Security audit
Security checks for vulnerabilities and agentic risk
This skill is a plain Markdown guide for multi-agent AI workflows; its main risks are broad activation and privacy-conscious logging, not hidden or malicious behavior.
Install this if you want structured help coordinating AI agents, prompts, and evals. Use it deliberately for orchestration tasks, review any Bash or write/edit actions before execution, and avoid retaining full interaction logs unless sensitive data is redacted and retention is controlled.
The implicit activation criteria are broad enough that this skill could be invoked for many generic complex tasks, causing over-selection of a high-privilege orchestration skill. Because the skill allows Bash, Write, and Edit, unintended activation can expand tool access and route user data or repository context through multi-agent workflows unnecessarily.
The instruction to 'Log all interactions' encourages retention of prompts, outputs, and possibly code, secrets, credentials, or personal data in plain language. In an AI orchestration context, aggregated multi-agent logs can centralize sensitive material from several steps, increasing breach impact and creating privacy, compliance, and insider-risk exposure.
Several explicit trigger phrases are still generic and may overlap with ordinary development or prompt-writing requests. This can lead to accidental activation of the orchestration skill in situations that do not need broad tooling or sub-agent coordination, increasing the chance of unnecessary context exposure or risky actions.
No suspicious patterns detected.
