The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides installation instructions in SKILL.md that download and execute a script from a remote URL using curl -sSL https://raw.githubusercontent.com/pimalaya/himalaya/master/install.sh | PREFIX=~/.local sh. This allows for arbitrary code execution from a third-party source.
[CREDENTIALS_UNSAFE]: The skill requires access to ~/.config/himalaya/config.toml in SKILL.md and references/configuration.md, which contains sensitive IMAP and SMTP credentials. The documentation also provides examples of potentially unsafe plain-text passwords.
[DATA_EXFILTRATION]: The skill provides the ability to read and send emails, which allows the agent to access private messages and transmit data to external recipients as part of its core functionality.
[PROMPT_INJECTION]: The skill processes untrusted email content from the himalaya message read command, creating a surface for indirect prompt injection.
Ingestion points: Raw email bodies are read into the agent's context in SKILL.md.
Boundary markers: None provided to separate untrusted email content from agent instructions.
Capability inventory: The skill can send emails, download files, and manage folders via the himalaya CLI.
Sanitization: There is no filtering or validation of the email content before it is processed by the agent.

himalaya