APort Privacy Policy

Last Updated:October 12, 2026
Version:1.1

This Privacy Policy explains how APort ("we", "our", "us") collects, uses, and shares information when you use our websites, dashboards, APIs, SDKs, and related services (the "Service").

We build a neutral trust rail for AI agents. Transparency is foundational: we describe what we collect and why.

Information We Collect

1) Account & Organization Info

Contact details (name, email), authentication identifiers (e.g., GitHub user ID).

Organization profile (name, domain, membership; where you connect GitHub "read:org" we receive org and membership metadata).

2) Passport Content

Template and Instance Passport data you submit (e.g., owner/controller, capabilities, limits, regions, assurance evidence, links, logos).

Public "About" pages you enable. You control what is public.

3) Verification & Policy Calls

Request metadata (timestamps, IP, user agent, headers), agent IDs, policy context fields (e.g., refund amount, export row count), allow/deny results, and reason codes.

We intentionally avoid processing sensitive payload data (e.g., raw PII) beyond the minimally necessary context. You should never send confidential data fields beyond what a policy requires.

4) Attestations

Evidence you submit for assurance (e.g., email challenge, GitHub org membership, domain TXT or /.well-known file). We store facts and timestamps; we do not store your GitHub password or OAuth tokens beyond the minimum needed scope.

5) Webhooks & Integrations

Endpoint URLs and delivery logs (success/failure, signatures). We store HMAC secrets to sign events.

6) Cookies & Analytics

Functional cookies (session/auth).

Basic analytics (page views, feature usage) to improve the Service. We avoid invasive tracking.

Task Delivery

Task delivery processes the task description and answers you provide, the selected repository and source context, generated requirements, readiness and review decisions, execution identifiers, usage observations, logs, files and pull request references. APort stores task records and recovery data to show progress, reconnect to saved work, verify results and avoid dispatching the same work twice.

Self-service repository setup records the APort account and workspace, GitHub user and installation IDs, selected repository identity, passport ID, connection status and expiry in the workspace's configured regional database. The temporary GitHub authorization token is encrypted in a Secure, HttpOnly cookie scoped to the setup endpoint for 15 minutes. Execution uses short-lived GitHub App installation tokens scoped to the selected repository. A shared US control database records hashed APort and GitHub identities, reservation and binding identifiers, sponsored allowances and hosting usage to prevent repeat claims and account for the program's limits. Hashed identity records are not anonymous data; configured database region labels have the limitations described below.

When you authorize preparation or execution, relevant task instructions and source material may be sent to the configured model or execution provider. The current API runtime uses OpenAI. Cloudflare hosts APort's task endpoint, storage and runtime recovery snapshots. GitHub receives the branches, commits and pull requests created in the authorized repository. Each provider also processes data under its applicable terms and account settings; APort does not establish a different provider retention or model-training policy.

When you use a connected MCP host, such as ChatGPT, APort returns the requested task information to that host and its embedded workspace. The connection does not give APort blanket access to your conversations or all repositories in your GitHub account. Repository execution requires separately configured access and allowance. Avoid submitting credentials or unrelated personal records in task descriptions and source context.

To recover a request after a timeout without starting the same work twice, the task workspace saves its operation ID and exact submitted fields in your browser's local storage. These fields can include task text and answers. The recovery entry is removed after an acknowledged response is successfully processed. Unresolved entries can remain across reloads and sign-out until reconciled or cleared from the browser. They use ordinary browser storage, so protect access to your browser profile. Clearing site data removes this local recovery record; it does not cancel or delete work already received by APort or a provider.

An unacknowledged passport creation also leaves a marker with the requested passport name in browser session storage, scoped to the account and workspace. This requires reviewing the current passport list before another creation. Website task-setup analytics use the existing website analytics configuration and normal page context. Task-specific event fields contain allowlisted step names, task stage and elapsed browser-session time; they exclude task text, repository names, passport and task IDs, credentials and log contents. Viewing a blocker is recorded as a blocker view, not evidence that an operator had to intervene. Embedded task panels do not emit these website funnel events.

Task data can remain in APort's task records and recovery snapshots, and in the connected provider or GitHub repository. Disconnecting the plugin or cancelling a task does not delete those records or confirm that already-started provider work has stopped. Task delivery does not yet offer self-service deletion or a guaranteed retention period. Save needed artifacts while available and contact support to request export or deletion. We may need to verify your authority, reconcile active work, or retain records required for security or legal obligations. A request to APort does not itself delete copies held in your GitHub or provider account.

Configured region labels describe routing and access boundaries; they do not by themselves guarantee the physical location of task execution, recovery snapshots or provider processing. Confirm the required locations and retention arrangements with APort before using task delivery for sensitive or regulated work.

How We Use Information

  • Provide, secure, and improve the Service.
  • Verify Passports and evaluate policy packs.
  • Generate Verifiable Attestation and countersign attestations.
  • Detect abuse, investigate incidents, enforce rate limits.
  • Communicate with you (product updates, security notices, billing).
  • Comply with legal obligations.

We do not sell personal data. We do not train ML models on verification payloads.

  • Performance of a contract (to provide the Service).
  • Legitimate interests (security, abuse prevention, product improvement).
  • Consent (where required, e.g., certain emails).
  • Legal obligations.

Sharing

We may share with:

  • Service providers (hosting, email, monitoring) under contracts with confidentiality and security obligations.
  • Third-party services you connect (e.g., GitHub) per your authorization.
  • Legal: to comply with law, enforce Terms, or protect rights and safety.
  • Business transfers: in a merger, acquisition, or sale, per applicable law.

We do not publish private Passport data unless you mark it public (e.g., Agent Passport pages).

Data Retention

  • Account and org records: for the life of your account and as required by law.
  • Verify logs: typically 90–180 days for security and audit, unless you request a different retention or law requires otherwise.
  • Attestations: until expiry or revocation.

You can request deletion subject to our need to retain minimal logs for security and legal compliance.

Security

We use reasonable administrative, technical, and organizational measures (e.g., key rotation, HMAC signatures, TLS, access controls). No system is perfect; you are responsible for securing your credentials, API keys, and webhook endpoints.

International Transfers

Your data may be processed in countries other than your own. Where required, we implement appropriate safeguards (e.g., SCCs).

Your Rights

Depending on your location, you may have rights to access, correct, delete, or restrict processing of your personal data. Contact us at [email protected]. We may ask you to verify your identity.

Children

The Service is not intended for children under 16. We do not knowingly collect data from children.

Changes

We may update this Policy. We will notify you of material changes (e.g., email or in-product notice). Continued use after the effective date constitutes acceptance.

Contact

This document is part of the APort platform. For questions or clarifications, please contact our support team.