The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the bash interpreter to execute a local script located at .opencode/skills/task-management/router.sh for managing task routing and monitoring execution status.
[INDIRECT_PROMPT_INJECTION]: The instructions create an attack surface by directing subagents to read and implement logic from local files which could contain untrusted data. * Ingestion points: File content from .tmp/tasks/feature/subtask_01.json and markdown files within .opencode/agent/subagents/ are read in Step 2 and the Example section. * Boundary markers: The prompts provided to the CoderAgent lack delimiters or explicit instructions to treat file content as data rather than instructions. * Capability inventory: The agent possesses capabilities to execute shell scripts and perform file system operations through the CoderAgent subagent. * Sanitization: No content validation or sanitization processes are defined for the data ingested from the local file system.

parallel-execution