Metadata-Version: 2.4
Name: jaunt
Version: 0.2.0
Summary: Spec-driven code generation framework: write intent as decorated Python stubs, generate implementations and tests with LLMs.
Project-URL: Homepage, https://github.com/creatorrr/jaunt
Project-URL: Repository, https://github.com/creatorrr/jaunt
Project-URL: Issues, https://github.com/creatorrr/jaunt/issues
Keywords: cli,code-generation,developer-tools,llm,testing
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3 :: Only
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Software Development :: Code Generators
Classifier: Topic :: Software Development :: Testing
Classifier: Typing :: Typed
Requires-Python: >=3.12
Provides-Extra: all
Requires-Dist: anthropic<1,>=0.39.0; extra == 'all'
Requires-Dist: fastmcp>=2.0.0; extra == 'all'
Requires-Dist: openai<2,>=1.0.0; extra == 'all'
Requires-Dist: watchfiles>=1.0.0; extra == 'all'
Provides-Extra: anthropic
Requires-Dist: anthropic<1,>=0.39.0; extra == 'anthropic'
Provides-Extra: mcp
Requires-Dist: fastmcp>=2.0.0; extra == 'mcp'
Provides-Extra: openai
Requires-Dist: openai<2,>=1.0.0; extra == 'openai'
Provides-Extra: watch
Requires-Dist: watchfiles>=1.0.0; extra == 'watch'
Description-Content-Type: text/markdown

# Jaunt

Jaunt is a small Python library + CLI for **spec-driven code generation**:

- Write implementation intent as normal Python stubs decorated with `@jaunt.magic(...)`.
- Optionally write test intent as stubs decorated with `@jaunt.test(...)`.
- Jaunt generates real modules under `__generated__/` using an LLM backend (OpenAI or Anthropic).

## Installation

```bash
pip install jaunt[openai]      # for OpenAI
pip install jaunt[anthropic]   # for Anthropic/Claude
pip install jaunt[all]         # both providers
```

## Quickstart (This Repo)

Prereqs: `uv` installed.

```bash
uv sync
export OPENAI_API_KEY=...   # or ANTHROPIC_API_KEY for Claude
uv run jaunt --version
```

See `docs-site/` for rendered docs, or `DOCS.md` for a plain-text walkthrough.

All examples live under `examples/`. See `examples/README.md` for the full list.

### Hackathon Demo (JWT Auth)

Headline demo: **JWT auth** (the "wow gap" example: short spec, real generated glue + tests).

```bash
# Generate implementations for @jaunt.magic specs.
uv run jaunt build --root examples/jwt_auth

# Generate pytest tests for @jaunt.test specs and run them.
PYTHONPATH=examples/jwt_auth/src uv run jaunt test --root examples/jwt_auth
```

## Eval Suite

Run the built-in eval suite against your configured backend:

```bash
uv run jaunt eval
uv run jaunt eval --model gpt-4o
uv run jaunt eval --provider anthropic --model claude-sonnet-4-5-20250929
```

Compare explicit provider/model targets:

```bash
uv run jaunt eval --compare openai:gpt-4o anthropic:claude-sonnet-4-5-20250929
```

Eval outputs are written under `.jaunt/evals/<timestamp>/`.

Prompt snapshots:

```bash
uv run pytest tests/test_prompt_snapshots.py --snapshot-update
```

## Auto-Generate PyPI Skills (Build)

`jaunt build` includes a best-effort pre-build step that auto-generates “skills” for external libraries your project imports and injects them into the build prompt.

What happens:

- Scan `paths.source_roots` for `import ...` / `from ... import ...` (ignores stdlib, internal modules, and relative imports).
- Resolve imports to installed PyPI distributions + versions from the current environment.
- Ensure a skill exists per distribution at:
  - `<project_root>/.agents/skills/<dist-normalized>/SKILL.md`
- If missing/outdated, fetch the exact PyPI README for `<dist>==<version>` and generate `SKILL.md` using the configured LLM provider.
- Inject the concatenated skills text into the build LLM prompt.

Overwrite rules:

- Jaunt only overwrites a skill if it was previously Jaunt-generated (it has a `<!-- jaunt:skill=pypi ... -->` header) and the installed version changed.
- If the header is missing, the file is treated as user-managed and will never be overwritten.

Failure mode: warnings to stderr, and the build continues without missing skills.

## Docs Site (Fumadocs)

The repository includes a Fumadocs (Next.js) documentation site under `docs-site/`.

```bash
cd docs-site
npm run dev
```

## Publish to PyPI

If you keep your token in `.env` as `UV_PUBLISH_TOKEN=...`, load it into your shell first:

```bash
set -a
source .env
set +a
```

Build and validate artifacts:

```bash
uv build
uvx twine check dist/*
```

Upload to PyPI:

```bash
uv publish --check-url https://pypi.org/simple/
```

## Dev

```bash
uv run ruff check .
uv run ty check
uv run pytest
```
