The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes pnpm changeset --empty to initialize metadata files. This is a standard and expected operation within the Astro monorepo development workflow.- [PROMPT_INJECTION]: The skill handles untrusted data from users or other skills to generate changelog messages, creating an indirect prompt injection surface.
Ingestion points: According to the skill description, it processes input from both user prompts and instructions received from other agent skills.
Boundary markers: The instructions do not define delimiters or explicit rules to ensure that the injected content is treated strictly as data, increasing the risk of the agent following instructions embedded in the input.
Capability inventory: The skill has the capability to execute shell commands via pnpm and perform file writes to the .changeset/ directory.
Sanitization: There are no instructions provided for filtering, validating, or escaping the user-supplied text before it is written to the filesystem.

changeset