Your AI can't search your files.
Tessera fixes that.

Your workspace has thousands of documents — PRDs, decisions, session logs. Your MCP client can't search any of them. Until now.

Get Started View Source

Open source · v0.4.0 · 100% local

Claude Desktop tessera connected
What did we decide about the auth flow?
|
|
decisions/2025-01-15-auth.md
Ask about your docs...

Five minutes to set up.
Zero maintenance after.

01

Point

Pick your document folders. Markdown, CSV, session logs.

02

Index

Tessera embeds everything locally. No cloud. No API keys.

03

Connect

Add one config snippet to your MCP client. Done.

04

Ask

Your AI searches your workspace automatically via MCP.

Everything your workspace needs. Nothing it doesn't.

Hybrid Search

Semantic vector search + BM25 keyword matching. Finds what you mean, not just what you typed. Works across Markdown, CSV, and session logs.

MCP Client tessera
How did we handle rate limiting?
|
|
docs/api-design-v2.md

100% Local

No cloud, no API keys. fastembed (ONNX) + LanceDB. Your data never leaves your machine.

MCP Native

Any MCP client discovers and calls tools automatically. No copy-paste, no manual context window.

Auto Sync

Only re-indexes changed files. Runs automatically on MCP server startup.

PRD Audit

Section coverage, versioning, Mermaid syntax checks. Catches PRD sprawl automatically.

Smart Cleanup

Detects orphaned files, empty directories, backup clutter. Suggests before it acts.

15 tools. Auto-discovered via MCP.

Your client discovers what's available and uses the right tool for each query. No manual invocation.

Search & Read

search_documents

Semantic + keyword hybrid search across all your docs

read_file

Read any file's full content on demand

list_sources

See everything that's indexed

Memory

remember

Save knowledge for cross-session persistence

recall

Search past memories from previous sessions

learn

Auto-capture insights and index them immediately

Knowledge Graph

knowledge_graph

Build relationship maps between docs and concepts

explore_connections

Find related documents and shared topics

Workspace

project_status

See what's changed recently in each project

extract_decisions

Surface past decisions from session logs

audit_prd

Check PRD quality, versioning, and sprawl

organize_files

Move, rename, and archive files

suggest_cleanup

Detect backup files, empty dirs, clutter

Indexing

ingest_documents

Full index of all workspace documents

sync_documents

Incremental sync — only changed files

Three commands. That's it.

1 · install 
$ git clone https://github.com/besslframework-stack/project-tessera.git
$ cd project-tessera
$ python3 -m venv .venv && source .venv/bin/activate
$ pip install -e .
2 · setup 
$ tessera init
# Interactive setup walks you through everything
3 · connect 
# Add to claude_desktop_config.json:
{
  "mcpServers": {
    "tessera": {
      "command": ".venv/bin/python",
      "args": ["mcp_server.py"]
    }
  }
}

Requirements

  • Python 3.11+
  • MCP client (Claude Desktop, Cursor, etc.)
  • ~220MB disk (embedding model)

CLI

tessera init
tessera ingest
tessera sync
tessera status

Built with

LanceDBfastembedMCP SDKLlamaIndexONNX Runtime

Your docs, searchable by AI.
Locally.

Open source. Five minutes to set up. Free forever.

Get Tessera on GitHub

AGPL-3.0 · Free and open source · Commercial licensing available