Check my Liverpool workshop setup in this Codex desktop project. This is a read-only check, not a design or modelling run.

Read START_HERE.md, SETUP.md and PROJECT.md. Find and read these exact files, with their linked resources:
- .agents/skills/convoai-exploration-2d/SKILL.md
- .agents/skills/convoai-exploration-3d/SKILL.md

Inspect inputs/pfa-option3-concept.png. Report the actual skill paths and any missing resource. Reading the file and discovering a skill in the app are separate checks; say which you can verify.

Check the capabilities actually available in this chat:
1. Can you inspect images, and is a real image-generation/editing tool exposed? Do not generate an image yet.
2. Is there a live Rhino connection? If so, use a documented read-only call to report the actual active document identity, whether it is empty, units, tolerance and Rhino version when exposed. Do not change or close an existing document. If no live connection is exposed, say so without guessing a port, plugin or server.
3. Can you resolve this project folder and the outputs directory?

Use the model already selected in my app. Record its visible name and reasoning setting only if exposed; otherwise write “not exposed”. Do not infer my account plan or change models.

Return a short table: skill files, image input, image generation, live Rhino read, output folder. Use “confirmed by this check”, “available but not tested” or “missing”. List only the next action I need to take with the tutor. Do not install anything, edit global settings, run a CLI pilot or claim that model generation/save/reopen has passed.