v0.35.0 21873e8

resident_tools_test.go RawHistoryFolder

1package tools
2
3import (
4	"context"
5	"encoding/json"
6	"errors"
7	"strings"
8	"sync"
9	"testing"
10	"time"
11
12	"github.com/bnema/zut/packages/agent/subagents"
13	"github.com/bnema/zut/packages/core"
14	"github.com/bnema/zut/packages/provider"
15)
16
17func TestResidentStatusRetrievesFailedResultWithoutExecution(t *testing.T) {
18	failure := errors.New("worker failed")
19	manager := subagents.NewResidentManager(t.TempDir(), func(_ subagents.ResidentChildSpec, journal *subagents.ResidentJournal) (subagents.ResidentTurnRunner, error) {
20		return func(context.Context, string) error {
21			if err := journal.RecordAgentEvent(core.EvAssistantMessage{Message: provider.Message{Role: provider.RoleAssistant, Content: []provider.Content{provider.TextBlock{Text: "partial finding"}}}}); err != nil {
22				return err
23			}
24			return failure
25		}, nil
26	})
27	t.Cleanup(func() { _ = manager.Close(context.Background()) })
28	spec := subagents.ResidentChildSpec{ID: "failed-status", InitialTurnID: "initial", SessionID: "session", Provider: "openai", Model: "test", Required: true}
29	completed, cancel := manager.WatchCompletion(spec.ID, spec.InitialTurnID)
30	defer cancel()
31	if _, err := manager.Spawn(t.Context(), spec, "investigate"); err != nil {
32		t.Fatal(err)
33	}
34	select {
35	case <-completed:
36	case <-time.After(time.Second):
37		t.Fatal("no completion")
38	}
39	status := &SubagentStatusTool{ResidentManager: manager, Enabled: func() bool { return true }}
40	for _, include := range []bool{false, true} {
41		raw := json.RawMessage(`{"agent_id":"failed-status"}`)
42		if include {
43			raw = json.RawMessage(`{"agent_id":"failed-status","include_result":true}`)
44		}
45		result, err := status.Execute(t.Context(), raw, nil)
46		if err != nil || result.IsError {
47			t.Fatalf("status: %#v, %v", result, err)
48		}
49		response := result.Details.(subagentStatusResponse)
50		if response.Agent.State != subagents.ResidentFailed || (response.Result != nil) != include {
51			t.Fatalf("response = %#v", response)
52		}
53		if include && (response.Result.Handoff != "" || response.Result.ErrorCode != "turn_failed" || response.Result.Summary != "partial finding") {
54			t.Fatalf("result = %#v", response.Result)
55		}
56	}
57	result, err := status.Execute(t.Context(), json.RawMessage(`{"include_result":true}`), nil)
58	if err != nil || !result.IsError {
59		t.Fatalf("missing child ID must fail: %#v, %v", result, err)
60	}
61	if len(manager.UnmetRequired()) != 1 {
62		t.Fatal("reading a result satisfied required work")
63	}
64}
65
66func TestResidentToolsUseManagerOnly(t *testing.T) {
67	runs := make(chan string, 2)
68	manager := subagents.NewResidentManager(t.TempDir(), func(subagents.ResidentChildSpec, *subagents.ResidentJournal) (subagents.ResidentTurnRunner, error) {
69		return func(_ context.Context, prompt string) error { runs <- prompt; return nil }, nil
70	})
71	t.Cleanup(func() { _ = manager.Close(context.Background()) })
72	spawn := &SubagentSpawnTool{
73		ResidentManager: manager,
74		Enabled:         func() bool { return true },
75		DefaultProvider: func() string { return "openai" },
76		DefaultModel:    func() string { return "gpt-5" },
77		BuildResidentSpec: func(_ context.Context, request ResidentSpawnRequest) (subagents.ResidentChildSpec, error) {
78			return subagents.ResidentChildSpec{ID: "resident-tool", SessionID: "child-session", Provider: request.Provider, Model: request.Model, Required: request.Required}, nil
79		},
80	}
81	result, err := spawn.Execute(context.Background(), json.RawMessage(`{"task":"initial"}`), nil)
82	if err != nil {
83		t.Fatal(err)
84	}
85	if got := toolResultText(t, result); !strings.Contains(got, "owns the delegated scope") || !strings.Contains(got, "Do not repeat that work in the parent") {
86		t.Fatalf("spawn result lacks ownership guidance: %q", got)
87	}
88	if got := <-runs; got != "initial" {
89		t.Fatalf("initial prompt = %q", got)
90	}
91	status := &SubagentStatusTool{ResidentManager: manager, Enabled: func() bool { return true }}
92	result, err = status.Execute(context.Background(), json.RawMessage(`{"agent_id":"resident-tool"}`), nil)
93	if err != nil {
94		t.Fatal(err)
95	}
96	if details, ok := result.Details.(subagentStatusResponse); !ok || details.Agent == nil || details.Agent.State == "" {
97		t.Fatalf("status details = %#v", result.Details)
98	}
99	resume := &SubagentResumeTool{ResidentManager: manager, Enabled: func() bool { return true }}
100	if _, err := resume.Execute(context.Background(), json.RawMessage(`{"agent_id":"resident-tool","prompt":"follow up"}`), nil); err != nil {
101		t.Fatal(err)
102	}
103	if got := <-runs; got != "follow up" {
104		t.Fatalf("follow-up prompt = %q", got)
105	}
106}
107
108func TestSubagentSpawnGuidanceRequiresIndependentOwnership(t *testing.T) {
109	spawn := &SubagentSpawnTool{}
110	for _, want := range []string{"independent sidecar", "keep immediate blockers local", "never duplicate it in the parent", "end or yield the parent turn"} {
111		if got := spawn.Description(); !strings.Contains(got, want) {
112			t.Fatalf("description missing %q: %s", want, got)
113		}
114	}
115	var schema struct {
116		Properties map[string]struct {
117			Description string `json:"description"`
118		} `json:"properties"`
119	}
120	if err := json.Unmarshal(spawn.Schema(), &schema); err != nil {
121		t.Fatal(err)
122	}
123	task := schema.Properties["task"].Description
124	for _, want := range []string{"concrete, bounded scope", "explicit ownership", "does not overlap", "Shared isolation uses this working directory", "worktree isolation captures a patch without merging it"} {
125		if !strings.Contains(task, want) {
126			t.Fatalf("task schema missing %q: %s", want, task)
127		}
128	}
129}
130
131func TestResidentSpawnWaitReturnsInitialCompletion(t *testing.T) {
132	for _, tc := range []struct {
133		name   string
134		err    error
135		status string
136	}{
137		{"success", nil, "completed"},
138		{"failure", errors.New("worker failed"), "failed"},
139		{"cancellation", context.Canceled, "interrupted"},
140		{"cancellation with failure", errors.Join(context.Canceled, errors.New("worker failed")), "interrupted"},
141	} {
142		t.Run(tc.name, func(t *testing.T) {
143			manager := subagents.NewResidentManager(t.TempDir(), func(subagents.ResidentChildSpec, *subagents.ResidentJournal) (subagents.ResidentTurnRunner, error) {
144				return func(context.Context, string) error { return tc.err }, nil
145			})
146			t.Cleanup(func() { _ = manager.Close(context.Background()) })
147			spawn := &SubagentSpawnTool{
148				ResidentManager: manager,
149				Enabled:         func() bool { return true },
150				DefaultProvider: func() string { return "openai" },
151				DefaultModel:    func() string { return "gpt-5" },
152				BuildResidentSpec: func(_ context.Context, request ResidentSpawnRequest) (subagents.ResidentChildSpec, error) {
153					return subagents.ResidentChildSpec{ID: "wait-for-completion", SessionID: "child-session", Provider: request.Provider, Model: request.Model, Required: request.Required}, nil
154				},
155			}
156			result, err := spawn.Execute(t.Context(), json.RawMessage(`{"task":"finish now","wait":1,"required":true}`), nil)
157			if err != nil || result.IsError {
158				t.Fatalf("Execute = (%#v, %v)", result, err)
159			}
160			if got := toolResultText(t, result); !strings.Contains(got, "\nstate: "+tc.status+"\n") || !strings.Contains(got, "\nrequired: "+tc.status+"\n") {
161				t.Fatalf("wait result = %q, want %s state", got, tc.status)
162			}
163			if unmet := len(manager.UnmetRequired()); (unmet != 0) != (tc.err != nil) {
164				t.Fatalf("unmet required = %d, error = %v", unmet, tc.err)
165			}
166		})
167	}
168}
169
170func TestResidentSpawnWaitExpiryReportsQueuedChild(t *testing.T) {
171	started := make(chan struct{}, 1)
172	release := make(chan struct{})
173	var releaseOnce sync.Once
174	releaseBlocker := func() { releaseOnce.Do(func() { close(release) }) }
175	finished := make(chan string, 1)
176	manager := subagents.NewResidentManagerWithLimit(t.TempDir(), 1, func(subagents.ResidentChildSpec, *subagents.ResidentJournal) (subagents.ResidentTurnRunner, error) {
177		return func(ctx context.Context, prompt string) error {
178			switch prompt {
179			case "blocker":
180				started <- struct{}{}
181				select {
182				case <-release:
183					return nil
184				case <-ctx.Done():
185					return ctx.Err()
186				}
187			case "waiter":
188				finished <- prompt
189			}
190			return nil
191		}, nil
192	})
193	t.Cleanup(func() {
194		releaseBlocker()
195		_ = manager.Close(context.Background())
196	})
197	spawn := &SubagentSpawnTool{
198		ResidentManager: manager,
199		Enabled:         func() bool { return true },
200		DefaultProvider: func() string { return "openai" },
201		DefaultModel:    func() string { return "gpt-5" },
202		BuildResidentSpec: func(_ context.Context, request ResidentSpawnRequest) (subagents.ResidentChildSpec, error) {
203			return subagents.ResidentChildSpec{ID: request.Task, SessionID: request.Task, Provider: request.Provider, Model: request.Model}, nil
204		},
205	}
206	if _, err := spawn.Execute(context.Background(), json.RawMessage(`{"task":"blocker"}`), nil); err != nil {
207		t.Fatalf("spawn blocker: %v", err)
208	}
209	select {
210	case <-started:
211	case <-time.After(5 * time.Second):
212		t.Fatal("blocker did not start")
213	}
214	result, err := spawn.Execute(context.Background(), json.RawMessage(`{"task":"waiter","wait":1}`), nil)
215	if err != nil || result.IsError {
216		t.Fatalf("spawn waiter = (%#v, %v)", result, err)
217	}
218	if got := toolResultText(t, result); !strings.Contains(got, "state: queued") || !strings.Contains(got, "wait: timed out after 1 seconds") {
219		t.Fatalf("wait expiry result = %q, want queued timeout", got)
220	}
221	if state, ok := manager.State("waiter"); !ok || state != subagents.ResidentQueued {
222		t.Fatalf("waiter state = %q (found=%t), want queued", state, ok)
223	}
224	releaseBlocker()
225	select {
226	case got := <-finished:
227		if got != "waiter" {
228			t.Fatalf("finished task = %q, want waiter", got)
229		}
230	case <-time.After(5 * time.Second):
231		t.Fatal("queued waiter did not execute")
232	}
233}
234
235func TestResidentSpawnRejectsInvalidInputBeforeCreatingChild(t *testing.T) {
236	manager := subagents.NewResidentManager(t.TempDir(), func(subagents.ResidentChildSpec, *subagents.ResidentJournal) (subagents.ResidentTurnRunner, error) {
237		return func(context.Context, string) error { return nil }, nil
238	})
239	t.Cleanup(func() { _ = manager.Close(context.Background()) })
240	built := 0
241	spawn := &SubagentSpawnTool{
242		ResidentManager: manager,
243		Enabled:         func() bool { return true },
244		BuildResidentSpec: func(context.Context, ResidentSpawnRequest) (subagents.ResidentChildSpec, error) {
245			built++
246			return subagents.ResidentChildSpec{ID: "should-not-exist", SessionID: "child"}, nil
247		},
248	}
249	for _, raw := range []string{
250		`{"task":"x","unexpected":true}`,
251		`{"task":"x"}{}`,
252		`{"task":"x","model":"gpt-5"}`,
253		`{"task":"x","wait":0}`,
254		`{"task":"x","wait":301}`,
255		`{"task":"x","isolation":"outside"}`,
256		`{"task":"x","budget_ratio":0.5}`,
257		`{"task":"x","budget_tokens":1000}`,
258	} {
259		result, err := spawn.Execute(context.Background(), json.RawMessage(raw), nil)
260		if err != nil && !strings.Contains(err.Error(), "invalid args") {
261			t.Fatalf("Execute(%s) error = %v", raw, err)
262		}
263		if err == nil && !result.IsError {
264			t.Fatalf("Execute(%s) = %#v, want protocol error", raw, result)
265		}
266	}
267	if built != 0 || len(manager.Snapshot()) != 0 {
268		t.Fatalf("invalid input reached resident manager: built=%d snapshots=%#v", built, manager.Snapshot())
269	}
270}
271
272func TestResidentSpawnProfileAndExplicitOverridesReachFactory(t *testing.T) {
273	manager := subagents.NewResidentManager(t.TempDir(), func(subagents.ResidentChildSpec, *subagents.ResidentJournal) (subagents.ResidentTurnRunner, error) {
274		return func(context.Context, string) error { return nil }, nil
275	})
276	t.Cleanup(func() { _ = manager.Close(context.Background()) })
277	profileFast := false
278	var got ResidentSpawnRequest
279	spawn := &SubagentSpawnTool{
280		ResidentManager:  manager,
281		Enabled:          func() bool { return true },
282		DefaultModel:     func() string { return "host-model" },
283		DefaultProvider:  func() string { return "host-provider" },
284		DefaultReasoning: func() string { return "medium" },
285		ResolveSubagent: func(name string) (*subagents.Profile, error) {
286			return &subagents.Profile{Name: name, Model: "openai/gpt-5.6-sol", Thinking: "low", FastMode: &profileFast}, nil
287		},
288		BuildResidentSpec: func(_ context.Context, request ResidentSpawnRequest) (subagents.ResidentChildSpec, error) {
289			got = request
290			return subagents.ResidentChildSpec{ID: "profiled", SessionID: "child", Provider: request.Provider, Model: request.Model, Profile: request.Profile.Name}, nil
291		},
292	}
293	result, err := spawn.Execute(context.Background(), json.RawMessage(`{"task":"review","agent":"reviewer","reasoning":"high","fast_mode":true}`), nil)
294	if err != nil || result.IsError {
295		t.Fatalf("Execute = (%#v, %v)", result, err)
296	}
297	if got.Model != "gpt-5.6-sol" || got.Provider != "openai" || got.Reasoning != "high" || got.FastMode == nil || !*got.FastMode {
298		t.Fatalf("factory request = %#v", got)
299	}
300}
301
302func TestResidentToolsRejectUnknownFields(t *testing.T) {
303	for _, tool := range []interface {
304		Execute(context.Context, json.RawMessage, func(string)) (core.ToolResult, error)
305	}{
306		&SubagentStatusTool{ResidentManager: subagents.NewResidentManager(t.TempDir(), nil), Enabled: func() bool { return true }},
307		&SubagentStopTool{ResidentManager: subagents.NewResidentManager(t.TempDir(), nil), Enabled: func() bool { return true }},
308		&SubagentResumeTool{ResidentManager: subagents.NewResidentManager(t.TempDir(), nil), Enabled: func() bool { return true }},
309	} {
310		_, err := tool.Execute(context.Background(), json.RawMessage(`{"unknown":true}`), nil)
311		if err == nil || !strings.Contains(err.Error(), "unknown field") {
312			t.Fatalf("unknown field error = %v", err)
313		}
314	}
315}
316
317func toolResultText(t *testing.T, result core.ToolResult) string {
318	t.Helper()
319	if len(result.Content) != 1 {
320		t.Fatalf("content = %#v, want one text block", result.Content)
321	}
322	block, ok := result.Content[0].(provider.TextBlock)
323	if !ok {
324		t.Fatalf("content = %#v, want text block", result.Content)
325	}
326	return block.Text
327}
328
329func TestPublicResidentStatusOmitsBudget(t *testing.T) {
330	entry := publicResidentStatus(subagents.ResidentSnapshot{
331		ID: "resident", State: subagents.ResidentIdle, Provider: "openai", Model: "test",
332	})
333	encoded, err := json.Marshal(entry)
334	if err != nil {
335		t.Fatal(err)
336	}
337	if strings.Contains(string(encoded), "budget") {
338		t.Fatalf("status entry carries budget = %s", encoded)
339	}
340}
341
342func TestFindResidentStatusSnapshotRejectsAmbiguousPrefix(t *testing.T) {
343	_, ok := findResidentStatusSnapshot([]subagents.ResidentSnapshot{{ID: "resident-abcd"}, {ID: "resident-abef"}}, "resident-ab")
344	if ok {
345		t.Fatal("ambiguous resident prefix resolved")
346	}
347}

Keyboard shortcuts

Single-key shortcuts are on. ctrl k and esc work whatever this says.

ctrl k :
go to a section, branch, tag, file or repository
?
this list
j k
move the row cursor down / up
enter
open the row under the cursor; on Files, a directory opens or closes in place
↑ ↓ ← → home end
in the file tree (focus on a row): previous / next row, close / open a directory or go to its parent / first entry, first / last row. They never open a file
l h
on Files: open a directory or a file / close a directory or go up to its parent, then to the parent page. Elsewhere: into the side panel (on a merge request, its files) / back to the content; there j k move between its controls
g g G
first / last row
/
filter the list; on Files, find a file anywhere in the repository (esc clears, enter opens the best match)
b
switch branch or tag
g s g f g c g b g m g i g r g a
tabs: summary, files, commits, branches & tags, merge requests, issues, releases, actions (when available)
g h
home
w
wide mode: use the whole width of the screen, or not (remembered)
y
copy a permalink pinned to the commit
d u x
diff: split / unified / collapse or expand all
e E
diff: the whole file in view / every file, with the lines the hunks hide
n
merge request: next note in the diff
esc
close a dialog, leave a field

Branches 1

Tags 477

↑↓ moveenter openesc close