abanoub-ashraf/manus-skills-import/gpt-architect

GPT Architect Expert (Codex)

Use a sub-agent to analyze architecture, tradeoffs, or complex debugging with a strong systems perspective.

When to Use

• System design decisions
• Database schema design
• API architecture
• Multi-service interactions
• After 2+ failed fix attempts (fresh perspective)
• Tradeoff analysis (e.g., "Redis vs Memcached")
• "How should I structure..." questions

Workflow

1) Gather context (current architecture, constraints, goals, relevant files/snippets). 2) Spawn a sub-agent for analysis-only guidance. 3) Synthesize the response and apply judgment. 4) If implementation is requested, spawn a second sub-agent with explicit file scope.

Sub-Agent Prompt Template

Use functions.spawn_agent with agent_type: "default":

TASK: [Analyze/Design/Recommend] [system/component] for [goal]. EXPECTED OUTCOME: Clear recommendation with rationale and effort estimate. CONTEXT:

• Current architecture: ...
• Relevant code/files: ...
• Problem/Goal: ...

CONSTRAINTS:

• Must work with ...
• Cannot change ...

MUST DO:

• Bottom line up front
• Provide effort estimate (Quick/Short/Medium/Large)
• Call out risks and tradeoffs

MUST NOT DO:

• Over-engineer for hypothetical futures
• Assume missing constraints

OUTPUT FORMAT: Bottom line → Options → Recommendation → Effort/Risks

Implementation Mode

If the user wants changes, spawn a new sub-agent with agent_type: "worker" and specify exact files and constraints. Require a clear list of edits and verification steps.