A powerful local AI assistant for file exploration, code analysis, semantic search, and multi-turn conversations. Built on LangGraph, Ollama, and FastAPI — all processing stays on your hardware.
Sensei combines deep file system access, semantic search, code analysis, and conversational AI — all without sending a single byte to the cloud.
All processing happens on your machine. Your files, conversations, and data never leave your hardware. No telemetry, no cloud calls, no data collection.
Read PDFs, DOCX, code files, configs, images with EXIF data, and more. Regex search within files and across directories.
AST-based Python structure extraction, directory tree visualization, git change inspection, and log file analysis. Understands your codebase deeply.
Local FAISS-based embeddings for semantic search within files. Find content by meaning, not just keywords. Powered by Ollama embeddings.
Token-by-token response streaming via SSE for a smooth, real-time chat experience. Watch responses appear as they are generated.
Create, switch, rename, pin, and delete sessions. Background execution preserves in-flight work when switching. Full SQLite persistence across restarts.
Automatic summarization when context hits 75% capacity. Visual context ring indicator, conversation archive recall, and manual /compact command.
Per-response metrics: token counts, tok/s, time-to-first-token, latency, and tool execution durations. Session-level aggregate stats in the header.
A polished React UI with dark theme, real-time streaming, tool call visibility, and comprehensive session management.
Full conversation flow with transparent tool execution, arguments, results, and per-response performance stats.
Pin up to 5 sessions, rename, delete, and switch seamlessly. Background execution preserves in-flight work.
Configure agents, model preferences, memory settings, and UI options. Per-agent controls for the multi-agent system.
File exploration, tool execution, and streaming responses in action.
Session management, context tracking, and advanced tool chains.
Sensei is designed to scale to a multi-agent architecture with dynamic routing, approval-based handoffs, and team orchestration.
The core AI assistant. File exploration, code analysis, semantic search, document processing, and multi-turn conversations.
Project planning, task decomposition, milestone mapping, and execution strategy. Structures chaos into actionable plans.
Developer agent for code generation, refactoring, debugging, and implementation. Turns plans into working code.
Idea generation, option exploration, and structured creativity. Helps think through problems from multiple angles.
/researchDeep investigation & evidence-backed summaries/devProject progress snapshots & blockers/learnLearning paths & progress tracking/tasksTask scheduling & calendar management/googleGmail, Drive, Docs, Sheets workflows/noteNotion integration & note capture/blogEngineering content discovery & curation/browseBrowser automation via Playwright/jobsJob search & application support/sysdesML system design mock interviews/dsaDSA/LeetCode guided practice/enhanceProductivity optimization/chillMusic & content discovery/navNavigation & travel planning/socialWhatsApp/Instagram inbox managementEvery tool is transparently visible in the UI with execution timing and results.
read_local_file PDF, DOCX, code, configs, imagesget_file_metadata Size, dates, EXIF, dimensionssearch_file_regex Regex search within a filefind_file Locate files by glob patternsearch_directory_regex Grep across directory contentsrebuild_file_index Rebuild SQLite file path indexsearch_semantic FAISS-based semantic searchget_directory_tree Tree visualizationget_code_structure AST parsing (Python + regex)analyze_logs Log file parsing & analysisget_local_changes Git diff inspectionsummarize_large_file Context-saving file summariesadd subtract multiply Arithmeticdivide_decimal divide_integer_* Divisionstat_mean stat_median stat_mode Statisticscount_values sum_numbers Aggregationrecall_conversation Search archived messagesfinalize_answer_from_tools Direct output pass-throughA modern stack designed for local-first AI with production-grade observability.
Quick access to session management, context control, exports, and system status.
/compact
Force context summarization
/summary
Preview conversation summary
/context
View token usage details
/clear
Clear current session
/tools
List all available tools
/export
Export as JSON or Markdown
/search <query>
Search indexed files
/index rebuild
Rebuild the file index
/index status
File index statistics
/agent status
Agent and model status
/model
Active model information
/history
Recent session history
/help
Show all available commands
Rigorous evaluation ensures reliability from individual tool calls to final answers.
Component-level correctness, robustness, and performance for each individual tool.
Tool selection quality, argument correctness, retry behavior, and investigation chains.
Task completion, groundedness, faithfulness, and hallucination detection.
Six implementation phases from foundation to production-ready multi-agent system.
Agent registry, per-agent policies, skills/rules framework, SQLite config persistence.
Specialized subgraphs, explicit/dynamic/fallback routing, approval-based handoff protocol.
/plan, /build, /brain slash commands. Team orchestration with sequential execution and approval-gated handoffs.
Full settings page: enable/disable agents, model overrides, tool policies, skill/rule editors, and agent config APIs.
Bounded task decomposition with max depth 2, max 3 concurrent sub-agents, and hard timeouts.
Security validation, server-side policy enforcement, audit logging, and production trace tuning.
git clone https://github.com/VjayRam/project-sensei.git && cd project-sensei
uv sync
ollama pull gemma4:e2b
uv run uvicorn src.server:app --reload
cd frontend-react && npm install && npm run dev
Have questions, ideas, or want to contribute? Reach out through any of these channels.