55 lines
2.8 KiB
Markdown
55 lines
2.8 KiB
Markdown
# Judge Testing Guide
|
|
|
|
## Supported platform
|
|
|
|
- Windows 11 x64
|
|
- Windows WebView2 runtime (included with current Windows 11 installations)
|
|
- No Thinkloom account required
|
|
|
|
The current hackathon build is unsigned. Windows may show an unknown-publisher or SmartScreen warning. Signing is not required to inspect the source or use the portable release executable, but a judge should follow their organization's security policy.
|
|
|
|
## Install without rebuilding
|
|
|
|
Download the Windows installers from the public [Thinkloom v0.1.0 release](https://github.com/Labyricorn/thinkloom-openai-hackathon/releases/tag/v0.1.0):
|
|
|
|
- [Public v0.1.0 release](https://github.com/Labyricorn/thinkloom-openai-hackathon/releases/tag/v0.1.0)
|
|
- `Thinkloom_0.1.0_x64-setup.exe` (NSIS installer)
|
|
- `Thinkloom_0.1.0_x64_en-US.msi` (MSI installer)
|
|
|
|
Current local artifact checksums:
|
|
|
|
```text
|
|
E70F700C04E995685160DDA70FD811912A55DD362A28F63DFBE297040C6A0BE0 Thinkloom_0.1.0_x64-setup.exe
|
|
5B2F5072E8A0BB24E34E2AB3FB0EE81574E74D67FCEDB4BC03D0FD4DAEB918C3 Thinkloom_0.1.0_x64_en-US.msi
|
|
```
|
|
|
|
These checksums must be regenerated if either installer is rebuilt.
|
|
|
|
## Five-minute test path
|
|
|
|
1. Launch Thinkloom. The bundled sample project, **The Attention Commons**, opens without a login.
|
|
2. In **Ideation**, review the conversation and idea provenance. Accept, reject, edit, archive, or create a variant of a suggested idea.
|
|
3. Open **Drafting**. Select an idea, edit the structured manuscript directly, and save a named version.
|
|
4. Open **Finalization**. Inspect the release checklist and preview-first editorial actions.
|
|
5. Open **History** and **Provenance** to inspect the hash-linked creative-process record and contribution relationships.
|
|
6. Open **Export** to inspect publication, backup, and sanitized evidence-package options.
|
|
|
|
## Model-backed features
|
|
|
|
The sample project and direct editing/history/export surfaces can be inspected without a model credential. To exercise live generation, use one of these entrant-configured options:
|
|
|
|
- local Ollama at `http://127.0.0.1:11434` (default model `llama3.2`);
|
|
- OpenAI with a credential stored in the operating-system vault; or
|
|
- an authorized OpenAI-compatible endpoint.
|
|
|
|
The first cloud request for a project requires explicit in-app approval. Credentials are not included in the repository, project files, logs, or judge build.
|
|
|
|
## Known limitations
|
|
|
|
- The build is currently unsigned.
|
|
- Windows is the verified hackathon platform; macOS and Linux have not been release-tested.
|
|
- Voice input uses an ephemeral system/webview fallback. The planned bundled faster-whisper and Silero VAD pipeline is not included.
|
|
- Richer PDF typography, signed clean-install testing, and extended 20,000-word profiling remain release-hardening items.
|
|
|
|
These limitations do not affect the demonstrated typed ideation, structured drafting, provenance, versioning, or export workflow.
|