The Agent Skills Directory

[SAFE]: The skill focuses on infrastructure design and provides clear instructions for secure credential management, tenant isolation, and data redaction. No malicious patterns or suspicious behaviors were detected.\n- [EXTERNAL_DOWNLOADS]: The skill references official Together AI resources, including the Python SDK and the official skills repository on GitHub. These references are standard for implementing the described architecture and originate from a well-known service provider.\n- [INDIRECT_PROMPT_INJECTION]: The skill utilizes the WebFetch tool to ingest current Together AI capabilities and limits from their official documentation. This represents a low-risk data ingestion point consistent with the skill's primary purpose.\n
Ingestion points: WebFetch tool in SKILL.md used to access docs.together.ai.\n
Boundary markers: None explicitly defined for tool output processing, as the data is used for architectural guidance rather than code execution.\n
Capability inventory: The skill uses Read, Glob, Grep, WebFetch, Write, and Edit tools for mapping trust boundaries and generating documentation.\n
Sanitization: No specific content filtering or sanitization is described for the information fetched from documentation.

together-reference-architecture