askdocs

A typed Python CLI that takes a question, calls a deployed LLM with proper auth, and streams the answer grounded in financial documents. Project 1 of 6 in a regulated-document AI platform — the same skeleton as repostat with one endpoint swap: the GitHub call replaced by an LLM call.

Usage

uv run askdocs "What is a debt-to-income ratio?" --provider anthropic

uv run askdocs "Summarize this loan application" --provider azure

Auth

Authentication follows the same pattern as repostat — secrets are read from environment variables, never hardcoded.

# .env
AZURE_ENDPOINT=https://your-foundry-project.services.ai.azure.com/openai/v1
AZURE_DEPLOYMENT=gpt-4o-mini
AZURE_TOKENS_MAX=1024                  # optional, defaults to 1024

ANTHROPIC_API_KEY=your_key_here
ANTHROPIC_MODEL=claude-haiku-4-5-20251001  # optional, defaults to haiku
ANTHROPIC_TOKENS_MAX=1024              # optional, defaults to 1024

FILE_PATH=docs                         # directory containing .txt files to ground answers

Azure uses DefaultAzureCredential — run az login for local development. No raw API keys in Azure.

Setup

uv sync
az login
uv run askdocs --help

Development

uv run pytest          # run tests
uv run ruff check src  # lint
uv run pyright src     # type check

Stack

• openai — OpenAI-compatible client pointed at Azure AI Foundry
• azure-identity — DefaultAzureCredential for keyless auth
• anthropic — direct Anthropic SDK for multi-provider support
• typer — type-hint-driven CLI
• python-dotenv — .env file loading
• pytest — testing with mocked model clients

What this project adds

• Calling an LLM SDK and printing a completion
• DefaultAzureCredential instead of a raw API key
• Streaming completions — tokens print as they are generated
• Lazy document chunking with yield
• Naive retrieval — keyword overlap used to ground answers in financial documents
• Multi-provider abstraction — --provider azure and --provider anthropic through the same code path

Part of a series

Project	What it adds
repostat	Python language fundamentals: CLI, REST, typed models, error handling, secrets, tests
askdocs	LLM SDK, streaming, naive RAG, multi-provider
agentcli	Tool-calling agents, memory, asyncio, MCP
ragservice	FastAPI, embeddings, vector + hybrid search, citations, PII handling
extractor	Document intelligence, vision, batch processing, structured validation
evalkit	Evals, observability, cost tracking, tracing, Docker