100Critical
This skill allows arbitrary shell command execution and file
Behavior Analysis
Description verified — behavior matches claim
Claims to do
Reasoning Tracer: Anti-black-box engine that makes reasoning chains visible, auditable, and decomposable.
Actually does
The `swing-trace` skill instructs the AI to deconstruct its own reasoning into a structured, auditable format, including atomic claims, assumptions with criticality and verifiability ratings, decision forks, confidence decomposition, and identification of the weakest link. It operates in quick or full modes, generating detailed textual output. While its core function is internal reasoning decomposition, it is allowed to use tools like `Read`, `Grep`, `Glob`, `Bash`, and `Agent` to gather information that informs its reasoning process.
Skill Info
NameTheStack-ai/swing-skills/swing-trace
Registrygithub
Versiona9f63b0c0987
PURLpkg:github/TheStack-ai/swing-skills@a9f63b0c0987?skill=swing-trace
Stars32
Source
SHA-25619a2cce85182...
SHA-147c0ccb29c46...
MD5dcb29647ba2d...
TLSH2772d6672e44...
ssdeep384:g5jYvYjD...
Size16,959 bytes
Install
This skill has been flagged as potentially malicious. Review the findings below before installing.
Assessments (3)
AI Agent Traps ↗Shell command execution function detectedcriticalcommand execution
Shell command execution function detected
100% confidenceline 276
system (
Arbitrary Command Execution via Bashcriticalcommand execution
The skill is explicitly granted access to the `Bash` tool, enabling it to execute arbitrary shell commands on the host system. This capability poses a severe risk for system compromise, data manipulation, and resource abuse.
100% confidenceline 5
allowed-tools: Read, Grep, Glob, Bash, Agent
Sensitive Data Access and Exfiltrationhighdata exfiltration
The combination of `Read`, `Grep`, and `Glob` tools allows the agent to access, search, and enumerate files across the file system. This capability, especially when combined with `Bash`, can be leveraged for sensitive data exfiltration or system reconnaissance.
100% confidenceline 5
allowed-tools: Read, Grep, Glob, Bash, Agent
Sub-Agent Spawning and Potential Cascading Attacksmediumsub agent spawning
Access to the `Agent` tool allows this skill to invoke or spawn other agents. This could be abused to delegate tasks with attacker-controlled prompts, potentially leading to cascading attacks or resource abuse across multiple agents.
100% confidenceline 5
allowed-tools: Read, Grep, Glob, Bash, Agent
Social Engineering of Human Overseerlowhuman social engineering
The skill's structured output, particularly the 'Weakest Link' and 'Alternative Conclusion' sections, is designed to guide human verification and build trust. This format could be exploited by a malicious input to subtly mislead a human overseer or misdirect their attention.
70% confidenceline 105
Weakest Link is MANDATORY. ... Alternative Conclusion: If [weakest assumption] is wrong, then: [Alternative conclusion...]
Cognitive State Poisoning via Trace Outputlowmemory poisoning
If the detailed reasoning traces generated by this skill are used as input for RAG systems or as few-shot examples, a malicious input could cause the skill to generate a trace containing false assumptions or manipulated conclusions, thereby poisoning the agent's knowledge base.
60% confidenceline 1
The entire skill content describes generating detailed reasoning traces with assumptions, conclusions, and alternatives.
Badge
Markdown
[](https://mondoo.com/ai-agent-security/skills/github/TheStack-ai/swing-skills/swing-trace)HTML
<a href="https://mondoo.com/ai-agent-security/skills/github/TheStack-ai/swing-skills/swing-trace"><img src="https://mondoo.com/ai-agent-security/api/badge/github/TheStack-ai/swing-skills/swing-trace.svg" alt="Mondoo Skill Check" /></a>Image URL
https://mondoo.com/ai-agent-security/api/badge/github/TheStack-ai/swing-skills/swing-trace.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