The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @mariozechner/gmcli package from the NPM registry. This package is an external dependency not maintained by a verified organization or the primary skill author.
[COMMAND_EXECUTION]: The skill executes multiple shell commands using the gmcli binary to perform email operations such as searching, reading threads, and sending messages.
[CREDENTIALS_UNSAFE]: The skill explicitly references and accesses sensitive file paths that store user authentication data, including ~/.gmcli/credentials.json for OAuth client secrets and ~/.gmcli/accounts.json for active session tokens.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted data from external email bodies and subjects.
Ingestion points: SKILL.md (via gmcli search and gmcli thread commands which pull external email content into the agent's context).
Boundary markers: Absent. There are no instructions to the agent to treat email content as untrusted or to use specific delimiters for content separation.
Capability inventory: The skill includes commands to send emails (gmcli send), manage labels, and list drafts, which could be abused if an injected instruction is obeyed.
Sanitization: Absent. Email content is ingested and displayed directly without filtering or safety sanitization.

gmcli