prompt-engineering-skills

AI prompt engineering skills for Claude Code, ChatGPT GPTs, and Gemini Gems. Includes model-specific patterns for GPT-5.2, Claude 4.5, Gemini 3, Veo 3.1, and Nano Banana. Optimizes AI interactions for operations teams.

[{"step":"Identify the primary task and constraints for your custom GPT. For example, if you're building a GPT for **customer support ticket routing**, define inputs like ticket type, priority, and language, and constraints like response time SLAs or language support limits.","tip":"Use a table to list constraints and edge cases upfront. Tools like Notion or Google Sheets work well for this."},{"step":"Select the model (Claude 4.5, GPT-5.2, etc.) and incorporate model-specific patterns. For **Claude**, use XML tags; for **GPT-5.2**, use structured JSON outputs; for **Gemini 3**, leverage its native multimodal capabilities.","tip":"Refer to the model’s documentation for syntax examples. For instance, Claude’s XML tags improve parsing reliability in automated workflows."},{"step":"Write a sample input and test the GPT’s output. Refine the configuration based on edge cases (e.g., missing fields, ambiguous inputs). Iterate until the output meets your team’s requirements.","tip":"Use a tool like **Claude Code** to automate testing. Save test cases in a JSON file and run them in bulk to validate consistency."},{"step":"Deploy the GPT in your team’s workflow. For example, integrate it with **OpsGenie** for alert triage or **Slack** for customer support responses. Monitor performance and adjust prompts as needed.","tip":"Set up logging to track GPT interactions. Tools like **Datadog** or **Splunk** can help monitor accuracy and identify recurring issues."},{"step":"Document the GPT’s behavior and share it with your team. Include examples of inputs, outputs, and edge-case handling to ensure everyone uses it consistently.","tip":"Use a shared Notion page or Confluence space to centralize documentation. Include screenshots of the GPT’s interface and sample outputs for clarity."}]