The Agent Skills Directory

[NO_CODE]: The skill consists entirely of instructional markdown and configuration YAML. It contains no executable scripts (e.g., Python, Bash, JavaScript) or external binaries.\n- [PROMPT_INJECTION]: The instructions are designed to restrict and control agent behavior through a structured orchestration graph. It explicitly forbids sub-delegation and limits graph depth to prevent uncontrolled agent spawning. There are no attempts to bypass safety filters or override system instructions.\n- [DATA_EXFILTRATION]: No network operations or sensitive file access patterns were found. The skill emphasizes minimizing data sharing by sending only essential, self-contained context packets to child agents.\n- [EXTERNAL_DOWNLOADS]: The skill does not reference or download any external packages, libraries, or scripts.

harness-orchestrate