halogen-flash-server
AGENTS.md1.3k starsChanged 18 days agoLicence not identifiedView on GitHub
The short form of this repository for an AI agent that is deploying, driving or debugging halogen-flash-server. README.md is the full account and is canonical wherever the two disagree. An…
- Reads credentials
Trending
More talk lately.about the whole repository
Who's talking about peonist-ai/halogen-flash-server
- gufo-org/gufo“Support halogen's HGN checkpoint format (clean-room spec available)”GitHub · Sep 24, 2026 · 7 reactions
- A Hacker News story“Halogen – Fastest Way to Run Qwen3.8-Flash-Next on AMD Strix Halo”Hacker News · Sep 30, 2026 · 3 points
- @jimsgarage.co.uk“Anyone with a strix halo running qwen3.8-flash should check this out. Over 2x prefill …”Bluesky · Sep 16, 2026 · 3 likes
- @frank8465.bsky.social“have y'all both seen github.com/peonist-ai/h... I'm getting some wild 100k ctx deep prefill/decode speeds …”Bluesky · Sep 12, 2026 · 2 likes
- kyuz0/ai-toolbox-cockpit“"latest" pin for halogen toolbox with HALOGEN_CHECKPOINT set swallows v2 upgrade switch advice on server start”GitHub · Sep 30, 2026 · 1 reactions
The biggest 5 of 19 mentions linking to the repository that our crawl found. Each line is their own headline.
See all 19
Talked about on GitHub, Bluesky and Hacker News (19 times).
On GitHub · 16
- Sep 24, 2026Support halogen's HGN checkpoint format (clean-room spec available) — gufo-org/gufo7 reactions · 3 comments
- Sep 30, 2026"latest" pin for halogen toolbox with HALOGEN_CHECKPOINT set swallows v2 upgrade switch advice on server start — kyuz0/ai-toolbox-cockpit1 reaction · 1 comment
- Oct 31, 2026Deploy and test halogen flash model via llmkube — firefly2442/my-flux-kubernetes0 reactions
- Oct 24, 2026Wire AMD NPU userspace through nix-amd-ai on the Strix Halo nodes — shikanime-labs/machines0 reactions · 2 comments
- Oct 24, 2026Wire the halogen-flash StatefulSet for NPU models — shikanime-labs/manifests0 reactions · 1 comment
11 more on GitHub
- Oct 24, 2026Vendor XRT userspace for the Strix Halo NPUs — shikanime-labs/machines0 reactions · 1 comment
- Oct 23, 2026標本抽出の要求で speculative sampling（STRATA_SPEC_PROB）を使う — ruc1j/XeStrata-890M0 reactions · 1 comment
- Oct 23, 2026上流 0.1.46〜0.1.47 と rulith の未移植の decode スイッチを移す（第 2 弾） — ruc1j/XeStrata-890M0 reactions · 1 comment
- Oct 22, 2026Point the halogen liveness probe at the API, not the engine — shikanime-labs/manifests0 reactions · 1 comment
- Oct 22, 2026halogen-flash-0 crash-loops on sashina: the liveness probe kills a recovering engine — shikanime-labs/manifests0 reactions · 2 comments
- Oct 18, 2026[Feature Request] Ability to configure the alias name by the user -a or --alias (for llama.cpp) --served-model-name (for Gufo) — kyuz0/ai-toolbox-cockpit0 reactions
- Sep 30, 2026halogen-flash metrics are exposed but never scraped; ship the app's monitoring block — shikanime-labs/manifests0 reactions · 6 comments
- Sep 29, 2026Evaluate halogen engine family for serving qwen3.8-27b and qwen3.8-flash-next on the Strix Halo nodes — shikanime-labs/manifests0 reactions · 4 comments
- Sep 15, 2026Determine the Home Manager lifecycle and Pi provider integration for Halogen — dejanr/dotfiles0 reactions · 1 comment
- Sep 15, 2026Establish Halogen kernel and memory feasibility on Framework — dejanr/dotfiles0 reactions · 1 comment
- Sep 15, 2026Run Halogen locally on Framework through an opt-in Home Manager module — dejanr/dotfiles0 reactions · 1 comment
On Bluesky · 2
- Sep 16, 2026Anyone with a strix halo running qwen3.8-flash should check this out. Over 2x prefill github.com/peonist-ai/h... — @jimsgarage.co.uk3 likes and reposts
- Sep 12, 2026have y'all both seen github.com/peonist-ai/h... I'm getting some wild 100k ctx deep prefill/decode speeds on qwen flash next - like, 30t/s+… — @frank8465.bsky.social2 likes and reposts
On Hacker News · 1
- Sep 30, 2026Halogen – Fastest Way to Run Qwen3.8-Flash-Next on AMD Strix Halo3 points
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
The file
We split this AGENTS.md into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
The licence could not be identified — read it at the source. We count its rules but don't quote them. Read them on GitHub.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.