1. Conduid
  2. Files
  3. Fituna
MCP server · Files

Fituna

by · github.com/leeyunseokarchive/fituna

Stop guessing your llama.cpp config. Give it a target tok/s and a quality budget, and it measures quantization levels and GPU offload on your own machine to find the lightest setup that actually hits the target.

Unclaimed benchmarkingcliggufllama-cppllmllm-inferencelocal-llmmcppythonquantization
InstallView source
34Low

Scored 6 days ago · breakdown

About Fituna

Fituna is an MCP server published by leeyunseokarchive in the Files category: stop guessing your llama.cpp config. Give it a target tok/s and a quality budget, and it measures quantization levels and GPU offload on your own machine to find the lightest setup that actually hits the target.

The repository has 2 stars and 0 forks, with the last commit 11 days ago. Recent commit activity is the strongest signal that reported issues will get attention.

Install

From source
git clone https://github.com/leeyunseokarchive/fituna

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Fituna

Powered by Claude · Grounded in docs

I know everything about Fituna. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

6 checks have not been run yet.

Releases

v0.3.6v0.3.6 — quit support, English support, refined quickstart messaging · 7 Oct 2026Added fituna quickstart can now be exited. The last option in every selection menu is now “Quit”. Entering q (quit, exit) at any prompt exits immediately. The exit code is 0 (Ctrl+C and EOF still exit with 1 as before). Previously, there…
v0.3.5v0.3.5 — quote printed commands, clean Ctrl+C exit · 6 Oct 2026What's Changed release 0.3.5 — quote printed commands, clean Ctrl+C exit by @leeyunseokarchive in https://github.com/leeyunseokarchive/fituna/pull/67 Full Changelog**: https://github.com/leeyunseokarchive/fituna/compare/v0.3.4...v0.3.5
v0.3.4v0.3.4 — search ngl against every --ctx candidate · 6 Oct 2026What's Changed test: cover quantized output publication and caching by @nawaaaaaAaar in https://github.com/leeyunseokarchive/fituna/pull/65 release 0.3.4 — search ngl against every --ctx candidate by @leeyunseokarchive in…
v0.3.3v0.3.3 — input validation, doctor launch checks, --hf --json fix · 5 Oct 2026Patch release. Bad input is now rejected up front instead of being searched, `fituna doctor` checks that the llama.cpp build can actually run and use your GPU, and `--hf --json` produces valid JSON. Fixes Out-of-range targets were…
v0.3.2v0.3.2 — full offload at n_layers + 1, CPU-only benches at -ngl 0 · 5 Oct 2026Patch release. Speed results from 0.3.1 and earlier could be off in two ways; this fixes both, plus two KLD-mode issues. Fixes Full offload left one layer on the CPU.** llama.cpp counts the output layer as one more layer, so full offload…
Questions

About Fituna

How do I install Fituna?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Fituna safe to use with an AI agent?

Its trust score is 34 out of 100 (low). Conduid hasn't run static security checks on this repository yet, so review the source yourself before granting it credentials. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Fituna still maintained?

Yes — the latest release is v0.3.6 (7 Oct 2026), and the last commit was 11 days ago. The repository has 2 stars.