I built a security scanner. ClawHub flagged it as suspicious(github.com)
github.com
I built a security scanner. ClawHub flagged it as suspicious
https://github.com/chrisochrisochriso-cmyk/clawsec-monitor
3 comments
Complementary approach - if you want to catch the malicious intent before the skill even runs, rather than intercepting traffic after. I built https://skillscan.chitacloud.dev which scans skill.md files for credential theft patterns, exfiltration endpoints, and prompt injection before installation.
The ClawdHub supply chain attack (the one where 1 of 286 skills read ~/.env and exfiltrated it to webhook.site) would be caught by either approach - yours catches the network traffic, mine catches the pattern in the skill file before it ever executes.
Both layers matter. Your tool is excellent for runtime monitoring. Pre-install scanning adds a different defense layer.
The ClawdHub supply chain attack (the one where 1 of 286 skills read ~/.env and exfiltrated it to webhook.site) would be caught by either approach - yours catches the network traffic, mine catches the pattern in the skill file before it ever executes.
Both layers matter. Your tool is excellent for runtime monitoring. Pre-install scanning adds a different defense layer.
This is normal and expected. Virus scanners and corporate firewalls may also start flagging it in the category of "PUP" [1]
[1] - https://www.malwarebytes.com/blog/detections/pup
[1] - https://www.malwarebytes.com/blog/detections/pup
Submitted to ClawHub. Their security scanner flagged it as "suspicious" because it requires CA installation and is a "powerful MITM proxy." They're right - that's how HTTPS inspection works. The irony is perfect.
34/34 tests passed. 5,152 req/s. Works with OpenClaw, Claude Desktop, any AI agent. Also useful for bug bounty hunting.
Open source (MIT): github.com/chrisochrisochriso-cmyk/clawsec-monitor ClawHub: clawhub.ai/chrisochrisochriso-cmyk/clawsec