector lets you work in your AccountAI books directly from Claude. Ask questions and get answers from your real, posted numbers: - Trial balance, profit and loss, balance sheet and cash position - VAT/tax return summaries in each framework's own format - Aged receivables and payables, customer and supplier statements - Stock by warehouse, inventory valuation, fixed assets and depreciation, loans, payroll runs Get work done (with a "Read and write" connection): - Create customers, suppliers, invoices, credit notes and bills; post them; record and void payments - Manual journals, batch invoicing, stock adjustments and transfers, PO to bill - Asset acquisition, disposal and depreciation runs, loan repayments, period locks Built for safety: - Claude always acts as the signed-in user, inside their own companies, role and permissions. A view-only accountant gets a view-only Claude. - You choose "Read only" or "Read and write" when you connect. - There are no delete tools. Posted records are corrected only by reversal, and every write uses an idempotency key, so retries never duplicate. - Every call is logged in the company's activity log, and you can disconnect at any time from Settings → Connect to Claude. - Company deletion, users and roles, billing, framework and sign-in settings stay human-only in the app. Works for multi-company groups: Claude names the company on every call, and each company keeps its own currency, tax rates and chart of accounts.

Ace Website MCP lets Claude users explore the public Ace catalog, read product and support pages, and begin a secure Stripe-hosted checkout. It provides six public tools: site information, sitemap search, page reading, live product and price listing, checkout-link creation, and checkout status lookup. Checkout creation never charges a buyer; it returns a Stripe-hosted link where the buyer reviews and completes payment.

Most companies have no idea how their own work gets done. Processes live in people's heads, SOPs go stale the day they're written, and nobody logs the workarounds that keep things running. Ace Work fixes that. Connect it to Claude and your workflow documentation becomes something you talk to instead of a wiki nobody reads. What you can do: - Capture new workflows in conversation. Claude interviews you about how a process really works and files it into your workspace, no forms, no blank templates. - Run any documented SOP step by step, with Claude walking you through the checklist and pulling in your other connected tools to do the actual work. - Log reality as it happens: how long a run really took, where you deviated from the documented process, and the friction points that slow you down. Everything lands as pending suggestions for human review, never silent edits. - Search and read your company's workflow library, scoped by company, division, and department. Every run makes your documentation more honest. That's the point: SOPs that reflect how work actually happens, not how someone hoped it would. Built by Ace Workflow. We fix work. https://aceworkflow.io

Encontre reuniões, leia transcrições e peça cortes, legendas, títulos e zooms nos vídeos do Acelere Screen. O aplicativo precisa estar aberto no computador autorizado. A conexão é opcional e individual; os trechos consultados passam pelo Claude. Os arquivos de áudio e vídeo e o histórico de ditados do VoiceText permanecem no computador. As edições aparecem no editor e podem ser desfeitas.

O Acervo Sonarview reúne relatórios, pesquisas, estudos de caso e regulatórios curados para planejamento estratégico, organizados por setor e por método. Com o conector, o Claude: - busca no acervo por assunto, com até seis perguntas de uma vez, e diz com franqueza o que o acervo cobre, o que cobre em parte e o que não cobre; - cita cada documento com o link para abri-lo na Vitrine da Sonarview; - sugere uma busca externa quando o acervo não tem a resposta; - lê o documento escolhido, por páginas, para você aprofundar. O acesso é só ao acervo público: o conector não vê planos, documentos privados nem dados de nenhuma conta, e não escreve nada. Para usar, basta um cadastro gratuito na biblioteca da Sonarview, por Google, LinkedIn ou e-mail, feito na primeira conexão. Há limites de uso por pessoa (por minuto, por dia e por mês), e as mensagens dizem quando eles voltam.

Açık Silivri Havadis gives Claude read-only access to sourced journalistic reporting on the Istanbul Metropolitan Municipality trial, including hearings, roster, transcript search, accusations, defenses, evidence, statements, expert opinions, and cross-examinations.

aclipp is a European PR analytics platform that helps communication teams and PR agencies show what their work achieves, in traditional media and in AI answers. The aclipp connector gives Claude access to your aclipp workspace, so you can analyze coverage, track AI visibility and build reports right in the conversation. What you can do with aclipp in Claude: - Analyze your media coverage by period, outlet, channel, country, tag or sentiment, including reach and media value - See how visible your brand is in AI answers across providers, topics and prompts, with share of voice, average position, sentiment and citation rate - Find out which domains and URLs AI answers cite as sources - Compare your brand or your clients with competitors - Add articles to a project by URL and update clippings, outlets and authors - Manage AI visibility prompts, topics, competitors and tags - Create and update reports and dashboards from your data Built for in-house communication teams and PR agencies, including agencies that manage several clients in one workspace. Requirements: an aclipp account. You sign in with your aclipp login via OAuth and choose which permissions Claude gets. Read access is the default, write access is optional and granted per scope. Claude only sees the projects you can access in aclipp. aclipp is developed in Germany and GDPR compliant. Setup guide and documentation: https://aclipp.com/docs/mcp

Acornonaut turns YouTube videos and playlists into spaced-repetition flashcards. Add a video by URL and have cards generated from its transcript, review the cards that are due today and record how well you knew each one, search, create and edit cards and playlists, check your study stats, and export a deck as JSON, CSV, Markdown or an Anki package.

AcornReply is a support inbox for small teams who answer their customers themselves: AI drafts a reply, a person edits it and hits send. This connector opens the knowledge base behind those replies: the same FAQs and help-center articles your team and your AI drafts both read from. Connect it and Claude can search the knowledge base semantically, read any entry in full, and create or update FAQs and articles, including publishing and unpublishing them. It manages the category tree the help center is organized by, reorders it, and uploads the images article bodies use. A formatting-guide tool returns the exact content contract entries must follow, so what Claude writes renders correctly in the help center instead of needing cleanup afterwards. It also reads per-article view statistics, which is what makes pruning practical: ask for the articles nobody has opened since a given date and Claude can find the dead content, rewrite what is worth keeping, and unpublish the rest in one conversation. You connect by signing in to AcornReply; there's no API key to create or paste. You choose the workspace Claude works in when you connect, and you can ask Claude to switch to another workspace your account can access later. Claude only ever reaches the knowledge base, never your inbox, customers, or billing, and only with your own account's access. All 20 tools are annotated read-only or write, so Claude asks before anything that changes content, and deleting an entry or category takes an explicit confirmation step. AcornReply is free to start, and the knowledge base and this connector are available on every plan including the free one.

The Acoustic MCP connects Claude directly to our platform, giving it the ability to run analytics and build marketing campaigns using your live customer data. It's built around four data pillars: Contacts (profiles, identities, preferences), Products (catalog, pricing, inventory), Behaviors (events, engagement, purchase history), and Campaigns (sends, segments, performance). With it, Claude can query across these pillars to answer analytics questions in plain language, and it can act: building audience segments, drafting messages, and assembling cross-channel campaigns without you touching the Connect UI directly.

Acquirepad helps professional commercial real estate investors turn market activity into deal intelligence. This connector lets Claude search recent deals, leases, newsletters, companies, locations, map markers, and sanitized enrichment details. It is a free read-only entry point into Acquirepad’s broader AI platform for buy-side investment teams: deal inflow, document extraction, data enrichment, acquisition profile matching, CRM context, and transaction workflows connected into one agentic intelligence layer.

Act Security is the action-first cloud security platform built to secure the cloud at its foundation. Rather than just surfacing findings, Act eliminates the conditions that make risk exploitable in the first place. By analyzing both network and identity layers in a single access graph, Act enforces precise boundaries for humans, workloads, and AI agents across multi-cloud infrastructure. Grounded in real business context and enforced through cloud-native controls, Act helps organizations eliminate attack paths, safely scale AI, and maintain continuous, audit-ready compliance.

Acting Resources is a free education site for actors. This connector lets an AI assistant look up answers from the site's own pages. It has three tools. No account or login is needed. Rules Q&A: answers questions about acting career rules (agents, unions, contracts, child-performer law, and which casting platforms a market uses). The answer is taken from the site's market guides, union lessons and scams lesson, and it ends with a link to the source page. Headshot, self-tape and profile checker: you paste a written description of a headshot brief, self-tape setup, bio or casting profile. It returns specific fixes from the site's checklists. It reads text only. It does not look at photos or video. Free course finder: given a career stage and an optional goal, it returns the two or three closest free courses on the site, with links. Every tool is read-only. The connector does not write data, send messages, make purchases or call other services. Answers are short excerpts and matches from actingresources.com, not legal, financial or career advice. When a question is not covered by the site, the connector says so instead of guessing.

The Actionstep MCP Server connects Claude directly to your firm's Actionstep environment. Fee earners, paralegals, and partners can handle day-to-day matter work from Claude instead of switching into the Actionstep UI - while Actionstep stays the system of record. Every write action Claude takes is logged and auditable exactly as if it were done in-app. What you can do • Document filing: Upload and download files on matter folders, download from contact folders, list what's already on file, and create matter folders as needed — all from a Claude conversation. • File notes: Turn an email thread or call summary into a file note on the right matter, and read back what's already been logged. • Task management: Create, update, and check the status of tasks tied to a matter, right from where you're working. • Intake and custom fields: Populate matter- and participant-level custom fields — including dropdown values — from an intake form, scanned ID, or email, individually or in bulk. • Contact management: Look up, create, and update contact records, and resolve contact types and roles. • Matter and staff lookup: Identify the current user, find a colleague to assign work to, and search for matters by ID and look up matter participants.

ActiveLayer stops spam on website forms, comments, registrations and reviews without CAPTCHAs. Connect your ActiveLayer account to review recent spam checks with their verdicts, scores and submitted fields, see your plan allowance and usage, and correct a false positive or false negative. Test a sample message against the detector, or create a project for a WordPress site, a Lovable app or a custom application and follow its setup instructions. What the connector can read: your projects and their API key status, your plan usage, and your detection records, which can include submitted message text, sender names and email addresses, website URLs and IP addresses. What it can change: it can create projects and submit or replace feedback on a detection. It cannot delete anything, change billing, or return API keys. Keys are generated in your ActiveLayer dashboard and copied directly into your site configuration, never into chat. Demo checks send the supplied fields to ActiveLayer's detection service under a separate first-party demo account. They are limited to 30 per account per hour and do not use your plan allowance or appear in your detection history. A demo result does not verify that your website integration is installed or working. Requires an ActiveLayer account with the short account setup completed. A free plan is enough.

Win more deals with Actively inside Claude by directly accessing your always-on per account agents that help you drive the next best action. Actively AI's per-account agents are synthesizing across all of your internal context (ex. CRM data, call transcripts, emails) and external signals to drive actionable intelligence. Designed for SDRs, AEs, AMs, and revenue leaders who need deep, contextual account knowledge, from meeting prep and deal strategy to territory prioritization, directly inside Claude.

Read and analyse data in ActivityInfo databases. Lists databases, forms and fields, retrieves records, create reports, re-code data.

Actually Asked is a FAQ for websites and Shopify stores that grows from what visitors ask. Visitors ask their own question under the FAQ, or highlight a sentence and ask about it, and answers come only from the site's own pages, products, policies and files. This connector lets Claude manage that FAQ in your Actually Asked account: - See what visitors asked, what your pages couldn't answer, and the visitors waiting for a reply. - Review suggested FAQ entries, check each answer against your pages, fix the wording and publish them. - Add, edit, pin, group and remove FAQ entries, and choose which pages each group shows on. - Search everything Actually Asked has read on your site, so answers are written from your pages rather than general knowledge. - Change what the FAQ section says and how it looks, get the install snippet for your platform, and check that it's live. It keeps the dashboard's rules: it can do only what your plan allows, entries you wrote are never rewritten by the system, removed questions stay removed, and it asks before deleting a group or page rule or discarding a question. It never sends email: you reply to visitors from your own inbox.

Ask what changed in Meta (Facebook and Instagram) or Google Ads this week, whether a feature is live, or why costs moved, and get dated summaries of platform announcements, trade press and practitioner reports, each with its source link and how well it's verified. Refreshed daily from Sum Digital's ad platform research. No account or login, and nothing you type is stored.

Ad Radar is a research board of Meta ads (Facebook and Instagram) from DTC and e-commerce brands. It keeps only ads that are already winning: ads with at least $50k in estimated spend, modeled on engagement from the public Meta Ad Library. Spend is always an estimate, useful for ranking ads against each other, not a competitor's real budget. Every ad is tagged on ten dimensions: hook pattern, hook mechanism, narrative angle, reader awareness, proof level, production format, video type, landing page type, niche and CTA type. Headlines, body copy and video transcripts are searchable. With the connector, Claude can: - Search winning ads by niche, tag, format, days live or a line from the script (search_ads) - Read one ad in full: copy, transcript, tags, days live, variants and links to the original (get_ad) - See which hooks, angles, formats and proof levels reach the top tier more often than the rest of the board, with the number of ads behind each (winning_patterns) - Read the hook and angle mix one brand is running (advertiser_hooks) - Get a brief that maps a winning ad's structure onto the product you describe (get_swipe_prompt) - List the valid tag values, so filters use the exact words (list_taxonomy) Answers come back with the brand, the opening line, the estimated spend and the days live, so you can open the ad and check it. Built for e-commerce owners, media buyers, creative strategists and agencies who ship Meta ads every week. Limits: Meta only (Facebook and Instagram). Ads under $50k in estimated spend are not on the board. All six tools are read-only: they read the Ad Radar board of competitor ads and never connect to your Meta ad account, so nothing in your campaigns can be seen or changed. Requires an active Ad Radar plan (Membership or Agency). You sign in to Ad Radar and approve the connection when you connect: there is no token to copy.