The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local binary (fluxctl) within the workspace's .opencode/bin/ directory to perform task management operations. It utilizes standard system utilities such as git, jq, and sed to interact with the environment and process command output.
[PROMPT_INJECTION]: The skill accepts user-supplied input through the $ARGUMENTS parameter. To protect against malicious input or command injection, the instructions enforce a specific format validation (e.g., identifiers must begin with the fn- prefix) before the data is utilized in shell commands.

flux-opencode-sync