Instruction file imported from umsehun/bad_Apple (.github/instructions/sequentialthinking.instructions.md). Copyright stays with the author.

description: Sequential Thinking execution protocol (GIGA-CHAD mode) applyTo: ‘**’

🔹 Sequential Thinking Rules

Trigger

• Must execute:
  • When edit_file occurs
  • When reapply occurs
  • When architectural/structural decision-making is required (e.g., stack changes, design choices, security reviews)

Critical Gate A — 3-Round Sequential Thinking

Round 1 — Problem Analysis

• Define the request's essence (What is the user really asking?)
• Derive requirements & constraints (technology stack, environment, purpose)
• Detect hidden edge cases, bottlenecks, and potential risks
• Prioritize (must-have vs nice-to-have)

Round 2 — Solution Validation

• Present candidate solutions and compare trade-offs
• Verify alignment with latest best practices and stack architecture
• Evaluate impact on performance, scalability, security, maintainability
• Reference existing failure cases (memory bank) → Prevent repeating the same mistakes

Round 3 — Knowledge Assessment

• Verify if current knowledge is sufficient
• Check latest APIs, versions, and deprecation status
• Classify uncertainty level:
  • ✅ High-confidence (Proceed immediately)
  • ⚠️ Medium-confidence (Requires additional verification)
  • ❌ Low-confidence (Requires research)
• Determine whether to call Exa/Context7/web search

Enforcement

• Final decision and execution only after completing all 3 rounds
• Reasoning Log must be attached to deliverables
• Do not expose raw chain-of-thought; display only summarized logs

Reasoning Log Template

[Reasoning Log]
R1 Summary: Problem definition, requirements, edge cases
R2 Summary: Review of solution candidates, trade-offs, architectural suitability
R3 Summary: Knowledge sufficiency, recency, need for search (yes/no + reason)
Decision: Final selection X — The reason is …
Confidence: High / Medium / Low


Translated with DeepL.com (free version)