my-project-public

repository

loading code, commits, and activity

repositories

loading repo index

#1	<p align="center">
#2	<img src="https://img.shields.io/badge/claude--code-skill-9945FF?style=for-the-badge&logo=data:image/svg+xml;base64,PHN2ZyB3aWR0aD0iMjQiIGhlaWdodD0iMjQiIHZpZXdCb3g9IjAgMCAyNCAyNCIgZmlsbD0ibm9uZSIgeG1sbnM9Imh0dHA6Ly93d3cudzMub3JnLzIwMDAvc3ZnIj48Y2lyY2xlIGN4PSIxMiIgY3k9IjEyIiByPSIxMCIgZmlsbD0iIzk5NDVGRiIvPjwvc3ZnPg==&logoColor=white" alt="Claude Code Skill">
#3	<img src="https://img.shields.io/badge/solana-blockchain-14F195?style=for-the-badge&logo=solana&logoColor=white" alt="Solana">
#4	<img src="https://img.shields.io/badge/license-MIT-blue?style=for-the-badge" alt="MIT License">
#5	</p>
#6
#7	<h1 align="center">
#8	ULTRATHINK × BLOCKCHAIN
#9	</h1>
#10
#11	<p align="center">
#12	<strong>A Claude Code skill for building production Solana systems with deep reasoning.</strong>
#13	</p>
#14
#15	<p align="center">
#16	Turn Claude Code into a blockchain-native development partner that thinks like a senior Solana engineer — with MEV awareness, transaction atomicity, retry logic, and production hardening baked into every prompt.
#17	</p>
#18
#19	---
#20
#21	## What This Is
#22
#23	Ultrathink Blockchain is an open-source [Claude Code skill](https://docs.anthropic.com/en/docs/claude-code) that teaches Claude how to reason about blockchain development — specifically Solana — at a production level.
#24
#25	It combines two ideas:
#26
#27	1. Ultrathink — Claude Code's extended thinking mode, focused on the areas that matter most for chain development (transaction structure, MEV, state races, security)
#28	2. A structured prompting protocol — context dump, interview, plan, constraints — that prevents generic web2 patterns from leaking into your on-chain code
#29
#30	The result: Claude Code writes blockchain code that actually ships to mainnet.
#31
#32	---
#33
#34	## The Formula
#35
#36	```
#37	CONTEXT + INTENT + INTERVIEW + ULTRATHINK + PLAN + CONSTRAINTS → PRODUCTION CODE
#38	```
#39
#40	\| Phase \| What Happens \|
#41	\|-------\|-------------\|
#42	\| Context Dump \| Prime Claude with chain, stack, and constraints \|
#43	\| Intent \| Dimensional goal with success criteria \|
#44	\| Interview \| Requirements extraction (3-5 questions at a time) \|
#45	\| Ultrathink \| Deep reasoning on TX, MEV, races, security \|
#46	\| Plan \| Accounts, instructions, error taxonomy, tests \|
#47	\| Constraints \| Retries, simulation, Jito, timeouts \|
#48	\| Execute \| Full files, production patterns \|
#49	\| Iterate \| Review, steer, continue \|
#50
#51	---
#52
#53	## Quick Start
#54
#55	### Install the Skill
#56
#57	```bash
#58	# Clone the repo
#59	git clone https://github.com/YOUR_USERNAME/ultrathink-blockchain.git
#60
#61	# Install the skill into Claude Code
#62	cp -r ultrathink-blockchain/skill ~/.claude/skills/ultrathink-blockchain
#63	```
#64
#65	Or use the one-liner:
#66
#67	```bash
#68	curl -fsSL https://raw.githubusercontent.com/YOUR_USERNAME/ultrathink-blockchain/main/install.sh \| bash
#69	```
#70
#71	### Use It
#72
#73	Once installed, start a Claude Code session and the skill activates automatically for blockchain work. Here's the one-shot production prompt:
#74
#75	```
#76	Context: Solana mainnet-beta production environment.
#77	Stack: TypeScript, Helius RPC + websockets, Birdeye data, Jito execution.
#78
#79	I want [your goal with success metrics].
#80
#81	Before writing code, interview me about requirements, constraints,
#82	existing infrastructure, and edge cases. Ask 3-5 questions at a time.
#83
#84	After the interview:
#85	1. Reflect requirements back for confirmation
#86	2. ultrathink about:
#87	- Transaction structure and instruction ordering
#88	- Account validation and PDA derivation
#89	- MEV exposure and protection strategy
#90	- Failure modes and retry logic
#91	- State race conditions
#92	3. Present plan (accounts, instructions, error taxonomy, tests)
#93	4. Wait for my approval
#94
#95	Write production code. I'm shipping this.
#96	```
#97
#98	---
#99
#100	## What's Inside
#101
#102	```
#103	ultrathink-blockchain/
#104	├── README.md # You are here
#105	├── LICENSE # MIT
#106	├── CONTRIBUTING.md # How to contribute
#107	├── install.sh # One-line installer
#108	├── skill/ # The Claude Code skill
#109	│ ├── SKILL.md # Core skill definition
#110	│ └── references/ # Reference docs the skill loads
#111	│ ├── antipatterns.md # Common blockchain antipatterns
#112	│ └── templates.md # Domain-specific prompt templates
#113	├── docs/ # Extended documentation
#114	│ ├── formula.md # The complete Ultrathink formula
#115	│ ├── depth-ladder.md # Extended thinking depth guide
#116	│ └── blockchain-patterns.md # Solana-specific patterns
#117	└── site/ # Visual reference (landing page)
#118	└── index.html # Interactive formula visualization
#119	```
#120
#121	---
#122
#123	## The Depth Ladder
#124
#125	\| Invocation \| Token Budget \| Best For \|
#126	\|------------\|-------------\|----------\|
#127	\| `think` \| ~500 tokens \| Simple decisions, quick fixes \|
#128	\| `think step by step` \| ~1000 tokens \| Multi-step problems, debugging \|
#129	\| `think hard` \| ~2000 tokens \| Architecture decisions, complex logic \|
#130	\| `think harder` \| ~4000 tokens \| System design, security analysis \|
#131	\| `ultrathink` \| ~8000+ tokens \| Production systems, critical code \|
#132	\| `megathink` \| Maximum depth \| Novel problems, research-grade work \|
#133
#134	The tradeoff: deeper thinking = better reasoning but slower response and higher token cost. Use the right level for the task.
#135
#136	---
#137
#138	## Blockchain Focus Areas
#139
#140	When you tell Claude to `ultrathink` about blockchain problems, it focuses on:
#141
#142	- Transaction Atomicity — instruction ordering, PDAs, compute units, lookup tables
#143	- MEV Exposure — sandwich risk, backrun opportunities, Jito bundle decisions
#144	- State Races — stale blockhash, account contention, retry strategies
#145	- Security — attack vectors, authority checks, reentrancy paths
#146	- Failure Modes — what breaks at 3am with no one watching
#147
#148	---
#149
#150	## Templates
#151
#152	The skill includes production-ready prompt templates for common blockchain patterns:
#153
#154	- Token Sniper / Trading Bot — websocket monitoring, Jito execution, exit strategies
#155	- DeFi Protocol Integration — instruction sequences, account schemas, edge cases
#156	- Indexer / Analytics Pipeline — ingestion, schema design, reorg handling
#157	- Anchor Program (Rust) — account sizing, PDA derivation, security invariants
#158	- Multi-Agent Trading System — agent coordination, state machines, risk management
#159
#160	See [`docs/formula.md`](docs/formula.md) for the complete guide with all templates.
#161
#162	---
#163
#164	## Never Ship Without
#165
#166	- [ ] Retry logic on all RPC calls
#167	- [ ] Transaction simulation before send
#168	- [ ] Dynamic priority fees via Helius
#169	- [ ] Jito bundles for value transactions
#170	- [ ] Explicit timeouts on network ops
#171	- [ ] Graceful shutdown / position cleanup
#172	- [ ] Structured logging with correlation IDs
#173	- [ ] Circuit breakers for cascading failures
#174
#175	---
#176
#177	## Contributing
#178
#179	See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines. In short:
#180
#181	1. Fork the repo
#182	2. Create a feature branch
#183	3. Make your changes
#184	4. Submit a PR with a clear description
#185
#186	We especially welcome contributions for:
#187
#188	- New blockchain prompt templates (EVM, Sui, Aptos, etc.)
#189	- Antipattern documentation
#190	- Skill refinements based on real-world usage
#191	- Translations
#192
#193	---
#194
#195	## License
#196
#197	MIT — see [LICENSE](LICENSE) for details.
#198
#199	---
#200
#201	<p align="center">
#202	<strong>ultrathink. build on-chain. ship to mainnet.</strong>
#203	</p>
#204

z6Mkq5mY3JWtxoxUobWcfNHm7AkRubgSWEZTkBVqZXJviFZ5/my-project-public