polymath-cli 0.1.0

Ask anything about any codebase. Local, private, fast.

Polymath

Ask anything about any codebase. Instant answers, zero data leaks.

---

What is Polymath?

Polymath is a local-first codebase intelligence CLI. Point it at any repository — public or private — and ask questions in plain English. It indexes your code into a local vector database, retrieves the most relevant chunks, and answers with exact file citations and line numbers.

Your code never leaves your machine.

pm cd https://github.com/pallets/click
pm ask "how does argument parsing work?"

## Argument Parsing in Click

Click's argument parsing is handled in `src/click/core.py` lines 1089-1134.
When a command is invoked, `Command.make_context()` creates a `Context` object
and calls `Command.parse_args()` which delegates to each parameter's
`Parameter.consume_value()` method...

## How to use this in your codebase

from click import argument, command

@command()
@argument('filename')
def process(filename):
    click.echo(f'Processing {filename}')

---

Why Polymath?

Feature	Polymath	GitHub Copilot	ChatGPT
Works on private repos	✅ Local only	❌ Sends to GitHub	❌ Sends to OpenAI
No internet required	✅ Fully offline	❌ Cloud dependent	❌ Cloud dependent
Remembers conversation	✅ Per-repo history	❌ No memory	❌ Resets each chat
Any language	✅ All languages	✅ All languages	✅ All languages
Free	✅ Always	❌ $10/month	❌ $20/month
Cites exact lines	✅ Always	❌ No citations	❌ No citations

---

Installation

Prerequisites

• Python 3.11+
• Docker Desktop

Install

pip install polymath-cli

First time setup

pm init

This spins up a local Postgres and Qdrant instance via Docker. Run once, never again.

Health check

pm doctor

---

Commands

Indexing

pm cd .                                      # index current directory
pm cd /path/to/project                       # index any local path
pm cd https://github.com/user/repo           # clone and index a public repo
pm refresh                                   # re-index the active repo
pm rm <repo-name>                            # remove an indexed repo

Asking Questions

pm ask "how does authentication work?"
pm ask "where is the database connection set up?"
pm ask "what happens when a user signs up?"
pm explain src/auth.py                       # explain an entire file
pm diff                                      # explain the last git commit

Reading Code

pm cat src/auth.py                           # read file with syntax highlighting
pm cat src/auth.py:23-45                     # read specific lines
pm find "jwt"                                # find all chunks containing a keyword

Conversation

pm history                                   # show conversation history
pm clear                                     # clear conversation history
pm save my-session                           # save conversation to markdown
pm export                                    # copy conversation to clipboard

Repo Management

pm ls                                        # list all indexed repos
pm status                                    # show active repo and stats

Setup

pm init                                      # first time setup
pm doctor                                    # check Docker, Postgres, Qdrant health

---

How it works

pm cd <repo>
  → crawl all files (respects .gitignore)
  → chunk code at 50-line boundaries
  → embed each chunk with BGE-base-en-v1.5
  → store vectors in local Qdrant
  → store metadata in local Postgres

pm ask "question"
  → embed the question
  → search Qdrant for top 8 most relevant chunks
  → fetch full chunk content from Postgres
  → inject context + conversation history into prompt
  → stream answer from Gemini 2.5 Flash
  → save to conversation history

All vector search and storage happens locally. Only the question + relevant code snippets are sent to Gemini to generate the answer.

---

Architecture

polymath/
  pm/
    cli.py           # Typer commands — the user interface
    agent/
      ask.py         # retrieval + LLM + streaming answer
    indexer/
      crawler.py     # file discovery, respects .gitignore
      chunker.py     # 50-line chunks with overlap
      indexer.py     # orchestrates the full indexing pipeline
    db/
      models.py      # SQLAlchemy: repos, files, chunks, conversations
      queries.py     # all Postgres operations
      database.py    # engine + session factory
    vector/
      store.py       # Qdrant operations + BGE embeddings
    utils/
      init.py        # pm init — Docker setup
      doctor.py      # pm doctor — health checks
      state.py       # active repo tracking via ~/.polymath/.env

Data stores:

• ~/.polymath/ — config, docker-compose, .env
• ~/.polymath/repos/ — cloned GitHub repos
• Postgres (port 5433) — repo metadata, file records, chunks, conversation history
• Qdrant (port 6334) — vector embeddings for semantic search

---

Security

Polymath is designed with privacy as a first principle.

• Code stays local — indexing runs entirely on your machine
• Vectors stay local — Qdrant runs in Docker on your machine
• Only snippets leave — when you ask a question, the relevant code chunks (not your entire codebase) are sent to Gemini to generate the answer
• No accounts — no login, no tracking, no telemetry
• Private repos — clone them yourself, pm cd the local path. Nothing is uploaded

If you need 100% air-gapped operation, swap Gemini for a local Ollama model — only the ask.py file needs changing.

---

Configuration

Polymath stores all config in ~/.polymath/.env:

DATABASE_URL=postgresql://polymath:polymath@localhost:5433/polymath
QDRANT_HOST=localhost
QDRANT_PORT=6334
GEMINI_API_KEY=your-key-here
ACTIVE_REPO_ID=1
ACTIVE_REPO_NAME=myproject

Get a free Gemini API key at aistudio.google.com.

---

Roadmap

• Tree-sitter AST chunking — chunk at function/class boundaries for better context
• Multi-query RAG Fusion — generate query variations for better retrieval
• pm share — generate shareable links for public repos
• Web interface — browser-based chat UI
• Ollama support — 100% offline mode with local LLMs
• GitHub OAuth — index private repos without cloning manually
• VS Code extension

---

Contributing

git clone https://github.com/yourname/polymath
cd polymath
uv install
pm init
pm cd .
pm ask "how does the indexer work?"

---

License

MIT — do whatever you want with it.

---

Built with Python, Qdrant, PostgreSQL, BGE embeddings, and Gemini 2.5 Flash.