Engineering Dashboard Notebook
AI coding agents can write most of the code now.
The problem is when they also start owning the architecture and you slowly stop knowing why your own system works the way it does.
edn keeps a local engineering notebook in sync with your repo so you can actually stay on top of:
- architecture
- components
- dependencies
- data flow
- tradeoffs
- failure modes
- security boundaries
- current vs proposed changes
The notebook lives locally as:
engineering-notebook.html
and gets added to .gitignore automatically.
For normal work, the agent just builds.
If something meaningfully changes the architecture, it shows you the current setup, the proposed change, tradeoffs, evidence, and simpler alternatives before implementing it.
You make the call.
After the task is done, edn updates the notebook to match the actual code.
On first use, edn picks an explanation level: it reuses one already stored in the notebook, asks when an interactive user is around, and otherwise defaults to Junior without blocking work.
Intern
New Grad
Junior
Mid-Level
Senior
Staff
Principal
Distinguished
Same engineering rigor, just different levels of context.
So if the agent wants to throw Redis into your app, it should explain why it helps your system, what extra complexity it adds, and whether you've even measured a bottleneck yet.
npx skills add skcache/ednFor Codex:
npx skills add skcache/edn -a codexFor Claude Code:
npx skills add skcache/edn -a claude-codeFor OpenClaw:
openclaw skills install @skcache/ednFor OpenSkills:
npx openskills install skcache/ednThe skills CLI supports a bunch of coding agents and installs skills directly from GitHub.
To update an existing install:
npx skills update ednmetadata.version in SKILL.md is for human release tracking; the CLI finds updates from the source repo, not this field.
Inside your repo:
Use the edn skill.
Bootstrap the engineering notebook for this repository.
That's it.
The agent can own the typing.
You should still own the system.
MIT