Skip to main content

​
Overview

Artifact Miner implements consent-gated data processing with four privacy levels. You control whether analysis uses local-only processing, local LLM assistance, or cloud-based AI services. Choose the consent level that matches your privacy and feature requirements:

none

No data processingAnalysis is disabled. Use this to review consent options before proceeding.

local

Local processing onlyDeterministic analysis using file patterns, Git statistics, and repository metrics. No LLM integration.
  • File type detection
  • Framework identification
  • Git commit analysis
  • User contribution metrics

local-llm

Local LLM processingUses locally-hosted Ollama models for AI-powered summaries and insights. Data never leaves your machine.
  • All features from local
  • AI-generated project summaries
  • Natural language insights
  • Requires Ollama installation

cloud

Cloud LLM processingUses OpenAI API for advanced AI summaries. Code snippets and commit messages are sent to OpenAI.
  • All features from local-llm
  • More sophisticated summaries
  • Better contextual understanding
  • Requires OpenAI API key
1

Check current consent

Retrieve your current consent configuration:
curl "http://127.0.0.1:8000/consent"
Response:
{
  "consent_level": "none",
  "accepted_at": null
}
2

Update consent level

Set your preferred consent level:
curl -X PUT "http://127.0.0.1:8000/consent" \
  -H "Content-Type: application/json" \
  -d '{
    "consent_level": "local-llm"
  }'
Response:
{
  "consent_level": "local-llm",
  "accepted_at": "2026-03-05T14:22:30"
}
Consent is stored in the database and persists across sessions. You can change it at any time.

​
Local Processing (local)

When you select local consent, the system performs:
  • Language detection: Analyzes file extensions and content patterns
  • Framework identification: Detects dependencies in package.json, requirements.txt, etc.
  • Git statistics: Counts commits, analyzes commit frequency, calculates contribution percentages
  • Health scoring: Evaluates repository quality metrics (commit activity, documentation, testing)
  • Skill extraction: Maps detected languages and frameworks to skill categories
No data leaves your machine. All processing is deterministic and template-based.

​
Local LLM Integration (local-llm)

Requires Ollama to be installed and running locally. See Ollama installation guide.
With local-llm, the system additionally:
  • Generates natural language summaries of your contributions
  • Creates project descriptions from Git history
  • Produces insights about collaboration patterns
Setup Ollama: 
# Install Ollama (macOS)
brew install ollama

# Start Ollama service
ollama serve

# Pull a model (e.g., llama2)
ollama pull llama2
Configuration: Ensure your .env file contains: 
OLLAMA_BASE_URL=http://localhost:11434
OLLAMA_MODEL=llama2

​
Cloud LLM Integration (cloud)

Requires an OpenAI API key. Code snippets and commit messages will be sent to OpenAI’s servers.
With cloud consent, the system uses OpenAI’s GPT models for:
  • Enhanced project summaries with better context understanding
  • More sophisticated skill proficiency estimation
  • Advanced insight generation
Setup OpenAI: 
# Set your API key in .env
OPENAI_API_KEY=sk-your-api-key-here

​
Privacy Considerations

​
What Data is Processed?

Depending on your consent level, the system may process:
Data Typelocallocal-llmcloud
File paths and names✓✓✓
Git commit metadata✓✓✓
User email (from Git)✓✓✓
Code additions (diffs)✓✓✓
Sent to external LLM✗✗✓
Leaves your machine✗✗✓

​
User Email Collection

Before analysis, you must provide your email address (matching your Git commit author email): 
curl -X POST "http://127.0.0.1:8000/answers" \
  -H "Content-Type: application/json" \
  -d '{
    "answers": {
      "email": "[email protected]",
      "end_goal": "Build a technical portfolio for job applications"
    }
  }'
This email is used to:
  • Filter your contributions from collaborative repositories
  • Calculate your commit percentage
  • Attribute skills and evidence to your work
Your email is stored locally in the SQLite database and never transmitted externally (unless you use cloud consent, in which case it may appear in Git commit metadata sent to OpenAI).

​
Switching Between LLM Providers

The system automatically selects the LLM provider based on your consent level: 
# Automatically determined by consent_level
if consent_level == "cloud":
    # Uses OpenAI helper (src/artifactminer/helpers/openai_helper.py)
    llm_provider = "openai"
elif consent_level == "local-llm":
    # Uses Ollama helper (src/artifactminer/helpers/ollama_helper.py)
    llm_provider = "ollama"
else:
    # No LLM, template-based summaries only
    llm_provider = None

​
Configuration Questions

Beyond consent, the system collects answers to configuration questions:
1

List available questions

curl "http://127.0.0.1:8000/questions"
Response:
[
  {
    "id": 1,
    "key": "email",
    "question_text": "What is your email address (must match Git author email)?",
    "required": true,
    "answer_type": "email"
  },
  {
    "id": 2,
    "key": "end_goal",
    "question_text": "What is your goal for this analysis?",
    "required": true,
    "answer_type": "text"
  }
]
2

Submit answers

curl -X POST "http://127.0.0.1:8000/answers" \
  -H "Content-Type: application/json" \
  -d '{
    "answers": {
      "email": "[email protected]",
      "end_goal": "Showcase projects for internship applications"
    }
  }'

​
API Reference

​
GET /consent

Retrieve current consent state. Returns: 
{
  "consent_level": "local" | "local-llm" | "cloud" | "none",
  "accepted_at": "2026-03-05T14:22:30" | null
}

​
PUT /consent

Update consent level. Request Body: 
{
  "consent_level": "local" | "local-llm" | "cloud" | "none"
}
Returns: Updated consent state with timestamp

​
GET /questions

List all configuration questions. Returns: Array of question objects with id, key, question_text, required, and answer_type

​
POST /answers

Submit answers to configuration questions. Request Body: 
{
  "answers": {
    "email": "[email protected]",
    "end_goal": "Your analysis objective"
  }
}

​
Next Steps

After configuring consent:
  1. Upload projects - Submit ZIP files for analysis
  2. Analyze repositories - Extract intelligence from your code
  3. Generate outputs - Create portfolio artifacts

Build docs developers (and LLMs) love

Get started for freeTalk to us