The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (MEDIUM): The skill uses npx to download and run the skills package and facilitates the installation of further remote code via npx skills add. These downloads originate from inference.sh, which is not included in the 'Trusted External Sources' list, making them unverifiable dependencies.
[COMMAND_EXECUTION] (MEDIUM): The skill defines a broad execution scope Bash(npx skills *). This allows the agent to perform any subcommand, including adding, updating, or removing software on the system, which poses a risk if the agent is manipulated into installing a malicious package.
[INDIRECT_PROMPT_INJECTION] (LOW): This skill has a significant attack surface for indirect injection.
Ingestion points: Data enters the agent context through the output of npx skills search and npx skills list from a remote registry.
Boundary markers: No delimiters or warnings are present to prevent the agent from obeying instructions embedded in search results.
Capability inventory: The skill has the capability to write and execute code via the add and update commands.
Sanitization: There is no evidence of sanitization for registry data. An attacker could poison the registry with malicious skill descriptions to trick the agent into installing a backdoor.

related-skill