Skip to main content

Turn manual Excel processes into reviewable, reproducible marimo notebooks.

Project description

kedge

CI PyPI Python Licence

Turn manual Excel processes into reviewable, reproducible marimo notebooks, with an AI copilot that operates the notebook through a controlled tool surface.

Status: under active construction. See PLAN.md for the full brief.

What it does

  1. Analyses a workbook offline — formula regions, dependency graph, SQL connections, Power Query M, cached values, and a findings list (circular refs, volatile functions, IFERROR swallowing, inconsistent formulas within a region).
  2. Plans the conversion — an AI-proposed, user-approved ProcessPlan describing stages, open questions, and what it intends to drop.
  3. Scaffolds a marimo notebook from the approved plan.
  4. Reconciles the generated Python against the values Excel last cached, so the translation checks itself against evidence rather than declaring itself finished.

Quick start

uv tool install kedge
kedge hub                  # a browser landing page: every workbook kedge has seen
kedge open process.xlsx    # straight to one, if you already know which

kedge hub starts the server with nothing open and puts a browser on the list. Add a workbook by browsing the filesystem, pasting a path, or dropping one on the page; each row shows whether the file is still there, whether a notebook and an approved plan exist, how many findings the analysis turned up and how much of the workbook the plan believes it can convert. Opening one runs the same sequence kedge open runs — clean up, analyse, plan, scaffold, spawn marimo, bootstrap the session — with the progress streamed into the page, and lands you in the chat-plus-notebook view. Where a marimo kedge started is still running, the hub offers to reattach rather than starting a second.

Or, offline and standalone — the analyser is useful on its own:

kedge inspect process.xlsx --out analysis.json --report report.html

Design notes

  • Generated code is polars, never pandas — enforced in the validation gate.
  • Excel's semantics do not match polars' (rounding mode, empty-vs-null, divide-by-zero). kedge.xl is a registered polars namespace that makes each compatibility choice explicit and greppable: col("amount").xl.round(2).
  • Single-user, local, loopback-bound. No accounts, no server deployment.

Contributing

CONVENTIONS.md is binding — read it before opening a pull request. CLAUDE.md is the shortest useful orientation to the codebase, and RELEASING.md covers how a version tag becomes a release.

uv sync
uv run pytest                       # unit + corpus
uv run pytest -m contract           # live-kernel tests; spawns a real marimo
uv run ruff check . && uv run ruff format --check .
uv run python scripts/guardrails.py

Apache-2.0.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

kedge-0.0.1.tar.gz (822.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

kedge-0.0.1-py3-none-any.whl (517.8 kB view details)

Uploaded Python 3

File details

Details for the file kedge-0.0.1.tar.gz.

File metadata

  • Download URL: kedge-0.0.1.tar.gz
  • Upload date:
  • Size: 822.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.14

File hashes

Hashes for kedge-0.0.1.tar.gz
Algorithm Hash digest
SHA256 1e17cc2d3491ac7aea96d922225bc6691f4e478396f561be0f7600234a862143
MD5 fd6f9ab02920c001e3d247e3d8870d52
BLAKE2b-256 6ebe0938d37d6f4c6d86ad91020b050dd7c4f02dd35cd748999f5b1900d58fa9

See more details on using hashes here.

Provenance

The following attestation bundles were made for kedge-0.0.1.tar.gz:

Publisher: release.yml on OpenAfterHours/kedge

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file kedge-0.0.1-py3-none-any.whl.

File metadata

  • Download URL: kedge-0.0.1-py3-none-any.whl
  • Upload date:
  • Size: 517.8 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.14

File hashes

Hashes for kedge-0.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 9ad7207b1a77afb5ce8d30e31e33354aa1108d93c568c5b0f44373d9011c267f
MD5 643e3d13afdc1101310e4fe53e929969
BLAKE2b-256 d97d0892c64e47cb4687b16a285a2f73bc0defe799172327e9d384da34a6c6fa

See more details on using hashes here.

Provenance

The following attestation bundles were made for kedge-0.0.1-py3-none-any.whl:

Publisher: release.yml on OpenAfterHours/kedge

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page