Use this skill when Claude needs an outside perspective on complex problems, is stuck on a technical challenge, needs validation of an approach, or wants consultation on architectural decisions. Provides access to Strategist, Codex, and Gemini for getting unbiased second opinions and fresh perspectives.
/plugin marketplace add qduc/claude-skills/plugin install advisor-skills@qduc-marketplaceThis skill inherits all available tools. When active, it can use any tool Claude has access to.
This skill enables Claude to consult other AI models and specialized agents when facing challenging problems that benefit from an outside perspective.
Use the consult skill when:
Best for: Unbiased, objective advice on complex problems and strategic decisions
How to use:
Use the Task tool with subagent_type name "advisor-skills:strategist".
Use when:
Best for: Deep reasoning problems, complex logical challenges, multi-step analysis
Command:
codex e "your query here" 2>/dev/null
Use when:
Best for: Online research, current information, real-world data, web-based queries
Command:
gemini "your query here" 2>/dev/null
Use when:
Before consulting external AI models, clearly identify:
Craft a clear, focused query that:
Use the Bash tool with appropriate timeout (10 minutes = 600000ms):
codex e "How can I efficiently find duplicate entries in a large dataset with 100M+ records while minimizing memory usage?" 2>/dev/null
or
gemini "I'm deciding between microservices and monolith architecture for a medium-sized team. What factors should guide this decision?" 2>/dev/null
After receiving advice:
DO:
DON'T:
Always use the Bash tool with these parameters:
Bash tool parameters:
- command: codex e "query" 2>/dev/null OR gemini "query" 2>/dev/null
- description: "Consulting [Codex/Gemini] about [brief problem description]"
- timeout: 600000 (10 minutes in milliseconds)
- dangerouslyDisableSandbox: true (REQUIRED - see note below)
You MUST set dangerouslyDisableSandbox: true when calling codex or gemini.
These tools need to write session files outside the default sandbox-allowed directories. Without this parameter, they'll fail with:
Read-only file system (os error 30)
This is safe because these are user-installed, whitelisted tools that require filesystem access for session management.
Situation: Need unbiased, objective perspective on a complex architectural decision.
Use Strategist Agent:
runSubagent tool with:
- agentName: "strategist"
- description: "Architectural decision consultation"
- prompt: "I'm deciding between microservices and monolith architecture for a new project. Our team has 5 developers, the product is still finding product-market fit, and we expect high growth if successful. However, our team is excited about microservices and has been pushing for it. I'm worried we might be choosing it for the wrong reasons. Can you provide an unbiased analysis of this decision?"
Situation: Need to deeply analyze trade-offs between multiple architectural approaches.
Query to Codex:
codex e "I need to design a notification system that handles 100K notifications/day with delivery guarantees. Compare event-driven (Kafka), queue-based (RabbitMQ), and polling approaches. Consider: failure handling, scaling, operational complexity, latency requirements, and team expertise (3 backend devs, no DevOps). What would you recommend and why?" 2>/dev/null
Situation: Need to know current best practices for a technology or framework.
Query to Gemini:
gemini "What are the current best practices for implementing authentication in Next.js 14 apps in 2024? I need to understand popular libraries, session vs JWT trade-offs, and security considerations." 2>/dev/null
Situation: Complex algorithmic problem requiring multi-step reasoning.
Query to Codex:
codex e "I need to implement a job scheduler that: 1) respects dependencies between jobs, 2) handles priority levels, 3) supports job cancellation mid-execution, 4) distributes across multiple workers. Walk through the data structures and algorithm design. What are the edge cases and how would you handle them?" 2>/dev/null
Situation: Need current information about tools, libraries, or frameworks.
Query to Gemini:
gemini "I need a Python library for parsing PDFs with table extraction. Research current options (2024), compare their accuracy, performance, and ease of use. What do developers recommend?" 2>/dev/null
Situation: Complex bug requiring careful logical reasoning through multiple possibilities.
Query to Codex:
codex e "I have a distributed transaction that occasionally commits on service A but fails on service B, leaving inconsistent state. The retry logic sometimes creates duplicates. Given: idempotency keys, 2PC is not an option, compensation logic exists but races with retries. Reason through the failure scenarios and design a robust solution." 2>/dev/null
Situation: Need to find specific API documentation or examples.
Query to Gemini:
gemini "Find documentation and examples for using PostgreSQL's LISTEN/NOTIFY feature with connection pooling (pgbouncer). What are the gotchas and how do production systems handle this?" 2>/dev/null
The consult skill complements Claude's normal problem-solving:
Remember: External consultants provide valuable perspectives, but you remain responsible for the final implementation and ensuring it fits the specific context and requirements.
This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.
This skill should be used when the user asks to "create a slash command", "add a command", "write a custom command", "define command arguments", "use command frontmatter", "organize commands", "create command with file references", "interactive command", "use AskUserQuestion in command", or needs guidance on slash command structure, YAML frontmatter fields, dynamic arguments, bash execution in commands, user interaction patterns, or command development best practices for Claude Code.
This skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "implement prompt-based hooks", "use ${CLAUDE_PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStop, SessionStart, SessionEnd, UserPromptSubmit, PreCompact, Notification). Provides comprehensive guidance for creating and implementing Claude Code plugin hooks with focus on advanced prompt-based hooks API.