The Agent Skills Directory

[SAFE]: The skill operates entirely within the project's local file system (specifically the brand/ directory) and uses standard, non-malicious methods for competitive research.- [PROMPT_INJECTION]: No attempts to override system safety guidelines or extract system prompts were found. The 'Safety Rules' section explicitly reinforces data integrity and forbids fabrication or cross-project data contamination.- [DATA_EXFILTRATION]: No exfiltration patterns detected. The skill reads local brand files and writes to a specific competitors.md file within the user's workspace. It utilizes the external Exa MCP for web research, which is a standard tool for this purpose and not a hidden data leak vector.- [REMOTE_CODE_EXECUTION]: No remote code execution patterns, shell commands, or package installations were identified. The skill focuses on natural language processing and information synthesis.- [OBFUSCATION]: Analysis of all text, including metadata and reference files, revealed no hidden characters, Base64 encoding of commands, or homoglyph substitutions.- [CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or private file path access (like .ssh or .aws) were found. Recommendations for secret management (like .env files) were not present but also not violated.

competitive-intel