home / skills / nickcrew / claude-cortex / multi-llm-consult

multi-llm-consult skill

safe

This skill consults external LLMs to provide second opinions, plan comparisons, and delegated reviews, then consolidates results for informed decisions.

npx playbooks add skill nickcrew/claude-cortex --skill multi-llm-consult

Review the files below or copy the command above to add this skill to your agents.

Files (3)

SKILL.md

1.9 KB

---
name: multi-llm-consult
description: Consult external LLMs (Gemini, OpenAI/Codex, Qwen) for second opinions, alternative plans, independent reviews, or delegated tasks. Use when a user asks for another model's perspective, wants to compare answers, or requests delegating a subtask to Gemini/Codex/Qwen.
---

# Multi-LLM Consult

## Overview

Use a bundled script to query external LLM providers with a sanitized prompt and return a concise comparison.

## Setup

- Configure API keys in the TUI: open the Command Palette (Ctrl+P) and run **Configure LLM Providers**.
- Keys are stored in `settings.json` under `llm_providers`.

## Workflow

1. Identify the **purpose** (`second-opinion`, `plan`, `review`, `delegate`).
2. Summarize the task and **sanitize sensitive data** before sending it out.
3. Run the consult script with the chosen provider.
4. Compare responses and reconcile with your own plan before acting.

## Consult Script

Always run `--help` first:

```bash
python scripts/consult_llm.py --help
```

Example: second opinion

```bash
python scripts/consult_llm.py \
  --provider gemini \
  --purpose second-opinion \
  --prompt "We plan to refactor module X. What risks or gaps do you see?"
```

Example: delegate a review

```bash
python scripts/consult_llm.py \
  --provider qwen \
  --purpose review \
  --prompt-file /tmp/review_request.md \
  --context-file /tmp/patch.diff
```

Example: plan check with Codex (OpenAI)

```bash
python scripts/consult_llm.py \
  --provider codex \
  --purpose plan \
  --prompt "Draft a 5-step plan for implementing feature Y."
```

## Output Handling

- Treat responses as advisory; verify against repo constraints and current state.
- Summarize the external response in 3-6 bullets before acting.
- If responses conflict, call out the differences explicitly and choose a path.

## References

- Provider defaults and configuration: `references/providers.md`

Overview

This skill consults external LLMs (Gemini, OpenAI/Codex, Qwen) to get second opinions, alternative plans, independent reviews, or to delegate discrete tasks. It runs a sanitized prompt against a chosen provider and returns concise, comparable outputs to help you decide or act.

How this skill works

You pick a purpose (second-opinion, plan, review, delegate), sanitize any sensitive data, and call the consult script with the provider and prompt or files. The script queries the external model, captures its response, and produces a short comparison and summary for easy reconciliation with your primary plan. Responses are advisory and require verification against your repository and constraints.

When to use it

You need an independent second opinion on a design, PR, or plan.
You want alternative implementation plans or step-by-step proposals.
You need an automated review of a patch, diff, or proposal.
You want to delegate a well-scoped subtask (drafting, tests, docs).
You need to compare outputs from multiple LLM providers before deciding.

Best practices

Sanitize prompts to remove secrets, tokens, or private data before sending.
Choose purpose explicitly (second-opinion, plan, review, delegate) to frame the model’s response.
Run the script with --help first to confirm available flags and provider names.
Summarize external responses in 3–6 bullets and call out conflicts or gaps.
Verify any suggested code or changes against the current repo and CI before merging.

Example use cases

Ask Gemini for a second opinion on risks and gaps before refactoring module X.
Request a 5-step implementation plan from Codex (OpenAI) for a new feature.
Send a patch.diff and prompt Qwen to perform an independent code review.
Delegate documentation drafting for a module to an external model, then edit the result.
Compare recommendations from two providers to resolve conflicting suggestions.

FAQ

How do I configure API keys?

Open the TUI command palette (Ctrl+P), run Configure LLM Providers, and add keys; they are stored in settings.json under llm_providers.

What should I do if providers give conflicting advice?

Summarize each response point-by-point, highlight conflicts, weigh them against repo constraints and tests, then choose or synthesize a path.