The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of Harness templates that can execute shell commands via user-provided runtime inputs (using the <+input> syntax).
Ingestion points: User-provided parameters and descriptions during the template generation process described in SKILL.md.
Boundary markers: The instructions lack specific guidance for the agent to use delimiters or security warnings when interpolating user input into template commands.
Capability inventory: The skill uses the harness_create and harness_update MCP tools to write these generated templates to the external Harness platform.
Sanitization: There is no mention of sanitization or validation of the content provided by users before it is placed into executable fields like command within the templates.

create-template