tasks

Orchestrates substantial MDE Linear tasks from planning through post-merge verification.

1|Updated Sep 15, 2026
One-click install
npx skills add https://github.com/amoai-tech/mdeai --skill tasks-amoai-tech
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tasks
Source: https://github.com/amoai-tech/mdeai/tree/main/.claude/skills/tasks
Command: npx skills add https://github.com/amoai-tech/mdeai --skill tasks-amoai-tech

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Substantial engineering tasks often lose scope, evidence, and resumability across long agent sessions and PR cycles. This Skill turns MDE Linear issues into dependency-ordered, evidence-backed execution runbooks that any agent can resume and verify. ## Core Features & Use Cases - Task orchestration: Classifies work as S0–S4 complexity, builds dependency-aware step graphs, and persists resumable state in Linear. - Execution standards: Enforces checkpoints, Definition of Done, pre-commit gates, risk-matched pre-merge tests, exact-head CI, and post-merge verification. - Domain routing: Directs Supabase, Mastra, CopilotKit, Cloudinary, and Next.js findings to the owning domain skill with the strongest available evidence. - Use Case: When starting a cross-system SAN task touching Supabase RLS and a Mastra workflow, use this Skill to produce the Linear runbook, execute each checkpoint with proof, resolve PR review comments, and certify the merged outcome. ## Quick Start Ask the agent to plan and execute a substantial MDE Linear task using the tasks skill, starting with a current-state audit and a Top Task Snapshot.

Frequently Asked Questions about tasks

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I structure a substantial Linear task for AI agent execution?

Use the mandatory task structure: Top Task Snapshot, current-state audit, user outcome, scope, dependencies, Definition of Done, and an ordered implementation runbook. Each file group ends with a named Checkpoint block containing success criteria, verification proof, and STOP conditions.

How does the tasks skill classify task complexity?

It classifies work as S0 trivial, S1 focused, S2 coordinated, S3 substantial cross-system, or S4 production-sensitive risk. S0 skips the orchestrator, S1 uses one execution path, and S2–S4 require a dependency-aware step graph with resumable Linear state.

When should a task be split into multiple Linear tickets?

Split using vertical tracer-bullet slices where each child ticket delivers one narrow, complete, independently verifiable outcome across all layers it needs. For wide mechanical refactors that cannot stay green vertically, use the expand, migrate, contract pattern instead.

Can a task be marked Done right after its PR merges?

No. Merge is not Done; post-merge verification must confirm the merge SHA on origin/main, main CI, deployment health, and the task's observable outcome. Only after all applicable post-merge checks pass can progress reach 100% and Linear move to Done.

How should PR review comments from bots be handled?

Treat every review comment as a hypothesis, not authority. Classify each as valid-blocker, valid-in-scope, stale, incorrect, out-of-scope, or needs-research, verify against current code and domain evidence, then fix, reply with proof, and resolve only after verification.