~/systems
system·project·render: lazy

DevTrails Utility

100% offline Chrome history developer profiler, local ML classifier, and temporal code-linked RAG vector generator.

status: activediagrams: mermaidgraphs: react-flowconsole: 3-panel
loading graph
less overview_&_developer_profiling_vision.md
md

DevTrails is a 100% offline, privacy-first developer history profiler and local RAG engine.

Instead of sending browsing logs to corporate cloud servers, DevTrails reads local Chrome history export JSONs (Google Takeout), strips out non-dev noise and social distraction logs, and groups browsing pathways into chronological Research Sessions.

It determines developer personas (The Bug Hunter, The Scholar, The Code Explorer), scores struggle-to-satisfaction ratios, trains a local scikit-learn text classifier, and correlates web searches directly with modified local code files to export RAG-ready vector datasets for local LLMs.

less implemented_features_&_local_rag_matrix.md
md
Subsystem FeatureImplementation StrategyLocal Privacy Guarantee
100% Offline ParsingNative JS historyParser.js grouping visits into 30-min session windows.🔒 Zero external network calls.
Trainable ML ClassifierPython scikit-learn TF-IDF Vectorizer + Logistic Regression (dev_classifier.pkl).🔒 Runs 100% locally via FastAPI.
Temporal Code Linkerproject_indexer.py scanning local file mtime timestamps against web search windows.🔒 Scans local projects directory only.
RAG Dataset ExporterGenerates devtrails_rag_dataset.json mapping search queries to edited files for ChromaDB.🔒 Stored on local disk.
Local LLM IntegrationCORS proxy to local Ollama (gemma2:9b) or optional Gemini API.🔒 Local inference on Apple Silicon.
Page Notes MemorizationOffline web_scraper.py extracting page summaries into scraped_cache.json.🔒 Local JSON cache.
cat devtrails_system_architecture_&_data_pipeline
mermaid
Architecture mapping Chrome Takeout ingestion, noise filtering, ML classification, temporal code indexing, local Ollama LLM, and React dashboard.
loading diagram