bias-detector
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists exclusively of markdown documentation, instructions, and examples. It does not contain any code blocks, shell commands, or executable scripts.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided text to identify cognitive biases. While this technically creates a surface for indirect prompt injection (where a user might provide text containing instructions), the skill is not granted any dangerous capabilities or tools (such as network access, file system writes, or code execution) that could be leveraged maliciously. The guidelines also explicitly instruct the agent to validate inputs before processing.
- [EXTERNAL_DOWNLOADS]: The skill contains a reference to a GitHub repository under the author's namespace (inbharatai/claude-skills). This is a static link for documentation purposes and does not involve automated downloads or execution of remote code.
- [METADATA_POISONING]: The metadata fields (name, description, tags) accurately reflect the stated purpose of the skill. Although the author field 'Claude Code' differs from the vendor account name 'inbharatai', this appears to be a naming convention or template artifact rather than an attempt at deception.
Audit Metadata