The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and import external skill definitions (SKILL.md files) from a cloud library, which are then processed as instructions by the agent. This creates a surface for indirect prompt injection where the external content could potentially influence agent behavior.
Ingestion points: cloud_browse_skills(action="import_skill", ...) fetches content to a local path specified in SKILL.md.
Boundary markers: None specified; the agent is instructed to read the imported file and follow its instructions directly.
Capability inventory: Tools include search_skills, cloud_browse_skills, and cloud_auth_flow.
Sanitization: No explicit validation or sanitization of the imported markdown content is defined.
[COMMAND_EXECUTION]: The skill provides documentation for the user to execute the openspace-cloud-auth command-line utility for manual agent bootstrapping.
[CREDENTIALS_UNSAFE]: The documentation references an authentication flow requiring a password but follows security best practices by instructing the agent to only collect passwords via secure secret input mechanisms and providing a local CLI alternative to avoid passing secrets through the LLM context.

skill-discovery