Files
aitbc/.windsurf/skills/ollama-gpu-provider/SKILL.md
oib 9b9c5beb23 ```
chore: enhance .gitignore and remove obsolete documentation files

- Reorganize .gitignore with categorized sections for better maintainability
- Add comprehensive ignore patterns for Python, Node.js, databases, logs, and build artifacts
- Add project-specific ignore rules for coordinator, explorer, and deployment files
- Remove outdated documentation: BITCOIN-WALLET-SETUP.md, LOCAL_ASSETS_SUMMARY.md, README-CONTAINER-DEPLOYMENT.md, README-DOMAIN-DEPLOYMENT.md
```
2026-01-24 14:44:51 +01:00

1.2 KiB

name, description, version, author, tags
name description version author tags
ollama-gpu-provider End-to-end Ollama prompt payment test against the GPU miner provider 1.0.0 Cascade
gpu
miner
ollama
payments
receipts
test

Ollama GPU Provider Test Skill

This skill runs an end-to-end client → coordinator → GPU miner → receipt flow using an Ollama prompt.

Overview

The test submits a prompt (default: "hello") to the coordinator via the host proxy, waits for completion, and verifies that the job result and signed receipt are returned.

Prerequisites

  • Host GPU miner running and registered (RTX 4060 Ti + Ollama)
  • Incus proxy forwarding 127.0.0.1:18000 → container 127.0.0.1:8000
  • Coordinator running in container (coordinator-api.service)
  • Receipt signing key configured in /opt/coordinator-api/src/.env

Test Command

python3 cli/test_ollama_gpu_provider.py --url http://127.0.0.1:18000 --prompt "hello"

Expected Outcome

  • Job reaches COMPLETED
  • Output returned from Ollama
  • Receipt present with a receipt_id

Notes

  • Use --timeout to allow longer runs for large models.
  • If the receipt is missing, verify receipt_signing_key_hex is set and restart the coordinator.