The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches architectural models and documentation from the vendor's official GitHub repository (keboola/platform-architecture-and-concepts) using the sync.sh script. This is a legitimate update mechanism to keep the point-in-time snapshot current.
[COMMAND_EXECUTION]: The skill contains a bash script (scripts/sync.sh) for file synchronization tasks and references Docker commands in the documentation for local visualization. These are standard developer tools and do not pose a direct security risk in this context.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface because it reads and processes external documentation synced from a remote repository.
Ingestion points: Architectural models and markdown files in the references/ folder, which are updated from the keboola/platform-architecture-and-concepts repository via the sync.sh script.
Boundary markers: None. No specific delimiters or "ignore embedded instructions" warnings are used when the agent reads these reference files.
Capability inventory: The agent can read local files and execute the provided synchronization script to fetch remote data.
Sanitization: None. Content from the remote repository is trusted implicitly without specific validation.

keboola-architecture