Agent Outcome Rubrics Are Just Prompts: How to Write Grading Criteria That Actually Fail Bad Output
Anthropic shipped the grading loop -- the rubric is now the hardest prompt you'll write
Brian Gagne
CTO of Kief Studio. 20+ years technology consulting. Cisco Certified Ethical Hacker, UPenn AI for Business, Perplexity AI Business Fellow. Builds the tools Qurtoo teaches you to use.
https://kief.studio65 articles
Uncertainty Markers Double LLM Reasoning Accuracy: The Self-Correction Prompting Technique Google Doesn't Want You to Sleep On
Embedding hesitation cues like 'wait, let me verify' into your prompts activates built-in self-correction that chain-of-thought never could
Your Agent Card Is a Prompt: Writing A2A Descriptions That Win Delegation From Other Agents
ADK 1.0 and the A2A protocol turned your agent's metadata into the most important prompt you're not optimizing
Stop Copy-Pasting Prompts Between Models: Format-Native Prompting for Claude 4.7, GPT-5.5, and Gemini 3.1
Each frontier model parses structure differently -- XML, JSON, or markdown. Match the format, and output quality jumps measurably.
Prompt Architecture for Managed Agent Runtimes: What Changes When You Stop Owning the Orchestration Loop
Anthropic's Managed Agents, Google ADK, and Microsoft Agent Framework 1.0 all abstract away the runtime -- your prompts need to catch up
GPT-5.5's Seven-Part Prompt Schema: Why OpenAI Wants You to Delete Every Prompt You've Ever Written
OpenAI's new prompting guide kills personas, step-by-step, and emotional primers -- here's the minimal structure that actually works on reasoning models
Prompting for Supply Chain Risk: Extract Red Flags From Vendor Documents Faster Than Any Analyst
Turn contract PDFs and compliance docs into structured risk assessments with three prompts
Writing gh skill Definitions That Work Across Claude Code, Copilot, and Cursor
The new GitHub CLI skill system lets you write one agent instruction set that runs everywhere -- here's how to prompt it right
Prompt Patterns That Catch MCP Tool Poisoning Before Your Agent Executes It
Defensive system prompts and validation chains for the attack class that hits harder the smarter your model is
Sub-Agent Delegation Prompts for Claude 4.7's 1M Context: When to Spawn and When to Stay
Stop burning your million-token window on grep output. Prompt patterns that decide what gets delegated.
Tuning Thinking Budgets: The Prompt Technique That Replaced Chain-of-Thought in Claude 4.7 and GPT-5.4
Stop writing 'think step by step'. Start setting token budgets that match the task.
Prompting for CPAs After Tax Day: Extension Workflows, IRS Notice Responses, and Amended Returns That Don't Hallucinate Code Sections
A post-April-15 playbook for tax pros using Claude and GPT to handle the messy 6-month tail of the 2026 filing season