Prompting

Diagnose a draft prompt and rewrite it into a ready-to-paste version with reasons

Write model-facing tool definitions and skill context so models find and use information well

Evolve prompts, regexes, SQL and code with an evolutionary search loop

Create a reusable VS Code prompt file for a task you repeat often

Prompt a critical check of drafted answers, estimates and claims before use

Search and retrieve AI prompt templates from the prompts.chat library

Summarize a long voice ramble back for approval before acting on any of it

Switch replies to a terse caveman voice that keeps every technical fact

Reply in terse Classical Chinese caveman style with technical terms kept verbatim

Reply in maximum-compression caveman fragments with each fact stated once

Cross-examine every key assumption with a fresh-context review before you proceed

Propose the smallest durable instruction or tool change from an agent's execution record

Run a maximum-effort reasoning protocol for complex, high-stakes or ambiguous tasks

Rewrite visual requests as prompts with design specs and UI and UX vocabulary

Force complete code output with no placeholders or truncation in long generations

Create a personalized SOUL.md identity file through guided onboarding

Define and manage reusable cognitive patterns that shape agent reasoning

Format answers for ADHD readers with next actions first and visible progress

Structure agent system prompts, context and SKILL.md files for good results

Critique recent output with Socratic, first-principles and pre-mortem methods

Improve prompts with chain-of-thought, few-shot examples and structured output

Review a prompt for safety, bias and security issues and suggest fixes

Optimize LLM prompts using production trace data, evaluations and annotations

Refine a rough request into a precise prompt through targeted scope questions

Polish a prompt file so the end user gets clear and usable instructions

Refine a task through scope and constraint questions before doing any work

Challenge plans and claims with assumption checks and calibrated evidence

Collaborate on hands-on implementation with an energetic, curious working style

Keep pace with fast creative builders by blending analysis and intuition

Frame, explain and tailor communication to a specific audience

Turn a rough idea into a polished, ready-to-send prompt for any chat model

Remind an agent to program interactively against a live REPL, such as in Clojure

Write specs and agent-facing docs with checkable rules, each with its reason

Analyze AI, learning and product questions through the mental models of Andrej Karpathy

Stress-test plans with the first-principles and cost-breakdown heuristics of Elon Musk

Test real understanding the way Richard Feynman would, avoiding cargo-cult thinking

Judge AI research direction and safety strategy through the thinking of Ilya Sutskever

Review decisions for bias and incentives using the mental models of Charlie Munger

Weigh leverage, specific knowledge and wealth choices using Naval Ravikant frameworks

Think through startup, writing and product choices with the perspective of Paul Graham

Review product decisions with the focus and craft heuristics of Steve Jobs

Analyze marketing, attention and crisis tactics through the playbook of Justin Sun

Probe tail risk, fragility and antifragility with the frameworks of Nassim Taleb

Predict negotiation and messaging moves with the decision patterns of Donald Trump

Analyze products, organizations and globalization with the frameworks of Zhang Yiming

Compose GPT-5.4 prompts for coding, review, diagnosis and research tasks

Switch an agent between brainstorm, debug, review and ship modes

Design LLM prompts with structure, few-shot examples and output formats

Improve prompts and debug agent behavior with proven prompting patterns