The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install an external CLI tool from the npm registry (npm i -g opencode-ai@latest) or via a third-party Homebrew tap (anomalyco/tap/opencode).
[REMOTE_CODE_EXECUTION]: The skill executes the opencode binary to perform autonomous coding tasks. This agent can generate and execute code on the host system based on model-generated instructions, representing a dynamic execution risk.
[COMMAND_EXECUTION]: The skill relies on the terminal() and process() tools to run various shell commands for setup, authentication, and task execution.
[DATA_EXFILTRATION]: The instructions suggest attaching project files, including potentially sensitive configuration or example environment files (e.g., .env.example, config.yaml), to the opencode CLI, which transmits this content to external AI service providers.
[PROMPT_INJECTION]: The skill processes untrusted external data, creating a surface for indirect prompt injection. Ingestion points: Pull request data via opencode pr and local files attached via the -f flag. Boundary markers: None are specified to prevent the agent from executing instructions embedded within the code or PRs. Capability inventory: The skill can execute shell commands and modify the filesystem via the opencode CLI. Sanitization: No validation or sanitization of the external content is performed before processing.

opencode