From olam
Dispatch a task prompt to a running Olam world's Claude Code agent. Use when the user says "dispatch", "send task to world", "tell world X to do Y", "run this prompt in world X", or wants the agent inside a specific world to pick up new work. Operates on already-running worlds; for creating a new world use `olam-create`.
How this skill is triggered — by the user, by Claude, or both
Slash command
/olam:dispatch [world-id] [prompt][world-id] [prompt]This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Send a task prompt to a running world's Claude Code agent.
Send a task prompt to a running world's Claude Code agent.
Parse arguments: first token is the world ID, rest is the prompt.
If no world ID provided, call olam_list to show available worlds and ask which to dispatch to.
If no prompt provided, ask the user what task to dispatch.
Call olam_dispatch with:
worldId — the world IDprompt — the task descriptionPresent results showing:
For CF-hosted worlds, dispatch requires in-container auth to have
completed first (OAuth Claude or device-code OpenAI, depending on
agent). If the caller hits a 401 from the container's /dispatch,
the operator must open $OLAM_WORKER_URL/sandbox/<id>/ and complete
auth, then retry.
The task field on olam_create triggers auto-dispatch after auth
completes, which avoids this roundtrip on world creation. Reference:
docs/CF_WORLDS_SPEC.md §4.3 + §5.
401 from container /dispatch (CF mode) — in-container auth incomplete. Tell the operator: open $OLAM_WORKER_URL/sandbox/<id>/, complete the OAuth flow, retry the dispatch.olam_dispatch returns "world not running" — call olam_status to confirm state. The world may have been destroyed or its container exited.olam_status and the world's dashboard.cost value or turn count — surface this to the operator so they can interrupt early via olam_destroy if the world is in a loop.olam-create. Dispatch requires an existing running world.olam-observe (separate verb; dispatch returns when the agent finishes a turn, observe streams the trace).olam-crystallize.olam-create — for new worlds.olam-list — find a world to dispatch to.olam-status — confirm a world is dispatch-ready.olam-crystallize — save the agent's thought graph after dispatch.npx claudepluginhub pleri/olam-cliProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.