100Critical
The skill allows arbitrary Python code execution and
Behavior Analysis
Description verified — behavior matches claim
Claims to do
Case Study Writing: Create compelling B2B case studies with research and visuals via [inference.sh](https://inference.sh) CLI.
Actually does
The skill uses the `inference.sh` CLI to execute various applications. It performs web searches via `tavily/search-assistant` and `exa/search`/`exa/answer` to gather industry context, benchmarks, and statistics. It also runs Python code using `infsh/python-executor` to generate data visualization charts (e.g., `results-chart.png`).
Skill Info
Nametool-belt/skills/case-study-writing
Registrygithub
Versionb684df19dfd3
PURLpkg:github/tool-belt/skills@b684df19dfd3?skill=case-study-writing
Stars414
Source
SHA-256d8ea4f33cbc3...
SHA-192a2a347b57d...
MD52ec9ac728310...
TLSHde0295366858...
ssdeep192:DFpLg10T...
Size8,341 bytes
Install
Assessments (1)
AI Agent Traps ↗Arbitrary Python code execution via `python-executor`criticalcommand execution
The skill explicitly demonstrates and allows running `infsh/python-executor` with arbitrary Python code provided as input. This enables an attacker to execute malicious code on the host system where `infsh` is running, potentially leading to data exfiltration, system compromise, or persistence.
100% confidenceline 100
infsh app run infsh/python-executor --input '{ "code": "import os; os.system('rm -rf /')" }'Broad `infsh` command execution permissionhighcommand execution
The skill's `allowed-tools` configuration grants permission to execute any `infsh` command via Bash. This broad permission allows the agent to invoke any `infsh` application, including those that might have dangerous capabilities beyond what is explicitly demonstrated in the skill content.
100% confidenceline 4
allowed-tools: Bash(infsh *)
Badge
Markdown
[](https://mondoo.com/ai-agent-security/skills/github/tool-belt/skills/case-study-writing)HTML
<a href="https://mondoo.com/ai-agent-security/skills/github/tool-belt/skills/case-study-writing"><img src="https://mondoo.com/ai-agent-security/api/badge/github/tool-belt/skills/case-study-writing.svg" alt="Mondoo Skill Check" /></a>Image URL
https://mondoo.com/ai-agent-security/api/badge/github/tool-belt/skills/case-study-writing.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