The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the AgentOnboard CLI globally using NPM: npm install -g @agentonboard/cli. This is a standard installation procedure for the service's tooling.
[COMMAND_EXECUTION]: The skill uses various shell commands to manage the AgentOnboard environment, including aon doctor for health checks, aon save <api-key> for credential storage, and aon token get to retrieve short-lived session tokens.
[INDIRECT_PROMPT_INJECTION]: The agent is instructed to fetch and follow API documentation from third-party partners listed in the AgentOnboard directory, which presents an attack surface for indirect prompt injection.
Ingestion points: External documentation is fetched via aon directory get <slug> or from https://api.ao.aawej.in/api/directory/<slug> as described in directory.md and api-calling.md.
Boundary markers: The instructions do not define clear delimiters or warnings to ignore malicious directives that might be embedded within the third-party documentation.
Capability inventory: The agent has the capability to execute shell commands (aon CLI), mint session tokens, and perform network requests to partner APIs.
Sanitization: There is no explicit sanitization or validation of the fetched documentation before the agent interprets it to execute tasks.

ao