Artifacts, Python & media
Find persistent deliverables, distinguish scratch files, and inspect generated media.
Deliverables and scratch work
| Area | Purpose |
|---|---|
| Artifacts & Deliverables | Persistent outputs such as plots, CSV exports, reports, audio, and video in artifacts/. |
| Sandbox workspace | Working scripts, intermediate data, and execution scratch files in .sandbox/. |
The storage base follows the application’s resolved user-data/workspace configuration. Do not assume a packaged installation writes into its install directory. Ask for a final output to be saved as an artifact if you need it in the deliverables view.
Find and inspect a file
- Open Artifacts & Workspace.
- Use Refresh after a task creates or changes files.
- Choose Newest First, Oldest First, Name A → Z, Name Z → A, Largest First, or Smallest First.
- Open an image preview, code/file viewer, or audio/video player from the file card.
Date sorting uses the file’s modified timestamp. Video delivery supports byte-range responses for seeking; successful playback still depends on the file’s codec. Use the external-player option when the built-in player cannot decode a video.
Use Python for a verifiable output
In Agent mode, use Python to create a simple bar chart from values 3, 5, and 8. Save the PNG as an artifact and report the filename.textPython resolution prefers an explicit PYTHON_PATH, then a bundled runtime, then the sandbox virtual environment, then a system interpreter. Dependencies may need installation before a script can run.
Implementation sources
Paths in your AI Plate source checkout:
core/python-engine.tselectron/main.tselectron/ipc-bridge.tsui/app.jsui/index.html
