Prepare a self-contained dissemination kit under dissemination/kakeyalattice/
so the FluffyAIcode/LLM-KV--Cache-compress (KakeyaLattice) repo can move
from 'only discoverable by exact name' to 'natural-language discoverable'
on the five channels that matter: GitHub topics, arXiv, vLLM issue tracker,
HuggingFace Spaces, Papers with Code.

Each of the five tasks is scripted to one command or one copy-paste:

- github_topics/  : gh-CLI script that sets 20 curated topics + description
- arxiv/          : LaTeX tarball builder + metadata.yaml + endorsement template + SUBMIT walkthrough
- vllm_issue/     : pre-written issue TITLE + BODY (mirrors NexusQuant vllm#39241 format) + OPEN guide
- huggingface/    : full Gradio Space scaffold (app.py + requirements + YAML frontmatter) + deploy.sh + model-card edit snippet
- paperswithcode/ : entry.json (source of truth) + SUBMIT walkthrough + pre-filled SOTA leaderboard tables

DISSEMINATION_PLAN.md is the top-level 5-step checklist.
README_PATCH.md contains badges + a 'Dissemination' section ready to paste
into KakeyaLattice's own README, plus one-command re-adoption instructions.

None of this touches the KakeyaLattice repo directly (this agent has no
write access to it); the kit is designed to be copied into
FluffyAIcode/LLM-KV--Cache-compress with a single git checkout command.

Co-authored-by: FluffyAIcode <FluffyAIcode@users.noreply.github.com>