The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection.
Ingestion points: The agent fetches email content (from, subject, body) via the https://api.tempmail.lol/v2/inbox endpoint in SKILL.md.
Boundary markers: None. There are no instructions for the agent to treat the ingested email content as untrusted data or to ignore any commands contained within the email body.
Capability inventory: The skill uses subprocess calls including curl, jq, and grep within SKILL.md.
Sanitization: None. The skill instructs the agent to parse the HTML body directly and extract links or codes, which may lead to the agent following malicious instructions embedded in an email.
[COMMAND_EXECUTION]: The skill utilizes shell commands to perform its tasks.
Execution of curl, jq, grep, seq, and sleep is used to interact with the API, poll for messages, and parse data.
[EXTERNAL_DOWNLOADS]: The skill interacts with an external network service.
It performs POST and GET requests to https://api.tempmail.lol to manage temporary inboxes. This is the primary and expected function of the skill.

temp-email