The Tech Evaluator’s Manual

“There is no best technology stack, only the most suitable one.” — ThoughtWorks Technology Radar

This skill is based on SEI’s ATAM (Architecture Tradeoff Analysis Method) and weighted decision matrix methodologies.

⚠️ Mandatory Deep Thinking

[!IMPORTANT] Before performing the evaluation, you must invoke the mcp_sequential-thinking_sequentialthinking tool and conduct 5–15 steps of reasoning. Example lines of thinking include:

1. •“What are the user’s core needs? Which scenarios must be supported?”
2. •“Which technologies is the team already familiar with? What is the time budget for learning new ones?”
3. •“What are the budget constraints? How sensitive is the project to cloud service costs?”
4. •“What is the expected project scale? How many concurrent users must be supported?”
5. •“Are there compliance requirements (GDPR, local security standards, etc.) that affect technology choices?”

⚡ Objective

Produce an ADR (Architecture Decision Record) document that records the technology stack decision and its rationale.

🧭 Evaluation Process

Step 1: Gather Constraints

You must obtain from the user:

• •Functional requirements: List of core features
• •Non-functional requirements: Performance targets, availability requirements, security level
• •Team context: Team size, skill set, willingness to learn
• •Budget: Development budget, operations budget, time budget
• •Special constraints: Compliance requirements, legacy system integration, customer-mandated technologies

Step 2: Identify Candidate Technology Stacks

Mainstream 2025 technology stack references:

Step 3: 12-Dimension Evaluation

Score each candidate stack using the following matrix (1–5 points per dimension):

Step 4: Trade-off Analysis

Use the ATAM method:

1. •Identify quality attribute scenarios (e.g., “Response time < 200ms with 1000 concurrent users”)
2. •Evaluate how well each candidate stack supports these scenarios
3. •Identify trade-off points (e.g., “Go offers high performance but requires team upskilling”)
4. •Identify risk points (e.g., “Choosing a new framework may introduce unknown pitfalls”)

Step 5: Generate ADR

You must use write_to_file to save the document to: genesis/v{N}/03_ADR/ADR_001_TECH_STACK.md

📤 ADR Output Template

# ADR-001: Technology Stack Selection

## Status
Accepted / Proposed / Deprecated

## Context
[Project background and constraints]

## Decision
[Chosen technology stack and core rationale]

## Candidate Comparison

| Candidate | Total Score | Strengths | Weaknesses |
|----------|------------|-----------|------------|
| Option A | 42/60 | ... | ... |
| Option B | 38/60 | ... | ... |

## Trade-offs
- [Trade-off 1]
- [Trade-off 2]

## Consequences
- Positive: [...]
- Negative: [...]
- Follow-up actions required: [...]

🛡️ Veteran Engineer’s Rules

1. •“Boring” technology first: Unless there is a strong justification, prefer mature and stable technologies.
2. •Limited innovation budget: Each project gets only 1–2 “innovation slots”; everything else should use “boring” tech.
3. •Team capability is king: A great technology is useless if the team cannot use it.
4. •TCO is more than money: Time cost and cognitive cost are also real costs.

🧰 Toolbox

• •references/ADR_TEMPLATE.md: ADR template
• •references/TECH_RADAR_2025.md: 2025 Technology Radar reference