Practical Python tools, document workflows, and AI automation.
Maintained by Rolly Calma (@Ghraven). We build small tools with readable code, reproducible examples, and focused tests. Core workflows stay lightweight; optional AI features explain their setup and limitations.
Project gallery · Portfolio and services · Contributor guide
| If you want to… | Start here | What to expect |
|---|---|---|
| Clean spreadsheet exports | Data Cleaner Tool | Browser-based CSV cleaning; data processing stays in your browser. |
| Check a website | Website QA Auditor | Standard-library Python CLI with Markdown reports for common SEO, accessibility, and page-hygiene checks. |
| Understand document retrieval | Mini-RAG | Pure-Python TF-IDF retrieval, small evaluation fixtures, and optional API-based answers. |
| Explore local AI conversations | Agent Meeting Room | Flask and Ollama chat, optional cloud agents, streaming discussions, and saved meeting notes. |
| Reuse Python building blocks | Python Utils Toolkit | Standard-library helpers for retries, rate limits, caching, file I/O, and automation. |
| Practise typing | WPM Trainer | Browser typing practice, weak-key drills, and locally stored progress. |
For a quick review, open a project's README, try its bundled sample, and inspect its tests and recent Actions runs. Hosted demos can have cold starts; local setup remains documented in each repository.
Contributions are welcome from people who enjoy making small tools more dependable. Useful starting points include:
- Reproducing a bug with a small, non-sensitive example.
- Testing setup instructions on Windows, Linux, or macOS.
- Improving keyboard usability, error messages, documentation, or regression coverage.
Open an issue in the relevant project with the problem and your proposed scope. For a general collaboration idea, start a conversation here. Mention the project and the kind of contribution you want to make; no personal contact details are needed.
You do not need organization membership to open an issue or submit a pull request. Ongoing collaborators can discuss a scoped role with the maintainer after working together. Membership and repository access are considered individually.
- Website audit results are heuristics, not accessibility certification or guaranteed search rankings.
- Retrieval scores and citations do not guarantee a correct AI-generated answer. Check the source passages.
- Optional cloud-model features send inputs to their configured provider. Use synthetic or public examples when sharing reports and bug reproductions.
- AI assistance is part of development; maintenance decisions and responsibility remain with the human project owner.
See the visitor guide for a short review path.