Keeping your data secure.
It’s fundamental to how we build.
This page describes the security controls Pointerly has in place today: the infrastructure we rely on, how we protect data, and who can access it.
Last updated: October 16, 2026
Infrastructure
Pointerly runs on managed cloud infrastructure. We do not operate our own data centres. Our core providers are:
Supabase (on AWS)
Database, authentication, encrypted secret storage (Vault), and file storage for workspace uploads and legacy link images. Enforces row-level security at the database layer. Also stores the records of our MCP server for connected apps.
Vercel
Application hosting, serverless functions, and content delivery network for the website, web app, short links, and our MCP server for connected apps (mcp.pointerly.io). Its firewall rate-limits requests to the MCP server.
Cloudflare
DNS and CDN for Pointerly hostnames, Turnstile bot protection on public forms, R2 object storage for public link and automation images (media.pointerly.io), and Browser Rendering for storefront imports.
Stripe
Payments. Card details are entered on Stripe and stored by Stripe, which is certified as a PCI DSS Level 1 service provider. Pointerly never receives card numbers.
Production is separated from development and preview environments, and changes reach production through a controlled build and deployment process. Access to production infrastructure is limited to authorized personnel. The full list of providers that process data is on our Subprocessors page.
Encryption and secrets
- In transit: connections to Pointerly and between Pointerly and its providers use TLS.
- At rest: the database and file storage (Supabase Storage and Cloudflare R2 for public link images) are encrypted at rest by our infrastructure providers.
- Connection tokens and keys: OAuth tokens for connected accounts, Telegram bot tokens, AI provider keys you add, and affiliate partner credentials are stored encrypted in Supabase Vault and decrypted only on our servers when needed. They are not returned to the browser.
- Platform secrets: our own API keys and signing secrets are kept in encrypted environment configuration, not in source code or client-side bundles.
Workspace isolation and permissions
- Row-level security: workspace data is protected by PostgreSQL row-level security policies, so a signed-in user can only read data from workspaces they belong to, even if application code made a mistake.
- Server-side authorization: server actions and API routes check who is calling and which workspace and resource they are acting on before reading or changing data.
- Roles: workspace members have roles, and owners and admins control membership, billing, and connected accounts.
- API keys are scoped to the workspace that created them and can be revoked at any time.
Authentication
- Sign-in is handled by Supabase Auth. Passwords are stored as salted hashes, never in plain text. You can also sign in with Google, Apple, Facebook, and other providers.
- You can turn on multi-factor authentication with an authenticator app.
- Sign-up, waitlist, and contact forms are protected against bots with Cloudflare Turnstile.
- Sessions use secure cookies, described in our Cookie Policy, and you can sign out of sessions from account settings.
Staff and operator access
- Pointerly staff access customer data only when needed to provide support you ask for, keep the service secure, investigate abuse, or comply with law.
- Access to administrative tools requires a staff account with the specific permission for that tool and a session verified with multi-factor authentication.
- Delegated Workspace operators can enter only the workspaces assigned to them, must use a Pointerly company email address and multi-factor authentication, and work within an operator role that excludes billing, members, API keys, and the browser extension.
- Staff access changes and operator assignments are recorded in audit logs.
Application security
- Input is validated on the server before it is processed or stored.
- Database access uses parameterized queries.
- User-provided HTML is sanitized before it is displayed, and React escapes output by default.
- We send security headers such as
X-Content-Type-Options,Referrer-Policy,Permissions-Policy, and framing controls (X-Frame-Optionsor aframe-ancestorspolicy for bio pages). - Actions that the Poppy assistant proposes in your workspace need your confirmation through a signed, single-use token before they run.
- Dependencies are reviewed and updated to address known vulnerabilities.
Integrations and webhooks
- Integrations use OAuth with the permissions each feature needs. You can disconnect them at any time.
- Incoming webhooks, for example from Stripe and Meta, are verified with the provider's signature or a shared secret before they are processed.
- Scheduled jobs and internal endpoints require a secret that is not exposed to browsers.
Connected apps (MCP)
AI apps and other MCP clients, such as Claude or ChatGPT, connect to Pointerly through our MCP server at mcp.pointerly.io. It runs on Vercel and keeps its records in our Supabase database. What an app receives and what we store is described in our Privacy Policy.
- OAuth 2.1: apps connect with the authorization code flow and PKCE (S256 only), and a resource indicator that limits each access token to our MCP server. Apps are public clients without a client secret, and their return addresses must use https or point to the user's own computer. Apps can register themselves (dynamic client registration); registration is rate-limited.
- Consent in Pointerly: the MCP server never shows its own sign-in or consent screen. Every authorization opens a Pointerly page where you sign in with your own account, including two-factor authentication if you use it, and choose the workspace and what the app may do. The app can never do more than you can in that workspace.
- Tokens: access tokens are signed with ES256 and expire after 15 minutes. Refresh tokens expire after 30 days and are replaced every time they are used. Each refresh checks the connection again, so a paused or revoked connection gets no new token and reduced permissions apply to the next one.
- Calls into Pointerly: the MCP server never forwards the app's access token to Pointerly. Each call carries a service key and a delegation signed with ES256 that expires within 60 seconds, is bound to that one request, and can be used only once. On every call Pointerly checks that the connection is active and that the person who connected it is still a member of the workspace with the permissions the call needs, within what the connection allows.
- Changes: in approval mode, a change runs only after the person who connected the app approves it in Pointerly while signed in. Some actions, such as deleting a link or publishing a post, always need approval. Automatic mode runs only the operations you picked, within per-run and per-period limits, for at most 90 days. Each workspace also has a daily limit on changes from connected apps for each module, set by its plan, which owners and admins can lower.
- Pausing and revoking: once you pause or revoke a connection in Settings → Connected apps, Pointerly refuses the app's requests and no new token is issued. The MCP server itself stops answering the app within 15 seconds.
- Edge protection: requests to mcp.pointerly.io pass through Vercel's firewall, which applies per-IP rate limits to app registration, authorization, token, and MCP requests.
- Storage and secrets: the MCP server connects to the database over TLS and verifies the database's certificate. Its records are in a separate database schema that is not exposed through Pointerly's public data API. Its signing keys and secrets are kept only in encrypted environment configuration, and Pointerly stores only a hash of the MCP server's service key.
- Logging: the MCP server logs each tool call's name, outcome, and duration, never its arguments, results, prompts, or tokens. Pointerly keeps an activity log for each connection (changes the app made, blocked requests, and settings changes) that you can review in Settings → Connected apps.
Link traffic and visitor data
- Short links record the data our customers need for analytics, described in our Privacy Policy. We store only the major version of a visitor's operating system, to limit fingerprinting.
- Clicks from user agents that look automated are flagged as likely bots.
- Short links do not set tracking cookies and do not load our website analytics tools.
Restricted partner data
Amazon Ads API and Creator Connections data is classified as restricted partner data. It is not sold, it is excluded from default automated third-party exports, and any allowed export must be started by the customer to their own connected destination and is recorded with the user, workspace, destination, data type, and time. Partner credentials are stored encrypted and never exposed to the browser. If a security incident involves Amazon information, we report it to Amazon as our agreements with Amazon require.
Logging and audit
- We log sign-in, security, billing, export, and administrative events to detect, investigate, and respond to problems.
- Logs are access-restricted and used for security, support, debugging, and compliance.
- We avoid logging secrets and access tokens.
Incident response
When we learn of a security incident, we:
- contain it and preserve evidence;
- work out which systems, data, customers, and partners are affected;
- rotate or revoke affected credentials;
- notify affected customers, regulators, and partners where the law or our contracts require it;
- fix the cause and review what we should change.
Account deletion
You can delete your account from account settings. Access is revoked immediately and your personal account data is deleted from active systems after a 30-day grace period. Workspaces you own are transferred to a member you choose or archived, and an archived workspace is permanently deleted 12 months later. Copies can remain in encrypted backups for up to 8 days. See our Privacy Policy for details.
Your part
Security is shared. Use a unique password, turn on multi-factor authentication, give workspace members only the access they need, remove people who no longer need access, keep API keys secret, and disconnect integrations you no longer use. Tell us right away at security@pointerly.io if you suspect someone has accessed your account.
Certifications
Pointerly does not currently hold its own security certification, such as SOC 2 or ISO 27001. Our core infrastructure providers publish their own certifications and audit reports. We will update this page if that changes.
Responsible disclosure
If you believe you have found a security vulnerability in Pointerly, email security@pointerly.io with a description, the steps to reproduce it, and its likely impact. We aim to acknowledge reports within 5 business days and will keep you informed while we fix the issue.
When you research vulnerabilities, please:
- use only accounts and workspaces you own or have permission to test;
- not access, change, or delete other people's data, and stop and tell us if you reach it by accident;
- not run denial-of-service tests, spam, social engineering, or physical attacks, and not send messages through connected platforms to people who have not agreed to it;
- give us reasonable time to fix the issue before telling anyone else.
If you follow these rules and act in good faith, we will not take legal action against you for your research and will consider it authorized under our Terms. We do not currently run a paid bug bounty program.
Version history
- October 16, 2026: Added Connected apps (MCP): OAuth 2.1 with PKCE and resource indicators, consent on a Pointerly page, token lifetimes, a signed single-use delegation for every call into Pointerly, live permission checks, approval mode and daily limits, how fast pausing or revoking takes effect, firewall rate limits, database and secret handling, and what the MCP server logs. Noted that Vercel and Supabase also run the MCP server.
- October 10, 2026: Documented Cloudflare for DNS/CDN, Turnstile, and R2 public image storage; clarified Supabase Storage scope for workspace uploads and legacy link images.
- October 10, 2026: Rewritten to describe only controls that are in place today, including encrypted token storage, staff and operator access, and a responsible disclosure policy with safe harbour.
- April 30, 2026: Previous version.