Skip to content

implementation_plan.md

Latest commit

 

History

History
98 lines (79 loc) · 4.65 KB

implementation_plan.md

File metadata and controls

98 lines (79 loc) · 4.65 KB
Raw

Kodra.ai Implementation Plan

Goal Description

Transform the existing "CareerConnect" (Ruvva) microservices project into Kodra.ai, an Autonomous Career Agent for the WeMakeDevs Hackathon. The goal is to build a platform that analyzes a user's GitHub code and generates personalized learning "Missions" to improve their skills.

Key Pivot:

  • From: Generic Career Counseling (Assessment -> Career Path)
  • To: Coding Mentorship (Code Analysis -> Learning Mission)

User Review Required

Important

Data Loss Warning: This plan involves deleting existing career-specific features (College database, RIASEC assessments, Mentor bookings) to focus on the Code Analysis features. Sponsor Requirements: We must strictly adhere to the technical requirements for Cline, Kestra, Oumi, Vercel, and CodeRabbit as detailed in we make dev.md.

Proposed Changes

1. Cleanup & Refactoring

We will first slim down the existing application to remove noise.

[DELETE] / Modify

  • Remove College, Career, Booking, Mentor entities and controllers.
  • Rename generic Assessment entity to CodeAnalysis or similar, or adapt it.

2. Microservices Architecture Updates

[MODIFY] Backend (Spring Boot)

  • Database Management:
    • Switch to Supabase (PostgreSQL).
    • Remove Firestore dependencies.
    • Create Tables: users, missions, code_analysis, skill_history.
  • GitHub Integration:
    • Add GitHubService to handle OAuth and API calls.
    • Add WebhookController to listen for GitHub and CodeRabbit events.
  • New Entities:
    • Mission (Title, Description, Status, RelatedRepo, ExpectedFix).
    • CodeAnalysis (Stores logic/security/performance scores).
    • SkillHistory (Tracks progress over time).
  • Logic:
    • Implement MissionService to generate tasks based on low scores from the AI service.

[MODIFY] AI Service (Python/Flask)

  • Code Analysis Engine:
    • Implement code_analyzer.py using AST/Regex to detect:
      • Exposed Secrets (Security)
      • console.log overuse (Best Practices)
      • SQL Injection patterns (Security)
      • Missing Error Handling (Logic)
  • Integration:
    • Oumi: Use Oumi logic (mock or real) to refine pattern detection scoring.
    • Kestra: Expose endpoints for Kestra to trigger batch analysis.

[MODIFY] Frontend (React/Kodra)

  • New Pages:
    • GitHubConnect.js: OAuth entry point.
    • Dashboard.js: Replacement for the old Home page. Shows Code Health graphs.
    • MissionList.js & MissionDetail.js: The core learning UI.
  • Chat:
    • Update the AI Chat interface to offer "Pair Programming" context (inject file context into prompt).

3. Sponsor Integrations Plan

Cline ($5k)

  • Plan: We will use Cline to generate the MissionService logic and Frontend Components. We will document this usage.
  • Feature: Build a "Teacher Mode" assistance prompt that Cline can use.

Kestra ($4k)

  • Plan: Define a Kestra Flow (YAML) that orchestrates the "Diagnosis" phase.
    1. Trigger: User Connects GitHub.
    2. Step 1: Clone Repo (Python).
    3. Step 2: Run Analysis (Python).
    4. Step 3: Summarize Findings (AI Agent).
    5. Step 4: Post results to Backend.

Oumi ($3k)

  • Plan: Integrate Oumi in the Python service.
  • Feature: "Feedback Loop". When a user marks a Mission as "Too Hard", use Oumi optimized model to adjust the difficulty of future mission pattern matching.

CodeRabbit ($1k)

  • Plan: Setup a repo-level webhook.
  • Flow: User pushes Mission fix -> CodeRabbit reviews PR -> Webhook hits Backend -> Update Mission Status to "Verified" if review is clean.

Vercel ($2k)

  • Plan: Deploy kodra (frontend) directory to Vercel. Ensure proxy settings in package.json are handled via Environment Variables for production.

Verification Plan

Automated Tests

  • Backend: Unit tests for GitHubService and MissionGenerator.
  • AI Service: Test code_analyzer.py against a known "Vulnerable Repo" to ensure it catches all 5 patterns.

Manual Verification (The Demo Flow)

  1. Login: Sign in with Firebase (reuse existing) - Correction: Will migrate to Supabase Auth if possible, or keep Firebase Auth and sync to Supabase. Plan implies full switch, but Auth might stick to Firebase for ease if just DB is needed. Let's assume full switch or Supabase for DB.
  2. Connect: Click "Link GitHub".
  3. Analyze: Verify Dashboard updates with "Real" data from the linked repo.
  4. Mission: Click a mission, see the details.
  5. Fix: Push a fix to the repo.
  6. Verify: Check if CodeRabbit webhook updates the dashboard score.