BearWisdom — Structural Code Intelligence

Core Engine

What BearWisdom Does

A structural search engine — not an LLM wrapper. Fast, deterministic, queryable.

Parse 31 Languages

Tree-sitter extracts every symbol, reference, and signature. Classes, functions, interfaces, enums — with full qualified names and scope paths.

Knowledge Graph

SQLite database with symbols, edges, and concepts. Call hierarchy, blast radius, cross-file references — all pre-computed and queryable.

6 Search Modes

FTS5 symbols, fuzzy matching, content trigram, regex grep, hybrid vector + text, and CodeRankEmbed semantic search.

4 Interfaces

CLI with 30+ JSON commands. MCP server for LLM agents. Web explorer with D3 force graph. Claude Code agent for conversational analysis.

Integrations

Built for AI-Native IDEs

BearWisdom isn't an LLM optimization layer. It's a structural search engine that serves both the editor and the AI.

The Editor

Go-to-definition, find-references, symbol search, file explorer. The same APIs power IDE features and agent queries.

The AI

MCP server feeds structural context to any LLM. Blast radius before refactoring. Call chains before debugging. Concepts for onboarding.

The Developer

Web explorer for visual architecture review. Force-directed knowledge graph. Concept sidebar. Symbol detail with code preview.

Knowledge Graph

Explore Your Architecture

Symbol Search

Concept Filtering

Real screenshots from the BearWisdom web explorer running on Microsoft's eShop reference architecture.

Switch to the Flow tab to see a Sankey diagram of execution flows — from entry points through services to data stores. Click paths to pin them, hover to explore, and trace from specific symbols.

Flow tab — Sankey diagram showing execution flows across an entire project. Nodes are colored by symbol kind (class, method, interface, constructor).

Tracing from CategoryApiController — shows the full call chain from controller methods through DI-injected services to repository operations.

Command Line

30+ Commands, One Tool

Every query returns structured JSON. Pipe into jq, scripts, or your editor.

Index & Query

$ bw open ./eShop

675 files, 3451 symbols, 2909 edges (2.8s)

$ bw blast-radius Entity --depth 3

62 symbols affected across 24 files

$ bw calls-in PlaceOrder

8 callers: OrderController, OrderService, ...

$ bw architecture

9 languages, 12 hotspots, 4 entry points

Search & Navigate

$ bw hybrid "payment retry logic"

3 results (FTS5 + vector RRF)

$ bw fuzzy-symbols "CatServ"

CatalogService (0.92), CatalogServiceTest (0.87)

$ bw grep "throw.*Exception" --lang rust

14 matches across 7 files

$ bw symbol OrderService --refs

Kind: class · 18 refs · 4 methods

v0.2 Commands

full-trace <path> [symbol] — End-to-end execution trace across call graph and flow edges

smart-context <path> <task> — Smart context selection for LLM prompts

investigate <path> <symbol> — Combined deep-dive: info, callers, callees, blast radius

complete-at <path> <file> <line> — Scope-aware symbol completion

diagnostics <path> <file> — Unresolved refs and low-confidence edges for a file

quality-check --baseline <file> — Regression testing against a quality baseline

import-scip <path> <scip-file> — Import SCIP index for high-confidence edges

Performance

Benchmarks

v0.2 benchmarks test three conditions — MCP tools, CLI tools, and native Read/Grep/Glob — across 4 real-world projects spanning .NET, Go, and TypeScript. Composite score combines precision, recall, and efficiency (penalizing tool call count).

Project	MCP	CLI	Native
eShop (.NET)	0.456	0.626	0.404
SimplCommerce (.NET)	0.627	0.574	0.433
go-gitea (Go)	0.294	0.414	0.302
react-calcom (TS)	0.534	0.549	0.380

Project	MCP Tokens	CLI Tokens	Native Tokens
eShop (.NET)	29,812	21,843	37,538
SimplCommerce (.NET)	61,306	50,084	45,895
go-gitea (Go)	52,267	35,128	36,572
react-calcom (TS)	49,567	38,410	40,244

Key Findings

CLI wins 3 of 4 projects on composite score. Both BearWisdom conditions (MCP and CLI) consistently outperform native tools on graph-heavy tasks — blast radius, call hierarchy, cross-file references. Native tools use 2–4× more tool calls to reach the same answers. CLI's slim JSON output keeps token cost lower than MCP in most cases.

Run your own: bw-bench full --project /path/to/project --output results/

What's Next

Roadmap

v0.2 — Current Release

v0.2

Flow Intelligence

Full Trace Engine

End-to-end execution flow tracing that walks the call graph AND cross-framework flow edges (DI, HTTP, events). Traces from specific symbols or auto-detected structural roots.

Smart Context (fixed)

Natural language task descriptions now return relevant symbols. Multi-strategy seeding: FTS5 raw, per-keyword, and LIKE fallback. Previously returned empty for anything that wasn't an exact symbol name.

Sankey Flow Visualization

New Flow tab in the web explorer. D3-Sankey diagram shows architecture flows from entry points through services to data stores. Click to pin paths, hover to explore, depth slider, symbol-specific tracing.

7 New CLI Commands

full-trace smart-context investigate complete-at diagnostics quality-check import-scip — bringing the total to 30+.

3-Condition Benchmarks

MCP, CLI, and Native conditions tested across 4 projects (eShop, SimplCommerce, go-gitea, react-calcom). CLI wins 3/4 on composite score. BearWisdom uses 2-4x fewer tool calls than native.

Integration Test Suite

43 new integration tests covering all query commands. 5 SCIP import tests. 6 quality-check tests. Total: 885 tests, 0 failures.

Cross-Language Sampler Fix

Benchmark sampler now supports TypeScript (NULL visibility) and Go (struct/type_alias kinds). Ground truth deduplication across all categories.

v0.3

Depth

Resolution engine

Per-language scope rules for 100% edge confidence — no more low-confidence cross-file edges.

RAG layer

Retrieve chunks, synthesize natural language answers via LLM.

Workspace / monorepo

Index multiple roots with cross-project references.

Git-aware indexing

Use git diff to scope re-indexing — faster than a full walk.

v0.4

Production

Plugin system

Custom connectors without forking.

Publish to crates.io

cargo install bearwisdom-cli

Streaming results

Stream search results as found for large codebases.

Setup

Get Started in 60 Seconds

Rust is the only prerequisite. No runtime, no daemon, no config file.

Build

Compile the CLI from source with a single cargo command.

cargo build --release -p bearwisdom-cli

Index

Point BearWisdom at any directory. Index builds in seconds.

bw open /path/to/your/project

Explore

Use the CLI, launch the web explorer, or register the MCP server with your IDE.

bw web # or bw mcp-serve

Internals

Inside BearWisdom

A thin set of consumer crates over a single intelligence library.

Parser

31 languages · Tree-sitter

Graph DB

SQLite · rusqlite

Search Engine

FTS5 · vector · hybrid

bearwisdom

core library · Rust

CLI · 30+ cmds

bw-mcp

MCP Server

bw-web

Web Explorer

bw-bench

Benchmarks

Your codebase,understood.

What BearWisdom Does

Built for AI-Native IDEs

Explore Your Architecture

30+ Commands, One Tool

Benchmarks

Roadmap

Get Started in 60 Seconds

Inside BearWisdom

Your codebase,
understood.