The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill manages sensitive authentication material (tokens and E2EE keys) stored in the ~/.linejs directory. The instructions include methods for retrieving these secrets into the agent's context using commands like line auth token and line storage list --values.
[EXTERNAL_DOWNLOADS]: The skill recommends installing the @evex-dev/linejs-cli package from the NPM registry or executing it via npx. These resources are official packages provided by the skill author.
[COMMAND_EXECUTION]: The skill relies on executing the line CLI binary to perform various messenger operations and includes an MCP server tool, execute_linejs_cli, that allows executing command arguments in-process.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources including LINE chat messages, square events, and Timeline posts. While it utilizes structured JSON output to facilitate parsing, it lacks explicit boundary markers or sanitization requirements for message text. The agent possesses capabilities to write messages and perform raw API calls, creating a surface where malicious content from incoming messages could influence subsequent tool use.

linejs-cli