Getting Started

Getting started

PB is local-first and intentionally manual about model servers. You start Text Generation WebUI yourself. PB verifies and guides you.

1) Start Proworkbench

Run PB locally (server on 127.0.0.1:8787, UI in your browser).

2) Start Text Generation WebUI (API mode)

Start Text WebUI manually on 127.0.0.1:5000:


./start_linux.sh --api --api-port 5000 --listen-host 127.0.0.1

Then open http://127.0.0.1:5000 and load a model.

3) Load a model

Known-good baseline:

models/quen/qwen2.5-coder-7b-instruct-q6_k.gguf

4) Use Doctor

Doctor checks reachability, models, approvals, MCP templates, and safety invariants.

Troubleshooting

  • If Text WebUI is not reachable: confirm API server on port 5000 and bound to 127.0.0.1.
  • If no models are available: load a model in Text WebUI, then refresh PB.
  • If tool execution is blocked: check tool policy/approval state and retry.
  • Browser note: Chrome/Chromium recommended for preview builds.