trentclaw
Security assessment for your OpenClaw agent environment. Flags misconfigurations and risky skills — prompt injection, permission escalation, data exfiltration — and the chained attack paths between them, across gateway config, tool permissions, MCP servers, and plugins.
- from
- Static Analysis & Code Quality
- added
- 2026-10-10
- likes
- 0
similar
-
Clutch Security clutch.security
MCP A View From The Trenches (2025) - Investigates the security risks of Model Context Protocol server deployments across enterprise environments, revealing a trend of silent, explosive adoption characterized by total security blindness. Critical findings highlight that…
-
threatcl github.com
Documenting your Threat Models with HCL
-
thClaws/thClaws github.com
Native Rust AI agent workspace with multi-provider LLM support, skills system, MCP servers, knowledge bases, and agent orchestration. Features desktop GUI, CLI REPL, and non-interactive modes.
-
Malicious code analysis: Abusing SAST (mis)configurations to hack CI systems medium.com
Misconfigured SAST tools can be exploited to execute malicious code on CI systems, allowing attackers to steal credentials or deploy malicious artifacts.
-
Clankerusecase clankerusecase.com
Threat-led library of 7,800+ MITRE ATT&CK-mapped detections in Defender/Sentinel KQL, Sigma, Splunk SPL, Datadog, Falcon LogScale, and AWS CloudWatch, generated from public threat-intel reporting and schema-validated before publish.
-
ToxicSkills: Snyk Finds Malware and Prompt Injection in 36% of AI Agent Skills snyk.io
Feb 2026 Snyk research across the ClawHub AI agent skills registry: 36% of audited skills contained security flaws, 1,467 malicious payloads found, and 2.9% used curl | bash remote instruction loading to evade static analysis. Covers indirect injection via poisoned web content…
Static Analysis & Code Quality › Other: “Security assessment for your OpenClaw agent environment. Flags misconfigurations and risky skills — prompt injection, permission escalation, data exfiltration — and the chained attack paths between them, across gateway config, tool permissions, MCP servers, and plugins.”