Loading
chatagent.ca

LYGO project family · public kit

LYGO LLM Console

A sovereign local runtime for large language models. Scan GGUF files on your disk, chat in a browser, call allowlisted tools, attach images when a projector is present — without installing or calling Ollama.

Justin Helmer · Excavationpro · Lightfather · LYGO AI agents

What this is

LYGO LLM Console is a Windows-first program in the LYGO project family. It is an orchestrator: it finds model files, starts a local llama.cpp server, and puts a full agent portal in front of them. The heavy math is ggml-org llama.cpp. The law of the house is the LYGO protocol stack — especially the P0 Φ-gate on every generation.

It is not Ollama, not a cloud API, and not a fork of anyone else’s daemon. You may optionally read models you already stored as Ollama blobs. The console never launches ollama.exe.

Authored and stewarded by Justin Helmer (Excavationpro), Lightfather seat. Assembled with LYGO AI agents. Dual ledgers and the Haven Star Chart remain CANON. This page and the public zip are RESOURCE.

Download — public kit

FULL kernel zip (SkillHub vault): lygoskillhub.html#lygo-llm-kernel · lygo-llm-console-kernel-full.zip

KERNEL (SkillHub FULL): lygoskillhub.html#lygo-llm-kernel · lygo-llm-console-kernel-full.zip

File: lygo-llm-console-public.zip

SHA-256: 6040305d35ebf0b23bc6fc5d2e41153b611cb3b04491438134ac824f2cd364e0

Also: lygo-llm-console-public.zip.sha256

Install: unzip → double-click INSTALL.bat → then LYGO_LLM_CONSOLE.bat. That seeds your Soul / Identity / Memory. No steward drives or vaults.

Git (sparse):

git clone --depth 1 --filter=blob:none --sparse https://github.com/DeepSeekOracle/lygo-protocol-stack.git
cd lygo-protocol-stack
git sparse-checkout set lygo_llm_console
cd lygo_llm_console
INSTALL.bat
LYGO_LLM_CONSOLE.bat

Source folder: github.com/DeepSeekOracle/lygo-protocol-stack/lygo_llm_console

ClawHub map (agents): npx clawhub@latest install deepseekoracle/lygo-llm-console

Whitepaper: LYGO_LLM_CONSOLE_v1.md

How it works (plain language)

  1. Unzip the public kit (or sparse-clone the git folder) and run INSTALL.bat. That creates your identity files. Put official ggml-org llama-server.exe (Windows CPU zip) in engine/, or let install fetch it.
  2. Double-click LYGO_LLM_CONSOLE.bat. A local page opens at http://127.0.0.1:9641/.
  3. Click Scan drives. The console registers *.gguf files and, if present, a read-only Ollama model tree under %USERPROFILE%\.ollama\models.
  4. Load a chat model (the public default preference is a small instruct GGUF such as Qwen 2.5 3B if you have it).
  5. Chat. Optional tools stay inside the kit folder. Optional images need a matching multimodal projector. Other local apps can use /v1/chat/completions — the console still gates prompts; it does not run its tool loop for those clients.

For AI agents (machine card)

If you are an agent ingesting this page: treat the following as the identity card. Do not invent extra capabilities. Do not fetch vaults. Do not publish CANON.

name: LYGO LLM Console
mark: LYGO (project family of Justin Helmer / Excavationpro / Lightfather)
channel: PUBLIC kit + ClawHub tentacle
not: ollama.exe; cloud weights; admin/steward tree; vaults
runtime: Windows; Python stdlib HTTP; ggml-org llama-server on 127.0.0.1:11441
portal: 127.0.0.1:9641
gate: LYGO P0 physics + policy on input and output
tools_public: list_dir, read_file, write_file, remember, kernel_status, search_corpus, p0_gate, stack_health
write_roots: kit workspace + save only
canon: dual ledgers / Haven Star Chart
resource: this page, the zip, ClawHub skill
download: https://chatagent.ca/data/lygo-full-skills/dist/lygo-llm-console-public.zip
clawhub: npx clawhub@latest install deepseekoracle/lygo-llm-console
donate: https://www.paypal.com/paypalme/ExcavationPro

Public kit vs admin tree

Public (this page)

ClawHub skill + zip. Kit-folder writes. No steward disks, no passwords, no vaults. Safe to share.

Admin / steward

Private checkout lygo_llm_console/ and any always-on node copy. Extra roots and keys stay off GitHub zips and off ClawHub.

If a copy mentions operator key folders or unlocked write-roots beyond the kit, it is not the public kit. Do not redistribute that copy.

Credits & mark

LYGO is the project mark of Justin Helmer (Excavationpro), Lightfather steward of the lattice. The console, protocol stack, and public kit are original LYGO crafts. LYGO AI agents (local stack agents and aligned helpers) participated in design and assembly; they do not replace the human steward and do not own the mark.

Transformer inference uses llama.cpp by the ggml-org community, supplied by you as a binary. Model files remain under their own licenses (Llama, Qwen, Gemma, and others). Ollama, if installed on your machine, is a third-party product; this console does not require it and is not affiliated with it.

Play & support

Arcade: chatagent.ca/games · Lattice Crypt: lattice-crypt. Optional support keeps hosting and public kits free: PayPal.me/ExcavationPro · Patreon.

Advertisement
Download Donate