Openclaw’s Claws Prove Rather Dull, Certik Purrs with Disdain

by

Darling, gather round! Certik, those darling sentinels of the digital realm, have uncovered a most amusing flaw in Openclaw, the open-source AI platform. Apparently, its “skill scanning” is about as effective as a sieve at a champagne party-utterly porous, my dear.

The Clawhub Farce: A Moderation Pipeline in Tatters

In a report that’s as scintillating as a gossip column, cybersecurity darlings Certik have revealed that Openclaw’s security is more hole than substance. Its reliance on “skill scanning” is, frankly, a joke-a feeble attempt to shield users from malicious third-party extensions. How très embarrassing!

Published on the 16th of March, 2026, the report drips with condescension, noting that Openclaw’s security model is as flimsy as a socialite’s alibi. It depends far too heavily on detection and warnings, rather than the robust runtime isolation one might expect from a platform of its pretensions.

According to the report, Openclaw’s marketplace, the oh-so-grandly named Clawhub, employs a “layered moderation flow” to review “skills”-third-party applications that purport to grant the AI agent capabilities such as system automation or cryptocurrency wallet operations. This includes the use of Virustotal for scanning known malware, the Static Moderation Engine (a recent addition, darling, introduced on March 8, 2026), and an “incoherence detector”-a tool so quaint it’s almost charming in its ineffectiveness.

However, Certik’s researchers, ever the spoilsports, pointed out that static rules searching for “red flags” can be circumvented with the simplest of code rewrites. And the AI review layer? Oh, it’s splendid at spotting the obvious, but utterly hopeless at identifying exploitable vulnerabilities hidden in otherwise plausible code. How utterly tedious.

The ‘Pending’ Farce: A Gap Wider Than a Society Matron’s Hat

One of the most delightful flaws-if one can call it that-is the treatment of pending scan results. Imagine, darling, a skill remaining active and installable while Virustotal dithers about for hours or days. In practice, these pending skills are treated as benign, allowing users to install them without so much as a raised eyebrow. How utterly risqué!

To illustrate this vulnerability, Certik’s researchers concocted a proof-of-concept (PoC) skill called “test-web-searcher.” It appeared as harmless as a kitten but contained a hidden bug that allowed for arbitrary command execution on the host machine. When invoked via Telegram, it bypassed Openclaw’s optional sandboxing and “popped a calculator” on the researcher’s machine-a classic demonstration of full system compromise. How delightfully old-school!

The report concludes, with a tone of exasperated elegance, that detection can never be a substitute for a true security boundary. Certik is urging Openclaw’s developers to run third-party skills in isolated environments by default, rather than leaving it to the whims of user configuration. They also suggest implementing a model where skills must declare specific resource needs up front, much like modern mobile operating systems. How very forward-thinking!

For users, Certik offers a warning as sharp as a well-timed barb: A “benign” label on Clawhub is not proof of security. Until stronger isolation is the default, the platform should only be used in low-value environments, far from sensitive credentials or assets. How dreadfully inconvenient!

FAQ ❓

  • What security issue did Certik find in Openclaw? Darling, Certik reported that Openclaw’s reliance on “skill scanning” is as effective as a chocolate teapot-utterly useless at protecting users from malicious third-party extensions.
  • How does Openclaw’s moderation flow function? It employs a “layered moderation flow,” including tools like Virustotal and an incoherence detector, to review third-party “skills.” How quaint!
  • What is the critical flaw regarding pending scan results? Skills can remain active and installable while scan results are pending, leaving users at risk of installing malicious extensions without a whisper of warning. How très awkward!
  • What should users do to protect their data on Openclaw? Users are advised to only use Openclaw in low-value environments until stronger isolation measures are implemented. How dreadfully inconvenient, but necessary, my dear!

Read More

2026-03-18 08:57