Works with the AI Agents you already use
Luca
Get a working understanding of technical topics or codebases fast using a structured, ladder-based learning method.
Free
Luca
Example session with this skill installed
Help me understand how the auth flow works in this Next.js project. I need to be able to debug it if it fails.
- Read your context and instructions
- Compiled the luca
I have analyzed the repository. The auth flow uses NextAuth.js with a custom JWT callback and a PostgreSQL adapter. Since you need to be able to debug it, we will aim for the
Vouch rung.
I've created the bottega/auth-flow/ directory and initialized your learning map.
Learning Map: Next.js Auth Flow
| Region | Items | Status | Key Question |
|---|---|---|---|
| Client Side | useSession, <SessionProvider> | 🧩 Scout | How does the hook trigger a refresh? |
| Middleware | middleware.ts, Protected Routes | 🧩 Scout | Where exactly does the redirect happen? |
| API Layer | [...nextauth].ts, Providers | 🧩 Scout | How are provider secrets injected? |
| Database | Adapter, User/Session Schemas | 🧩 Scout | What is the cleanup policy for sessions? |
Route
- Stop 1: The Handshake. Predict the lifecycle of a login request. (Target: Scout)
- Stop 2: JWT Mutation. Modify the token payload in the callback. (Target: Wield)
- Stop 3: Session Expiry. Simulate a database connection failure during auth. (Target: Vouch)
Level Contract: At
Vouch, you will be able to debug the flow under pressure and review changes to the auth logic. This takes significantly more effort than just using the useSession hook.
Calibration Question: If you delete a user from the database, does their active browser session terminate immediately?
Next steps
- Review the generated
bottega/auth-flow/MAP.mdfor the full architecture diagram. - Answer the calibration question above.
- Run the first experiment in
bottega/auth-flow/lab/01_lifecycle_trace.ts.
Connects securely to your tools. The creator never sees your data.
About this skill
Technical debt and codebase complexity often force developers into a "shipping blindly" trap where they commit code they don't fully understand. Luca stops the cycle by delivering a working understanding of a technical topic or codebase through just-in-time theory and hands-on validation.
What it does
- Mental Mapping identifies the regions of a topic or codebase, highlighting unknown unknowns and potential theory blind spots.
- Rung-Based Learning tracks progress through a four-stage ladder: Scout, Wield, Reforge, and Vouch, ensuring the depth of study matches the stakes of the work.
- Predictive Exercises forces the learner to write predictions for code behavior before execution, turning wrong guesses into captured insights.
- Knowledge Persistence maintains a
bottega/directory with a live map, runnable labs, and "click lines" that record exactly when a concept finally clicked. - Adaptive Scaffolding provides fill-in-the-blank code that fades over time, transitioning the learner from guided exercises to writing from memory.
How it works
- Scope and Resume looks for an existing
bottega/folder to pick up where a previous session left off or reads the target codebase to map its shape. - Calibration checks prior knowledge via 2-3 targeted questions to avoid re-teaching known concepts.
- Route Construction creates a checkbox list of stops in
MAP.md, each with a target rung and time estimate. - Lab Exercises generates small, runnable experiments in a
lab/directory, focusing on code execution over verbal quizzes.
Frameworks & tools
The skill operates within a local file system, creating and managing a bottega/ directory. It uses Mermaid for architectural diagrams and Git for managing throwaway experiment branches.
Why this beats prompting it yourself
Generic prompts often lead to "wall of text" explanations that feel clear but fail to stick in long-term memory. Luca enforces a structured pedagogical approach that prioritizes prediction, active recall, and evidence-based progress over passive reading.
Use cases
- Codebase Onboarding to map a large repository and reach the Vouch rung for critical systems.
- Technical Deep Dives to internalize complex patterns like Rust ownership or distributed consensus without watching hours of video.
- AI-Generated Code Review to verify that you actually understand the logic an AI assistant just proposed.
Known limitations
Does not handle simulated-experience training like incident response drills. Defers to repository-specific learning skills if they already exist in the codebase.
Not for high-pressure bug fixing or quick factual lookups. Requires a local filesystem to manage the bottega directory. Does not support incident response simulations.
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
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 8 days ago
- Free with an Agensi account
Requires