Legal · Privacy policy

Privacy policy

Last updated: 20 November 2026.

MeshArc is operated by Discover Webtech Pvt Ltd, India ("MeshArc", "we"). This policy says what we collect when you use the web app at https://mesharc.dev, the API at https://api.mesharc.dev, the mesharc SDKs and the MCP server, why we collect it, how long we keep it, and what you can ask of us. Questions go to [email protected].

There are two kinds of data here, and they are treated differently. Your account and workspace — who you are, what you configure, what you are billed — is data we hold about you. The pages you crawl are yours: you decide which sites to read, and we read them on your behalf. The first is covered in full below; the second has its own section, Pages you crawl.

What we collect#

Account. A work email address, a display name and — unless you only ever sign in with Google or GitHub — a password. The password is stored only as a salted PBKDF2-SHA256 hash; we cannot read it. Your profile also holds a time zone and your email preferences: whether you receive plan-limit emails, run emails and the weekly digest, and whether you want product news.

Signing in with Google or GitHub. If you choose to sign in or sign up with Google or GitHub, that provider tells us who you are and nothing more. From Google: your Google account's ID, your email address, whether Google has verified it, and your name. From GitHub: your GitHub account's ID, your username and name, and your primary email address with whether GitHub has verified it. We keep the provider's account ID and that email address with your MeshArc account, so your next sign-in finds it. We ask for no other permission — not your contacts, files, mail, calendar or repositories — and the access the provider grants is used once, to read those details, and is never stored. An account created this way has no password until you set one.

Workspace. Its name, its members and their roles, invitations (an email address and a role, valid for seven days), API keys (stored hashed, shown to you once), webhook secrets, and the connections you configure to your own databases, buckets and vector stores — including the credentials those need, which are encrypted at rest and never returned by the API.

Sessions. When you sign in we record the time, the IP address and the user agent of the browser, so Profile → Sessions can show you where you are signed in and let you end any of them. A session lasts 30 days and ends after 14 idle days.

Billing. Your plan, credit balance and usage. Card, UPI and bank details never reach us: payments are taken by Razorpay, and we keep only the Razorpay customer and subscription ids, the subscription state, and the invoices we issue.

Usage and logs. Every API request is logged with its time, path, status, the workspace and key it was made with, the credits it cost, and a request id. Logs are kept for 180 days, as Indian law asks of service providers, for operations, security and abuse prevention.

Web searches. When you search the web through the app or the API, the query, its settings (country, language, freshness, sites to include or leave out) and the results are kept with your workspace for seven days (ninety for a search a monitor ran). That is so you can open past searches, and so an equal search within the hour is answered from the last one at no charge. Searches are visible to your workspace's members and API keys, and never to another workspace.

Support mail. What you write to us at [email protected], kept as long as the conversation is useful.

Newsletter. If you sign up for the newsletter on the site: your email address, where on the site you signed up, the sentence you agreed to and when, and whether and when you confirmed or left. You need no account for it. Nothing is sent until you click the link in the confirmation email we send, and every newsletter carries a link to leave, which works at once. With an account you can also turn product news on or off under Profile → Email: the newsletter then goes to your account's verified address, and turning the switch on is your consent.

Where you came from. The first time you visit the site, a cookie can note where that visit came from: the campaign tags in the link (utm_source and the like), the address of the site that sent you — its domain only — and the page you landed on, without anything after the ?. If you then create an account, that note is kept with it, so we can tell which pages and campaigns bring people who stay. Where the law asks for your consent first (see below), the note is only made after you say yes; it is never made when your browser sends Global Privacy Control.

Site analytics. The site loads Google Tag Manager, which runs Google Analytics: pages viewed, how you arrived, the browser and device type, an approximate location derived from your IP address, and a few events we send ourselves — a sign-up, a free-tool run, a purchase — without your name, email or anything you typed. Where consent is needed Analytics waits for a yes; you can change your answer under Cookie settings at the foot of every page.

Errors. When something breaks, an error report goes to Sentry with the stack trace, the release, the browser or server involved and the URL without its query string. We do not send session replays, form contents or your identity with an error.

If you are in the European Economic Area, the United Kingdom or Switzerland — or we cannot tell where you are — a banner asks before any analytics or first-visit cookie is set, and saying no leaves only the cookies the site needs. Elsewhere they are on, and Cookie settings at the foot of every page turns them off. A browser that sends Global Privacy Control is treated as a no. We tell where you are from the country our network provider, Cloudflare, attaches to each request; we do not look up anything more precise.

CookieWhat it is forKept
mesharc_sessionKeeps you signed in. Needed for the app to work30 days, or until you sign out
mesharc_ssoTies a Google or GitHub sign-in to the browser that started it. Needed, and only set when you use one10 minutes
mesharc_sso_backThe page a Google or GitHub sign-in started from, so you return to it if it does not go through. Needed, and only set when you use one10 minutes
mesharc_mfaHolds a Google or GitHub sign-in while you enter your two-factor code. Needed, and only set thenUp to 15 minutes
mesharc_regionThe country code of your visit, so the banner knows whether to ask. Needed30 days
mesharc_consentYour answer to the banner. Needed12 months
mesharc_ftWhere your first visit came from, as described above. Only with consent where consent is needed90 days, or until you sign up
_ga, _ga_*Google Analytics: tells one visit from the next. Only with consent where consent is neededUp to 2 years

The app remembers your theme choice in your browser's local storage, which never leaves your browser. There is no advertising pixel, no session recording and no product analytics inside the app. We do not buy data about you from anyone, and we do not sell or share yours.

How we use it#

  • To run the service: sign you in, keep your workspace, schedule and run your crawls, deliver pages to your destinations, meter credits and bill you.
  • To tell you about your own account: verification codes, password resets and invitations; billing notices, including when your workspace's credits or the free plan's stealth trial run low or run out; a notice when pages of a run needed more than your plan allows (at most once a week); run emails and the weekly digest if you have them on; and notices about the service or this policy. Plan-limit emails, run emails and the digest can each be turned off under Profile → Email or from the link at the foot of any of them. Product news reaches you only if you turn it on. We send no other marketing mail, and we never give your address to anyone to send theirs.
  • To stop mail that cannot be delivered: when our mail provider tells us an address bounced for good, or that its owner marked our mail as spam, we stop sending to it, except the codes and resets you ask for yourself.
  • To understand what works: which pages, articles and campaigns bring people to MeshArc, and which of them go on to use it — as aggregate numbers, not to profile you.
  • To keep the service safe: rate limits, abuse detection and the investigation of incidents.
  • To meet legal obligations, such as tax and accounting records.

What Google or GitHub tells us when you sign in with them is used only to sign you in, to create or find your account, and to stop someone else from claiming your email address. It is never used for advertising, never sold, and never shared. MeshArc's use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.

We make no decision about you by automated means alone, and we do not profile you.

Under the GDPR and UK GDPR the legal bases are performance of our contract with you (running the service, billing), our legitimate interest in keeping it secure and working (logs, error reports, abuse prevention), and legal obligation (financial records). Plan-limit emails rest on our legitimate interest in telling you why work you asked for was not done, and you can turn them off as above. Where consent is the basis — run emails and the digest, product news and the newsletter, and, in the EEA, the UK and Switzerland, analytics and the first-visit note — you can withdraw it at any time: in Profile for email, by the link in any newsletter for the newsletter, under Cookie settings for cookies. Elsewhere, analytics and the first-visit note rest on our legitimate interest in knowing which of our pages are useful, and you can object the same way.

Pages you crawl#

When you scrape, crawl or map a site, MeshArc fetches pages from that site and stores what came back — bodies, headers, extracted fields, screenshots if you asked for them, and a record of what changed since the last run — in your project. Whatever those pages contain, including any personal data on them, is collected on your instructions and for your purposes.

  • You decide. You choose the sites, the scope, the schedule, the formats and where the results go. For that data you are the controller and we are the processor, acting only on your instructions. The data processing addendum sets this out and is part of the terms for every workspace; on the Growth plan and above we will also sign a copy.
  • Retention is yours to set. Each project keeps what its pages said — their HTML, text and markdown, and screenshots — for the retention you choose: 30 days, 90 days, a year, or until you delete the project. Past it, that content is deleted, except in the project's two most recent runs and in any older run whose pages a newer run still reuses. Each page's URL, headers, title and meta tags, links and extracted fields, and the record of what changed, stay with the project until you delete it. A webhook you set to send pages' markdown keeps it in the project's last 200 webhook messages until you delete the project; a crawl's own per-page messages are kept for 7 days. Deleting a project deletes its runs and pages. A one-off scrape, crawl or map that is not kept as a project is deleted after a day.
  • Who can see it. The members of your workspace, according to their roles, and the API keys you issue. Our staff do not read your pages except to investigate an incident you report, an abuse report, or a legal demand, and then as little as the task needs.
  • Where it goes. Destinations you configure — your warehouse, your database, your bucket, your vector index — receive rows because you told us to write there. What happens to the data after it lands is governed by your agreement with that provider.
  • What we do with it. Nothing but store it, diff it, and deliver it to you. We do not train models on your pages, do not read them for our own purposes, and do not combine them across workspaces.

If we crawled your site#

MeshArc reads public web pages on behalf of its customers. If our crawler has visited your site:

  • It obeys robots.txt and rate limits by default. A customer can override them, and it is the customer's responsibility to have the right to do so.
  • One exception: to answer a customer's web search, MeshArc reads a search engine's results page (Google's, and DuckDuckGo's and then Bing's when Google refuses or reads short), although those engines' robots.txt disallows it. Only the organic results are read (title, link, snippet). The pages a search then goes on to read obey robots.txt as usual.
  • Requests identify themselves by user agent unless the customer chose otherwise, and come from addresses we operate or from residential exits we lease.
  • To ask that your site not be crawled, or that pages already fetched be removed, write to [email protected] with the domain. We will identify the workspace, pass the request to it, and remove pages where the law requires it.

Who else sees it#

The app, the API and the database run on our own servers in India. Beyond those, we use a small number of providers to run the service. Each processes only what its job needs, under a contract that holds it to this policy.

ProviderWhat forWhat it sees
RazorpayPayments and subscriptionsYour email, name, plan, and the card, UPI or bank details you give Razorpay directly
CloudflareDelivers the site and the API, and protects them from attackYour IP address, the country it resolves to, and the requests you make
Google (Tag Manager, Analytics)Site analytics, where allowedThe pages you view, the events above, your browser and an approximate location
SentryError reportsStack traces, release, browser, URL without query string
Zoho ZeptoMailVerification codes, resets, invitations, billing, credit and plan-limit notices, run emails, digests, the newsletter's confirmation email; it tells us when an address bounces or reports our mail as spamYour email address and the message
Proxy and unblocking providers (Bright Data, ScrapingBee, ZenRows)Reaching pages that refuse a plain requestThe URLs the crawl fetches through them, never your account
OpenAIThe model behind an AI agent run, when you have not connected your ownYour instructions to the agent and the text of the pages it read, never your account

The subprocessors page lists each of them with where it processes data, and we post a new one there 30 days before it starts.

We disclose data beyond this only when the law requires it, to protect the service or someone's safety, or as part of a sale or merger, in which case this policy continues to apply and you will be told.

Where it is kept#

Data is stored on our own servers in India. If you are in the EEA, the UK or Switzerland and data is transferred elsewhere, the transfer is covered by standard contractual clauses or an adequacy decision.

How long we keep it#

DataKept
Account and workspaceUntil you delete the account or the workspace
A Google or GitHub account linked to yours (its account ID, its email address, and when it was linked and last used)Until you remove it under Profile → Sign-in methods, or your account is deleted
Sessions30 days, or 14 idle days, or until you end them
InvitationsSeven days
Verification codes and reset links30 minutes
What crawled pages said (their HTML, text and markdown) and screenshotsThe project's retention setting; the project's two most recent runs, and any older run whose pages a newer run still reuses, keep theirs
Runs, page URLs and headers, extracted fields, and change recordsUntil you delete the project
Webhook messages that carried pages' markdownThe project's last 200 webhook messages, until you delete the project; a crawl's own per-page messages, 7 days
One-off scrapes, crawls and maps not kept as a projectOne day
Web searches (query, settings and results)Seven days; ninety for a search a monitor ran
Pages a web search scrapedOne day
Agent runs (your instructions, the run's trace, its saved progress, its notes, and the pages it read)Seven days
Monitors (a search or agent request you keep on a schedule, and what each of its runs found)While the monitor exists: its last 200 runs, and, for one that follows records, those records and their history; deleting the monitor deletes them
Triggers (an action a change starts, such as an agent prompt, and a record of each time it fired)The trigger while it exists; each record of a firing, 90 days; deleting the trigger deletes them
Agent proposals (a project or monitor an agent proposed for you to approve, with the start of the question that led to it)30 days, whatever you decide; dismissing one deletes it
Request logs180 days
Sign-in and security events (time, IP address, device, outcome)180 days
Requests counted per workspace, day and route400 days
Where an account came fromAs long as the account
A newsletter sign-up never confirmed30 days
A newsletter subscriptionUntil you unsubscribe; after that, your address and the dates, as the record of your choice, until you ask us to erase them
An address that bounced for good or reported our mail as spam, the reason and the dateUntil it is lifted, by you writing to us or by us; it is the record that keeps us from sending to it
Site analytics in Google Analytics2 months (Google Analytics' standard setting for event data)
Error reports90 days
Invoices and financial recordsAs long as tax law requires, usually seven years

Deleting your account removes your profile and ends your sessions at once. A workspace whose last owner deletes their account is deleted with its projects, pages, keys and connections within 30 days, except for the financial records above.

Your rights#

Wherever you are, you can see and correct what we hold about you in Profile and Settings. If you are in the EEA, the UK, Switzerland, California or another place with a data protection law, you also have the right to ask for a copy of your data, to have it corrected or erased, to restrict or object to its processing, to take it with you, and to complain to your supervisory authority. To exercise any of these, including deleting your account, write to [email protected] from the address on the account and we will answer within 30 days. You may also name someone to exercise these rights for you if you are unable to. We will never treat you differently for exercising a right.

Our grievance officer for data protection is Siddharth Mishra, reachable at [email protected] and at the address below.

Security#

Everything travels over TLS. Passwords are hashed, API keys are hashed and shown once, destination credentials are encrypted at rest, and every query is scoped to the workspace that owns it. Keys carry scopes, an optional project list, an expiry and a rate limit. How we run security, and how to report a vulnerability, is on the security page.

If something goes wrong#

If personal data is lost, exposed or taken, we tell the workspace owners affected without undue delay, saying what happened, what data was involved, and what we are doing about it. We report the incident to CERT-In within six hours of noticing it, to the Data Protection Board of India as its rules require, and to the supervisory authority in the EEA or the UK within 72 hours where the law requires it.

Children#

MeshArc is a tool for work and is not directed at anyone under 18. We do not knowingly hold data about children; tell us if we do and we will delete it.

Changes#

When this policy changes we update the date at the top. For a change that affects what we collect or how we use it, we email every workspace owner before it takes effect.

Contact#

[email protected] · Discover Webtech Pvt Ltd, BH 16B, Bapat Square, next to Suryoday Ashram, HIG, Sukhliya, Indore, Madhya Pradesh 452010, India.