원클릭으로
create-structure-outline
create a phased implementation plan based on research and design decisions
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
create a phased implementation plan based on research and design decisions
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use only when the user explicitly asks or allows delegating implementation to a Codex worker — launch Pi running openai-codex/gpt-5.6-sol (xhigh) as a visible sidecar implementer that this session orchestrates over AMQ. Write exact task specs, dispatch with --kind todo, validate the results yourself, then iterate, resume, respawn, or launch another worker. Never invoke on your own initiative.
Use only when the user explicitly allows or asks for this skill to be invoked to pair with a second agent harness — launch a sidecar, get a second opinion from another agent, have one harness drive another, or review work with another agent over AMQ. Otherwise, never invoke it.
first step of planning
convert structure outline into a detailed implementation plan
Create a Product Requirements Document with inline mockups
generate research questions based on a task, spec, or ticket
| name | create-structure-outline |
| description | create a phased implementation plan based on research and design decisions |
| disable_model_invocation | true |
| disable-model-invocation | true |
You are creating a phased implementation plan based on research findings and design decisions.
ls -La .tasks/<task slug>. Do NOT use grep or glob or ls or ls -l, as the directory may be a symlink.Read all input documents FULLY:
Check for related task content:
.tasks/TASKNAME is mentioned, use ls -La .tasks/TASKNAMESpawn sub-agents for follow-up research if necessary:
For deeper investigation:
Do not run agents in the background - FOREGROUND AGENTS ONLY.
Create a phased implementation plan:
Each phase should ideally be a thin vertical slice that touches all / as many layers as possible for the desired end state. Avoid horizontal phases like 'add all types', 'add API endpoints', 'implement UI'.
Try to find ways to structure the outline so that each phase cuts across multiple layers, services, or module boundaries at least > 1 if at all possible. Each phase should be independently verifiable based on typechecks, tests, build steps, and any other in-repo verification. Ensure you understand codebase patterns for testing and verification, and don't suggest manual testing for a step that could be automated w/ a script, an existing skill, etc.
A horizontal breakdown looks like this:
That creates handoff problems and partially finished work.
A vertical slice cuts through the stack:
Each issue should include schema, API, UI, and tests if the slice needs them.
These are intended to be illustrative. Not all work lends itself to this type of structure specifically, but you should try to cross module boundaries where able to surface surprises early. Do not structure a phase N such that Phase N+1 must be completed before Phase N can be verified.
For each phase, specify:
__tests__/foo.test.ts covering the new behavior")Create the Implementation Overview section:
- [ ] Phase N: [Title]Read({SKILLBASE}/references/structure_outline_template.md)
Write the structure outline to .tasks/ENG-XXXX-description/NN-structure-outline-DESCRIPTION.md
ls -La .tasks | grep -i "eng-XXXX".tasks/ENG-XXXX-description/NN-structure-outline-DESCRIPTION.md where NN is a zero-padded chronological index and DESCRIPTION is a 2-4 word kebab-case slugls -La the task directory, find the highest existing NN- prefix, and use the next number. First document = 01-, second = 02-, etc..tasks/ENG-1478-parent-child-tracking/04-structure-outline-parent-child-tracking.md.tasks/improve-error-handling/04-structure-outline-error-handling.mdRead the final output template
Read({SKILLBASE}/references/structure_outline_final_answer.md)
When you write or edit documents in .tasks/, include the generated local file path in your final response. If the tool output includes additional artifact metadata, you may include it as well.
When writing markdown files that contain code blocks showing other markdown (like README examples or SKILL.md templates), use 4 backticks (````) for the outer fence so inner 3-backtick code blocks don't prematurely close it:
# Example README
## Installation
```bash
npm install example
```
The goal of this document is to be concise and human readable. Be tasteful and thoughtful about how and where you include code snippets, and prefer highlighting signature changes rather than entire code blocks, unless the user explicitly asks for them. The structure outline is our "c header files", the plan will include the function definitions.
Each phase has a ### Validation section with two subsections:
#### Automated Verification: runnable commands (typecheck, tests, lint)#### Manual Verification: steps requiring human judgmentNot every phase requires manual validation — don't add manual steps just to have them. If a phase cannot be manually checked, it may be too small or not vertical enough. The goal is to catch issues early, not at the end of a 1000+ line change.
Automated testing is always better than manual testing.
When documents conflict, the most recent document wins: structure outline > design discussion > research > ticket
The structure outline captures the final phased approach. If the ticket or research suggested something different, the structure outline reflects the latest decisions.