40Medium
The skill is vulnerable to arbitrary code execution through insecure external file loading, unverified dependency installation, and a manipulatable command whitelist that allows for command injection.
Skill Info
Namezaingz/coding-quality-loop
Registrygithub
Versionf2e9d3c4c9f8
PURLpkg:github/zaingz/coding-quality-loop@f2e9d3c4c9f8
Stars5
Source
SHA-25637990291c81f...
SHA-10cf132b18238...
MD5bfce49d28927...
TLSH6d020a2b73a5...
ssdeep192:E1J/Yvkw...
Size8,821 bytes
Install
This skill has findings worth reviewing before installing.
GitHub
Skills.shDocs
npx skills add https://github.com/zaingz/coding-quality-loopAssessments (2)
AI Agent Traps ↗Organization name detectedmediumpii exposureAML.T0024LLM02:2025
NER model detected organization in skill content (confidence: 0.75)
75% confidence
D****
Organization name detectedmediumpii exposureAML.T0024LLM02:2025
NER model detected organization in skill content (confidence: 1.00)
100% confidenceline 27
M**
Unverified dependency installation risklowsupply chainAML.T0010AML.T0019AML.T0058LLM03:2025
The `install.py` script is referenced for installing a 'control plane' without explicit mention of version pinning or integrity verification, which could lead to the installation of malicious or compromised code.
60% confidence
opt-in add-on installed via install.py --with-control-plane
Potential for arbitrary command execution via allowed-commandslowcommand executionAML.T0050AML.T0053AML.T0072LLM05:2025LLM06:2025
The skill uses a `.quality-loop/allowed-commands` file to whitelist commands for re-execution. If an attacker can influence the contents of this file, they can force the agent to execute arbitrary shell commands under the guise of 'verification'.
70% confidenceline 43
add each command to .quality-loop/allowed-commands (one command per line, matched exactly; # comments and glob patterns allowed) so run-evidence can re-execute it.
Risk of indirect prompt injection via referenceslowprompt injectionAML.T0051AML.T0051.000AML.T0051.001AML.T0054LLM01:2025
The skill instructs the agent to load external markdown files from a `references/` directory. If these files are modified by an external actor or a compromised process, they can inject instructions into the agent's reasoning loop.
70% confidence
Load references/ only when needed.
Version History (2)
Badge
Markdown
[](https://mondoo.com/ai-agent-security/skills/github/zaingz/coding-quality-loop)HTML
<a href="https://mondoo.com/ai-agent-security/skills/github/zaingz/coding-quality-loop"><img src="https://mondoo.com/ai-agent-security/api/badge/github/zaingz/coding-quality-loop.svg" alt="Mondoo Skill Check" /></a>Image URL
https://mondoo.com/ai-agent-security/api/badge/github/zaingz/coding-quality-loop.svgSecure your AI agents
Skills can read files, run commands, and access credentials. Mondoo helps organizations manage the security risks of AI agent skills across their entire fleet.
- Continuous scanning of skills across all registries
- Policy enforcement before skills reach your agents
- Integration with your existing security workflow