The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill directs the execution of a remote shell script using the pattern curl -fsSL https://temps.sh/deploy.sh | bash. This is a high-risk behavior as it runs unverified code from a remote source directly in the system shell.\n- [COMMAND_EXECUTION]: The instructions frequently include sensitive credentials such as GitHub tokens, Cloudflare API tokens, and AWS access keys as plain-text CLI arguments (e.g., in the temps setup and temps login commands).\n- [CREDENTIALS_UNSAFE]: Passing sensitive tokens as command-line arguments results in their exposure within the shell's command history files (e.g., .bash_history), which is a significant security risk for credential theft.\n- [PROMPT_INJECTION]: The skill provides a platform for processing untrusted external data, which creates a vulnerability surface for indirect prompt injection.\n
Ingestion points: The agent reads untrusted data from temps logs (streaming deployment output), git clone (external repository content), and temps env import (environment files).\n
Boundary markers: The instructions lack markers or warnings to ignore instructions that might be embedded in the processed data.\n
Capability inventory: The skill allows for various powerful actions, including executing shell commands (temps, docker, npm, git, curl) and performing network and file system operations.\n
Sanitization: There are no instructions for sanitizing or validating external content before it is processed or displayed in the agent's context.

temps-platform-setup