The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx to download the alignfirst package from the npm registry. This package is a resource associated with the author 'paleo'.
[REMOTE_CODE_EXECUTION]: The skill executes the alignfirst package via npx to retrieve the aad protocol guide. This involves running code fetched from a remote registry.
[COMMAND_EXECUTION]: The skill executes the shell command npx -y alignfirst guide aad to generate instructions for the agent.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to follow instructions generated by an external tool, which introduces a vulnerability to tool output poisoning.
Ingestion points: The output of the npx -y alignfirst guide aad command is ingested into the agent's context as instructions.
Boundary markers: None identified; the skill does not use delimiters to wrap the external instructions or provide warnings to ignore malicious commands embedded in that output.
Capability inventory: The skill directs the agent to "follow" the generated guide, which may trigger any tool or action available to the agent depending on the guide's content.
Sanitization: There is no evidence of validation or sanitization of the tool's output before the agent acts upon it.

al