Semantic codebase search

Your grep can't find
what it doesn't understand.

Giddyanne indexes the meaning of your code, not just text patterns. Search the way you think — completely offline, completely private.

Get Started View on GitHub
100% Local & Private Open Source (MIT) v1.6.0 
$ giddy find "authentication logic"

src/auth/login.py:45-67 (0.89)
  def verify_user_credentials(username, password):
      hashed = bcrypt.hash(password, user.salt)
      return secrets.compare_digest(hashed, user.password)

src/middleware/session.py:12-38 (0.82)
  class AuthenticationMiddleware:
      async def __call__(self, request, call_next):

grep / ripgrep

$ rg "auth"
437 matches across 89 files
...good luck.

giddyanne

$ giddy find "authentication logic"
src/auth/login.py (0.89)
src/middleware/session.py (0.82)
src/api/tokens.py (0.76)

How It Works

1

Index

Scans your project, splits code at function and class boundaries, and creates embeddings for every chunk.

2

Search

Query in natural language. Hybrid search combines semantic understanding with keyword matching.

3

Stay in sync

File watcher re-indexes changed files automatically. No manual rebuilds, no stale results.

Get started in 30 seconds

# Install
$ git clone --depth 1 --branch v1.6.0 \
    https://github.com/upship-ai/giddyanne.git
$ cd giddyanne && make install

# Search any git repo — no config needed
$ giddy find "error handling"

Works out of the box in any git repo. Add a .giddyanne.yaml for optional path descriptions that improve search relevance.

$ giddy -h

giddy - Semantic code search CLI

Usage:
  giddy find <query> [options]    Run semantic search
  giddy up [options]              Start the server
  giddy down                      Stop the server
  giddy bounce [options]          Restart the server
  giddy status                    Server status
  giddy health [options]          Diagnostic information
  giddy sitemap [options]         List indexed files
  giddy log                       Stream server logs
  giddy init                      Print setup prompt for new projects
  giddy install                   Download embedding model (~90MB)
  giddy mcp                       Run MCP server (for Claude Code)
  giddy embed <start|stop|status> Manage shared embedding server
  giddy drop                      Remove search index
  giddy clean [options]           Remove all .giddyanne data
  giddy completion <shell>        Generate shell completions

Ready to search smarter?

Get set up in under a minute. No accounts, no API keys, no cloud.

Every command abbreviates — giddy f for find, giddy st for status.

CLI Reference Configuration Integrations Changelog

Private

Your code never leaves your machine. No cloud APIs, no data collection, no telemetry.

Fast

Millisecond searches after initial indexing. Pre-computed embeddings mean instant results.

Offline

Runs entirely on your hardware using local models. No internet, no API keys, no rate limits.

Smart

Understands intent. "Where do I format dates?" finds formatTimestamp() even when the words don't match.

Integrated

CLI, Emacs, VSCode, MCP for AI assistants, or HTTP API for custom workflows.

Live

File watcher re-indexes on save. Your search results stay current without manual rebuilds.