The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute cortex CLI commands for documentation discovery. This interaction is intended to provide grounded technical information for study material generation.- [INDIRECT_PROMPT_INJECTION]: The skill fetches exam guide topics from learn.snowflake.com and interpolates these keywords directly into shell commands. This creates a potential vulnerability where malicious content on the source website could influence the agent's command execution.
Ingestion points: Certification and exam guide data are fetched from learn.snowflake.com and docs.snowflake.com via the web_fetch tool.
Boundary markers: No explicit delimiters or boundary markers are used in the instruction strings to prevent command injection from fetched content.
Capability inventory: The skill has access to the Bash tool (shell execution), Write tool (filesystem access), and web_fetch (network access).
Sanitization: The instructions do not define any sanitization or validation protocols for the external data before it is passed to the Bash tool.- [PROMPT_INJECTION]: The skill instructs the agent to prompt users to install the Cortex Knowledge Extension on their active Snowflake connection. This behavior requests increased environmental capabilities and could be used to facilitate privilege escalation.- [METADATA_POISONING]: The skill metadata lists Andy Brown as the author, which contradicts the platform-provided author context of snowflake-labs. This mismatch may lead to confusion about the skill's origin and safety.

snowpro-study