The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct the agent to fetch content from external URLs using curl and to clone remote repositories using git to use as reference material.
[COMMAND_EXECUTION]: The instructions involve the use of system command-line tools including curl for network requests and git (clone, status) for repository management.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection because it is designed to ingest, read, and 'follow' logic from untrusted external sources provided at runtime.
Ingestion points: User-specified URLs and remote repositories defined in the instructions (SKILL.md).
Boundary markers: The instructions do not provide delimiters or specific warnings to ignore instructions that might be embedded within the external source code.
Capability inventory: The skill assumes the agent has capabilities for file system access, network operations (curl), and version control (git).
Sanitization: No sanitization or validation steps are defined for content retrieved from external sources before the agent incorporates its structure or behavior.

build-from-example