NewApp Monitoring is liveSign up free
AgentbeamBeta
AI agent vulnerability · OpenClaw

CVE-2026-26326: OpenClaw skills.status could leak secrets to operator.read clients

mediumCVSS 4.0 5.3GHSA-8mh7-phf8-xgfm

In OpenClaw 2026.2.13 and earlier, the gateway method `skills.status` returns raw config values in `configChecks`, so a client with only `operator.read` scope can read secrets such as a Discord bot token. Upgrade `openclaw` to 2026.2.14 and rotate any Discord tokens that read-scoped clients could have seen.

Published Feb 17, 2026 · Updated Feb 20, 2026 · Analysis Nov 1, 2026

What an attacker can do

skills.status reports whether each skill's requirements are met. For every requires.config entry it included the resolved value in configChecks[].value. When a skill required a broad subtree, such as channels.discord, the report carried everything under it, including the bot token.

The method is callable with operator.read. A client holding that scope therefore got secrets that normally need operator.admin or config.* access. The scope split exists so that dashboards, monitors and read-only integrations can look at state without seeing credentials. This bug let them see credentials.

The attacker needs a connection to the gateway with a read-scoped credential. No user interaction is needed, and the CVSS 4.0 vector shows low privileges. What they get is limited to config values that some skill's requires.config path points at. The advisory names the Discord token as the example. With that token they can act as the bot in Discord (CWE-200). Other channel secrets under a broad required path would leak the same way.

Who is affected

The openclaw npm package, version 2026.2.13 and earlier. Exposure needs a skill whose requires.config points at a path holding secrets, and a client with operator.read. The Discord skill's broad requirement was the known case. Check:

npm ls openclaw
openclaw --version

openclaw

Ecosystem
npm
Affected
< 2026.2.14
Fixed in
2026.2.14
Check
npm ls openclaw

How to fix it

Upgrade to openclaw 2026.2.14:

npm install -g [email protected]

The fix, in commits d342805 and ebc6886, changes requirement checks to return only { path, satisfied } with no values. It also narrows the Discord skill requirement to the token key rather than the whole channels.discord subtree.

The advisory's mitigation is to rotate any Discord tokens that read-scoped clients may have seen. Treat other channel credentials the same way if a custom skill required a parent path.

How to tell if you were hit

Review which clients hold operator.read on your gateway, and whether any of them called skills.status before the upgrade. Check gateway logs for those calls. If a client you do not fully trust had read scope, assume the Discord bot token and any secret under a skill's required config path was exposed. Look in the Discord developer portal and audit log for bot activity you did not start, then rotate the token.

Original advisory

From the GitHub Advisory Database (CC BY 4.0).

Summary

skills.status could disclose secrets to operator.read clients by returning raw resolved config values in configChecks for skill requires.config paths.

Affected Packages / Versions

  • Package: openclaw (npm)
  • Affected: <= 2026.2.13
  • Patched: 2026.2.14

Details

The gateway method skills.status returned a requirements report that included configChecks[].value (the resolved value for each requires.config entry). If a skill required a broad config subtree (for example channels.discord), the report could include secrets such as Discord bot tokens.

skills.status is callable with operator.read, so read-scoped clients could obtain secrets without operator.admin / config.* access.

Fix

  • Stop including raw resolved config values in requirement checks (return only { path, satisfied }).
  • Narrow the Discord skill requirement to the token key.

Fix commit(s):

  • d3428053d95eefbe10ecf04f92218ffcba55ae5a
  • ebc68861a61067fc37f9298bded3eec9de0ba783

Mitigation

Rotate any Discord tokens that may have been exposed to read-scoped clients.

Thanks @simecek for reporting.

---

Fix commits d3428053d95eefbe10ecf04f92218ffcba55ae5a and ebc68861a61067fc37f9298bded3eec9de0ba783 confirmed on main and in v2026.2.14. Upgrade to openclaw >= 2026.2.14.

Credited to simecek, stanislavfortaisle.

Questions

Do I need admin access to read the secrets?

No. That is the flaw. operator.read alone was enough to call skills.status and see the resolved values, without operator.admin or config.* access.

Which secrets leak?

Any config value under a path a skill lists in requires.config. The advisory's example is the Discord bot token, exposed because the Discord skill required all of channels.discord.

Is 2026.2.14 enough, or must I do more?

Upgrade, then rotate any Discord token that read-scoped clients could have received. Upgrading stops new leaks but does not undo past exposure.

Is it exploited in the wild?

No source reports exploitation, and it is not in CISA's KEV catalog.

References

How this page is made: the data panel comes from the GitHub Advisory Database, CISA’s KEV catalog, the NVD and FIRST EPSS, and updates twice a day. The analysis is written by an AI model from the sources listed above and checked by automated rules before it is published. Check versions against the vendor's advisory before you act. Methodology

More OpenClaw vulnerabilities

All OpenClaw entries