The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents an attack surface where the agent processes third-party manifests (alien-agent-id.json) which contain descriptions and property data.
Ingestion points: node CLI discover, node CLI capabilities, and node CLI support fetch remote JSON/HTML data.
Boundary markers: The 'Trust boundary' section explicitly instructs the agent that manifests are 'third-party data, not instructions' and to treat descriptions as 'third-party prose'.
Capability inventory: The skill allows Bash execution of a local Node.js CLI, curl, and jq to perform network operations and data parsing.
Sanitization: The skill warns the agent NOT to pass any manifest fields as shell arguments and notes that the CLI itself rejects cross-authority URL fetching.
[COMMAND_EXECUTION]: The skill relies on executing shell commands via Bash for its primary operations, including calling a local Node.js CLI script and system utilities like curl. These are scoped to the intended purpose of the skill.

agent-id-auth