This skill should be used when the user mentions "ClickUp task", "CU-xxx", "task workflow", "clickup integration", or when working with task plugin commands (/task:start, /task:done, /task:merge, /task:status, /task:new). Provides workflow patterns for ClickUp task management integrated with kas.
From tasknpx claudepluginhub kastheco/claude-plugins --plugin taskThis skill uses the workspace's default tool permissions.
references/setup.mdreferences/subagent-prompts.mdreferences/task-templates.mdDesigns and optimizes AI agent action spaces, tool definitions, observation formats, error recovery, and context for higher task completion rates.
Implements structured self-debugging workflow for AI agent failures: capture errors, diagnose patterns like loops or context overflow, apply contained recoveries, and generate introspection reports.
Compares coding agents like Claude Code and Aider on custom YAML-defined codebase tasks using git worktrees, measuring pass rate, cost, time, and consistency.
This skill provides workflow patterns for managing ClickUp tasks integrated with kas workflow commands. The core principle: delegate all ClickUp API calls to the clickup-task-agent to keep the main conversation context clean.
NEVER call ClickUp MCP tools directly in main context.
All ClickUp operations delegate to the clickup-task-agent subagent which returns concise summaries. This prevents verbose API responses from polluting the conversation.
| /task Command | Skills/Commands Used | ClickUp Actions |
|---|---|---|
/task:start CU-xxx | superpowers:brainstorming → kas:review-plan (loop) → superpowers:subagent-driven-development OR superpowers:executing-plans | Fetch task, set "in progress", assign |
/task:done | /kas:verify → /kas:done → PR | Set "ready for review", comment PR |
/task:merge | /kas:merge | Set "complete" |
/task:status [id] | None | Fetch and display status |
/task:new | None | Create task via interview |
/task:start uses superpowers-driven planning:
superpowers:brainstorming - explores, asks questions, outputs designkas:review-plan - max 5 iterations until APPROVED or user overrides/abortssuperpowers:subagent-driven-development (same session)superpowers:executing-plans (separate session with worktree)Activate this skill when detecting:
https://app.clickup.com/t/...CU-xxx, #xxx, task xxxWhen needing ClickUp data, spawn the clickup-task-agent:
Task tool with subagent_type="task:clickup-task-agent":
"[Operation description]
1. [MCP tool call 1]
2. [MCP tool call 2]
Return ONLY: [concise format]"
The agent uses haiku model for speed and returns formatted summaries.
feat/CU-<id>-<slug> or fix/CU-<id>-<slug>CU-<task-id> for ClickUp linking[CU-xxx] prefix| Scenario | Action |
|---|---|
| Fetch fails at start | Stop, show error, suggest retry |
| Status update fails after kas success | Warn user, show manual command |
| kas command fails | Stop, do NOT update ClickUp |
Partial failures: If kas succeeds but ClickUp API fails, warn user and provide manual update command. Don't fail the entire operation for ClickUp sync issues.
For detailed patterns and prompts:
references/subagent-prompts.md - Copy-paste subagent prompts for each operationreferences/setup.md - ClickUp authentication setupreferences/task-templates.md - Task description templates