The Agent Skills Directory

[SAFE]: The skill is purely instructional and defines a coordination logic for multi-agent systems. It does not contain any prompt injections, obfuscation, or unauthorized access patterns.- [DATA_EXPOSURE]: The skill references local paths for storing reports (~/.dh/projects/{slug}/reports/), but this is part of a structured project management workflow (SAM) and does not involve sensitive system files or credential exfiltration.- [INDIRECT_PROMPT_INJECTION]: The skill defines a process where the orchestrator agent reads and synthesizes summaries and reports produced by sub-agents. While this represents a data ingestion surface from external (sub-agent) sources, the instructions are focused on logical flow and status checking rather than unsafe execution or interpolation of untrusted strings into privileged commands.

dispatch