GHSA-77q5-rr5v-x43q: OpenClaw: Trusted retry endpoint checks could match hostname prefixes

The retry endpoint check compared the start of the hostname instead of the full origin. A hostname such as a trusted name followed by extra characters or a longer domain starts with the trusted string, so it passed. This is an incomplete comparison (CWE-1023) in input validation.

The attacker must be able to influence the retry endpoint URL through lower-trust input and must hold low privileges in the Gateway. They register or control a lookalike hostname and supply it as the retry endpoint. When OpenClaw retries, it sends authentication material to that host. The vector scores high confidentiality impact only: credentials leak, which the attacker can then reuse against the real service. It does not by itself give code execution.

The advisory keeps the trusted-operator model unchanged and says real impact depends on configuration and on whether lower-trust input can reach the path. It does not name the feature or the credentials involved.

The npm package openclaw before 2026.5.7, where the trusted retry endpoint feature is enabled and a lower-trust party can influence the endpoint URL.

npm ls openclaw

Use npm ls -g openclaw for a global install. Anything below 2026.5.7 is affected.

Upgrade to [email protected], the first stable patched version:

npm install [email protected]

Until then, the advisory says to pin retry endpoints to exact trusted origins. It also advises narrow channel and tool allowlists, not sharing one Gateway between mutually untrusted users, and disabling the feature when unneeded. If a lookalike endpoint could have been used, rotate the credentials that OpenClaw sends on retries.

Review configured retry endpoints and any URLs supplied through channels or tools for hostnames that start with a trusted host but continue past it. Check outbound connection logs and DNS logs from the Gateway host for requests to unfamiliar domains that resemble your trusted ones. The advisory gives no log line. Rotate credentials if you find a match.

Summary

Trusted retry endpoint checks could match hostname prefixes. In affected versions, a retry endpoint URL chosen by lower-trust input could pass validation by using a hostname prefix that resembled a trusted host.

This advisory is scoped to the named feature and configuration. It does not change OpenClaw's trusted-operator model: authenticated Gateway operators, installed plugins, and intentional local execution surfaces remain trusted unless a separate policy, approval, allowlist, sandbox, or auth boundary is crossed.

Impact

When the affected feature is enabled and reachable, this could send authentication material to an endpoint outside the intended trust target. Practical impact depends on the operator's configuration and whether lower-trust input can reach that path.

Patched Versions

The first stable patched version is 2026.5.7.

Mitigations

pin retry endpoints to exact trusted origins until patched. As general hardening, keep channel and tool allowlists narrow, avoid sharing one Gateway between mutually untrusted users, and disable the affected feature when it is not needed.