The Agent Skills Directory

[PROMPT_INJECTION] (HIGH): The skill is designed to ingest untrusted external data and transform it into executable agent skills, a classic Indirect Prompt Injection surface.
Ingestion points: The skill accepts arbitrary URLs via the /create-llms-for-skills command as seen in SKILL.md.
Boundary markers: None. There are no instructions to the agent to ignore or delimit instructions found within the fetched documentation.
Capability inventory: The workflow culminates in the /create-skills-via-llms command which writes persistent skill files to the filesystem at ~/.claude/skills/ (noted in SKILL.md).
Sanitization: No sanitization or validation logic is defined to prevent the scraped documentation from containing malicious instructions that would then be saved as a new skill.
[EXTERNAL_DOWNLOADS] (HIGH): The skill facilitates downloading and processing content from any URL provided by the user or constructed from crate names.
While documentation sites like docs.rs are common, the workflow explicitly supports 'Custom URLs' which allows an attacker to point the agent to a malicious server hosting a tailored llms.txt payload.
[COMMAND_EXECUTION] (MEDIUM): The skill instructs the agent to execute custom slash commands (/create-llms-for-skills and /create-skills-via-llms) that perform network operations and local file modifications.

rust-skill-creator