- Home
- Skills
- Code Quality & Review
- code walkthrough
code walkthrough
Follow a traceable execution path through source code with file references and verified snippets.
Free
Works with the AI Agents you already use
code walkthrough
Example session with this skill installed
Explain how a user login request is handled in this repo, starting from the auth controller.
- Read your context and instructions
- Compiled the code walkthrough
I have traced the login flow from the initial request through to the session creation. This walkthrough identifies the validation logic, the database query, and the specific error handler for invalid credentials.
Login Request Walkthrough
| Step | File | Symbol/Reference | Description |
|---|---|---|---|
| 1. Entry | src/controllers/auth.ts | AuthController.login | Receives POST request, extracts email and password. |
| 2. Validate | src/middleware/validate.ts | loginSchema | Joi schema validation. Rejects if email format is invalid. |
| 3. Query | src/services/user.ts | UserService.findByEmail | Queries Postgres. Returns null if user not found. |
| 4. Verify | src/utils/crypto.ts | verifyPassword | Compares hashed password using bcrypt.compare. |
| 5. Token | src/lib/jwt.ts | generateToken | Signs a new JWT with user ID and 24h expiry. |
Source Snippet: Validation Logic
// src/middleware/validate.ts line 12
export const loginSchema = Joi.object({
email: Joi.string().email().required(),
password: Joi.string().min(8).required(),
});
Example Trace
- Input:
{ "email": "dev@example.com", "password": "password123" } - Transformation:
auth.tspasses payload tologinSchema. - Database:
UserServicereturns user object with IDusr_99. - Outcome: HTTP 200 with
Authorization: Bearer <token>.
Key Invariants
- Password comparison only occurs if the user record exists in the database.
- JWT secret is loaded from environment variables during the sign step.
Unknowns
src/lib/jwt.tsuses an external library for signing; the internal implementation ofjsonwebtokenwas not inspected.
Next steps
- Review
src/services/user.tsto see how connection pooling is handled. - Inspect
tests/integration/auth.test.tsto see how the invalid password case is tested.
Connects securely to your tools. The creator never sees your data.
About this skill
Onboarding to a new codebase is slow and error-prone. This skill prevents the "wall of text" explanation by forcing a traceable execution path through actual source files. It stops the agent from guessing how code works and requires it to cite specific files, symbols, and line references.
What it does
- Traceable execution follows the logic flow from entry points like requests, commands, or events in chronological order.
- Source verification extracts actual snippets from the repository rather than generating hallucinations from memory.
- Static analysis separation distinguishes between observed implementation details and inferred developer intent.
- Context mapping provides a flow map and identifies key invariants that must hold true across the path.
- Evidence-based tracing tracks inputs through transformations and error handling, linking back to existing tests.
How it works
- Identify the entry point and inspect the repository structure to locate relevant files.
- Follow the execution path, documenting dependencies and outputs at each step.
- Extract and cite code snippets for every critical logic gate or transformation.
- Produce a flow map including a concrete example trace and identified unknowns.
Frameworks & tools
This skill is language-agnostic and works with any repository reachable via the agent's file system tools. It relies on static analysis of source files and test suites.
Why this beats prompting it yourself
Standard prompts often result in high-level summaries that skip over edge cases or hallucinate function names. This skill enforces a strict linear walkthrough pattern that requires the agent to prove its logic with line-level evidence.
Use cases
- Onboarding developers to complex legacy modules by tracing a single successful request.
- Debugging logic by following an input path to identify exactly where validation or transformation fails.
- Audit preparation by mapping how sensitive data moves from an API endpoint to a database.
- Code review of large PRs to understand the actual runtime impact of new logic flows.
Known limitations
This skill does not execute code and relies on static analysis. It is not designed to perform code edits or refactors. Snippets are filtered to avoid leaking secrets.
Only performs static analysis, not execution. Cannot be used to modify code. Requires access to source files. Scope is limited to the provided repository context.
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 24 days ago
- Free with an Agensi account