Open source, local first

Two public repos. Your data stays on your machine. The code is yours to read, improve, and share.

Why open source matters here

core-skills processes your meeting transcripts -- some of the most sensitive data in your organization. You should be able to read every line of code that touches it. No black boxes, no cloud sync, no telemetry. Everything runs locally and every file is plain text.

We also believe the best tools get built when real users contribute. If you find a better way to extract insights, structure a meeting summary, or display task data -- we want that improvement in the main repo so everyone benefits.

core-claude-skills

Claude Code skills

The 9 skills that process meetings, track tasks, prepare briefings, and extract knowledge. Pure text files -- each skill is a SKILL.md that Claude Code reads as instructions.

Stack Markdown + YAML
Dependencies None (Claude Code only)
Output .md + .yaml files

What's inside

/ops Meeting processor with org config
/transcript Lightweight call summarizer
/preparation Meeting briefing generator
/tasks Cross-project task tracker
/daily-dashboard Daily overview builder
/insights Knowledge extraction backfill
/update-skills Repo and symlink manager
ops-base Shared formats and standards
ops-config Schema and base defaults

core-claude-skills-visualisation

Companion web app

A local Flask web app that reads your vault files in real-time and turns them into interactive dashboards, kanban boards, and insight timelines. No database -- it reads your Markdown and YAML directly.

Stack Python / Flask / Chart.js
Database None (reads files directly)
Runs on localhost:5050

Pages and features

Dashboard with stats and charts
Kanban task board with filters
Activity heatmap and trends
Insight browser with timeline
Recent files with preview
Privacy toggle for sensitive data
Help page with live docs from repos

How the repos connect

Meeting happens
core-skills processes transcript
.md + .yaml files in your vault
Visualisation app reads them

The visualisation app is entirely optional. core-skills works perfectly on its own -- you can browse the files in Obsidian, VS Code, or any text editor.

Contribute

Both repos welcome contributions. Here's what helps the most:

Share how you use it

Your workflow is different from ours. Telling us how you use the skills -- what works, what's awkward, what's missing -- is the most valuable contribution you can make.

Start a discussion

Improve the skills

Found a better prompt structure? A smarter way to extract insights? Skills are Markdown files -- you can read, edit, and test them locally before submitting a PR.

Fork the skills repo

Extend the visualisation

The web app is Flask + vanilla JS + Chart.js. Add a new chart, improve the task board, build a better insights view. The parsers make it easy to work with vault data.

Fork the vis repo

Report issues

Something broken or unexpected? Open an issue. Include what you expected vs what happened. Screenshots and error logs help us fix things faster.

Open an issue

Our principles

Local first

All data stays on your machine. Plain Markdown and YAML. No cloud, no sync, no accounts. You own your files completely.

Fully transparent

Every skill is a readable Markdown file. Every parser is straightforward Python. No compiled code, no obfuscation. Read it all.

No lock-in

Stop using core-skills tomorrow and your files are still there -- standard Markdown, standard YAML. No proprietary formats, no export needed.

Community driven

The best features came from real users solving real problems. We build what people actually need, not what looks good in a demo.