Security

Venom holds your organization’s projects, tasks, logs, artifacts, skills, and tool connections, and your agents read and write them through the Venom MCP server. This page describes how that data is stored and protected. For what personal data we collect and why, see the Privacy Policy.

Where your data is stored

• The web app, the MCP server, and the services behind them run on Amazon Web Services in the EU (Ireland, eu-west-1). Our AWS organization policy blocks the use of other AWS regions for these services.
• Account and organization data is stored in Amazon Aurora DSQL, projects, tasks, logs, and artifacts in Amazon DynamoDB, and skill files in Amazon S3. Sign-in is handled by Amazon Cognito. All of these are in eu-west-1.
• The marketing website and documentation contain no customer data and are served through Amazon CloudFront’s global network.
• Product analytics (page views, clicks, usage events, and web app error reports) are stored in PostHog’s EU cloud in Frankfurt, Germany. The Privacy Policy lists what they contain.

Encryption

• In transit. Connections to the Service use HTTPS with TLS 1.2 or 1.3. Plain HTTP requests are redirected to HTTPS. Traffic from our load balancer to our services, and between our own services, stays inside our AWS virtual private network and is not encrypted with TLS inside that network.
• At rest. Stored files are encrypted in S3 with AES-256, and our databases are encrypted at rest with keys owned and managed by AWS.
• Third-party credentials. API keys and personal sign-in tokens for MCP servers your organization connects are encrypted with a dedicated AWS Key Management Service key that rotates automatically. Of our services, only the one that stores a credential can encrypt it, and only the one that uses it can decrypt it.
• Tokens we issue. MCP access and refresh tokens are stored only as one-way SHA-256 hashes, so they can’t be read back.

Access control

• Sign-in. Accounts are managed by Amazon Cognito. Passwords must be at least 12 characters and include upper-case and lower-case letters and a number. You can also sign in with Google.
• Every request is checked. The web app verifies your sign-in token on each request and checks your live membership and role in the organization before any data is read or written. Organization roles are owner, admin, member, and viewer.
• Agents. AI clients connect to the MCP server with OAuth 2.1 and PKCE. You approve each connection for one organization. Access tokens last 1 hour; refresh tokens are replaced each time they are used (an old one stays valid for 60 seconds so that a retried request doesn’t fail) and expire after 30 days without use.
• Network. Only the web app and the MCP server accept traffic from the internet, through our load balancer. The services that hold account and project data accept connections only from our own services.

How our team accesses production

• Production runs in its own AWS account, separate from development. People sign in through AWS single sign-on with time-limited sessions, and the root user of the production and development accounts is blocked.
• AWS CloudTrail records administrative activity in every account to a separate security account, with log file validation, and keeps it for 400 days.
• Deployments run from our code repository through GitHub Actions with short-lived AWS credentials (OpenID Connect), not stored access keys. Changes are normally made through pull requests, which run our automated checks and a Terraform plan against our development environment. Production deploys are started by hand from the main branch.

Retention and recovery

• Deleted or replaced log entries can be recovered for 30 days, then they are removed; backup copies expire within a further 35 days.
• Artifact revisions are kept for the lifetime of the organization.
• Earlier versions of files stored in S3, such as skills, are kept for 90 days after they are replaced or deleted.
• Records of agent activity (which operation ran, for whom, and whether it succeeded, without the content) are kept for 30 days.
• Server and security logs (IP addresses, the addresses of pages and API calls, and error details) are kept for 90 days for application logs and 30 days for firewall logs.
• Product analytics events, which carry the same activity details without content, are kept in PostHog for 1 year.
• In production, the project and task database has point-in-time recovery enabled, and the account database and project database are protected against accidental deletion.
• When an owner deletes an organization, deletion of its workspace content, teams, members’ roles and the third-party API keys stored for it starts straight away and normally completes within minutes. Members’ personal sign-ins to third-party MCP servers made through the organization are deleted when they disconnect them, or on request to support. Copies in our backups and earlier file versions are removed within 100 days. The organization’s billing records stay with Stripe and in our accounts for as long as tax and accounting law requires.
• Personal accounts are deleted on request to support@venomlabs.dev, within one month; copies in our backups expire within a further 35 days.

Payments

Paid plans are billed through Stripe. Cards are entered in Stripe’s Payment Element, so card numbers go straight to Stripe and never reach Venom’s servers. We store only the Stripe customer and subscription identifiers and the subscription’s plan, seat count, status, and billing period. Billing updates from Stripe are accepted only when Stripe’s signature on them is valid.

AI models and your data

Venom uses your workspace content only to provide, secure and support the Service for you, and as the law requires. Venom does not use your workspace content to train or improve AI or machine-learning models, does not sell it, does not use it for advertising, and does not use it for any other purpose of its own. Venom does not run AI models. Your content reaches an AI model only when an AI client that you connect (for example Claude) reads it through the MCP server. That client handles it under its provider’s own terms.

Our position under the EU AI Act: Venom and the EU AI Act.

Subprocessors

• Amazon Web Services: hosting, databases, file storage, sign-in, email delivery, and encryption keys. EU (Ireland).
• Stripe: payment processing for paid plans.
• Google: optional “Sign in with Google”, and Google Workspace for our own support email.
• Attio: our customer relationship manager. It receives what you send through our contact sales form, and for each person who signs up their name, email address, how they signed in, and the names of organizations they create or are invited to; never workspace content.
• Slack: our team’s internal messaging. It receives a message for each person who signs up with the same details as Attio; never workspace content.
• Cloudflare: DNS for venomlabs.dev, and Turnstile, which checks that email sign-ups come from a person. Traffic to the Service does not pass through Cloudflare.
• PostHog: product analytics for the website, the web app, and our services. It receives page views, clicks, usage events, and web app error reports; we don’t send it workspace content. EU (Frankfurt, Germany).

Third-party MCP servers that your organization chooses to connect receive your agents’ requests to them, under those services’ own terms.

Certifications

Venom does not yet hold a SOC 2 report or ISO 27001 certification. We’ll list certifications here when we have them.

Reporting a vulnerability

If you believe you have found a security issue in Venom, email support@venomlabs.dev with “Security” in the subject line. Our vulnerability disclosure policy explains what’s in scope, how we respond, and our safe-harbour commitment for good-faith research. Our security.txt is at venomlabs.dev/.well-known/security.txt.

Questions about security for your organization? Contact sales.