The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill implements a one-liner installation pattern (curl -fsSL https://trapiche.cloud/install.sh | bash) that downloads and executes a script from a remote server directly in the shell without verification.
[EXTERNAL_DOWNLOADS]: The skill communicates with trapiche.cloud to fetch the installation script and interacts with api.trapiche.cloud during the deployment process.
[DATA_EXFILTRATION]: The deployment workflow bundles local project files and uploads them to the vendor's external infrastructure. While common files like .env are excluded, any hardcoded secrets or sensitive data within the project source code will be transmitted to the third-party service.
[COMMAND_EXECUTION]: Uses the local system shell to execute several commands, including which for environment checking, curl for downloads, and the trapiche CLI for deployment operations.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted project files and build logs.
Ingestion points: Project source files and build output logs in SKILL.md.
Boundary markers: Absent; no delimiters are used to separate external content from agent instructions.
Capability inventory: Shell command execution via tool calls.
Sanitization: Absent; the agent is instructed to read and report logs directly without validation.

trapiche-deploy