What shipped today: Windows, .docx/.xlsx, Elixir, and more (v0.3.3 - v0.3.6)

[safishamsi]

A lot landed today. Here is a quick rundown of everything that shipped.

---

Windows support (v0.3.3)

Graphify now works on Windows. Running graphify install on a Windows machine auto-detects the platform and installs a PowerShell-compatible skill. All the bash-specific commands that used to break (which, head, rm -f, mkdir -p) are replaced with PowerShell equivalents. You can also force it with graphify install --platform windows if needed.

Windows encoding fix (v0.3.4)

Chinese Windows and other systems with non-UTF-8 default locales were hitting a UnicodeDecodeError on graphify install. Fixed by adding explicit encoding="utf-8" to every file read and write in the installer. If you were running 0.3.3 and hit this, upgrade to 0.3.4 or later.

.docx and .xlsx support (v0.3.5)

Office documents now work in graphify. Install the extra and drop your Word and Excel files into the folder alongside everything else:

pip install graphifyy[office]

.docx files are converted to markdown (headings, lists, and tables preserved). .xlsx files get each sheet turned into a markdown table. The converted files land in graphify-out/converted/ so you can inspect them. No changes to the skill or the pipeline needed.

This one came from a user in legal who has hundreds of badly scanned PDFs and Word documents. Felt like a real gap.

Elixir support (v0.3.6)

.ex and .exs files are now supported in the AST extraction pipeline. The extractor pulls out defmodule, def/defp, alias/import/require/use, and the call graph between functions. Same honest EXTRACTED/INFERRED tagging as every other language.

pip install --upgrade graphifyy

---

Other things

• Sponsor button is live if you want to support the project
• Discussions are open now (you are here)
• Two community PRs were incorporated and credited: the GBK encoding fix and the Elixir extractor

If you run into anything broken or have ideas for what to add next, open an issue or start a thread here.