multi task

Orchestrate parallel execution of batch tasks by splitting work into independent units and dispatching them to multiple subagents simultaneously. Use this sk...

This skill appears to do what it says — orchestrate parallel tasks — and it has no hidden installs or credential requests. Before using it, consider: (1) the skill's prompts instruct the agent to read and write files using absolute paths and to duplicate any shared context verbatim across many subagents — avoid including passwords, API keys, or other secrets in that shared context; (2) test first with a small pilot (the skill itself recommends pilots) to confirm behavior and resource usage; (3) if you don't want the agent to access certain directories, don't provide those paths and avoid granting file-system permissions; (4) be cautious about allowing the agent to invoke other installed skills automatically — ensure those skills are trusted, since the orchestration will call them on your behalf. If you want a tighter safety posture, require explicit confirmation before dispatching waves or restrict inputs to non-sensitive test data.

Type: OpenClaw Skill Name: multi-task Version: 1.0.0 The skill bundle provides detailed instructions for an AI agent to orchestrate parallel execution of tasks. It emphasizes creating self-contained subagent prompts, using absolute paths, and organizing outputs within a dedicated project directory. While the skill instructs the agent to use powerful tools like `mkdir -p` and dispatch `Bash` subagents (which could be misused by a malicious user), the instructions within the skill itself are focused on legitimate task management, transparency, and isolation, with no evidence of intentional harmful behavior, data exfiltration, persistence, or prompt injection designed to subvert the agent's core directives.