CLAUDE.md (or AGENTS.md) is not good place to put all code change rules, because these rules would dillute the context and "distract" the agent e.g. during bugs triage or business analysis.
Instead modularize the knowledge with skills and specialized MD files. Agent should lazy load what is needed to do focused work.
Skills have usage description metadata, but with free files you can simply instruct agent with CLAUDE.md to load them, e.g.: "Before you attempt to change any frontend code first load and follow `docs/{JS|HTML|CSS}_coding_rules.md`".
I think whats funny is that employees were most likely already covering the cost for these tools because they are useful. Companies didn't believe employees were using these tools and now have forced their usage and no longer have the costs subsidized.
Similarly companies seem to reward high token usage as a sign of someone willing to play ball with AI and again have forced higher costs on themselves for people reward hacking or using tokens out of spite.
There is no world where I can put my company’s data through an external site without their express consent and security sign off. I suspect at most companies there’s zero path for people to have been paying for it themselves.
None of the 5 places I have worked is this possible, but they are also all highly regulated industries. Firewalls block virtually everything by default.
Fair, but I assume everything on my work laptop is key logged. Surely they would notice Claude phoning home from my company laptop? I suspect a network rule to look for that traffic is trivial?
My employer doesn't specifically block this stuff, but does put up a warning when you visit it to review our AI usage policy. There isn't detection for using things in ways we shouldn't, but they have an audit trail and can review it if there is suspicion.
An increasingly disturbing trend from Github and I only see this getting worse.
I wouldn't rule out them moving away from offering the free tier to stop the all the code pushes. I think new code mostly written by AI isn't that appealing of a data set to train on.
You could fork and reference it in the issue if you still wanted to contribute.
reply