- Home
- Skills
- Web3 & Crypto
- solana sensei
Works with the AI Agents you already use
solana sensei
A Socratic mentor for Solana and Anchor that refuses to write code, forcing developers to learn through guided inquiry.
Free
solana sensei
Example session with this skill installed
I'm trying to implement a PDA for a user profile in Anchor but I'm getting a seeds constraint error. Can you just write the correct account struct for me?
- Read your context and instructions
- Compiled the solana sensei
I cannot write the full account struct for you, as that would bypass the learning process. Instead, let's look at your design so we can figure out why the seeds are failing.
To start, please share the code snippet for your current account struct and the seeds you are trying to use. Once you've shared that, tell me:
- What specific pieces of data (seeds) are you using to derive this PDA?
- Why did you choose those specific seeds for a user profile?
- What is the exact error message you are seeing in your terminal or test logs?
Security Check
While you gather that, consider this: If you use only a static string like "profile" as a seed, what happens when a second user tries to create their profile?
Next steps
- Paste your current
#[account]struct and the instruction where you initialize it. - Identify which line in your test file or program code triggers the error.
- Explain your hypothesis for why the seeds might not be matching during initialization.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop leaking learning potential by letting agents write your code for you. This skill transforms your AI agent from a code generator into a Socratic mentor for Solana, Anchor, and Rust development. It forces the developer to engage with the logic, security, and architecture of their programs rather than outsourcing the thinking.
What it does
- Enforces Socratic dialogue by refusing to write core solution code and instead asking targeted questions about the developer's intent.
- Validates mental models by requiring the student to explain account structures, instruction flows, and PDA derivations in their own words.
- Debugs through hypothesis by guiding the user to locate error origins and test their own fixes before providing conceptual hints.
- Surfaces security risks using a built-in 8-point Anchor checklist covering signer validation, PDA seeds, and reentrancy.
- Verifies comprehension through "teach-back" loops where the user must explain the solution before the session concludes.
How it works
- Trigger a workflow by asking for help with an assignment, a Rust error, or a specific Solana concept like PDAs.
- Submit an attempt by providing your current code and your reasoning, which the agent uses to identify specific logic gaps.
- Analyze security during the review process as the agent prompts you to find vulnerabilities like integer overflows or missing constraints.
- Demonstrate mastery by explaining the final implementation back to the agent to prove you can replicate the logic without assistance.
Frameworks & tools
This skill is optimized for the Solana blockchain ecosystem, specifically utilizing the Anchor framework and Rust programming language.
Why this beats prompting it yourself
Standard LLMs default to "helpful assistant" mode, providing full code blocks that bypass the learning process. This skill embeds strict behavioral rules that prevent the agent from solving the problem for you, ensuring you actually learn how to build secure programs.
Use cases
- Classroom assignments where a student needs guidance on Anchor instructions without violating academic integrity.
- Junior developer onboarding for teams that want new hires to learn Solana security best practices through guided discovery.
- Complex debugging when a PDA derivation or account constraint is failing and you need to systematically isolate the root cause.
- Security self-audits to verify that every instruction account has been properly validated against common attack vectors.
Known limitations
The agent will refuse to complete TODOs or write full function implementations. It is not a code-writing tool and requires an active initial attempt from the user.
Refuses to write core logic or complete TODOs. Requires the user to provide their own code attempts and reasoning first.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 26 days ago
- Free with an Agensi account