לפני השיעורים

Set up ASM in this project and leave it in a working state.

ASM (Agent Skill Manager) is a project-local skill system. It installs skills into `.asm/`, builds `.asm/main_asm.md`, groups skills into expertises, and syncs the router into the active agent config.

Do this in order:

1. Check whether `asm` is already installed. If not, install it with:
   `curl -LsSf https://raw.githubusercontent.com/gil-kapel/asm/main/install.sh | sh`

2. Check whether this repo already has ASM initialized by looking for `asm.toml`.
   - If ASM is not initialized, run `asm init`.
   - If ASM is already initialized, do not recreate it.

3. Inspect the current project briefly and choose the most relevant skill search queries for this repo.
   - Prefer curated, high-signal skills.
   - Do not install a large pile of generic skills.

4. Search and install only the most relevant skills.
   - Use `asm search "<query>"`.
   - Use `asm add skill <source>` for the best matches.

5. Configure expertise routing for the current project task or stack.
   - Run `asm expertise auto "<task description>"` with a concrete description of what this project needs.

6. Sync ASM into the active agent config.
   - Run `asm sync`.

7. If `ASM_CLOUD_API_URL` is configured and there is at least one local skill worth checking, run one cloud analysis:
   - `asm skill analyze <skill-name> --cloud`

8. At the end, report:
   - whether ASM was installed or already present
   - whether the workspace was initialized or already initialized
   - which skills were installed
   - which expertise was selected or created
   - which agent config(s) were synced
   - whether cloud analysis was run, and where the scorecard was saved

Success criteria:
- `asm.toml` exists
- `.asm/main_asm.md` exists
- at least one relevant skill is installed
- at least one expertise is active or created
- agent sync completed successfully