Twenty-one prompts against the live assistant found one fabricated success,
a model that believed it was 2025, and a describe_garden that was ~450 plop
entries per turn. This is the set of fixes, each traceable to a finding:
- The gardener's LOCAL day travels with the turn (`today` on POST /agent/chat,
sent by the UI like plantedAt) into the system prompt and every dated tool
default. Left to guess, the model dated journal entries a year back; left to
the server, a 9 pm fill landed on UTC's tomorrow.
- describe_garden groups plops by plant — count, where, planted date, days to
maturity — and lists ids only for groups of ≤ 8; list_plantings spells a big
group out on demand and remove_plantings acts on one plant in a bed ("take
the beets out, leave the garlic"), which used to mean 116 single removals.
- New tools: move_planting (keeps the planting date; across beds via the new
MovePlanting, which is why the store's UPDATE now writes object_id),
update_plant, read_history, copy_garden (the "<garden> — <year>" plan
convention). fill_region takes an explicit local rectangle and a seedLotId;
place_planting's radius defaults to one plant (spacing/2) instead of a guess.
- The system prompt states the date and the gardener's units, forbids claiming
a change no tool made, says it cannot undo and points at the Undo button,
asks before clearing beds on an ambiguous sentence, and stops narrating its
own plantings into the journal.
- A mutation aimed at ANOTHER garden inside a turn is recorded under that
garden as its own change set, not filed into the open scope.
- UI: the thread scrolls inside the Assistant panel so the composer stays
put; every tool has a step label; wide tables stay inside the bubble.
Co-Authored-By: Claude Fable 5 <[email protected]>
729 lines
28 KiB
Go
729 lines
28 KiB
Go
package agent
|
||
|
||
import (
|
||
"context"
|
||
"encoding/json"
|
||
"strings"
|
||
"testing"
|
||
|
||
"gitea.stevedudenhoeffer.com/steve/majordomo/llm"
|
||
|
||
"gitea.stevedudenhoeffer.com/steve/pansy/internal/config"
|
||
"gitea.stevedudenhoeffer.com/steve/pansy/internal/domain"
|
||
"gitea.stevedudenhoeffer.com/steve/pansy/internal/service"
|
||
"gitea.stevedudenhoeffer.com/steve/pansy/internal/store"
|
||
)
|
||
|
||
// TestToolboxScenario is the #19 demo harness: it drives the DESIGN scenario
|
||
// through the tool layer (JSON args → DefineTool → service) rather than a live
|
||
// model — fill the NE corner with garlic, the NW with basil, the south with
|
||
// beans — and checks describe_garden reports three groups in the right places.
|
||
// It also confirms the toolbox inherits pansy's ACL (a viewer is refused).
|
||
//
|
||
// Build/run with: go test -tags majordomo ./internal/agent/
|
||
func TestToolboxScenario(t *testing.T) {
|
||
ctx := context.Background()
|
||
svc, ownerID := newAgentTestService(t)
|
||
box := NewToolbox(svc, ownerID, "")
|
||
|
||
call := func(name string, args any) llm.ToolResult {
|
||
t.Helper()
|
||
raw, _ := json.Marshal(args)
|
||
return box.Execute(ctx, llm.ToolCall{ID: "1", Name: name, Arguments: raw})
|
||
}
|
||
|
||
// Garden + plants are set up directly (there are no create_garden/plant tools);
|
||
// the agent-facing bits — object + fills + describe — go through the toolbox.
|
||
g, err := svc.CreateGarden(ctx, ownerID, service.GardenInput{Name: "Plot", WidthCM: 2000, HeightCM: 2000})
|
||
if err != nil {
|
||
t.Fatalf("garden: %v", err)
|
||
}
|
||
garlic := mustPlant(t, svc, ownerID, "Garlic", 15, "🧄")
|
||
basil := mustPlant(t, svc, ownerID, "Basil", 25, "🌿")
|
||
beans := mustPlant(t, svc, ownerID, "Beans", 10, "🫘")
|
||
|
||
// create_object → a 400×400 bed.
|
||
res := call("create_object", map[string]any{
|
||
"gardenId": g.ID, "kind": "bed", "name": "Bed 1", "xCm": 1000, "yCm": 1000, "widthCm": 400, "heightCm": 400,
|
||
})
|
||
if res.IsError {
|
||
t.Fatalf("create_object: %s", res.Content)
|
||
}
|
||
var bed struct {
|
||
ID int64 `json:"id"`
|
||
}
|
||
if err := json.Unmarshal([]byte(res.Content), &bed); err != nil {
|
||
t.Fatalf("decode created object: %v", err)
|
||
}
|
||
|
||
// fill_region for each corner/half.
|
||
for _, f := range []struct {
|
||
region string
|
||
plantID int64
|
||
}{{"ne", garlic.ID}, {"nw", basil.ID}, {"south", beans.ID}} {
|
||
if r := call("fill_region", map[string]any{"objectId": bed.ID, "region": f.region, "plantId": f.plantID}); r.IsError {
|
||
t.Fatalf("fill_region %s: %s", f.region, r.Content)
|
||
}
|
||
}
|
||
|
||
// describe_garden → parse and check locations.
|
||
res = call("describe_garden", map[string]any{"gardenId": g.ID})
|
||
if res.IsError {
|
||
t.Fatalf("describe_garden: %s", res.Content)
|
||
}
|
||
var desc service.DescribeResult
|
||
if err := json.Unmarshal([]byte(res.Content), &desc); err != nil {
|
||
t.Fatalf("decode describe: %v", err)
|
||
}
|
||
if len(desc.Objects) != 1 {
|
||
t.Fatalf("objects = %d, want 1", len(desc.Objects))
|
||
}
|
||
// Plantings come grouped by plant: a group's Where names the region when the
|
||
// whole group sits in one, and a small group also lists its plops.
|
||
seen := map[string]map[string]bool{}
|
||
for _, g := range desc.Objects[0].Plantings {
|
||
if seen[g.Plant] == nil {
|
||
seen[g.Plant] = map[string]bool{}
|
||
}
|
||
seen[g.Plant][g.Where] = true
|
||
for _, p := range g.Each {
|
||
seen[g.Plant][p.Location] = true
|
||
}
|
||
}
|
||
if !seen["Garlic"]["NE corner"] {
|
||
t.Errorf("garlic at %v, want NE corner", seen["Garlic"])
|
||
}
|
||
if !seen["Basil"]["NW corner"] {
|
||
t.Errorf("basil at %v, want NW corner", seen["Basil"])
|
||
}
|
||
if len(seen["Beans"]) == 0 {
|
||
t.Error("beans produced no plops")
|
||
}
|
||
for loc := range seen["Beans"] {
|
||
if loc == "north" || loc == "NE corner" || loc == "NW corner" || loc == "center" {
|
||
t.Errorf("beans at %q, want only southern locations", loc)
|
||
}
|
||
}
|
||
|
||
// ACL: a viewer's fill_region is refused (the toolbox runs as that actor).
|
||
viewerUser, err := svc.Register(ctx, service.RegisterInput{Email: "[email protected]", DisplayName: "V", Password: "password123"})
|
||
if err != nil {
|
||
t.Fatalf("register viewer: %v", err)
|
||
}
|
||
if _, err := svc.AddShare(ctx, ownerID, g.ID, "[email protected]", domain.RoleViewer); err != nil {
|
||
t.Fatalf("share: %v", err)
|
||
}
|
||
viewerBox := NewToolbox(svc, viewerUser.ID, "")
|
||
vr := viewerBox.Execute(ctx, llm.ToolCall{ID: "2", Name: "fill_region", Arguments: mustJSON(t, map[string]any{
|
||
"objectId": bed.ID, "region": "all", "plantId": garlic.ID,
|
||
})})
|
||
if !vr.IsError {
|
||
t.Errorf("viewer fill_region succeeded, want a permission error")
|
||
}
|
||
}
|
||
|
||
func mustJSON(t *testing.T, v any) json.RawMessage {
|
||
t.Helper()
|
||
b, err := json.Marshal(v)
|
||
if err != nil {
|
||
t.Fatalf("marshal: %v", err)
|
||
}
|
||
return b
|
||
}
|
||
|
||
func mustPlant(t *testing.T, svc *service.Service, owner int64, name string, spacing float64, icon string) *domain.Plant {
|
||
t.Helper()
|
||
p, err := svc.CreatePlant(context.Background(), owner, service.PlantInput{
|
||
Name: name, Category: domain.CategoryVegetable, SpacingCM: spacing, Color: "#4a7c3f", Icon: icon,
|
||
})
|
||
if err != nil {
|
||
t.Fatalf("create plant %s: %v", name, err)
|
||
}
|
||
return p
|
||
}
|
||
|
||
// TestGarlicBedToCucumbers is the flagship interaction from #58, driven end to
|
||
// end through the tool layer: "change the garlic garden bed to instead be
|
||
// cucumbers this year".
|
||
//
|
||
// The sequence is the one a model would actually run — describe to find the bed,
|
||
// find_plant to turn the word "cucumber" into an id, clear, refill — and until
|
||
// find_plant existed step three had no way to get its plantId, which blocked the
|
||
// whole thing on one missing tool.
|
||
func TestGarlicBedToCucumbers(t *testing.T) {
|
||
ctx := context.Background()
|
||
svc, owner := newAgentTestService(t)
|
||
box := NewToolbox(svc, owner, "")
|
||
|
||
call := func(name string, args any) llm.ToolResult {
|
||
t.Helper()
|
||
raw, _ := json.Marshal(args)
|
||
return box.Execute(ctx, llm.ToolCall{ID: "1", Name: name, Arguments: raw})
|
||
}
|
||
|
||
g, err := svc.CreateGarden(ctx, owner, service.GardenInput{Name: "Plot", WidthCM: 2000, HeightCM: 2000})
|
||
if err != nil {
|
||
t.Fatalf("garden: %v", err)
|
||
}
|
||
garlic := mustPlant(t, svc, owner, "Garlic", 15, "🧄")
|
||
mustPlant(t, svc, owner, "Cucumber", 45, "🥒")
|
||
|
||
bed, err := svc.CreateObject(ctx, owner, g.ID, service.ObjectInput{
|
||
Kind: domain.KindBed, Name: "Garlic bed", XCM: 1000, YCM: 1000, WidthCM: 400, HeightCM: 400,
|
||
})
|
||
if err != nil {
|
||
t.Fatalf("bed: %v", err)
|
||
}
|
||
if _, err := svc.FillNamedRegion(ctx, owner, bed.ID, "all", garlic.ID, nil, service.FillClump, nil); err != nil {
|
||
t.Fatalf("seed the garlic: %v", err)
|
||
}
|
||
|
||
// 1. describe_garden — "the garlic bed" resolves to an object id, because the
|
||
// description carries the NAMES of what's planted in each object.
|
||
res := call("describe_garden", map[string]any{"gardenId": g.ID})
|
||
if res.IsError {
|
||
t.Fatalf("describe_garden: %s", res.Content)
|
||
}
|
||
if !strings.Contains(res.Content, "Garlic") {
|
||
t.Fatalf("describe_garden didn't name what's planted: %s", res.Content)
|
||
}
|
||
|
||
// 2. find_plant — the step that used to be impossible.
|
||
res = call("find_plant", map[string]any{"query": "cucumber"})
|
||
if res.IsError {
|
||
t.Fatalf("find_plant: %s", res.Content)
|
||
}
|
||
var matches []struct {
|
||
ID int64 `json:"id"`
|
||
Name string `json:"name"`
|
||
}
|
||
if err := json.Unmarshal([]byte(res.Content), &matches); err != nil {
|
||
t.Fatalf("decode find_plant: %v (%s)", err, res.Content)
|
||
}
|
||
if len(matches) == 0 || matches[0].Name != "Cucumber" {
|
||
t.Fatalf("find_plant(cucumber) = %+v", matches)
|
||
}
|
||
|
||
// 3. clear_object, 4. fill_region.
|
||
if r := call("clear_object", map[string]any{"objectId": bed.ID}); r.IsError {
|
||
t.Fatalf("clear_object: %s", r.Content)
|
||
}
|
||
if r := call("fill_region", map[string]any{
|
||
"objectId": bed.ID, "region": "all", "plantId": matches[0].ID,
|
||
}); r.IsError {
|
||
t.Fatalf("fill_region: %s", r.Content)
|
||
}
|
||
|
||
// The bed is cucumbers, and no garlic is still growing in it.
|
||
full, err := svc.GardenFull(ctx, owner, g.ID, nil)
|
||
if err != nil {
|
||
t.Fatalf("GardenFull: %v", err)
|
||
}
|
||
if len(full.Plantings) == 0 {
|
||
t.Fatal("the bed ended up empty")
|
||
}
|
||
for _, p := range full.Plantings {
|
||
if p.PlantID == garlic.ID {
|
||
t.Errorf("garlic is still active in the bed")
|
||
}
|
||
if p.PlantID != matches[0].ID {
|
||
t.Errorf("unexpected plant %d in the bed", p.PlantID)
|
||
}
|
||
}
|
||
}
|
||
|
||
// TestFindPlantReturnsCandidatesNotAGuess — "garlic" against a catalog holding
|
||
// both "Garlic" and "German Red Garlic" is genuinely ambiguous, and silently
|
||
// picking one is how the agent plants the wrong thing.
|
||
func TestFindPlantReturnsCandidatesNotAGuess(t *testing.T) {
|
||
ctx := context.Background()
|
||
svc, owner := newAgentTestService(t)
|
||
box := NewToolbox(svc, owner, "")
|
||
|
||
mustPlant(t, svc, owner, "German Red Garlic", 15, "🧄")
|
||
|
||
raw, _ := json.Marshal(map[string]any{"query": "garlic"})
|
||
res := box.Execute(ctx, llm.ToolCall{ID: "1", Name: "find_plant", Arguments: raw})
|
||
if res.IsError {
|
||
t.Fatalf("find_plant: %s", res.Content)
|
||
}
|
||
var matches []struct {
|
||
Name string `json:"name"`
|
||
}
|
||
if err := json.Unmarshal([]byte(res.Content), &matches); err != nil {
|
||
t.Fatalf("decode: %v", err)
|
||
}
|
||
names := map[string]bool{}
|
||
for _, m := range matches {
|
||
names[m.Name] = true
|
||
}
|
||
// The built-in catalog seeds a plain "Garlic"; the custom one is ours.
|
||
if !names["Garlic"] || !names["German Red Garlic"] {
|
||
t.Errorf("find_plant(garlic) = %+v, want both the built-in and the custom variety", matches)
|
||
}
|
||
// Exact match ranks first, so a caller taking [0] gets the least surprising one.
|
||
if len(matches) > 0 && matches[0].Name != "Garlic" {
|
||
t.Errorf("first match = %q, want the exact name", matches[0].Name)
|
||
}
|
||
}
|
||
|
||
// TestCreatePlantIsUserScoped — the catalog belongs to the user, not to any
|
||
// garden, so someone with no editable garden can still name a new variety. The
|
||
// issue asks for this to be asserted rather than assumed.
|
||
func TestCreatePlantIsUserScoped(t *testing.T) {
|
||
ctx := context.Background()
|
||
svc, owner := newAgentTestService(t)
|
||
other, err := svc.Register(ctx, service.RegisterInput{Email: "[email protected]", DisplayName: "B", Password: "password123"})
|
||
if err != nil {
|
||
t.Fatalf("register: %v", err)
|
||
}
|
||
box := NewToolbox(svc, other.ID, "")
|
||
|
||
raw, _ := json.Marshal(map[string]any{
|
||
"name": "Painted Mountain Corn", "category": "vegetable",
|
||
"spacingCm": 30.0, "color": "#c08a3f", "icon": "🌽",
|
||
})
|
||
res := box.Execute(ctx, llm.ToolCall{ID: "1", Name: "create_plant", Arguments: raw})
|
||
if res.IsError {
|
||
t.Fatalf("create_plant: %s", res.Content)
|
||
}
|
||
var created struct {
|
||
ID int64 `json:"id"`
|
||
OwnerID *int64 `json:"ownerId"`
|
||
Name string `json:"name"`
|
||
}
|
||
if err := json.Unmarshal([]byte(res.Content), &created); err != nil {
|
||
t.Fatalf("decode: %v", err)
|
||
}
|
||
if created.OwnerID == nil || *created.OwnerID != other.ID {
|
||
t.Errorf("ownerId = %v, want the acting user %d", created.OwnerID, other.ID)
|
||
}
|
||
// And it stays theirs: the other user's catalog doesn't gain it.
|
||
ownerPlants, err := svc.ListPlants(ctx, owner)
|
||
if err != nil {
|
||
t.Fatalf("ListPlants: %v", err)
|
||
}
|
||
for _, p := range ownerPlants {
|
||
if p.Name == "Painted Mountain Corn" {
|
||
t.Error("a plant created by one user showed up in another's catalog")
|
||
}
|
||
}
|
||
}
|
||
|
||
// TestJournalToolWritesADatedObservation — "note that the west bed has mildew"
|
||
// is squarely the kind of thing you say out loud while walking around.
|
||
func TestJournalToolWritesADatedObservation(t *testing.T) {
|
||
ctx := context.Background()
|
||
svc, owner := newAgentTestService(t)
|
||
box := NewToolbox(svc, owner, "")
|
||
|
||
g, err := svc.CreateGarden(ctx, owner, service.GardenInput{Name: "Plot", WidthCM: 2000, HeightCM: 2000})
|
||
if err != nil {
|
||
t.Fatalf("garden: %v", err)
|
||
}
|
||
bed, err := svc.CreateObject(ctx, owner, g.ID, service.ObjectInput{
|
||
Kind: domain.KindBed, Name: "West bed", XCM: 500, YCM: 500, WidthCM: 200, HeightCM: 200,
|
||
})
|
||
if err != nil {
|
||
t.Fatalf("bed: %v", err)
|
||
}
|
||
|
||
raw, _ := json.Marshal(map[string]any{
|
||
"gardenId": g.ID, "objectId": bed.ID,
|
||
"body": "Powdery mildew on the west bed", "observedAt": "2026-08-14",
|
||
})
|
||
res := box.Execute(ctx, llm.ToolCall{ID: "1", Name: "add_journal_entry", Arguments: raw})
|
||
if res.IsError {
|
||
t.Fatalf("add_journal_entry: %s", res.Content)
|
||
}
|
||
|
||
entries, _, err := svc.ListJournal(ctx, owner, g.ID, service.JournalQuery{ObjectID: &bed.ID})
|
||
if err != nil {
|
||
t.Fatalf("ListJournal: %v", err)
|
||
}
|
||
if len(entries) != 1 {
|
||
t.Fatalf("got %d entries, want 1", len(entries))
|
||
}
|
||
if entries[0].Body != "Powdery mildew on the west bed" || entries[0].ObservedAt != "2026-08-14" {
|
||
t.Errorf("unexpected entry: %+v", entries[0])
|
||
}
|
||
if entries[0].AuthorID != owner {
|
||
t.Errorf("author = %d, want the acting user %d", entries[0].AuthorID, owner)
|
||
}
|
||
}
|
||
|
||
// TestCorrectiveTools covers the #85 gaps: the agent can now read the journal it
|
||
// could only write, resize and delete an object it could only create and move,
|
||
// pull a single plop instead of clearing the whole bed, and record/read seed
|
||
// lots. Each is driven through the tool layer the way a model would run it.
|
||
func TestCorrectiveTools(t *testing.T) {
|
||
ctx := context.Background()
|
||
svc, owner := newAgentTestService(t)
|
||
box := NewToolbox(svc, owner, "")
|
||
|
||
var gid int64 // set once the garden exists; the describe closure reads it.
|
||
call := func(name string, args any) llm.ToolResult {
|
||
t.Helper()
|
||
return box.Execute(ctx, llm.ToolCall{ID: "1", Name: name, Arguments: mustJSON(t, args)})
|
||
}
|
||
describe := func() service.DescribeResult {
|
||
t.Helper()
|
||
res := call("describe_garden", map[string]any{"gardenId": gid})
|
||
if res.IsError {
|
||
t.Fatalf("describe_garden: %s", res.Content)
|
||
}
|
||
var d service.DescribeResult
|
||
if err := json.Unmarshal([]byte(res.Content), &d); err != nil {
|
||
t.Fatalf("decode describe: %v", err)
|
||
}
|
||
return d
|
||
}
|
||
|
||
g, err := svc.CreateGarden(ctx, owner, service.GardenInput{Name: "Plot", WidthCM: 2000, HeightCM: 2000})
|
||
if err != nil {
|
||
t.Fatalf("garden: %v", err)
|
||
}
|
||
gid = g.ID
|
||
basil := mustPlant(t, svc, owner, "Basil", 25, "🌿")
|
||
bed, err := svc.CreateObject(ctx, owner, g.ID, service.ObjectInput{
|
||
Kind: domain.KindBed, Name: "Bed", XCM: 1000, YCM: 1000, WidthCM: 400, HeightCM: 400,
|
||
})
|
||
if err != nil {
|
||
t.Fatalf("bed: %v", err)
|
||
}
|
||
|
||
// update_object: "make that bed 100cm wider" — read the version, pass a new width.
|
||
d := describe()
|
||
if r := call("update_object", map[string]any{
|
||
"objectId": bed.ID, "version": d.Objects[0].Version, "widthCm": 500.0,
|
||
}); r.IsError {
|
||
t.Fatalf("update_object: %s", r.Content)
|
||
}
|
||
if w := describe().Objects[0].WidthCM; w != 500 {
|
||
t.Errorf("width = %v after update_object, want 500", w)
|
||
}
|
||
|
||
// place a plop, then remove_planting it by id+version — one plop, not the bed.
|
||
if r := call("place_planting", map[string]any{
|
||
"objectId": bed.ID, "plantId": basil.ID, "xCm": 0, "yCm": 0, "radiusCm": 30,
|
||
}); r.IsError {
|
||
t.Fatalf("place_planting: %s", r.Content)
|
||
}
|
||
d = describe()
|
||
if len(d.Objects[0].Plantings) != 1 || len(d.Objects[0].Plantings[0].Each) != 1 {
|
||
t.Fatalf("want 1 plop before removal, got %+v", d.Objects[0].Plantings)
|
||
}
|
||
plop := d.Objects[0].Plantings[0].Each[0]
|
||
if r := call("remove_planting", map[string]any{"plantingId": plop.ID, "version": plop.Version}); r.IsError {
|
||
t.Fatalf("remove_planting: %s", r.Content)
|
||
}
|
||
if n := len(describe().Objects[0].Plantings); n != 0 {
|
||
t.Errorf("want 0 active plops after remove_planting, got %d", n)
|
||
}
|
||
|
||
// add then read the journal — the write/read asymmetry the issue flagged.
|
||
if r := call("add_journal_entry", map[string]any{
|
||
"gardenId": g.ID, "objectId": bed.ID, "body": "aphids", "observedAt": "2026-06-01",
|
||
}); r.IsError {
|
||
t.Fatalf("add_journal_entry: %s", r.Content)
|
||
}
|
||
res := call("read_journal", map[string]any{"gardenId": g.ID, "objectId": bed.ID})
|
||
if res.IsError {
|
||
t.Fatalf("read_journal: %s", res.Content)
|
||
}
|
||
var jr struct {
|
||
Entries []struct {
|
||
Body string `json:"body"`
|
||
} `json:"entries"`
|
||
}
|
||
if err := json.Unmarshal([]byte(res.Content), &jr); err != nil {
|
||
t.Fatalf("decode read_journal: %v (%s)", err, res.Content)
|
||
}
|
||
if len(jr.Entries) != 1 || jr.Entries[0].Body != "aphids" {
|
||
t.Errorf("read_journal = %+v, want the one aphids entry", jr.Entries)
|
||
}
|
||
|
||
// record then list a seed lot — the detail behind find_plant's "remaining".
|
||
if r := call("record_seed_lot", map[string]any{
|
||
"plantId": basil.ID, "quantity": 2.0, "unit": "packets", "vendor": "Johnny's",
|
||
}); r.IsError {
|
||
t.Fatalf("record_seed_lot: %s", r.Content)
|
||
}
|
||
res = call("list_seed_lots", map[string]any{"plantId": basil.ID})
|
||
if res.IsError {
|
||
t.Fatalf("list_seed_lots: %s", res.Content)
|
||
}
|
||
var lots []struct {
|
||
Quantity float64 `json:"quantity"`
|
||
Unit string `json:"unit"`
|
||
}
|
||
if err := json.Unmarshal([]byte(res.Content), &lots); err != nil {
|
||
t.Fatalf("decode list_seed_lots: %v (%s)", err, res.Content)
|
||
}
|
||
if len(lots) != 1 || lots[0].Quantity != 2 || lots[0].Unit != "packets" {
|
||
t.Errorf("list_seed_lots = %+v, want one lot of 2 packets", lots)
|
||
}
|
||
|
||
// delete_object: the counterpart to create_object.
|
||
if r := call("delete_object", map[string]any{"objectId": bed.ID}); r.IsError {
|
||
t.Fatalf("delete_object: %s", r.Content)
|
||
}
|
||
if n := len(describe().Objects); n != 0 {
|
||
t.Errorf("want 0 objects after delete_object, got %d", n)
|
||
}
|
||
}
|
||
|
||
// newAgentTestService spins up an in-memory pansy with one registered user.
|
||
func newAgentTestService(t *testing.T) (*service.Service, int64) {
|
||
t.Helper()
|
||
ctx := context.Background()
|
||
db, err := store.Open(":memory:")
|
||
if err != nil {
|
||
t.Fatalf("open: %v", err)
|
||
}
|
||
t.Cleanup(func() { db.Close() })
|
||
if err := db.Migrate(ctx); err != nil {
|
||
t.Fatalf("migrate: %v", err)
|
||
}
|
||
svc := service.New(db, &config.Config{Registration: config.RegistrationOpen, LocalAuth: true})
|
||
owner, err := svc.Register(ctx, service.RegisterInput{Email: "[email protected]", DisplayName: "A", Password: "password123"})
|
||
if err != nil {
|
||
t.Fatalf("register: %v", err)
|
||
}
|
||
return svc, owner.ID
|
||
}
|
||
|
||
// TestToolsFromTheLiveSweep covers what a day of driving the live assistant
|
||
// asked for: grouped describes, whole-group removal, moves that keep the
|
||
// planting date, fills by rectangle, seed attribution, catalog edits, history
|
||
// reads, plan copies — and every date stamped the gardener's local day rather
|
||
// than the server's (UTC) or the model's (a year from its training data).
|
||
func TestToolsFromTheLiveSweep(t *testing.T) {
|
||
ctx := context.Background()
|
||
svc, owner := newAgentTestService(t)
|
||
const today = "2026-08-22"
|
||
box := NewToolbox(svc, owner, today)
|
||
call := func(name string, args any) llm.ToolResult {
|
||
t.Helper()
|
||
return box.Execute(ctx, llm.ToolCall{ID: "1", Name: name, Arguments: mustJSON(t, args)})
|
||
}
|
||
ok := func(name string, args any) string {
|
||
t.Helper()
|
||
r := call(name, args)
|
||
if r.IsError {
|
||
t.Fatalf("%s: %s", name, r.Content)
|
||
}
|
||
return r.Content
|
||
}
|
||
decode := func(raw string, into any) {
|
||
t.Helper()
|
||
if err := json.Unmarshal([]byte(raw), into); err != nil {
|
||
t.Fatalf("decode %v: %s", err, raw)
|
||
}
|
||
}
|
||
|
||
g, err := svc.CreateGarden(ctx, owner, service.GardenInput{Name: "Plot", WidthCM: 2000, HeightCM: 2000, UnitPref: domain.UnitImperial})
|
||
if err != nil {
|
||
t.Fatalf("garden: %v", err)
|
||
}
|
||
garlic := mustPlant(t, svc, owner, "Garlic", 15, "🧄")
|
||
beet := mustPlant(t, svc, owner, "Beet", 10, "🌱")
|
||
tomato := mustPlant(t, svc, owner, "Cherokee Purple", 60, "🍅")
|
||
bed, err := svc.CreateObject(ctx, owner, g.ID, service.ObjectInput{Kind: domain.KindBed, Name: "South bed", XCM: 1000, YCM: 1000, WidthCM: 240, HeightCM: 120})
|
||
if err != nil {
|
||
t.Fatalf("bed: %v", err)
|
||
}
|
||
other, err := svc.CreateObject(ctx, owner, g.ID, service.ObjectInput{Kind: domain.KindBed, Name: "North bed", XCM: 1000, YCM: 300, WidthCM: 240, HeightCM: 120})
|
||
if err != nil {
|
||
t.Fatalf("other bed: %v", err)
|
||
}
|
||
lot, err := svc.CreateSeedLot(ctx, owner, service.SeedLotInput{PlantID: beet.ID, Quantity: 500, Unit: "seeds"})
|
||
if err != nil {
|
||
t.Fatalf("lot: %v", err)
|
||
}
|
||
|
||
// fill_region by rectangle (the middle third of the bed's width), in grid mode,
|
||
// charged to the lot, dated today by default.
|
||
ok("fill_region", map[string]any{
|
||
"objectId": bed.ID, "plantId": beet.ID, "mode": "grid", "seedLotId": lot.ID,
|
||
"x0Cm": -40.0, "y0Cm": -60.0, "x1Cm": 40.0, "y1Cm": 60.0,
|
||
})
|
||
// Neither a region nor a full rectangle is a mistake the model can read.
|
||
if r := call("fill_region", map[string]any{"objectId": bed.ID, "plantId": beet.ID, "x0Cm": -40.0}); !r.IsError || !strings.Contains(r.Content, "x0Cm, y0Cm, x1Cm, y1Cm") {
|
||
t.Errorf("half a rectangle: %+v, want a readable refusal", r)
|
||
}
|
||
if r := call("fill_region", map[string]any{"objectId": bed.ID, "plantId": beet.ID}); !r.IsError {
|
||
t.Error("fill_region with nowhere to fill succeeded")
|
||
}
|
||
// place_planting without a radius → one plant at half the spacing; two garlic
|
||
// cloves along the north edge, dated today.
|
||
ok("place_planting", map[string]any{"objectId": bed.ID, "plantId": garlic.ID, "xCm": -100, "yCm": -50})
|
||
ok("place_planting", map[string]any{"objectId": bed.ID, "plantId": garlic.ID, "xCm": 100, "yCm": -50})
|
||
// And a tomato planted back in May, with an explicit date.
|
||
ok("place_planting", map[string]any{"objectId": bed.ID, "plantId": tomato.ID, "xCm": 0, "yCm": 0, "plantedAt": "2026-05-20"})
|
||
|
||
// describe_garden: one group per plant, dated; only the small ones listed.
|
||
var d service.DescribeResult
|
||
groups := func() map[string]service.DescribeGroup {
|
||
t.Helper()
|
||
decode(ok("describe_garden", map[string]any{"gardenId": g.ID}), &d)
|
||
out := map[string]service.DescribeGroup{}
|
||
for _, o := range d.Objects {
|
||
if o.ID == bed.ID {
|
||
for _, gr := range o.Plantings {
|
||
out[gr.Plant] = gr
|
||
}
|
||
}
|
||
}
|
||
return out
|
||
}
|
||
gs := groups()
|
||
beets := gs["Beet"]
|
||
if beets.Plops <= 8 || beets.Each != nil {
|
||
t.Errorf("beets: %d plops, each=%v; want a large group with no per-plop listing", beets.Plops, beets.Each)
|
||
}
|
||
if beets.PlantedAt != today || beets.Plants != beets.Plops {
|
||
t.Errorf("beets plantedAt %q plants %d; want today and one plant per grid plop", beets.PlantedAt, beets.Plants)
|
||
}
|
||
if !strings.Contains(beets.Where, "cm from the centre") {
|
||
t.Errorf("beets where = %q, want the bounding box of a middle-third fill", beets.Where)
|
||
}
|
||
cloves := gs["Garlic"]
|
||
if cloves.Plops != 2 || len(cloves.Each) != 2 || cloves.Where != "north half" || cloves.PlantedAt != today {
|
||
t.Errorf("garlic group = %+v, want 2 listed plops in the north half, dated today", cloves)
|
||
}
|
||
if r := cloves.Each[0].RadiusCM; r != 7.5 {
|
||
t.Errorf("a clove placed without a radius got %v, want spacing/2 = 7.5", r)
|
||
}
|
||
tom := gs["Cherokee Purple"]
|
||
if tom.Plops != 1 || tom.Where != "center" || tom.PlantedAt != "2026-05-20" {
|
||
t.Errorf("tomato group = %+v, want one plop at the center dated 2026-05-20", tom)
|
||
}
|
||
|
||
// The lot counts the beets as used.
|
||
var lots []struct {
|
||
Used float64 `json:"used"`
|
||
Remaining float64 `json:"remaining"`
|
||
}
|
||
decode(ok("list_seed_lots", map[string]any{"plantId": beet.ID}), &lots)
|
||
if len(lots) != 1 || lots[0].Used != float64(beets.Plants) || lots[0].Remaining != 500-float64(beets.Plants) {
|
||
t.Errorf("lots = %+v, want %d used of 500", lots, beets.Plants)
|
||
}
|
||
|
||
// list_plantings spells the big group out, narrowed to one plant.
|
||
var listed []service.DescribePlanting
|
||
decode(ok("list_plantings", map[string]any{"objectId": bed.ID, "plantId": beet.ID}), &listed)
|
||
if len(listed) != beets.Plops {
|
||
t.Errorf("list_plantings: %d beets, want %d", len(listed), beets.Plops)
|
||
}
|
||
|
||
// move_planting: the tomato to the north bed, date kept; a within-bed move too.
|
||
var moved domain.Planting
|
||
decode(ok("move_planting", map[string]any{
|
||
"plantingId": tom.Each[0].ID, "version": tom.Each[0].Version, "toObjectId": other.ID, "xCm": 10.0, "yCm": -20.0,
|
||
}), &moved)
|
||
if moved.ObjectID != other.ID || moved.PlantedAt == nil || *moved.PlantedAt != "2026-05-20" {
|
||
t.Errorf("moved tomato = %+v, want it in the north bed with its May date", moved)
|
||
}
|
||
decode(ok("move_planting", map[string]any{
|
||
"plantingId": cloves.Each[0].ID, "version": cloves.Each[0].Version, "xCm": -110.0, "yCm": -55.0,
|
||
}), &moved)
|
||
if moved.ObjectID != bed.ID || moved.XCM != -110 {
|
||
t.Errorf("within-bed move = %+v, want the same bed at x=-110", moved)
|
||
}
|
||
|
||
// remove_plantings: the beets out, the garlic stays — dated today.
|
||
var removed struct {
|
||
Removed int `json:"removed"`
|
||
}
|
||
decode(ok("remove_plantings", map[string]any{"objectId": bed.ID, "plantId": beet.ID}), &removed)
|
||
if removed.Removed != beets.Plops {
|
||
t.Errorf("remove_plantings removed %d, want the %d beets", removed.Removed, beets.Plops)
|
||
}
|
||
gs = groups()
|
||
if _, still := gs["Beet"]; still || gs["Garlic"].Plops != 2 {
|
||
t.Errorf("after remove_plantings the bed has %+v, want the garlic only", gs)
|
||
}
|
||
var pulled domain.Planting
|
||
decode(ok("remove_planting", map[string]any{"plantingId": gs["Garlic"].Each[0].ID, "version": gs["Garlic"].Each[0].Version}), &pulled)
|
||
if pulled.RemovedAt == nil || *pulled.RemovedAt != today {
|
||
t.Errorf("remove_planting dated the removal %v, want today %s", pulled.RemovedAt, today)
|
||
}
|
||
|
||
// read_history sees all of that, newest first, and marks what was undone.
|
||
var hist struct {
|
||
Entries []historyEntry `json:"entries"`
|
||
HasMore bool `json:"hasMore"`
|
||
}
|
||
decode(ok("read_history", map[string]any{"gardenId": g.ID, "limit": 3}), &hist)
|
||
if len(hist.Entries) != 3 || !hist.HasMore {
|
||
t.Fatalf("read_history = %d entries, hasMore=%v; want 3 and more", len(hist.Entries), hist.HasMore)
|
||
}
|
||
if e := hist.Entries[1]; !strings.HasPrefix(e.Summary, "Removed Beet from South bed") || !strings.Contains(e.Changes, "planting") || e.Undone {
|
||
t.Errorf("entry = %+v, want the beet removal, not undone", e)
|
||
}
|
||
if _, conflicts, err := svc.RevertChangeSet(ctx, owner, hist.Entries[1].ID, domain.SourceUI); err != nil || len(conflicts) != 0 {
|
||
t.Fatalf("undo: err=%v conflicts=%+v", err, conflicts)
|
||
}
|
||
decode(ok("read_history", map[string]any{"gardenId": g.ID, "limit": 3}), &hist)
|
||
if hist.Entries[0].Undo == nil || !hist.Entries[2].Undone {
|
||
t.Errorf("after an undo: newest = %+v, undone = %+v; want the revert to point at the removal, and the removal marked undone", hist.Entries[0], hist.Entries[2])
|
||
}
|
||
|
||
// update_plant on the user's own plant; a built-in is refused.
|
||
var matches []struct {
|
||
ID int64 `json:"id"`
|
||
Version int64 `json:"version"`
|
||
}
|
||
decode(ok("find_plant", map[string]any{"query": "cherokee"}), &matches)
|
||
var updated domain.Plant
|
||
decode(ok("update_plant", map[string]any{"plantId": matches[0].ID, "version": matches[0].Version, "daysToMaturity": 75}), &updated)
|
||
if updated.DaysToMaturity == nil || *updated.DaysToMaturity != 75 || updated.Name != "Cherokee Purple" {
|
||
t.Errorf("update_plant = %+v, want days 75 and the name untouched", updated)
|
||
}
|
||
decode(ok("find_plant", map[string]any{"query": "basil"}), &matches)
|
||
if r := call("update_plant", map[string]any{"plantId": matches[0].ID, "version": matches[0].Version, "daysToMaturity": 60}); !r.IsError {
|
||
t.Error("update_plant changed a built-in")
|
||
}
|
||
|
||
// add_journal_entry is dated today unless told otherwise.
|
||
ok("add_journal_entry", map[string]any{"gardenId": g.ID, "body": "aphids on the beets"})
|
||
entries, _, err := svc.ListJournal(ctx, owner, g.ID, service.JournalQuery{})
|
||
if err != nil {
|
||
t.Fatalf("ListJournal: %v", err)
|
||
}
|
||
if len(entries) != 1 || entries[0].ObservedAt != today {
|
||
t.Errorf("journal = %+v, want one entry observed %s", entries, today)
|
||
}
|
||
|
||
// copy_garden makes next year's plan: a whole copy under the plan name.
|
||
var plan domain.Garden
|
||
decode(ok("copy_garden", map[string]any{"gardenId": g.ID, "name": "Plot — 2027"}), &plan)
|
||
if plan.Name != "Plot — 2027" || plan.ID == g.ID {
|
||
t.Errorf("copy_garden = %+v, want a new garden named for the plan", plan)
|
||
}
|
||
decode(ok("describe_garden", map[string]any{"gardenId": plan.ID}), &d)
|
||
if len(d.Objects) != 2 {
|
||
t.Errorf("the plan copy has %d objects, want the source's 2", len(d.Objects))
|
||
}
|
||
}
|
||
|
||
// TestToolsDefaultToTheServiceDayWithoutOne — a toolbox built with no local day
|
||
// (a bare API caller) still dates everything: the service's UTC today.
|
||
func TestToolsDefaultToTheServiceDayWithoutOne(t *testing.T) {
|
||
a := &adapter{today: ""}
|
||
if d := a.day(""); d != nil {
|
||
t.Errorf("no day at all → %q, want nil (the service default)", *d)
|
||
}
|
||
if d := a.day(" 2026-01-02 "); d == nil || *d != "2026-01-02" {
|
||
t.Errorf("an explicit day → %v, want it trimmed", d)
|
||
}
|
||
a.today = "2026-08-22"
|
||
if d := a.day(""); d == nil || *d != "2026-08-22" {
|
||
t.Errorf("the gardener's day → %v, want 2026-08-22", d)
|
||
}
|
||
if d := a.day("2026-05-20"); d == nil || *d != "2026-05-20" {
|
||
t.Errorf("an explicit day beats the default: %v", d)
|
||
}
|
||
}
|