Showcase

Use the showcase folder in your environment repo to store replay JSON exported from platform runs — for demos, documentation, or sharing benchmark trajectories.

Folder from mesocosm init

When you scaffold a project:

mesocosm init

you get:

showcase/
  README.md              # Brief notes on showcase usage
  replay.example.json    # Example shape (not a real run)

Keep exported replays here (or anywhere in your repo) and commit them if you want them in version control.

Export a run

After a platform benchmark completes:

mesocosm run export RUN_ID -o showcase/my-replay.json

Without -o, JSON is written to stdout (pipe to a file or another tool).

Argument	Description
RUN_ID	Platform run id from run create or run get
-o, --output	Output file path (optional)

The export includes run data suitable for replay or analysis (episodes, traces, replay payload as provided by the API).

Requirements

• You must be authenticated (member or guest, depending on run visibility)
• RUN_ID must be a run you can access on the bench API

Typical workflow

1. Run a platform benchmark and wait for completion
2. mesocosm run get RUN_ID — confirm status and scores
3. mesocosm run export RUN_ID -o showcase/replay.json
4. Review JSON locally; commit to GitHub if sharing with reviewers
5. Reference the file in your repo README or docs

Replay JSON

The exact schema follows the bench API export format. Use replay.example.json from mesocosm init as a structural hint; always treat a real run export as the source of truth for your run.

For episode-level detail without a full export:

mesocosm run episodes RUN_ID --traces

Visibility

When creating runs, you can set visibility for gallery features:

mesocosm run create ... --visibility gallery_public

See Running benchmarks for run create flags.

Related

• Running benchmarks
• Command reference — run export
• Local development — local runs do not produce showcase exports; export is for platform RUN_IDs