Security Model
Last updated: 19 September 2026
PoolPort is a non-custodial tool. Your seed phrases and private keys are the highest-value secret in the flow, and the design of the Service assumes they will be treated as such: they never leave your browser, and no server we operate ever receives them.
1. What runs where
All key material — mnemonics, private keys, and derived signing objects — exists only in your browser's memory for the duration of a preview or migration. Nothing is written to disk, localStorage, sessionStorage, or IndexedDB. Refreshing the page clears everything.
The only things that leave your browser are:
- Public wallet addresses, sent to RPC endpoints to read balances.
- Signed transaction payloads, broadcast to the appropriate chain.
- Public migration receipt data (addresses, hashes, fee amounts) sent to our worker for operator reconciliation.
2. How to verify
You do not need to trust our word for this. The Service is designed to be independently verifiable:
- DevTools → Network. Open the Network tab before clicking Preview. Watch every request. No request body contains your mnemonic or any private key.
- DevTools → Sources. The entire client bundle is served as a single file. You can read it, search it, and confirm that no code path calls home with your phrase.
- Bundle integrity. The script tag includes an SRI hash. Your browser verifies the file hasn't been tampered with before executing it.
- Release metadata. The current bundle's SHA-256 is published at
/releases/latest.txt. You can fetch it and compare against a locally-computed hash of the served file.
3. Threat model
We defend against the following:
- Server compromise. If our worker, hosting, or any third-party RPC is compromised, the attacker never sees a mnemonic, private key, or unsigned transaction from your side.
- Supply-chain tampering. SRI hashes on the client bundle, plus the published release SHA-256, let you detect if a compromised CDN serves altered code.
- Malicious RPC providers. Because every transaction is signed locally, a malicious RPC can only refuse to serve, not lie about the contents of a signed transaction.
We do not defend against:
- Compromised endpoints. If your computer is running malware that reads browser memory or intercepts keystrokes, no web tool can protect you. Use a clean device.
- Browser extensions. Malicious extensions can read page content and modify the DOM. Use an isolated browser profile or a browser with no extensions for high-value migrations.
- Phishing. Confirm you are on the correct domain before pasting any phrase. Bookmark the site; do not reach it through search results or DMs.
- User error. Migration is irreversible. If the destination phrase you paste is wrong, the assets are gone. Verify before executing.
4. Sponsorship key handling
To sponsor gas on wallets that cannot pay for their own migration, the operator holds funded wallets on several chains. Those keys live in Cloudflare's encrypted secret store, are only accessible to the worker at request time, and are never included in the client bundle or served to a browser.
The sponsor wallets are operationally separate from the fee wallets that receive the migration fee. Sponsor keys are hot wallets with a small float, topped up periodically and never used to hold revenue.
5. Reporting a vulnerability
If you believe you have found a security issue, please email security@poolport.xyz. Do not open a public issue, and do not test against wallets holding funds you do not control.
We will acknowledge reports within 72 hours and aim to have a fix or mitigation deployed as soon as reasonably possible. If you would like credit in a future release note, tell us how you want to be named.
6. Operational practices
- All traffic is served over HTTPS. HSTS is enabled.
- The client bundle is rebuilt from source on every push; the deploy pipeline is public on GitHub.
- Sponsor and fee wallet addresses are public and can be audited on-chain.
- The operator dashboard is gated by a shared secret, rate-limited per IP, and requires HTTPS to reach.