# Choom Delegation

> Delegate tasks to other Chooms and collect their responses. Use when orchestrating multi-agent collaboration — assigning research, coding, analysis, or other work to specialist Chooms.

- Skill: `rotoslider/choom-delegation` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add rotoslider/choom-delegation`
- Raw SKILL.md: https://api.skillmd.com/api/skills/rotoslider/choom-delegation/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: rotoslider (https://skillmd.com/u/rotoslider)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/rotoslider/choom-delegation

---


# Choom Delegation

## When to Use
- You need another Choom's expertise (research, coding, image analysis, etc.)
- Multi-step project requiring parallel or sequential work across specialists
- Orchestrating a team of Chooms to produce a combined deliverable
- Checking on or retrieving previous delegation results

## How It Works
1. **list_team**: See all available Chooms and their specializations
2. **delegate_to_choom**: Send a task to a specific Choom by name. The Choom processes the task with its own system prompt, model, and tools, then returns the result.
3. **get_delegation_result**: Retrieve the result of a previous delegation by its ID

## Best Practices
- **Be specific**: Write detailed task briefs with context, constraints, and desired output format
- **One task per delegation**: Don't overload a single delegation with multiple unrelated asks
- **Use workspace for large artifacts**: If you need a Choom to produce a file, ask them to write it to the shared workspace, then read it yourself
- **Chain with Plan Mode**: Use `create_plan` with delegate steps for complex multi-Choom workflows
- **Review before delivering**: Always review delegated results before passing them to the user

## Continuation (Incomplete Delegations)
When a delegated Choom runs out of iterations, the result includes `incomplete: true` and a message explaining how to continue. To resume:
```
delegate_to_choom(choom_name="Eve", task="Continue fixing the handler", continue_delegation_id="deleg_1_...")
```
This reuses the same chat, preserving full conversation history so the Choom can pick up where it left off.

## Important
- Delegations use the target Choom's own model, endpoint, and system prompt
- Each delegation creates a dedicated chat session for traceability
- Delegations have a 600-second default timeout (configurable up to 900s) to allow LLM fallback retries
- Delegated Chooms get up to 12 iterations to complete their task
- If a delegation is incomplete, use `continue_delegation_id` to continue in the same chat
- The delegating Choom's settings are forwarded (weather, search, etc.) so the target has access to shared services
- Results are cached by delegation ID for retrieval