원클릭으로
engineering-architecture-review
Conduct a structured architecture review for a proposed design, system, or technical approach
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Conduct a structured architecture review for a proposed design, system, or technical approach
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
Draft an offer letter with compensation details and terms
Create structured interview plans with competency-based questions and scorecards. Triggers on "interview plan for [role]", "interview questions for [role]", "scorecard for [role]"
Headcount planning, org design, and team structure optimization. Triggers on "org planning", "headcount plan", "team structure", "reorg", "who should we hire next"
Track and manage recruiting pipeline from sourcing through offer. Triggers on "recruiting update", "candidate pipeline", "how many candidates", "hiring status for [role]"
Generate a prioritized daily plan. Triggers on "plan my day", "what's on my plate", "start my day", "morning plan", "daily priorities"
Quick meeting prep for any meeting — context, agenda, and talking points. Triggers on "prep for my meeting", "what's my next meeting about", "meeting context for [topic]"
| name | engineering-architecture-review |
| description | Conduct a structured architecture review for a proposed design, system, or technical approach |
| metadata | {"openclaw":{"tags":["engineering","architecture","design-review"]}} |
When the user asks to review an architecture proposal, evaluate a design doc, or assess a technical approach:
Use capability_execute with the following parameters:
Locate the design document, RFC, or proposal under review. If the user provides a direct link or document name, search for it specifically.
Use capability_execute with the following parameters:
Gather existing architecture decision records and prior designs for the same system or adjacent systems. Understanding what was decided before is essential for evaluating whether a new proposal is consistent or deliberately diverging.
Use capability_execute with the following parameters:
Find related tickets, epics, or tasks that provide requirements context, known constraints, or stakeholder feedback on the design.
Use capability_execute with the following parameters:
Find any discussions, debates, concerns, or alternative approaches that team members have raised about this proposal.
Use capability_execute with the following parameters:
Research how similar architectures have performed at scale, what common pitfalls exist, and what the industry consensus is on the proposed approach.
Use capability_execute with the following parameters:
Understand the operational context: current SLAs, monitoring coverage, deployment procedures, and infrastructure constraints that the proposed architecture must work within.
Format the output as follows:
## Architecture Review
### Proposal Summary
- **Document:** <title and link>
- **Author(s):** <design_doc_authors>
- **System/Component:** <what is being designed or changed>
- **Review date:** <current_date>
- **Reviewer:** <current_user>
### Problem Statement
<restate the problem being solved in 2-3 sentences, as understood from
the design doc>
### Proposed Solution Summary
<concise summary of the proposed architecture, highlighting the key
design decisions>
### Review Dimensions
#### Correctness & Completeness
- Does the proposal solve the stated problem?
- Are there requirements or edge cases not addressed?
- **Assessment:** <meets/partially meets/does not meet>
- **Notes:** <specific observations>
#### Consistency with Existing Architecture
- How does this fit with existing systems and patterns?
- Are there conflicts with prior ADRs?
- **Prior decisions:** <relevant ADRs found and their relationship>
- **Assessment:** <consistent/divergent with justification/conflicting>
- **Notes:** <specific observations>
#### Scalability
- How does the design handle growth in data, traffic, or users?
- What are the scaling limits?
- **Assessment:** <adequate/concerns/insufficient>
- **Notes:** <specific scaling observations>
#### Reliability & Failure Modes
- What happens when components fail?
- Are there single points of failure?
- How does the design handle partial failures?
- **Assessment:** <robust/adequate/fragile>
- **Notes:** <specific failure scenarios to consider>
#### Operability
- Can this be deployed incrementally or does it require a big-bang migration?
- How is it monitored and debugged?
- What is the rollback strategy?
- **Current SLAs:** <relevant SLAs for affected services>
- **Assessment:** <operable/concerns/difficult>
- **Notes:** <specific operational observations>
#### Security
- What is the trust boundary model?
- How is authentication and authorization handled?
- Are there data privacy implications?
- **Assessment:** <adequate/needs review/concerns>
- **Notes:** <specific security observations>
#### Cost & Complexity
- What is the implementation complexity relative to the problem?
- Are there simpler alternatives that were considered?
- What is the ongoing operational cost?
- **Assessment:** <proportional/over-engineered/under-invested>
- **Notes:** <observations>
### Industry Context
- <relevant patterns or anti-patterns from research>
- <lessons from similar architectures at other organizations>
- <emerging best practices that apply>
### Stakeholder Feedback Summary
- <key points raised in chat discussions>
- <unresolved debates or open questions from the team>
### Risks
| Risk | Likelihood | Impact | Mitigation |
|------|-----------|--------|------------|
| <risk> | <low/medium/high> | <low/medium/high> | <suggested mitigation> |
### Recommendations
1. **Must address before approval:**
- <blocking issue with specific suggestion>
2. **Should address:**
- <important improvement with rationale>
3. **Consider for future iteration:**
- <non-blocking suggestion for refinement>
### Verdict
<approve / approve with conditions / request revision / reject>
**Rationale:** <2-3 sentences explaining the overall assessment>
### Suggested Next Steps
1. <specific action with owner>
2. <specific action with owner>
3. <timeline recommendation>
If certain review dimensions cannot be fully assessed due to missing information, state what is missing rather than skipping the dimension.