Active development: 100+ commits since 2026-07-25 across 2 contributors
Code ActivitySource linked
18 active day(s) in runningoffcode/agent-wormhole since 2026-07-25; only the newest 100 commits were read.
Agentic Finance
ResearchedThe security layer for agentic finance.
Code ActivitySource linked
18 active day(s) in runningoffcode/agent-wormhole since 2026-07-25; only the newest 100 commits were read.
Github ReleaseSource verified
On-chain memos: a permissionless injection channel New wormhole memos command and readguard coverage for the one inbound channel that requires nothing of the agent. Every other vector needs the agent to go somewhere — fetch a page, clone a repo, install a skill. An on-chain memo needs none of that: anyone can pay a fraction of a cent to write arbitrary text into an agent's transaction history, unsolicited. The payload lands when the agent reads its own history, and it arrives as tool output. The worm case is why this ships beside the config scanners rather than in the payment guard: a m
View evidenceGithub ReleaseSource verified
Security hardening from the post-launch review, plus first-class facilitator support. Closed bypasses - guardSigner now guards every signing method a wallet exposes — signAllTransactions, signAndSendTransaction, and signAndSendAllTransactions previously routed around the check entirely. - The System program is now a strict allowlist (only AdvanceNonceAccount passes). TransferWithSeed, Assign, WithdrawNonceAccount, allocation instructions, and unclassifiable tags are refused (X402-007 / new X402-009). - Burn / BurnChecked riding beside a correct payment are refused (
View evidenceGithub ReleaseSource verified
Two packages, one thesis: an agent obeys instructions it did not write, so the places those instructions live need integrity checks. wormhole-guard (PyPI, Python 3.8+, zero dependencies) — integrity monitoring and hardening for agent instruction files (AGENTS.md, CLAUDE.md, .cursor/rules, and eleven more formats), MCP tool definitions, and agent-to-agent handoffs. Prevention first: harden drops the write bit and pre-creates absent config paths, guard inspects writes before they land, outbound refuses to pass a payload to another agent. Detection rules are triage on top, and eve
View evidenceWritten by HEY researchers and reviewed before publishing. Commentary with sources — not the project's own claims, and not evidence of shipping.
No published notes on Agent Wormhole yet.
Know something about Agent Wormhole? Sign in with GitHub to submit a research note. Notes are reviewed before they appear here.