# Wire the Spectrum Seed into Runtime Implementation Plan

> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.

**Goal:** Expose the existing idempotent `db::seed::seed_spectrum_cataloguing` as a `server seed` CLI subcommand (plus a `just seed` recipe and README note), so an operator can seed an instance's baseline cataloguing fields.

**Architecture:** Mirror the existing `create-user` one-shot exactly — add a `Seed` variant to the clap `Command` enum, dispatch it to a new `server::seed(database_url)` that connects with a tiny pool, applies migrations (idempotent, so it works on a fresh DB), runs the seed inside a transaction, commits, and exits. The seed content and its idempotency are already tested at the db layer; the new code is thin glue.

**Tech Stack:** Rust (clap derive, sqlx/Postgres, anyhow, tokio). Backend-only + docs.

**Conventions:** `cargo +nightly fmt`; `cargo clippy --workspace --all-targets -- -D warnings`; tests via `cargo nextest run`; never write the codename ("biggus"/"dickus"). Test infra: compose Postgres on host **5442**, Meili **7700**; `#[sqlx::test(migrations = "../db/migrations")]` provisions its own temp DB. Env for manual runs comes from `.env` via the justfile's `set dotenv-load`.

**Spec:** `docs/superpowers/specs/2026-06-05-spectrum-seed-wiring-design.md`

---

## File Structure

- `crates/server/src/main.rs` — add a `Seed` variant to the `Command` enum + a dispatch arm.
- `crates/server/src/lib.rs` — add `pub async fn seed(database_url: &str) -> anyhow::Result<()>` (modeled on `create_user`, but with a `db.migrate()` step).
- `crates/server/tests/seed.rs` (new) — a server-crate building-block regression test mirroring `crates/server/tests/create_user.rs` (seed twice via the test pool; assert a known seeded vocabulary + field).
- `justfile` — add a `seed` recipe.
- `README.md` — add a seed step to the "Running locally" setup sequence.

The seed *content* + idempotency stay covered by the existing `crates/db/tests/seed.rs` (unchanged).

---

## Task 1: `server seed` subcommand

**Files:**
- Modify: `crates/server/src/main.rs`
- Modify: `crates/server/src/lib.rs`
- Create: `crates/server/tests/seed.rs`

**Reference (the template to mirror) — `server::create_user` in `crates/server/src/lib.rs`:**
```rust
pub async fn create_user(database_url: &str, email: &str, role: Role) -> anyhow::Result<()> {
    // ...email parse + password hash...
    let db = Db::connect(database_url, 2).await.context("connecting to the database")?;
    let mut tx = db.pool().begin().await?;
    let id = db::users::create_user(&mut tx, AuditActor::System, &NewUser { /* ... */ }).await
        .context("creating the user (is the email already taken?)")?;
    tx.commit().await?;
    println!("created user {id} ({role:?})");
    Ok(())
}
```
`Db::connect(url, n)`, `db.migrate()`, `db.pool()` all already exist (`run` calls `db.migrate()` at `lib.rs:22`). The seed fn `db::seed::seed_spectrum_cataloguing(conn: &mut sqlx::PgConnection)` is idempotent and uses `AuditActor::System` internally — no actor plumbing needed.

- [ ] **Step 1: Write the server-crate building-block test.** Create `crates/server/tests/seed.rs`. Mirror the harness comment + pool approach from `crates/server/tests/create_user.rs` (the temp-DB URL isn't exposed, so we exercise the building block the command composes — `db::seed::seed_spectrum_cataloguing` — against the test pool, including a second run to prove idempotency):

```rust
use db::{Db, fields, seed, vocab};
use sqlx::PgPool;

// Note: `server::seed` opens its own DB connection by URL, but `#[sqlx::test]`
// provisions a temporary database whose URL is not directly exposed. This test
// exercises the building block the command composes — `db::seed::seed_spectrum_cataloguing`
// — against the test pool, run twice to prove the idempotency the command relies on.
#[sqlx::test(migrations = "../db/migrations")]
async fn seed_is_idempotent_via_building_block(pool: PgPool) {
    let db = Db::from_pool(pool);

    for _ in 0..2 {
        let mut tx = db.pool().begin().await.unwrap();
        seed::seed_spectrum_cataloguing(&mut tx).await.unwrap();
        tx.commit().await.unwrap();
    }

    // A representative seeded vocabulary and field definition are present after two runs.
    assert!(
        vocab::vocabulary_by_key(db.pool(), "material").await.unwrap().is_some(),
        "vocabulary 'material' should be seeded"
    );
    assert!(
        fields::field_definition_by_key(db.pool(), "title").await.unwrap().is_some(),
        "field definition 'title' should be seeded"
    );
}
```
(Confirm the seeded keys by reading `crates/db/src/seed.rs` — it seeds vocabularies `material`/`object_name`/`technique` and a field def `title`; adjust the asserted keys if they differ.)

- [ ] **Step 2: Run the test — it should PASS immediately.**
```
DATABASE_URL=postgres://postgres:postgres@localhost:5442/cms_dev cargo nextest run -p server -E 'test(seed_is_idempotent_via_building_block)'
```
Expected: PASS. (Unlike classic TDD, this guards an already-working building block the new command depends on — there is no failing-first state because `db::seed` already exists. The genuinely new code is the glue in Steps 3–4, verified by build + the manual smoke in Step 6.)

- [ ] **Step 3: Add the `Seed` command variant + dispatch** in `crates/server/src/main.rs`. Add to the `Command` enum (after `CreateUser { … }`):
```rust
    /// Seed the baseline Spectrum cataloguing vocabularies + field definitions (idempotent).
    Seed,
```
And add a match arm in `main` (the `match cli.command { … }`), after the `CreateUser` arm:
```rust
        Some(Command::Seed) => seed(&cli.config.database_url).await,
```
Update the import at the top of `main.rs` from `use server::{Config, create_user, run};` to:
```rust
use server::{Config, create_user, run, seed};
```

- [ ] **Step 4: Add the `seed` one-shot** in `crates/server/src/lib.rs`, next to `create_user`:
```rust
/// One-shot: apply migrations (idempotent), then seed the baseline Spectrum cataloguing
/// vocabularies + field definitions. Safe to re-run (the seed is idempotent).
pub async fn seed(database_url: &str) -> anyhow::Result<()> {
    // CLI one-shot: a tiny pool is plenty.
    let db = Db::connect(database_url, 2)
        .await
        .context("connecting to the database")?;

    // Apply migrations first so `server seed` works on a fresh DB without first
    // starting the server. Migrations are idempotent.
    db.migrate()
        .await
        .context("running database migrations")?;

    let mut tx = db.pool().begin().await?;

    db::seed::seed_spectrum_cataloguing(&mut tx)
        .await
        .context("seeding Spectrum cataloguing baseline")?;

    tx.commit().await?;

    println!("seeded Spectrum cataloguing baseline (idempotent)");

    Ok(())
}
```
(`Db`, `anyhow::Context`/`context` are already imported in `lib.rs` — verify the `use` lines; `create_user` already uses `.context(...)` and `Db::connect`, so the imports exist.)

- [ ] **Step 5: Build, fmt, clippy, and run the server tests.**
```
cargo +nightly fmt
cargo clippy --workspace --all-targets -- -D warnings
DATABASE_URL=postgres://postgres:postgres@localhost:5442/cms_dev cargo nextest run -p server
```
Expected: builds clean, clippy clean, all server tests pass (including the existing `create_user` + `config` + `serve` + `embed` tests and the new seed test). Also confirm the subcommand is wired:
```
cargo run -p server -- --help
```
Expected: the help output lists a `seed` subcommand alongside `create-user`.

- [ ] **Step 6: Manual smoke — verify the real command (connect + migrate + commit glue).** With compose up (`docker compose up -d`):
```
DATABASE_URL=postgres://postgres:postgres@localhost:5442/cms_dev cargo run -p server -- seed
DATABASE_URL=postgres://postgres:postgres@localhost:5442/cms_dev cargo run -p server -- seed
```
Expected: both print `seeded Spectrum cataloguing baseline (idempotent)` and exit 0 (the second run is a no-op). (This exercises the URL-connect + migrate + commit path that `#[sqlx::test]` can't.)

- [ ] **Step 7: Commit.**
```bash
git add crates/server
git commit -m "feat(server): 'seed' subcommand wiring the Spectrum cataloguing seed (#14)"
```

---

## Task 2: `just seed` recipe + README note

**Files:**
- Modify: `justfile`
- Modify: `README.md`

- [ ] **Step 1: Add the `seed` recipe** to `justfile`. Insert after the `run` recipe (keeping the existing comment style), before `test`:
```
# Seed the baseline Spectrum cataloguing vocabularies + field definitions (idempotent)
seed:
    cargo run -p server -- seed
```

- [ ] **Step 2: Verify just parses it.**
```
just --list
```
Expected: `seed` appears in the recipe list with its description.

- [ ] **Step 3: Add a seed step to the README "Running locally" setup sequence.** Open `README.md`, find the "Running locally" section and the step that creates the admin user (the `create-user` instruction). Immediately after it, add a step:
```markdown
4. Seed the baseline cataloguing fields (idempotent):

   ```bash
   just seed            # or: cargo run -p server -- seed
   ```
```
(Match the surrounding numbering/formatting of the existing steps — renumber subsequent steps if the section is numbered. Read the section first and adapt the wording to its style; the content is: run `just seed` once after creating the admin user to populate the baseline Spectrum vocabularies + field definitions.)

- [ ] **Step 4: Commit.**
```bash
git add justfile README.md
git commit -m "docs: 'just seed' recipe + README seed step (#14)"
```

---

## Task 3: Final verification

**Files:** none (verification only).

- [ ] **Step 1: Full suite + lints.**
```
cargo +nightly fmt --check
cargo clippy --workspace --all-targets -- -D warnings
DATABASE_URL=postgres://postgres:postgres@localhost:5442/cms_dev MEILI_URL=http://localhost:7700 MEILI_MASTER_KEY=masterKey cargo nextest run --workspace
```
Expected: all green.

- [ ] **Step 2: Codename scan + tree hygiene.**
```
git grep -in 'biggus\|dickus' -- crates README.md justfile || echo "CLEAN"
git status --short
```
Expected: `CLEAN`; working tree clean after the task commits.

---

## Self-Review (completed)

**1. Spec coverage:**
- `server seed` subcommand → Task 1 (main.rs variant + dispatch). ✓
- `server::seed` one-shot mirroring create_user, migrate-first → Task 1 Step 4. ✓
- Idempotent / safe to re-run → asserted in Task 1 Step 1 test + Step 6 smoke. ✓
- `just seed` recipe + README note → Task 2. ✓
- Testing: existing db-layer seed tests unchanged + new server-crate building-block test + manual glue smoke → Task 1. ✓
- Acceptance: nextest green / fmt / clippy / no codename → Task 3. ✓
- Out of scope (no `--seed` flag, no auto-boot, no provisioning, no term seeding, create_user unchanged) → respected; only the four files above change. ✓

**2. Placeholder scan:** No TBD/“handle errors”/“similar to”. The two “confirm the seeded keys / read the section first” notes are verification steps against real files, not deferred implementation; concrete code is given for every code step.

**3. Type consistency:** `seed(database_url: &str) -> anyhow::Result<()>` is defined in Task 1 Step 4 and imported/dispatched in Step 3 (`use server::{… seed}`, `Some(Command::Seed) => seed(&cli.config.database_url).await`). The test uses `db::seed::seed_spectrum_cataloguing(&mut tx)` + `vocab::vocabulary_by_key` + `fields::field_definition_by_key`, all existing signatures (mirrored from `crates/db/tests/seed.rs` and `create_user.rs`).

## Notes
- No new dependencies → no `Cargo.lock` churn expected.
- `Command::Seed` has no clap args; it reuses the flattened `Config.database_url`, exactly like `CreateUser` does.