The Agent Skills Directory

[Data Exposure & Exfiltration] (SAFE): The skill manages external communications with trusted domains (tabelog.com and api.openai.com). It uses standard python-dotenv patterns for managing the OpenAI API key, which is common practice for local development. No signs of hardcoded secrets or unauthorized exfiltration were found.
[Remote Code Execution] (SAFE): All project dependencies specified in pyproject.toml are standard, well-known packages. The skill does not download external scripts for execution or use dangerous functions like eval() or exec().
[Indirect Prompt Injection] (LOW): The skill serves as a data retrieval tool that ingests information from an external source (Tabelog). While this ingestion point is a potential surface for indirect prompt injection if the output is later processed by an LLM without sanitization, the skill itself does not perform any unsafe operations with the scraped data.
[Obfuscation] (SAFE): Code analysis confirmed no usage of multi-layer Base64, zero-width characters, homoglyphs, or other techniques designed to hide malicious intent.

mcp-builder