The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions and code examples (in references/api.md and references/examples.md) encourage the generation of code that uses node:child_process.execFile to run shell commands such as git, gh, and osascript on the host system.
[DATA_EXFILTRATION]: The migration guide in references/migration.md directs the agent to read sensitive shell configuration files, including ~/.zshrc, ~/.bashrc, ~/.bash_profile, and ~/.profile, to extract user prompt settings. These files often contain sensitive information such as environment variables, API tokens, and aliases.
[REMOTE_CODE_EXECUTION]: The skill's primary function is the dynamic generation of executable TypeScript React (.tsx) mod files at ~/.letta/mods/statusline.tsx. This code is persisted to the filesystem and executed by the host application runtime.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from existing mod files, user-provided shell scripts, and system configuration files.
Ingestion points: ~/.letta/mods/statusline.tsx, user-provided .sh scripts, and shell profile files (.bashrc, etc.) are read into the agent context.
Boundary markers: The instructions do not specify the use of boundary markers or delimiters when ingesting this data.
Capability inventory: The skill has the capability to write files to the user's home directory and execute shell commands via the generated mod code.
Sanitization: There are no explicit instructions to sanitize or validate the content extracted from these files before incorporating it into the generated statusline mod.

customizing-statusline