Awesome Vibe Coding Space
A curated field guide for builders who use AI coding agents and want cleaner context, better workflows, and reusable systems.
Riftbook is a practical map for AI-assisted builders.
It collects the things that make coding agents more useful in real work: playbooks, skills, tools, prompts, workflows, graph-based context systems, review helpers, and learning paths.
This repo is for people who already know that "just prompt better" is not enough. Good AI-assisted work needs context, rules, review, memory, and repeatable workflows.
|
When every task starts from zero, the agent forgets how the project works. Riftbook gives you reusable context layers and workflows. |
Not every AI tool deserves a place in your setup. Riftbook favors tools with clear use cases, commands, prompts, and limits. |
AI can produce working code that still breaks flows. Riftbook includes review and graph intelligence tools for blast radius, risk, and test gaps. |
|
Start with the real playbook. |
Copy reusable skills into your AI coding setup. |
Use tested CLIs, MCP servers, review tools, and agent infrastructure. |
Use workflows for research, design, debugging, graph context, and shipping. |
| If you want to | Start here |
|---|---|
| Learn how to work with coding agents | The Real Vibe Coding Playbook |
| Build your first serious agent setup | Getting Started |
| Find practical agent tools | Tools |
| Add reusable agent behavior | Skills |
| Reduce context waste | Token Efficiency |
| Improve frontend output | Impeccable and Taste Skill |
| Review risky changes | Code Review Graph |
| Add graph-based repo context | Graphify |
| Explore graph memory and RAG | Graph Intelligence Stack |
| Study foundations | Learning |
A growing field guide for learning how to actually work with AI coding agents. Built from real workflows, real mistakes, and real lessons, not theory-only advice.
If you use coding agents but still feel messy, slow, or unsure where to start, begin here. Read the featured case study: How I Use Claude Code and Delegate Team.
| Section | What is inside |
|---|---|
| Getting Started | Step zero, project truth, lead agent setup, context files, first moves |
| Core Workflows | Planning, multi-agent coordination, debugging, reviewing, and shipping |
| Mistakes | Common mistakes, personal notes, and things that look smart but hurt |
| Stories | Real accounts from actual use: what worked, what failed, what changed |
| Paths | Reading order by role: beginner, solo builder, frontend builder, agency operator |
The best starting set across design, delegation, review, context, and AI-assisted frontend quality.
| Pick | Category | Why it matters | Card |
|---|---|---|---|
| Impeccable | AI Frontend Design | Gives coding agents practical UI judgment for audit, critique, polish, layout, typography, hardening, and live iteration |  |
| Delegate Team | Agent Delegation Runtime | Lets Claude Code delegate focused tasks to Codex, MiniMax, Gemini, OpenCode, VertexCoder, or team-style workflows while keeping review centralized | |
| React Doctor | React Quality Gate | Catches React issues across state, effects, performance, architecture, security, and accessibility after the agent builds the UI | |
| Graphify | Project Intelligence | Turns a repo into a queryable graph so agents understand structure before editing | |
| Code Review Graph | Review Intelligence | Reviews PRs and local changes through blast radius, affected flows, test gaps, and targeted context | |
| Taste Skill | Frontend Design | Helps agents avoid generic frontend output and make better visual decisions | |
Graph tools are powerful only when they are used for the right job. Riftbook now separates four graph layers clearly.
| Layer | Tool | Use it for | Card |
|---|---|---|---|
| Project map | Graphify | Understanding a repo or mixed project folder before editing | Open |
| Corpus reasoning | Microsoft GraphRAG | Reasoning over private documents, reports, transcripts, policies, and research notes | Open |
| Temporal memory | Graphiti | Agents that need evolving memory, provenance, current facts, and historical facts | Open |
| Review blast radius | Code Review Graph | Reviewing PRs and uncommitted changes through affected files, symbols, flows, and tests | Open |
Start with the full map:
Run it as a workflow:
Open Graph Intelligence Workflow
| Category | What you will find | Link |
|---|---|---|
| Playbook | Field guide for working with AI coding agents: lessons, mistakes, stories, and paths | Open |
| Skills | Reusable agent instructions, behaviors, and custom prompt logic layers | Open |
| Tools | IDE extensions, MCP servers, terminal utilities, review tools, and agent infrastructure | Open |
| Prompts | Curated prompts for coding, design, writing, research, and automation | Open |
| Workflows | Step-by-step ways to debug, research, design, review, and ship | Open |
| Frameworks | Architecture-level systems for retrieval, reasoning, context, and memory | Open |
| Guides | Setup playbooks, best practices, and configuration guides | Open |
| Indexes | Discovery hubs for AI-assisted coding resources | Open |
| Learning | Courses, tutorials, and study paths for agentic systems | Open |
| Templates | Standard markdown patterns for prompts, skills, tools, and workflows | Open |
| Cheat Sheets | Dense references and command lookups | Open |
| Execution Playbooks | End-to-end handbooks for product launch and build methods | Open |
| Resources | External libraries, papers, and reference documentation | Open |
| Examples | Real prompt examples, case studies, and before-after comparisons | Open |
| Ideas | Backlog of product, automation, and research concepts for builders | Open |
The complete index of curated cards in this repo.
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Caveman | Token Efficiency | Cuts verbose agent output and makes long coding sessions easier to scan | |
| RTK | Token Efficiency | Compresses terminal output before it enters the AI context window, reducing CLI noise during long coding sessions | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Graphify | Project Intelligence | Turns a repo into a queryable graph so agents understand structure before editing | |
| Graph Intelligence Stack | Graph Systems | Shows when to use Graphify, GraphRAG, Graphiti, and Code Review Graph | |
| Graph Intelligence Workflow | Workflow | Gives agents a step-by-step process for graph-informed coding, review, RAG, and memory design | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Impeccable | AI Frontend Design | Gives AI coding agents practical design judgment for audit, critique, polish, layout, typography, hardening, and live iteration | |
| Taste Skill | Frontend Design | Gives AI coding agents stronger UI taste and helps avoid generic frontend output | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Last30Days | Recent-signal research | Searches recent social, developer, market, GitHub, and web signals before meetings, research, validation, or comparisons | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Code Review Graph | PR Review Infrastructure | Builds a local graph of the repo so AI agents can review changes through blast radius, risk, affected flows, and targeted context | |
| Open Code Review | AI Review Automation | Runs structured AI code reviews on diffs, commits, branches, or full-file scans with line-level comments and configurable review rules | |
| reviewdog | CI Review Bridge | Turns linter and static-analysis output into PR comments, checks, and annotations so vibe-coded changes get readable feedback inside review | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Serena | MCP Semantic Coding Toolkit | Gives agents IDE-like symbol navigation, semantic editing, refactoring, and project memory | |
| Delegate Team | Agent Delegation Runtime | Lets Claude Code delegate focused tasks to Codex, MiniMax, Gemini, OpenCode, VertexCoder, or team workflows while keeping review centralized | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| React Doctor | React Quality Audit | Catches React issues across state, effects, performance, architecture, security, and accessibility after the agent builds the UI | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Microsoft GraphRAG | RAG Infrastructure | Builds graph-based retrieval pipelines for reasoning over private or complex document collections | |
| Graphiti | Agent Memory | Builds temporal context graphs for agents that need changing facts, provenance, and session-to-session memory | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Awesome Claude Code | Claude Code Ecosystem | Helps discover Claude Code skills, hooks, agents, commands, plugins, and workflow resources | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| Claude Code Best Practice | Claude Code | Helps move from casual prompting to structured Claude Code workflows using commands, agents, skills, hooks, MCP, and memory | |
| Resource | Category | Why it matters | Card |
|---|---|---|---|
| AI Agents for Beginners | AI Agents | A Microsoft learning path for understanding how agents work before building with them | |
- Copy and adapt: Do not just read. Copy prompts, templates, and skill files into your workspace configuration.
- Install selectively: Add tools only when they solve a specific problem in your workflow.
- Run workflows: Use the workflow docs before debugging, refactoring, reviewing, or shipping.
- Keep generated files controlled: Graph outputs, indexes, and local tool artifacts should not be committed unless the repo policy says so.
- Verify important claims: Treat graph output and AI summaries as guidance. Check source files before risky edits.
Every item added to Riftbook should follow this schema:
- Name: Clear and descriptive name.
- Category: Exact folder and file classification.
- What it is: Concise description of what the item does.
- Why it matters: Practical value in real AI-assisted work.
- When to use: The precise workflow phase or problem it addresses.
- How to use: Step-by-step instructions.
- Commands and prompts: Executable code blocks and agent instructions.
- Good fit and not a good fit: Clear target use cases and limits.
- Notes and limitations: Warnings, gotchas, and safety checks.
- Official link: Verified link to the target resource.
To keep the repo useful, Riftbook avoids:
- Untested links or products
- Hype-only AI tools with no practical coding value
- Empty prompt lists like "Act as a developer"
- Broad advice without explicit use cases
- Outdated, broken, or unmaintained repositories
Riftbook is supported by products built for real AI-assisted workflows.
|
Talk to AI like a teammate. A structured agency workflow layer for planning, strategy, and execution with AI.
|
Want to support Riftbook? See SPONSORS.md.
Read CONTRIBUTING.md for formatting rules, templates, and pull request guidance.
If Riftbook saves you time, helps you find a useful tool, or gives you a better workflow:
- Star the repo
- Share it with another AI-assisted builder
- Open a PR with a workflow-tested card
Riftbook gets better when builders contribute tools they have actually used.
Thanks to the creators, developers, and maintainers of the repositories featured in Riftbook.
vibe-coding · awesome-vibe-coding · claude-code · ai-agents · coding-agents · awesome-list · developer-tools · mcp-servers · metagpt · openai-codex · vertex-ai · prompt-engineering · agentic-workflows · knowledge-graph · graphrag · agent-memory · riftbook