The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains detailed instructions directing the agent to bypass security controls and permission prompts. It recommends using platform-specific flags to disable security guards and explicitly lists shell syntax patterns to avoid in order to evade detection by static security analyzers. Additionally, the skill's reliance on processing untrusted external OpenAPI documents creates an attack surface for indirect prompt injection.
[REMOTE_CODE_EXECUTION]: The skill uses npx to download and execute the latest version of external packages from the NPM registry at runtime. This introduces a supply chain risk where the agent runs code that was not part of the original skill package.
[COMMAND_EXECUTION]: The skill implements a workflow that involves the agent generating Python scripts at runtime and subsequently executing them on the local filesystem. This dynamic code generation and execution pattern is used to modify the target OpenAPI documents but presents a significant security risk if the generated code is influenced by untrusted input.
[EXTERNAL_DOWNLOADS]: The skill fetches files from arbitrary external URLs provided by the user and downloads Docker images from container registries during operation.
[DATA_EXFILTRATION]: The skill handles multiple sensitive API keys (including for Jentic and various AI providers) as environment variables and transmits portions of the input OpenAPI specifications to external LLM services as part of its improvement logic.

jentic-api-improve