Skip to content
This repository was archived by the owner on Aug 3, 2026. It is now read-only.

Rotate validator seal pubkey (post-compromise) - #81

Merged
bitzic merged 1 commit into
RalphLabsAI:mainfrom
karpabot:rotate-seal-key
Jun 29, 2026
Merged

Rotate validator seal pubkey (post-compromise)#81
bitzic merged 1 commit into
RalphLabsAI:mainfrom
karpabot:rotate-seal-key

Conversation

@karpabot

Copy link
Copy Markdown

Rotate validator seal pubkey (post-compromise)

The validator's X25519 seal keypair (dated Jun 23, embedded as DEFAULT_VALIDATOR_PUBKEY) was on the validator box during the Jun-27 root compromise → must be treated as exfiltrated. Because proof bundles sit encrypted on the public HF dataset repo, the leaked key can decrypt every bundle sealed to the old pubkey. Rotating is the fix.

Already done on the validator: clean cutover to a fresh keypair. The box now decrypts bundles sealed to the new pubkey only; old-pubkey bundles are rejected.

This PR: publishes the new DEFAULT_VALIDATOR_PUBKEY so miners seal to it.

⚠️ Action required — miners

Pull this change and re-seal your proof bundles to the new pubkey. Submissions sealed to the old pubkey are rejected until re-sealed. (The current king is unaffected.)

🤖 Generated with Claude Code

The Jun-23 validator X25519 seal keypair was present on the validator box
during the Jun-27 root compromise and must be treated as exfiltrated. The
validator has switched to a fresh keypair (clean cutover); this publishes the
new public key so miners seal proof bundles to it.

ACTION REQUIRED (miners): pull + re-seal submissions to the new pubkey. Bundles
sealed to the old pubkey are rejected by the validator until re-sealed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
@bitzic
bitzic merged commit 8860307 into RalphLabsAI:main Jun 29, 2026
4 checks passed
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants