Developer Tools & CLI Utilities

Standalone scripts and benchmark runners for MnemosyneC. No desktop app required — just Node.js and Ollama.


MnemosyneC Plow CLI — M5 Shard Bundle

Standalone Node.js benchmark runner. No UI required — runs against local Ollama.
Executes the M5 psychology + other domains shard (200 questions).

Requirements: Node 18+, Ollama installed with gemma2:2b pulled

Check Node version: node --version  ·  Pull model: ollama pull gemma2:2b

↓ Download ZIP (47.8 KB) →

Run after unzipping:

node plow-cli.js m5_shard.json

BP084 · June 2026 · Liana Banyan Platform


🛰️ MnemosyneC Plow CLI — LAN Mesh Bundle

For running the mesh benchmark across multiple LAN machines. Contains: plow-cli.js · aggregate.js · all 5 node shards · setup-helper.ps1 · README.md

Download MnemosyneC-Plow-CLI-Mesh-LAN.zip (~250 KB)

  1. Unzip on each LAN machine
  2. Run setup-helper.ps1 once per machine — checks Node + Ollama + recommends your shard
  3. Run node plow-cli.js shards\mX_shard.json --model gemma4:12b --out mX_results.jsonl
  4. Send mX_results.jsonl back to M0
  5. M0 runs aggregate.js over all returned files

→ Need paste cards for each LAN node? Open the LAN Mesh Cards page.

See Son’s bootstrap card for the single-node WAN pattern.


How to use

Step 1 — Install Ollama (if not already installed)

Download from ollama.com/download, then pull the benchmark model:

ollama pull gemma2:2b

Verify it runs:

ollama run gemma2:2b "Hello"

Step 2 — Download the CLI bundle

Click the Download ZIP button above, or go directly to:

https://mnemosynec.ai/download/MnemosyneC-Plow-CLI-Son-M5.zip

Unzip to any folder (e.g. C:\MeshTest\ or ~/meshtest/).


Step 3 — Run the benchmark

Open a terminal in the unzipped folder:

node plow-cli.js m5_shard.json

You’ll see progress like:

[ 10/200] psychology   · 85.0% · 0 quarantined · ETA 42m
[ 20/200] psychology   · 87.5% · 1 quarantined · ETA 38m
...
[200/200] · 174 correct · 5 quarantined · score 87.3%

If Ollama is on a different port:

node plow-cli.js m5_shard.json --ollama http://localhost:11434

Resume after a crash:

node plow-cli.js m5_shard.json --resume

ETA

ModelQuestionsEst. Time
gemma2:2b20030–50 min

Troubleshooting

“Cannot reach Ollama” → Make sure Ollama is running. On Windows it starts automatically after install. On Mac/Linux run ollama serve in a separate terminal.

“Model not found” → Run ollama pull gemma2:2b first.

Progress stops → Ctrl+C, then re-run with --resume to continue.


No npm install needed — the CLI uses only built-in Node modules. SSPL · BP084 · MnemosyneC