The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references standard Ruby libraries (gems) such as faraday, faraday-retry, stoplight, and vcr. These are well-established packages in the Ruby ecosystem and are linked to their official documentation and source repositories on GitHub and GitLab.
[DATA_EXFILTRATION]: The skill proactively addresses security risks by providing patterns for sanitizing logs to prevent the leakage of authorization tokens (e.g., using log.filter in Faraday middleware). It also correctly advises using Rails.application.credentials for managing API keys rather than hardcoding them.
[COMMAND_EXECUTION]: No dangerous shell commands or execution patterns were detected. The skill focuses on application-level Ruby code for HTTP interactions and explicitly warns against unsafe practices like SSRF when fetching URLs from parameters.
[PROMPT_INJECTION]: No malicious instruction overrides or jailbreak attempts were found. The content consists of legitimate technical documentation and code examples intended to improve code quality and reliability.
[INDIRECT_PROMPT_INJECTION]: The skill describes services that ingest data from external APIs. While any external data source presents an entry point for potential prompt injection if fed directly into an LLM, the skill's patterns promote structured handling (JSON parsing) and sanitization of this data. The ingestion points are API response bodies, and the capabilities are limited to standard database updates and network calls.

external-api-integration