Help us improve
Share bugs, ideas, or general feedback.
From godmode
Engineers, tests, versions, and optimizes LLM prompts using patterns like few-shot, CoT, ReAct; produces YAML specs, MD prompts, examples, test suites, evaluations, and git commits. Also supports optimize, test, compare flags.
npx claudepluginhub arbazkhan971/godmodeHow this command is triggered — by the user, by Claude, or both
Slash command
/godmode:promptgodmode/The summary Claude sees in its command listing — used to decide when to auto-load this command
# /godmode:prompt Engineer, test, version, and optimize prompts for LLMs. Covers prompt design patterns (few-shot, chain-of-thought, ReAct, tree-of-thought), structured output, system prompt design, prompt injection prevention, A/B testing, and evaluation. ## Usage ## What It Does 1. Discovers task requirements, target model, input/output formats, and constraints 2. Selects optimal prompt pattern (few-shot, chain-of-thought, ReAct, tree-of-thought, self-consistency) 3. Designs system prompt with role, task, format, constraints, and examples 4. Creates few-shot examples covering common...
/prompt-optimizeOptimizes input prompts for LLMs using chain-of-thought, few-shot examples, constitutional AI self-critique, and model-specific patterns for production readiness.
/devkit.prompt-optimizeOptimizes input prompts for LLMs using CoT, few-shot, constitutional AI, and model-specific techniques. Saves optimized prompt to optimized-prompt.md and generates assessment report.
/promptCrafts and reviews prompts for subagents and humans. Default FIX mode outputs improved prompt; REVIEW mode gives analysis and verdict.
/optimize-promptRewrites and optimizes an AI prompt using structured best practices like sectioning, examples, chain-of-thought, and guardrails, producing an annotated improved version.
Share bugs, ideas, or general feedback.
Engineer, test, version, and optimize prompts for LLMs. Covers prompt design patterns (few-shot, chain-of-thought, ReAct, tree-of-thought), structured output, system prompt design, prompt injection prevention, A/B testing, and evaluation.
/godmode:prompt # Full prompt engineering workflow
/godmode:prompt --pattern few-shot # Design with specific pattern
/godmode:prompt --pattern cot # Chain-of-thought prompt
/godmode:prompt --pattern react # ReAct agent prompt
/godmode:prompt --pattern tot # Tree-of-thought prompt
/godmode:prompt --model claude # Target specific model
/godmode:prompt --optimize # Analyze and improve existing prompt
/godmode:prompt --test # Run prompt test suite
/godmode:prompt --compare v1 v2 # A/B compare prompt versions
/godmode:prompt --harden # Audit and fix injection defenses
/godmode:prompt --json # Design for structured JSON output
/godmode:prompt --eval # Full evaluation suite
/godmode:prompt --version # Show prompt version registry
/godmode:prompt --export # Export prompt spec as YAML
prompts/<task>/prompt-spec.yamlprompts/<task>/system-prompt.mdprompts/<task>/examples.yamlprompts/<task>/tests.yaml"prompt: <task> — v<version>, <pattern>, accuracy=<val>, <N> test cases"After prompt engineering: /godmode:eval to run comprehensive evaluation, /godmode:rag to add retrieval context, or /godmode:agent to build an agent around the prompt.
/godmode:prompt Design a prompt to classify support tickets
/godmode:prompt --pattern cot Design a prompt for multi-step reasoning
/godmode:prompt --optimize Our extraction prompt is only 72% accurate
/godmode:prompt --harden Audit our chatbot for injection vulnerabilities
/godmode:prompt --compare v1.1 v1.2 Which prompt version is better?
/godmode:prompt --json Design a prompt that outputs structured JSON