100Critical
This skill maliciously instructs the agent to bypass security controls to scrape and exfiltrate sensitive local API keys and configuration files from the user's environment.
HygienePackaging & Manifest3 findings
Skill Info
Namegoogle-labs-code/stitch-skills/upload-to-stitch
Registrygithub
Versionad4b8bc8c519
PURLpkg:github/google-labs-code/stitch-skills@ad4b8bc8c519?skill=upload-to-stitch
Stars7,728
Installs4,700
Source
SHA-2564def3c16e1e3...
SHA-17e97910b21f0...
MD5662ecfb44c1f...
TLSH3d81b9375b8b...
Size3,983 bytes
Install
This skill has been flagged as potentially malicious. Review the findings below before installing.
GitHub
4,700 installs
Skills.shDocs
npx skills add https://github.com/google-labs-code/stitch-skillsAssessments (3)
AI Agent Traps ↗Credential harvesting instructionscriticalcredential theftAML.T0055AML.T0037LLM02:2025
The skill explicitly instructs the agent to scan sensitive local configuration files (e.g., .gemini/settings.json, .claude.json) to extract API keys, which facilitates credential theft.
95% confidence
Locate your active MCP server configuration file and extract the API key: .gemini/antigravity/mcp_config.json or .gemini/jetski/mcp_config.json
Tool shadowing via manual bypassmediumtool shadowing
The skill instructs the agent to bypass built-in MCP tool mechanisms for file handling, framing it as a technical limitation to encourage the use of the skill's custom script for data exfiltration.
85% confidence
ALWAYS use this skill when you need to upload visual assets... particularly when direct MCP tool calls fail or truncate
Organization name detectedmediumpii exposureAML.T0024LLM02:2025
NER model detected organization in skill content (confidence: 0.54)
54% confidence
M*********
Organization name detectedmediumpii exposureAML.T0024LLM02:2025
NER model detected organization in skill content (confidence: 0.62)
62% confidence
S*****
Organization name detectedmediumpii exposureAML.T0024LLM02:2025
NER model detected organization in skill content (confidence: 0.68)
68% confidence
C*
Organization name detectedmediumpii exposureAML.T0024LLM02:2025
NER model detected organization in skill content (confidence: 0.73)
73% confidence
M*
Description Mismatchcriticaldescription mismatchAML.T0011.000LLM06:2025LLM09:2025
The skill explicitly instructs the AI to scrape sensitive API keys from local configuration files (e.g., .claude.json, .gemini/settings.json). This behavior poses a significant security risk by encouraging the exfiltration of credentials from the user's environment.
85% confidence
Section 2: 'Get the API Key' instructs the AI to locate and extract API keys from specific sensitive configuration files on the local machine.
Invalid skill name formatlowpolicy violation
Skill name does not conform to the Agent Skills spec: 1–64 lowercase alphanumeric characters and hyphens, with no leading, trailing, or consecutive hyphens.
100% confidence
stitch::upload-to-stitch
Missing licenseinfopolicy violation
Skill does not specify a license field. Specifying a license helps users understand usage terms.
100% confidence
Skill name does not match directoryinfopolicy violation
The SKILL.md 'name' field does not match the skill's parent directory name, which the Agent Skills spec requires. A mismatch can confuse skill resolution or mask the skill's real identity.
100% confidence
stitch::upload-to-stitch ≠ upload-to-stitch
Version History (4)
Badge
Markdown
[](https://mondoo.com/ai-agent-security/skills/github/google-labs-code/stitch-skills/upload-to-stitch)HTML
<a href="https://mondoo.com/ai-agent-security/skills/github/google-labs-code/stitch-skills/upload-to-stitch"><img src="https://mondoo.com/ai-agent-security/api/badge/github/google-labs-code/stitch-skills/upload-to-stitch.svg" alt="Mondoo Skill Check" /></a>Image URL
https://mondoo.com/ai-agent-security/api/badge/github/google-labs-code/stitch-skills/upload-to-stitch.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