Coding

For engineers and product owners running AI agents in production — or almost there. Describe or paste your agent's architecture, task, tools, and what you're currently measuring. Get a systematic evaluation across five dimensions: task quality, failure modes, tool use integrity, token efficiency, and observability gaps. Trust but verify — especially verify.

You've been using Cursor, Copilot, or Claude to write most of your codebase. It runs. You understand it in broad strokes. But would it survive a senior engineer's review? A 2am incident? An audit? This prompt runs a systematic five-layer check on AI-generated code: security, correctness, maintainability, data integrity, and the failure modes specific to AI-generated code that look fine until they aren't.

A specialized debugging companion for tracing failures, bottlenecks, and logic errors in multi-step AI agent workflows — from tool calls to chain-of-thought breakdowns.

Paste any AI-generated output — image, text, code, or video — and reverse-engineer the exact prompt that would reproduce it. The ultimate meta-prompt for learning prompt craft by deconstruction.

Systematically reverse-engineer, document, and map unfamiliar or legacy codebases. Produces architecture diagrams, dependency maps, risk assessments, and onboarding guides from raw source code.

Design agent-to-agent communication systems using Google's A2A protocol and Anthropic's MCP — architect multi-agent workflows where specialized AI agents discover, negotiate, and collaborate autonomously.

Transform messy git diffs, commit logs, and PR descriptions into polished changelogs and user-facing release notes — with proper semantic versioning, audience-aware tone, and zero fluff.

Systematically identify, classify, and prioritize security threats for any application or system using STRIDE, attack trees, and trust boundary analysis — then output actionable mitigations ranked by risk.

Design step-by-step computer-use agent prompts that guide AI models with native desktop/browser control to complete real tasks — clicking, typing, navigating, and verifying outcomes like a skilled operator.

A meta-prompt that transforms any vague request into a precision-engineered prompt with XML/JSON output schemas, validation rules, and anti-hallucination guardrails for consistent, parseable AI responses.

Design and build Model Context Protocol (MCP) servers that expose your APIs, databases, and internal tools to AI agents — with proper auth, schema design, and context-window efficiency.