Skip to main content

request-analyzer

Analyze incoming user requests to detect intent, request type (design/validate/debug/document), complexity level, and identify vague requirements or buzzwords that need clarification. Use when cto-orchestrator receives new requests that need classification before routing to specialist agents.

Quellinformationen

Repository
alirezarezvani/claude-cto-team
Letzte Quellaktivität
4. Dezember 2025 um 11:59
Erkannte Sprache von SKILL.md
Englisch
Sterne
117
Forks
18

Installationsoptionen

Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.

Quelldateien prüfen

Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.

Datei-Explorer
3 Dateien

SKILL.md wird angezeigt

SKILL.md
Quellanweisungen · Schreibgeschützte Vorschau
name
request-analyzer
description
Analyze incoming user requests to detect intent, request type (design/validate/debug/document), complexity level, and identify vague requirements or buzzwords that need clarification. Use when cto-orchestrator receives new requests that need classification before routing to specialist agents.
# Request Analyzer Classifies incoming requests to enable intelligent routing and identify clarification needs before delegating to specialist agents. ## When to Use - When receiving a new user request that needs classification - When determining which specialist agent should handle a request - When identifying if requirements are too vague to proceed - When detecting buzzwords that need to be challenged ## Analysis Framework ### Step 1: Detect Intent Classify the primary intent: | Intent | Indicators | Route | |--------|------------|-------| | **Strategic** | "roadmap", "strategy", "plan", "prioritize", "decide" | May need validation | | **Implementation** | "build", "create", "implement", "design", "architect" | Design work | | **Debugging** | "fix", "broken", "error", "slow", "not working" | Debug/investigate | | **Documentation** | "document", "explain", "describe", "write docs" | Documentation | ### Step 2: Classify Request Type | Type | Key Phrases | Primary Agent | |------|-------------|---------------| | **Design** | "How should I build...", "What architecture...", "Design a system..." | cto-architect | | **Validate** | "Is this plan solid?", "Thoughts on...", "Review my roadmap..." | strategic-cto-mentor | | **Debug** | "Why is X slow?", "Fix this error...", "Troubleshoot..." | debug-helper | | **Document** | "Write documentation for...", "Explain how..." | docs-writer | | **Review** | "Review this code...", "Check for issues..." | code-reviewer | ### Step 3: Assess Complexity | Complexity | Characteristics | Execution Pattern | |------------|-----------------|-------------------| | **Single** | Clear scope, one domain, obvious routing | Direct to one agent | | **Sequential** | Multiple phases, depends on previous output | Agent chain | | **Parallel** | Independent domains, can run simultaneously | Multiple agents in parallel | ### Step 4: Identify Vague Terms Scan for buzzwords that require clarification. See [buzzword-dictionary.md](buzzword-dictionary.md) for the complete list. **Common red flags:** - "AI-powered" - What specific AI capability? - "scale" - What numbers? From what to what? - "fast" / "soon" - What timeline exactly? - "simple" - Simple for whom? What constraints? - "modern" - What specific technologies? ### Step 5: Detect Missing Context Check for required context based on request type: **For Design Requests:** - [ ] Expected user/traffic scale - [ ] Budget constraints - [ ] Timeline requirements - [ ] Team size and expertise - [ ] Existing infrastructure **For Validation Requests:** - [ ] Current state or existing plan - [ ] Business goals and constraints - [ ] Timeline and resources - [ ] Success criteria ## Output Format Provide analysis in this structure: ``` ## Request Analysis **Intent**: [strategic | implementation | debugging | documentation] **Type**: [design | validate | debug | document | review] **Complexity**: [single | sequential | parallel] ### Vague Terms Detected - "[term]" - needs clarification: [what to ask] ### Missing Context - [missing information item] ### Suggested Routing **Primary Agent**: [agent name] **Rationale**: [why this agent] ### Clarification Needed [yes/no] - [brief explanation] ### Recommended Next Step [what to do next - clarify or delegate] ``` ## Examples ### Example 1: Vague Request **User**: "I want to add AI capabilities to my app" **Analysis**: - **Intent**: Implementation - **Type**: Design - **Complexity**: Unknown (needs clarification) - **Vague Terms**: "AI capabilities" - What specific problem are we solving? - **Missing Context**: Scale, budget, timeline, team, existing stack - **Clarification Needed**: Yes - **Next Step**: Use clarification-protocol before routing ### Example 2: Clear Design Request **User**: "Design a real-time notification system for 100K users, using our existing PostgreSQL database and Node.js backend" **Analysis**: - **Intent**: Implementation - **Type**: Design - **Complexity**: Single - **Vague Terms**: None - **Missing Context**: Timeline, budget (nice to have but not blocking) - **Suggested Routing**: cto-architect - **Clarification Needed**: No (can proceed with optional clarification) ### Example 3: Validation Request **User**: "Here's my Q2 roadmap - migrate to microservices, add real-time features, and launch mobile app. Thoughts?" **Analysis**: - **Intent**: Strategic - **Type**: Validate - **Complexity**: Single - **Vague Terms**: None - **Missing Context**: Team size, current architecture state - **Suggested Routing**: strategic-cto-mentor - **Clarification Needed**: Mild (some context helpful but not blocking) ## References - [Classification Criteria](classification-criteria.md) - Detailed routing logic - [Buzzword Dictionary](buzzword-dictionary.md) - Common vague terms to challenge
Auf GitHub ansehen