everlier

7 ms·

31.
Interface Hall of Shame (1999)(hallofshame.gp.co.at)
4 points by everlier 6mo ago|0 comments
32.
A History of Local LLMs(av.codes)
4 points by everlier 6mo ago|0 comments
33.
by everlier 6mo ago
OpenCode is an awesome tool. Many folks from other tools are only getting exposed to the same functionality they got used to, but it offers much more than other harnesses, especially for remote coding. You can start a service via `opencode
34.
by everlier 6mo ago
What I feel is more like "too much of a good thing", and too many people that want quick riches concentrating on this field after the web3 fiasco. Both of these seems more like a systematic societal issues rather than the problem
35.
40 Months of Prompt Injection(openguard.sh)
4 points by everlier 6mo ago|1 comments
36.
by everlier 6mo ago
A dossier on major events in agentic security since the first release of ChatGPT on November 17th, 2022. Fun fact, first prompt injection was documented three days before that date.
37.
There Is No Firewall for English(openguard.sh)
7 points by everlier 6mo ago|0 comments
38.
by everlier 6mo ago
We call it a slip slop at work, it's ok to slip some slop if it's "our" slop :-)
39.
by everlier 6mo ago
Never thought about this, but it makes sense they don't want a better local search, just for users to rely more on their product. It's messed up - so much time and human potential wasted on poor search and ads.
40.
by everlier 6mo ago
Great observation! OpenGuard as a project aims to solve exactly this problem - a protocol-level "firewall" for LLM APIs.
41.
by everlier 6mo ago
Thank you for the feedback! I actually assure you that no deep research from any of the provider was used to create the article itself, but I used a custom-built research pipeline for creating a dossier on promt injections as a starting poi
42.
by everlier 6mo ago
Thank you for the feedback! It's very early days of the project, there's indeed a lot to improve in this aspect. OpenGuard is an OpenAI/Anthropic-compatible LLM proxy with middleware-style configuration for protocol-level ins
43.
by everlier 6mo ago
OpenAI and other labs are trying to do this within existing structure by explicitly training a specific chain of authority in the inputs: https://github.com/openai/model_spec/blob/main/model_spec.md... H
44.
The Webpage Has Instructions. The Agent Has Your Credentials(openguard.sh)
37 points by everlier 7mo ago|25 comments
45.
by everlier 7mo ago
You can do same with Harbor in a couple of commands :) # Pull Qwen 3.5 35B harbor pull unsloth/Qwen3.5-35B-A3B-GGUF:UD-Q4_K_XL # Set llama.cpp image to an optimised one for performance on Strix Halo harbor config set llamacpp.image_roc
46.
Show HN: A single CLI to manage llama.cpp/vLLM/Ollama models(github.com)
2 points by everlier 7mo ago|1 comments
47.
Show HN: OpenGuard(openguard.sh)
1 points by everlier 7mo ago|0 comments
48.
by everlier 7mo ago
I'm sure that there are more providers and harnesses that are better suited for lighter models. Please share the ones you know.
49.
Tell HN: Tips for (mostly) free agentic coding setup
2 points by everlier 7mo ago|1 comments
50.
by everlier 7mo ago
Most of these tools work with OpenAI-compatible APIs, register for OpenRouter, top up your account for $10 one time to avoid rate limits and use free models that are always there in one way or another.
51.
Rug – Repeat Until Good(github.com)
1 points by everlier 7mo ago|0 comments
52.
Prompt Mixer - real-time LLM steering UI(github.com)
1 points by everlier 8mo ago|1 comments
53.
by everlier 8mo ago
How do we express nuance when steering agents and fleets of agents in the future? Telling your agent to be "slightly more concise" or "less formal" - how can that be quantified? How do you balance between dozens of chara
54.
Planetary Roller Screws(humanityslastmachine.com)
49 points by everlier 8mo ago|10 comments
55.
Show HN: Remotion Bits – building blocks for your videos(remotion-bits.dev)
1 points by everlier 8mo ago|0 comments
56.
by everlier 8mo ago
Someone has a lot of money in all the investments, so continuing the hype cycle is cheaper to maintain the portfolio until the IPOs happen
57.
by everlier 8mo ago
I couldn't get past panning not moving the dots background correctly and mouse zoom scaling not being attached to be position of a cursor
58.
LongCat ZigZag Attention(k-a.in)
2 points by everlier 8mo ago|1 comments
59.
by everlier 9mo ago
https://av.codes/ To this day I think it remains the only landing page with extensive use of SVG metaballs as a splitter between sections
60.
Ministral 3 – pruning via Cascade Distillation(arxiv.org)
5 points by everlier 9mo ago|0 comments

More ›