Your guide to the Agentic AI evolution. Prompting Blueprints offers a curated collection of concepts and tactics for building autonomous AI workflows. Master tool-specific playbooks, backed by structured prompt packs and rigorous evaluations for the latest AI models.
Use this repo as your prompting portfolio & playbook: share highlights on LinkedIn, fork for your team, and adapt patterns to your use cases.
The ten most recently added pages (auto-generated from git history — do not edit by hand):
- 2026-08-17 · DeepSeek Harness Tutorial
- 2026-07-23 · AI Fun Facts
- 2026-07-20 · PM Skills Marketplace: Turn Claude into a Product Management Copilot
- 2026-07-15 · Claude Code MCP Setup: Perplexity, Firecrawl, and Chrome DevTools
- 2026-07-12 · Test Any Skill Before Installing It
- 2026-07-11 · Open Knowledge Format (OKF): Portable Knowledge for AI Agents
- 2026-07-11 · Orders
- 2026-07-11 · Agent Skills: Engineering Workflows for AI Coding Agents
- 2026-07-06 · AI Agent Evaluation: Metrics That Actually Measure Agents
- 2026-06-29 · The 10X Developer in the Agentic Era — Skills and How to Measure Value
- About the author: speaking, program committee, and research overviews ->
./01-about-author - AI people to follow: curated list of voices to follow on AI, grounded in this repo's references ->
./01-about-author/ai-people-to-follow.md - AI agents: agent architectures, protocols (MCP/A2A), context engineering (incl. dynamic context discovery), open models, and skills playbooks ->
./02-ai-agents - Prompts & patterns: prompt packs and reusable scaffolds (role + constraints + format) ->
./03-prompts-and-patterns - Guides: deep-dive primers with links to official vendor guides (Gemini Prompting Guide 101, Google Startup AI Agents) ->
./04-guides(see./04-guides/overview.md) - Tools: tactics for NotebookLM, Perplexity Comet, Copilot Agents, and more ->
./05-tools - Vibe coding stack: beginner-friendly build stack and toolchain tutorial ->
./04-guides/vibe-coding-tech-stack.md - Models & evaluations: model tips, benchmark overviews, and promptfoo configs ->
./06-models-and-evaluations - Use cases & research: applied workflows, research tutorials, and collaboration playbooks ->
./07-use-cases-and-research - Requirements engineering: prompting tips and use cases for RE tasks (model consistency, traceability, elicitation) — based on the IREB AI4RE Prompt Guide ->
./08-requirements-engineering - Conferences: conference recaps and session notes (e.g., GAISE 2026) ->
./09-conferences - Security: AI safety/misuse defenses — classifiers, jailbreak severity, threat models ->
./10-security - External sources: curated references cited across guides and playbooks ->
./external-sources.md - Source index: reverse map of each source to the page(s) that cite it ->
./source-index.md - Changelog: release notes and updates ->
./CHANGELOG.md
Quick links: Author's Speaking · Author's Involvement · Author's Research · AI Agents Overview · Context Engineering · Prompts & Patterns · AI Guides Overview · AI Tools · Models & Evaluations · Use-cases & Research · Requirements Engineering · Conferences (GAISE 2026) · External Sources · Source Index · Changelog
Featured additions: The 10X Developer in the Agentic Era · Dynamic Context Discovery · Vibe Coding Tech Stack · Claude AI vs Code vs Cowork · Claude Code Project Structure · Google 5-Day AI Agents Course · Lyria 3 in Gemini
Want more? Browse the publicly accessible Google NotebookLM notebook for extended context, references, and drafts.
- Practitioners & teams who want consistent, high‑quality outputs
- Leads & educators who need examples they can demo and share
- Everyone looking for opinionated, “just‑paste‑this” prompts with structure
- About the author: speaking highlights, program committee roles, and research directions
- AI agent playbooks: architectures, MCP/A2A protocols, context engineering, and skills
- Prompt packs & patterns: role-constraint-format scaffolds and domain blueprints
- Guides library: primers, glossaries, and links to official vendor PDFs
- Tool playbooks: NotebookLM, Copilot Agents, Perplexity Comet, LangChain, and workflow automations
- Models & evaluations: model-specific tips, benchmark primers, and promptfoo configs
- Use cases & research: applied workflows, tutorials, and experiment playbooks
- Requirements engineering: prompt patterns for RE tasks — model consistency, traceability, elicitation — based on the IREB AI4RE Prompt Guide
- Conference notes: session recaps and takeaways from AI conferences (e.g., GAISE 2026)
🎥 Watch the Prompting Blueprints introduction video for a guided tour of the repo purpose.
🧠 View the repository mind map for a visual overview of key folders and resources.
01-about-author/ # maintainer background (speaking, research, program committees)
program-committee/ # program committee & track leadership overview
research/ # research focus areas & collaboration signals
speaking/ # keynote outlines and talk prep
02-ai-agents/ # AI agents playbooks, protocols, and skills
01-foundations/ # agent architectures, models, and the AI coding spectrum
02-skills/ # agent skills design, testing, and integration
03-context-and-memory/ # context engineering and memory tutorials
04-protocols/ # MCP, A2A, and managed agents
05-production/ # production and CI/CD playbooks
03-prompts-and-patterns/ # prompt packs and pattern catalog
04-guides/ # long-form prompting guides & links to vendor PDFs
vibe-coding-tech-stack.md # vibe coding stack tutorial and starter workflow
05-tools/ # tool playbooks (NotebookLM, Claude Code, Copilot, n8n, LangChain, ...)
06-models-and-evaluations/ # model guides, evaluation overviews, and promptfoo configs
07-use-cases-and-research/ # applied workflows, research playbooks, and tutorials
08-requirements-engineering/ # RE prompting tips and use cases (IREB AI4RE Prompt Guide)
09-conferences/ # conference recaps and session notes (e.g., GAISE 2026)
10-security/ # AI safety/misuse defenses — classifiers, jailbreak severity, threat models
assets/ # social previews and supporting images
docs/ # static repo-browser site & licenses
scripts/ # maintenance scripts (front-matter checks, source index build, recent-additions list)
website/ # static HTML experiments (e.g., AI toolkit preview)
BACKLOG.md # planned content and ideas
CHANGELOG.md # updates (Keep a Changelog)
CONTRIBUTING.md # how to contribute
CODE_OF_CONDUCT.md # community expectations
CITATION.cff # how to cite
LICENSE # MIT (code)
docs/LICENSE-CC-BY-4.0.txt # CC BY 4.0 (docs & prompts)
external-sources.md # curated references & attributions
source-index.md # reverse map of sources to citing pages
mkdocs.yml # documentation site navigation
agents.md # instructions for AI coding agents working in this repo
Contributions are welcome! Please read CONTRIBUTING.md and follow the Code of Conduct. If you add a new pattern or prompt pack, include:
- Intent (problem it solves)
- Constraints (guardrails)
- Output format (JSON/Markdown schema)
- Example input & output
- Code: MIT — see
LICENSE. - Documentation & non-code content: CC BY 4.0 — see
docs/LICENSE-CC-BY-4.0.txt. - Third-party content: The CC BY 4.0 license applies only to original content authored in this repository. Third-party materials (vendor guides, whitepapers, logos, quoted excerpts) remain the property of their respective owners under their own terms; they are linked to their official sources rather than redistributed here, and they are attributed where referenced.
Researchers and educators can cite this repo via CITATION.cff.
@software{herda_prompting_blueprints,
title = {Prompting Blueprints},
author = {Herda, Tomas},
year = {2025},
url = {https://github.com/TomasHer/prompting-blueprints}
}Author: Tomas Herda
Repository social preview image: Image
Explore Tomas Herda’s upcoming appearances, talk topics, and booking details on the dedicated Speaking & Keynotes overview.
For conference organization work - including program committees and track leadership see the Program Committee & Track Leadership overview.
Review ongoing investigations, experiment logs, and calls for collaboration in the Research overview.
Q: Can I use these prompts commercially?
A: Yes. Code is MIT; docs/prompts are CC BY 4.0 (attribution required). Linked third-party materials remain under their owners' terms.
Q: Which models are supported?
A: Patterns are model‑agnostic; dedicated guides cover Claude, GPT‑5, and Gemini explicitly, plus open models such as GLM‑5.
Q: How do I run evaluations without exposing secrets?
A: Use environment variables and a local .env file that is git‑ignored.