Skip to main content

​
Overview

After uploading ZIP files, the analysis pipeline discovers Git repositories, extracts intelligence about technology stacks and collaboration patterns, and calculates user-specific contribution statistics.

​
Analysis Pipeline

The analysis process consists of four major stages:
1

Repository Discovery

Scans the extracted ZIP for directories containing .git folders.
# Automatically discovers all Git repos in extraction path
.extracted/1/
├── web-app/.git/          ← Discovered
├── mobile-app/.git/       ← Discovered
└── data-analysis/.git/    ← Discovered
2

Repository Intelligence

Analyzes each repository to extract:
  • Languages and percentages
  • Frameworks and dependencies
  • Commit history and windows
  • Collaboration metadata
  • Health score (0-100)
3

User Contribution Analysis

Calculates user-specific metrics:
  • Total commits by the user
  • Contribution percentage
  • Commit frequency (commits/week)
  • Activity breakdown
  • Inferred role (owner, contributor, etc.)
4

Skills and Evidence Extraction

Runs deep analysis to derive:
  • Skills detected from languages/frameworks
  • Proficiency levels
  • Project evidence and insights
  • Code quality signals

​
Starting Analysis

Use the POST /analyze/{zip_id} endpoint to begin the full pipeline:

​
Basic Analysis

Analyze all Git repositories in the entire ZIP: 
curl -X POST "http://127.0.0.1:8000/analyze/1" \
  -H "Content-Type: application/json"

​
Directory-Scoped Analysis

Analyze only selected directories: 
curl -X POST "http://127.0.0.1:8000/analyze/1" \
  -H "Content-Type: application/json" \
  -d '{
    "directories": ["web-app", "mobile-app"]
  }'
Use directory scoping to exclude irrelevant folders (e.g., node_modules/, vendor/) or to incrementally analyze large archives.

​
Analysis Response

The endpoint returns comprehensive results: 
{
  "zip_id": 1,
  "extraction_path": ".extracted/1",
  "repos_found": 3,
  "repos_analyzed": [
    {
      "project_name": "web-app",
      "project_path": ".extracted/1/web-app",
      "frameworks": ["React", "Express.js"],
      "languages": ["JavaScript", "TypeScript", "CSS"],
      "skills_count": 12,
      "insights_count": 8,
      "user_contribution_pct": 87.5,
      "user_total_commits": 147,
      "user_commit_frequency": 3.2,
      "user_first_commit": "2025-01-15T09:00:00",
      "user_last_commit": "2026-02-28T16:45:00"
    },
    {
      "project_name": "mobile-app",
      "project_path": ".extracted/1/mobile-app",
      "frameworks": ["React Native", "Expo"],
      "languages": ["TypeScript", "JavaScript"],
      "skills_count": 9,
      "insights_count": 6,
      "user_contribution_pct": 45.2,
      "user_total_commits": 68,
      "user_commit_frequency": 2.1,
      "user_first_commit": "2025-09-01T10:30:00",
      "user_last_commit": "2026-01-20T14:00:00"
    }
  ],
  "rankings": [
    {
      "name": "web-app",
      "score": 87.5,
      "total_commits": 168,
      "user_commits": 147
    },
    {
      "name": "mobile-app",
      "score": 45.2,
      "total_commits": 150,
      "user_commits": 68
    }
  ],
  "summaries": [
    {
      "project_name": "web-app",
      "summary": "Full-stack web application with React frontend and Express backend. Primary contributor implementing authentication, real-time updates, and responsive UI."
    }
  ],
  "consent_level": "local-llm",
  "user_email": "[email protected]"
}

​
Repository Intelligence Metrics

For each discovered repository, the system computes:

​
Technology Stack

Languages detected:
  • File extension analysis (.js, .py, .java, etc.)
  • Line-of-code percentages
  • Primary language identification
Frameworks identified:
  • package.json dependencies (Node.js projects)
  • requirements.txt / Pipfile (Python projects)
  • pom.xml / build.gradle (Java projects)
  • Configuration files (.eslintrc, tsconfig.json, etc.)

​
Collaboration Metrics

is_collaborative flag:
  • true if multiple Git authors detected
  • false if single author (solo project)
collaboration_metadata: 
{
  "total_authors": 4,
  "commit_distribution": {
    "[email protected]": 147,
    "[email protected]": 18,
    "[email protected]": 3
  }
}

​
Health Score

Repositories receive a health score (0-100) based on:
  • Commit recency: Recent activity scores higher
  • Commit frequency: Regular commits indicate active maintenance
  • Documentation: Presence of README, comments, docs/
  • Testing: Test files and coverage indicators
  • Code organization: Clear directory structure
Health scoring is deterministic and does not require LLM consent.

​
User Contribution Statistics

​
Filtering by Email

The analysis uses your configured email (from /answers) to filter contributions: 
# System matches your email against Git commit authors
user_email = "[email protected]"
user_commits = [c for c in commits if c.author.email == user_email]

​
Contribution Metrics

userStatspercentages:
  • Your commits / total repository commits × 100
  • Indicates ownership level (>70% suggests primary author)
commitFrequency:
  • Average commits per week during active development
  • Calculated as: user_total_commits / weeks_between_first_and_last
activity_breakdown: 
{
  "additions": 12543,
  "deletions": 3421,
  "files_changed": 287,
  "commits_by_month": {
    "2025-09": 12,
    "2025-10": 18,
    "2025-11": 15
  }
}
user_role: Automatically inferred based on contribution percentage:
  • Owner: >70% of commits
  • Core Contributor: 30-70%
  • Contributor: 10-30%
  • Minor Contributor: <10%
You can override the inferred role using PUT /projects/{project_id}/role.

​
Repository Ranking

After analyzing all repositories, the system ranks them by your contribution level: 
{
  "rankings": [
    {
      "name": "web-app",
      "score": 87.5,
      "total_commits": 168,
      "user_commits": 147
    },
    {
      "name": "mobile-app",
      "score": 45.2,
      "total_commits": 150,
      "user_commits": 68
    },
    {
      "name": "data-analysis",
      "score": 12.3,
      "total_commits": 89,
      "user_commits": 11
    }
  ]
}
Rankings help prioritize which projects to showcase in your portfolio.

​
AI-Generated Summaries

If you’ve enabled local-llm or cloud consent, the system generates natural language summaries: 
{
  "summaries": [
    {
      "project_name": "web-app",
      "summary": "Designed and implemented a full-stack web application for real-time collaboration. Led development of the React frontend with TypeScript, implemented WebSocket-based live updates, and built a RESTful API with Express.js. Contributed 87% of commits over 14 months."
    }
  ]
}
Summaries are generated for the top 3 ranked projects by default. Without LLM consent, the system uses template-based summaries instead.

​
Skills Extraction

The deep analyzer extracts skills from:
  • Languages: JavaScript, Python, Java, TypeScript, etc.
  • Frameworks: React, Django, Spring Boot, etc.
  • Infrastructure: Docker, CI/CD configs, cloud deployment files
  • Tools: Git, testing frameworks, build systems
Skills are stored with:
  • Proficiency level (0.0-1.0 scale)
  • Evidence (specific files or patterns that demonstrate the skill)
  • Category (Programming Language, Framework, Tool, etc.)
Access extracted skills via: 
curl "http://127.0.0.1:8000/skills"

​
Analyzing Individual Repositories

You can also analyze a single local repository without uploading a ZIP: 
curl -X POST "http://127.0.0.1:8000/repos/analyze" \
  -H "Content-Type: application/json" \
  -d '{
    "repo_path": "/Users/you/projects/my-app",
    "user_email": "[email protected]"
  }'
The POST /repos/analyze endpoint requires the repository to be accessible on the server’s filesystem.

​
Error Handling

​
No Git Repositories Found

{
  "detail": "No git repositories found in the uploaded ZIP file"
}
Solution: Ensure your projects contain .git directories. You can verify: 
ls -la project-directory/.git

​
User Has No Commits

{
  "repos_analyzed": [
    {
      "project_name": "team-project",
      "error": "User [email protected] has no commits in this repository"
    }
  ]
}
Solution: Verify your email matches your Git commit author email: 
cd project-directory
git log --author="[email protected]"

​
Performance Considerations

Large repositories: Analysis time scales with:
  • Number of commits (>1000 commits may take 30+ seconds)
  • Number of files (scans all file types)
  • LLM consent level (cloud LLM adds 5-10 seconds per project)

​
Next Steps

After analysis completes:
  1. View project details - Access extracted intelligence
  2. Generate portfolio - Create resume and portfolio artifacts
  3. Customize output - Reorder projects, highlight skills, edit summaries

​
API Reference

​
POST /analyze/

Orchestrates full analysis pipeline for an uploaded ZIP. Path Parameters:
  • zip_id (integer, required): Database ID from /zip/upload
Request Body (optional): 
{
  "directories": ["web-app", "mobile-app"]
}
Returns: AnalyzeResponse with repos analyzed, rankings, and summaries

​
POST /repos/analyze

Analyzes a single local repository. Request Body: 
{
  "repo_path": "/path/to/repo",
  "user_email": "[email protected]"
}
Returns: Repository stats and user contribution metrics

Build docs developers (and LLMs) love

Get started for freeTalk to us