The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a shell script that searches for a local utility (gsd-tools.cjs) in multiple platform-specific configuration directories (e.g., .claude, .cursor, .gemini, .cline) and executes it via node.
[EXTERNAL_DOWNLOADS]: If local tools are not found, the skill instructs the user to download the vendor's package using npx -y @opengsd/gsd-core@latest. This refers to an official package belonging to the skill's author context ('open-gsd').
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as its primary function is to ingest and process external plan files.
Ingestion points: External plan files specified by the --from argument enter the agent's context (SKILL.md).
Boundary markers: The instructions do not define explicit delimiters or warnings to treat the imported file content as untrusted data.
Capability inventory: The skill has access to powerful tools including Write, Edit, Bash, and Agent, which could be exploited by instructions hidden in imported files.
Sanitization: There is no mechanism described for sanitizing or validating the content of external plan files before they are processed by the agent.

gsd-import