inference-optimisation
A skill from Sybil Solutions for making a local large language model server faster, covering decode and prefill speed, speculative decoding, KV cache size, startup and time to first token on any hardware and quantisation.
What the skill teaches
A skill from Sybil Solutions for making a local large language model server faster, covering decode and prefill speed, speculative decoding, KV cache size, startup and time to first token on any hardware and quantisation. The agent works out theoretical ceilings from measured memory bandwidth and compute first, then changes one setting at a time against interleaved control runs, and every result must pass a long-context stability check. Bundled Python and shell scripts benchmark any OpenAI-compatible server, compute the ceilings, queue runs one after another and report a container's real memory use. Notes cover engines such as vLLM, SGLang, llama.cpp and ExLlama.
Capabilities
- Benchmark an OpenAI-compatible model server
- Compute theoretical speed ceilings
- Tune one serving setting at a time
- Check long-context stability after each change
About
- Publisher
- Sybil Solutions
Works with
Install
The project doesn't document install steps that we could confirm. See the project's own instructions
What's inside
- examples/ceiling-ornith-b390.json
- examples/config-existing-server.json
- examples/config-sglang-docker.json
- references/ceilings.md
- references/gotchas.md
- references/levers.md
- references/methodology.md
- references/publishing.md
- scripts/bench.py
- scripts/ceiling.py
- scripts/mem.sh
- scripts/queue.sh
MCP servers and skills can run code on your computer and act on your data. Review the source before installing.
This skill includes scripts that your agent may run.
Details
| Fact | Value | Source |
|---|---|---|
| Type | Skill | |
| Tool | Not stated in its documentation | |
| Publisher | Sybil Solutions | source |
| Official status | Not stated in its documentation | |
| Needs an API key or sign-in | Not stated in its documentation | |
| Includes scripts | Yes | source |
| Platforms | Not stated in its documentation | |
| AI clients | Claude Code | source |
| Licence | MIT | source |
| Pricing | Not stated in its documentation | |
| Activity | Last commit Nov 16, 2026198 GitHub starsDownload data not available (not on npm or PyPI) | source |
| Links | Repository |
Sources
- GitHub repositorygithub.com/sybil-solutions/inference-optimisation-skill · Backs: AI clients, Licence, Publisher, Includes scripts, ActivityNov 17, 2026
- Skill file (SKILL.md)github.com/sybil-solutions/inference-optimisation-skillNov 17, 2026
- Folder in the repository (inference-optimisation)github.com/sybil-solutions/inference-optimisation-skill · Backs: Includes scriptsNov 17, 2026