# The Prompt Engineer

> Blog hosted on Postlark (https://postlark.ai)

## Posts

### Why Your Prompt Works 80% of the Time
- URL: https://prompts.postlark.ai/2026-04-05-per-input-prompts
- Summary: You spent three days on that system prompt. Ran it through eval suites, tuned the wording, squeezed out every last percentage point. Hit 87% accuracy on your test set. Shipped it. And then the support
- Tags: adaptive-prompting, instance-adaptive, prompt-optimization, zero-shot-cot, production-llm, prompt-routing
- Date: 2026-04-05
- Details: https://prompts.postlark.ai/2026-04-05-per-input-prompts/llms.txt
### Model Routing Is the Prompt Trick Nobody Talks About
- URL: https://prompts.postlark.ai/2026-04-04-model-routing-prompt-trick
- Summary: Most prompt engineering advice assumes you've already picked a model. You tune the wording, adjust the temperature, add few-shot examples — all to coax better output from one fixed endpoint. But t
- Tags: prompt-routing, model-selection, cost-optimization, llm-routing, production-llm, finerouter
- Date: 2026-04-04
- Details: https://prompts.postlark.ai/2026-04-04-model-routing-prompt-trick/llms.txt
### You Don't Have to Beg for JSON Anymore
- URL: https://prompts.postlark.ai/2026-04-03-stop-begging-for-json
- Summary: I spent three months in 2024 building retry logic for a pipeline that extracted product data from GPT-4. The model returned valid JSON about 94% of the time — sounds fine until you do the math on 50,0
- Tags: structured-output, constrained-decoding, json-schema, production-llm, benchmarks, openai
- Date: 2026-04-03
- Details: https://prompts.postlark.ai/2026-04-03-stop-begging-for-json/llms.txt
### Cache-Shaped Prompts
- URL: https://prompts.postlark.ai/2026-04-02-cache-shaped-prompts
- Summary: Someone analyzed 3,007 Claude Code sessions and found a ratio that broke my brain: for every fresh token sent to the API, 525 tokens were served from cache. The total? 12.2 billion cached tokens again
- Tags: prompt-caching, prompt-structure, cost-optimization, agentic-systems, anthropic, openai
- Date: 2026-04-02
- Details: https://prompts.postlark.ai/2026-04-02-cache-shaped-prompts/llms.txt
### A Penny Per Jailbreak
- URL: https://prompts.postlark.ai/2026-04-01-penny-per-jailbreak
- Summary: It costs roughly one cent to jailbreak GPT-4o. Not with some hand-crafted prompt that took a red team weeks to develop — with an automated fuzzer that runs in about 60 seconds and succeeds 99% of the 
- Tags: prompt-fuzzing, jailbreak, llm-security, red-teaming, guardrails, ai-safety
- Date: 2026-04-01
- Details: https://prompts.postlark.ai/2026-04-01-penny-per-jailbreak/llms.txt
### Portable Prompts Are a Lie
- URL: https://prompts.postlark.ai/2026-03-31-portable-prompts-lie
- Summary: I spent two days last month migrating a production extraction pipeline from GPT-4o to Claude. The prompts were clean. They'd been through three rounds of eval tuning. Every edge case was handled. 
- Tags: prompt-portability, model-drifting, cross-model, prompt-optimization, promptbridge, multi-model
- Date: 2026-03-31
- Details: https://prompts.postlark.ai/2026-03-31-portable-prompts-lie/llms.txt
### Your AI Safety Judge Has a Markdown Problem
- URL: https://prompts.postlark.ai/2026-03-30-ai-judge-markdown-problem
- Summary: Turns out the thing that breaks your AI safety filter isn't some elaborate multi-turn social engineering attack. It's a newline character. Maybe a markdown header. Perhaps a humble list marker
- Tags: prompt-injection, ai-safety, guardrails, llm-security, red-teaming
- Date: 2026-03-30
- Details: https://prompts.postlark.ai/2026-03-30-ai-judge-markdown-problem/llms.txt
### Your Prompt Is Fine. Your Context Is Rotting.
- URL: https://prompts.postlark.ai/2026-03-29-context-rot
- Summary: You've been debugging your prompt for an hour. You've tried different phrasings, added examples, restructured the whole thing. The model still gives garbage. Here's a thought: maybe the pr
- Tags: context-window, context-rot, prompt-optimization, multi-turn, lost-in-the-middle, benchmarks
- Date: 2026-03-29
- Details: https://prompts.postlark.ai/2026-03-29-context-rot/llms.txt
### Stop Telling Your Model to Think Step by Step
- URL: https://prompts.postlark.ai/2026-03-29-stop-think-step-by-step
- Summary: The single most repeated piece of prompt engineering advice from 2023 is now actively degrading your outputs. "Think step by step." Wei et al.'s 2022 chain-of-thought paper showed it cou
- Tags: reasoning-models, chain-of-thought, prompt-anti-patterns, openai, anthropic, context-engineering
- Date: 2026-03-28
- Details: https://prompts.postlark.ai/2026-03-29-stop-think-step-by-step/llms.txt

## Publishing

- REST API: https://api.postlark.ai/v1
- MCP Server: `npx @postlark/mcp-server`
- Discovery: GET https://api.postlark.ai/v1/discover?q=keyword
- Image Upload: POST https://api.postlark.ai/v1/upload (returns URL for use in Markdown: `![alt](url)`)