axeldunkel

7 ms·

1.
by axeldunkel 11d ago
What I've done for years: "generate" email addresses on the fly like zz-<portal>@mydomain.com and have a catchall rule for zz* in my main inbox. So if I get any mail or spam I can see where it originates from (and by do
2.
by axeldunkel 5mo ago
Good question on Code Mode internals. In Code Mode the model sees only two tools by default: list_tools(pattern) and execute_code(code). list_tools takes a regex and returns TypeScript signatures for matching tools. execute_code runs JavaSc
3.
by axeldunkel 5mo ago
A bit more on DADL, since this is what people typically ask first - why ANOTHER standard? DADL is on purpose narrower than e.g. OpenAPI. It describes only the tool surface that an agent is allowed to call - not the full API contract that hu
4.
Show HN: ToolMesh – turn all your REST APIs into MCP tools via declarative YAML
2 points by axeldunkel 5mo ago|2 comments
5.
by axeldunkel 5mo ago
You are also right and I half agree - but the same issue arises with new employees. They already have some of this experience you are talking about - but usually there is something about the very environment they are in which causes them to
6.
by axeldunkel 5mo ago
Well the LLM might well have read all operating procedures etc. - just like the teenager - but it has not *experienced* issues himself/herself/itself. If you look at the published statements about "AI deleted my..." thos
7.
Ask HN: AI is like a hyper-motivated 16-year old genius – how to manage?
1 points by axeldunkel 5mo ago|5 comments
8.
by axeldunkel 5mo ago
Amazing, looks too good to be true :-) But no ToS, Imprint, whether it is commercial or open source?
9.
by axeldunkel 5mo ago
I only use a decay function to see how "hot" a chunk is - not for forgetting old ones. What concerns me more are memory chunks with errors in them - they need to be corrected/removed by some other mechanism, not by decay (sin
10.
by axeldunkel 5mo ago
Think of AI just like of a genius 16-year old. Accidents will happen - only let AI and the 16-year old access systems where you are sure you have a recovery plan.
11.
by axeldunkel 5mo ago
the better the tokenizer maps text to its internal representation, the better the understanding of the model what you are saying - or coding! But 4.7 is much more verbose in my experience, and this probably drives cost/limits a lot.
12.
by axeldunkel 5mo ago
why does this matter?
13.
by axeldunkel 5mo ago
Great site! I once created muelltonne.de (german for "trashcan") where users could send (spam) mails which they did not like - and got poems or jokes in return that were made from exactly the same letters (plus some remains that c
14.
by axeldunkel 5mo ago
Great idea, I do like the concept to give the LLM more information and context so it can decide what approach is better. But why do you implement it as a mcp server and not as a proxy to have the full context?
15.
by axeldunkel 5mo ago
Sounds like "security by obscurity" to me - if you think AI is so good at finding security issues - it will find them in compiled code as well. Why not using it in your favor and let it search for bugs you'd otherwise not fin
16.
by axeldunkel 5mo ago
I like the approach to bring in the context surgeon as a wrapper that sees the conversation and can attach ids to the lines. You might even use the surgeon to replace parts with summaries?
17.
by axeldunkel 5mo ago
Nice clean project - Python 3.14 as a requirement might be an issue for adoption...
18.
by axeldunkel 5mo ago
I'd bet this does work but how do you solve the problem with scoring the quality of a modified agent?
19.
by axeldunkel 5mo ago
I still have a 486 linux system from those days - has not been turned on in this century but I'll try some day together with a glas of whisky :-)