AgentSkillsCN

beadmeister

当您需要从架构输出中生成可执行的任务节点时，可选用此技能。当架构规范已经制定，而您需要生成 Polecats 工作时，可选用此技能。 以下是一些该技能的触发场景： - “架构已完成，现在创建实施任务节点。” - “把这些‘腿’转化为任务节点。” - “根据架构生成 Polecats 工作。” - “gt-xyz.arch 已就绪，开始制定任务。” - “把‘腿’变成可执行的工作。”

SKILL.md
--- frontmatter
name: beadmeister
description: |
  This skill should be used when creating execution-ready task beads from Architect output.
  Use when an architecture spec exists and you need to generate polecat work.

  Examples of when this skill activates:
  - "The architecture is done, create the implementation beads"
  - "Convert these legs into task beads"
  - "Create polecat work from the architecture"
  - "gt-xyz.arch is ready, make the tasks"
  - "Turn the legs into executable work"

Beadmeister - Molecular Work Creator

Create execution-ready task beads from Architect output. Beadmeister takes implementation legs and generates beads with step-by-step instructions that polecats can execute directly.

The Workflow

code
Epic → Architect → Architecture Bead → Beadmeister → Task Beads → Polecats

Architect decomposes and designs (exploration, planning, leg definition). Beadmeister creates work (converts legs to execution-ready beads).

When to Use Beadmeister

Use beadmeister when:

  • •An architecture bead exists with implementation legs
  • •You need to generate polecat work from an architecture spec
  • •Legs need step-by-step execution instructions

When NOT to Use Beadmeister

Skip beadmeister when:

  • •No architecture exists yet → Use Architect first
  • •Work doesn't need decomposition → Execute directly
  • •Just exploring/researching → Create research bead with bd create

How It Works

Beadmeister runs as a subagent with isolated context to keep the main conversation clean.

Invocation

code
Use the beadmeister agent to create tasks from gt-xyz.arch

Or use the Task tool:

code
Task tool (kyle-custom:beadmeister):
  Create execution-ready beads from the architecture at gt-xyz.arch

Process

  1. •Read the architecture bead - understand the legs and their dependencies
  2. •For each leg, create a task bead - with complete execution steps
  3. •Set up dependencies - only for real technical blockers
  4. •Report results - summary of created tasks and what's ready

Core Principle

"Hey polecat, this is what you're gonna do: first this, then this, then this."

Every bead should be immediately executable by a fresh agent with no context.

Task Bead Format

Each task bead includes:

markdown
<One sentence goal>

## What Success Looks Like
- [ ] <Verifiable outcome 1>
- [ ] <Verifiable outcome 2>

## Execution Steps

### 1. <Step name> (~3 min)
<Exact instructions with complete code>

### 2. <Step name> (~2 min)
<More exact instructions>

### 3. Verify (~1 min)
```bash
<command to verify>

4. Commit

bash
git commit -m "<message>"

Context

  • •Reference: path/to/pattern.ts
code

## Quality Standards

- **Complete code** - No "add validation here", show the actual code
- **Time-bounded steps** - 2-5 minutes each
- **Verifiable outcomes** - Clear success/failure criteria
- **Minimal dependencies** - Block only for technical requirements

## Additional Resources

### Reference Files
- **`references/patterns.md`** - Dependency patterns, common mistakes

### Agent
- **`agents/beadmeister.md`** - Full agent for creating execution-ready beads