detached a753677

fix(subagents): address trace review findings

commit
a753677bdd4c10a5ab010510d57ad263c76320e8
author
brice · · 2026-08-12 06:45 UTC
parent
895a89c
signature
GPG VerifiedCD88F9E564E0102646FB6077A1A07C7C7C3E349D · arch-gpg Verified for the committer email
14 files · +305 -112
Split
UnifiedSplit
M README.md +1 -1
@@ -419,7 +419,7 @@ Background subagents that run alongside your main session. Each one is a separat
419419
420420 **Session scoping** — each agent is stamped with the host session that spawned it and only shows up in that session's dashboard. Swap sessions with `/sessions` and the dashboard re-narrows accordingly. Agents from other sessions keep running in the background and reappear when you switch back.
421421
422-**Persistence across zut restarts** — every spawn writes durable metadata, a session file, and structured `result.json` under `$ZUT_HOME/subagents/agents/<id>/`. A local execution trace is an optional separate diagnostic bundle. On the next `zut` launch, retained session and result data can be used with `R` or `/subagents resume-session <id>` to continue the existing session. `/subagents restart-task <id>` is the explicit operation that intentionally replays the stored task.
422+**Persistence across zut restarts** — every spawn writes durable metadata and a session file under `$ZUT_HOME/subagents/agents/<id>/`; a completed or terminated worker writes structured `result.json`. A local execution trace is an optional separate diagnostic bundle. On the next `zut` launch, retained session and result data can be used with `R` or `/subagents resume-session <id>` to continue the existing session. `/subagents restart-task <id>` is the explicit operation that intentionally replays the stored task.
423423
424424 **Where state lives** — per-agent metadata, session files, results, and patches live under `$ZUT_HOME/subagents/agents/<id>/`; inbox sockets are runtime-only and permission-restricted. When enabled, trace bundles live below `ZUT_SUBAGENT_TRACE_DIR`; `zut debug trace <bundle>` reports the open operation, terminal fact, or last event. Shared-mode edits land directly in the repo. Worktree-mode edits are captured as durable patches and changed-file summaries before cleanup.
425425
@@ -419,7 +419,7 @@ Background subagents that run alongside your main session. Each one is a separat
419 419
420 **Session scoping** — each agent is stamped with the host session that spawned it and only shows up in that session's dashboard. Swap sessions with `/sessions` and the dashboard re-narrows accordingly. Agents from other sessions keep running in the background and reappear when you switch back.420 **Session scoping** — each agent is stamped with the host session that spawned it and only shows up in that session's dashboard. Swap sessions with `/sessions` and the dashboard re-narrows accordingly. Agents from other sessions keep running in the background and reappear when you switch back.
421 421
422-**Persistence across zut restarts** — every spawn writes durable metadata, a session file, and structured `result.json` under `$ZUT_HOME/subagents/agents/<id>/`. A local execution trace is an optional separate diagnostic bundle. On the next `zut` launch, retained session and result data can be used with `R` or `/subagents resume-session <id>` to continue the existing session. `/subagents restart-task <id>` is the explicit operation that intentionally replays the stored task.422+**Persistence across zut restarts** — every spawn writes durable metadata and a session file under `$ZUT_HOME/subagents/agents/<id>/`; a completed or terminated worker writes structured `result.json`. A local execution trace is an optional separate diagnostic bundle. On the next `zut` launch, retained session and result data can be used with `R` or `/subagents resume-session <id>` to continue the existing session. `/subagents restart-task <id>` is the explicit operation that intentionally replays the stored task.
423 423
424 **Where state lives** — per-agent metadata, session files, results, and patches live under `$ZUT_HOME/subagents/agents/<id>/`; inbox sockets are runtime-only and permission-restricted. When enabled, trace bundles live below `ZUT_SUBAGENT_TRACE_DIR`; `zut debug trace <bundle>` reports the open operation, terminal fact, or last event. Shared-mode edits land directly in the repo. Worktree-mode edits are captured as durable patches and changed-file summaries before cleanup.424 **Where state lives** — per-agent metadata, session files, results, and patches live under `$ZUT_HOME/subagents/agents/<id>/`; inbox sockets are runtime-only and permission-restricted. When enabled, trace bundles live below `ZUT_SUBAGENT_TRACE_DIR`; `zut debug trace <bundle>` reports the open operation, terminal fact, or last event. Shared-mode edits land directly in the repo. Worktree-mode edits are captured as durable patches and changed-file summaries before cleanup.
425 425
M docs/subagents.md +1 -1
@@ -164,7 +164,7 @@ When the relevant manager tool is permitted by launch-time policy, the model can
164164
165165 ## Local execution trace
166166
167-Set `ZUT_SUBAGENT_TRACE_DIR` to a private local directory before launching zut to retain an execution-trace bundle for each supervisor. Each bundle contains `manifest.json`, ordered `trace.jsonl`, and (only in detailed mode) private `payloads/` files. The default normal mode records redacted metadata. `ZUT_SUBAGENT_TRACE_MODE=detailed` retains unredacted prompts, tool arguments, and tool output in local trace files; enable it only for an appropriate investigation. Trace bundles are not uploaded or deleted automatically, and should be removed manually after the investigation.
167+Set `ZUT_SUBAGENT_TRACE_DIR` to a private local directory before launching zut to retain an execution-trace bundle for each supervisor. Each bundle contains `manifest.json`, ordered `trace.jsonl`, and (only in detailed mode) private `payloads/` files. The default normal mode records redacted metadata. `ZUT_SUBAGENT_TRACE_MODE=detailed` retains unredacted prompts, tool arguments, and tool output that can contain credentials, secrets, personal data, or private source code. Enable it only when the local bundle can be protected and manually deleted afterwards; do not commit, upload, or otherwise expose trace data. Bundles request restrictive local permissions, but access control remains platform-dependent.\n\nExplicit runtime `TraceDir` and `TraceMode` configuration takes precedence over these environment variables. Environment mode values are case-insensitive (`normal` or `detailed`); an invalid value is ignored with a stderr warning. Detailed mode requires a non-empty trace directory, otherwise tracing remains disabled.
168168
169169 Inspect a retained bundle with:
170170
@@ -164,7 +164,7 @@ When the relevant manager tool is permitted by launch-time policy, the model can
164 164
165 ## Local execution trace165 ## Local execution trace
166 166
167-Set `ZUT_SUBAGENT_TRACE_DIR` to a private local directory before launching zut to retain an execution-trace bundle for each supervisor. Each bundle contains `manifest.json`, ordered `trace.jsonl`, and (only in detailed mode) private `payloads/` files. The default normal mode records redacted metadata. `ZUT_SUBAGENT_TRACE_MODE=detailed` retains unredacted prompts, tool arguments, and tool output in local trace files; enable it only for an appropriate investigation. Trace bundles are not uploaded or deleted automatically, and should be removed manually after the investigation.167+Set `ZUT_SUBAGENT_TRACE_DIR` to a private local directory before launching zut to retain an execution-trace bundle for each supervisor. Each bundle contains `manifest.json`, ordered `trace.jsonl`, and (only in detailed mode) private `payloads/` files. The default normal mode records redacted metadata. `ZUT_SUBAGENT_TRACE_MODE=detailed` retains unredacted prompts, tool arguments, and tool output that can contain credentials, secrets, personal data, or private source code. Enable it only when the local bundle can be protected and manually deleted afterwards; do not commit, upload, or otherwise expose trace data. Bundles request restrictive local permissions, but access control remains platform-dependent.\n\nExplicit runtime `TraceDir` and `TraceMode` configuration takes precedence over these environment variables. Environment mode values are case-insensitive (`normal` or `detailed`); an invalid value is ignored with a stderr warning. Detailed mode requires a non-empty trace directory, otherwise tracing remains disabled.
168 168
169 Inspect a retained bundle with:169 Inspect a retained bundle with:
170 170
M packages/agent/modes/subagent_indicator.go +1 -1
@@ -40,7 +40,7 @@ func renderSubagentActivityLine(th tui.Theme, spinnerGlyph string, snapshot suba
4040 if spinnerGlyph == "" {
4141 spinnerGlyph = "."
4242 }
43- activity := strings.TrimSuffix(operation.Type, ".started") + " open"
43+ activity := operation.Label()
4444 age := formatSubagentActivityAge(operation.StartedAt, now)
4545 plain, layout := fitSubagentActivityLine(spinnerGlyph, name, activity, age, width-2)
4646 if plain == "" {
@@ -40,7 +40,7 @@ func renderSubagentActivityLine(th tui.Theme, spinnerGlyph string, snapshot suba
40 if spinnerGlyph == "" {40 if spinnerGlyph == "" {
41 spinnerGlyph = "."41 spinnerGlyph = "."
42 }42 }
43- activity := strings.TrimSuffix(operation.Type, ".started") + " open"43+ activity := operation.Label()
44 age := formatSubagentActivityAge(operation.StartedAt, now)44 age := formatSubagentActivityAge(operation.StartedAt, now)
45 plain, layout := fitSubagentActivityLine(spinnerGlyph, name, activity, age, width-2)45 plain, layout := fitSubagentActivityLine(spinnerGlyph, name, activity, age, width-2)
46 if plain == "" {46 if plain == "" {
M packages/agent/modes/subagents_dialog.go +11 -10
@@ -344,7 +344,7 @@ func (d *subagentsDialog) transcriptEditorCursorRow(width, popupRows, editorRowO
344344 }
345345
346346 row += popupRows // @-file-suggest popup, if active
347- if d.agentHasOpenOperation(a.ID) {
347+ if a.Status == subagents.StatusRunning && d.send != nil && d.agentHasOpenOperation(a.ID) {
348348 row += 2 // blank + spinner row inserted above the editor
349349 }
350350 row++ // blank above editor (appendTranscriptEditor)
@@ -1462,7 +1462,7 @@ func (d *subagentsDialog) traceSummary(id string) string {
14621462 return view.Terminal
14631463 }
14641464 if len(view.OpenOperations) != 0 {
1465- return strings.TrimSuffix(view.OpenOperations[0].Type, ".started") + " open"
1465+ return view.OpenOperations[0].Label()
14661466 }
14671467 if view.LastEvent.Type != "" {
14681468 return "last event " + view.LastEvent.Type
@@ -1618,13 +1618,7 @@ func (d *subagentsDialog) renderPromptEditor(th tui.Theme, width int, out []stri
16181618 return out
16191619 }
16201620
1621-// formatSupervisorRow is the one-line summary shown per agent.
1622-//
1623-// Layout (fixed-width columns, then free-form activity):
1624-//
1625-// OPERATION ID AGE TASK
1626-// tool open 3m fix-login-12345 3m inspect main.go
1627-// completed write-tests-67890 1h add coverage
1621+// traceView retrieves the latest shared factual trace projection for id.
16281622 func (d *subagentsDialog) traceView(id string) subagents.AgentTraceView {
16291623 if d != nil && d.traceViews != nil {
16301624 return d.traceViews()[id]
@@ -1632,13 +1626,20 @@ func (d *subagentsDialog) traceView(id string) subagents.AgentTraceView {
16321626 return subagents.AgentTraceView{AgentID: id}
16331627 }
16341628
1629+// formatSupervisorRow is the one-line summary shown per agent.
1630+//
1631+// Layout (fixed-width columns, then free-form task):
1632+//
1633+// OPERATION ID AGE TASK
1634+// tool open 3m fix-login-12345 3m inspect main.go
1635+// completed write-tests-67890 1h add coverage
16351636 func formatSupervisorRow(r subagents.AgentSnapshot, view subagents.AgentTraceView, maxWidth int) string {
16361637 fact := "no observable operation"
16371638 if view.Terminal != "" {
16381639 fact = view.Terminal
16391640 } else if len(view.OpenOperations) != 0 {
16401641 operation := view.OpenOperations[0]
1641- fact = strings.TrimSuffix(operation.Type, ".started") + " open " + formatAge(operation.StartedAt)
1642+ fact = operation.Label() + " " + formatAge(operation.StartedAt)
16421643 } else if view.LastEvent.Type != "" {
16431644 fact = "last " + view.LastEvent.Type + " " + formatAge(view.LastEvent.Timestamp)
16441645 }
@@ -344,7 +344,7 @@ func (d *subagentsDialog) transcriptEditorCursorRow(width, popupRows, editorRowO
344 }344 }
345 345
346 row += popupRows // @-file-suggest popup, if active346 row += popupRows // @-file-suggest popup, if active
347- if d.agentHasOpenOperation(a.ID) {347+ if a.Status == subagents.StatusRunning && d.send != nil && d.agentHasOpenOperation(a.ID) {
348 row += 2 // blank + spinner row inserted above the editor348 row += 2 // blank + spinner row inserted above the editor
349 }349 }
350 row++ // blank above editor (appendTranscriptEditor)350 row++ // blank above editor (appendTranscriptEditor)
@@ -1462,7 +1462,7 @@ func (d *subagentsDialog) traceSummary(id string) string {
1462 return view.Terminal1462 return view.Terminal
1463 }1463 }
1464 if len(view.OpenOperations) != 0 {1464 if len(view.OpenOperations) != 0 {
1465- return strings.TrimSuffix(view.OpenOperations[0].Type, ".started") + " open"1465+ return view.OpenOperations[0].Label()
1466 }1466 }
1467 if view.LastEvent.Type != "" {1467 if view.LastEvent.Type != "" {
1468 return "last event " + view.LastEvent.Type1468 return "last event " + view.LastEvent.Type
@@ -1618,13 +1618,7 @@ func (d *subagentsDialog) renderPromptEditor(th tui.Theme, width int, out []stri
1618 return out1618 return out
1619 }1619 }
1620 1620
1621-// formatSupervisorRow is the one-line summary shown per agent.1621+// traceView retrieves the latest shared factual trace projection for id.
1622-//
1623-// Layout (fixed-width columns, then free-form activity):
1624-//
1625-// OPERATION ID AGE TASK
1626-// tool open 3m fix-login-12345 3m inspect main.go
1627-// completed write-tests-67890 1h add coverage
1628 func (d *subagentsDialog) traceView(id string) subagents.AgentTraceView {1622 func (d *subagentsDialog) traceView(id string) subagents.AgentTraceView {
1629 if d != nil && d.traceViews != nil {1623 if d != nil && d.traceViews != nil {
1630 return d.traceViews()[id]1624 return d.traceViews()[id]
@@ -1632,13 +1626,20 @@ func (d *subagentsDialog) traceView(id string) subagents.AgentTraceView {
1632 return subagents.AgentTraceView{AgentID: id}1626 return subagents.AgentTraceView{AgentID: id}
1633 }1627 }
1634 1628
1629+// formatSupervisorRow is the one-line summary shown per agent.
1630+//
1631+// Layout (fixed-width columns, then free-form task):
1632+//
1633+// OPERATION ID AGE TASK
1634+// tool open 3m fix-login-12345 3m inspect main.go
1635+// completed write-tests-67890 1h add coverage
1635 func formatSupervisorRow(r subagents.AgentSnapshot, view subagents.AgentTraceView, maxWidth int) string {1636 func formatSupervisorRow(r subagents.AgentSnapshot, view subagents.AgentTraceView, maxWidth int) string {
1636 fact := "no observable operation"1637 fact := "no observable operation"
1637 if view.Terminal != "" {1638 if view.Terminal != "" {
1638 fact = view.Terminal1639 fact = view.Terminal
1639 } else if len(view.OpenOperations) != 0 {1640 } else if len(view.OpenOperations) != 0 {
1640 operation := view.OpenOperations[0]1641 operation := view.OpenOperations[0]
1641- fact = strings.TrimSuffix(operation.Type, ".started") + " open " + formatAge(operation.StartedAt)1642+ fact = operation.Label() + " " + formatAge(operation.StartedAt)
1642 } else if view.LastEvent.Type != "" {1643 } else if view.LastEvent.Type != "" {
1643 fact = "last " + view.LastEvent.Type + " " + formatAge(view.LastEvent.Timestamp)1644 fact = "last " + view.LastEvent.Type + " " + formatAge(view.LastEvent.Timestamp)
1644 }1645 }
M packages/agent/subagent_runtime.go +18 -6
@@ -2,6 +2,8 @@ package agent
22
33 import (
44 "context"
5+ "errors"
6+ "fmt"
57 "os"
68 "path/filepath"
79 "strings"
@@ -104,8 +106,19 @@ func newSubagentRuntime(cfg subagentRuntimeConfig) *subagentRuntime {
104106 if cfg.TraceDir == "" {
105107 cfg.TraceDir = strings.TrimSpace(os.Getenv("ZUT_SUBAGENT_TRACE_DIR"))
106108 }
107- if mode := strings.TrimSpace(os.Getenv("ZUT_SUBAGENT_TRACE_MODE")); mode == string(subagents.TraceModeDetailed) {
108- cfg.TraceMode = subagents.TraceModeDetailed
109+ if cfg.TraceMode == "" {
110+ switch strings.ToLower(strings.TrimSpace(os.Getenv("ZUT_SUBAGENT_TRACE_MODE"))) {
111+ case "", string(subagents.TraceModeNormal):
112+ cfg.TraceMode = subagents.TraceModeNormal
113+ case string(subagents.TraceModeDetailed):
114+ cfg.TraceMode = subagents.TraceModeDetailed
115+ default:
116+ fmt.Fprintf(os.Stderr, "zut: ignoring invalid ZUT_SUBAGENT_TRACE_MODE\n")
117+ cfg.TraceMode = subagents.TraceModeNormal
118+ }
119+ }
120+ if cfg.TraceDir == "" {
121+ cfg.TraceMode = subagents.TraceModeNormal
109122 }
110123 rt := &subagentRuntime{
111124 args: cfg.Args,
@@ -460,10 +473,9 @@ func (rt *subagentRuntime) Close(ctx context.Context) error {
460473 if rt.closed {
461474 return nil
462475 }
463- if err := rt.supervisor.StopAllContext(ctx); err != nil {
464- return err
465- }
466- if err := rt.supervisor.Close(); err != nil {
476+ stopErr := rt.supervisor.StopAllContext(ctx)
477+ closeErr := rt.supervisor.CloseContext(ctx)
478+ if err := errors.Join(stopErr, closeErr); err != nil {
467479 return err
468480 }
469481 rt.closed = true
@@ -2,6 +2,8 @@ package agent
2 2
3 import (3 import (
4 "context"4 "context"
5+ "errors"
6+ "fmt"
5 "os"7 "os"
6 "path/filepath"8 "path/filepath"
7 "strings"9 "strings"
@@ -104,8 +106,19 @@ func newSubagentRuntime(cfg subagentRuntimeConfig) *subagentRuntime {
104 if cfg.TraceDir == "" {106 if cfg.TraceDir == "" {
105 cfg.TraceDir = strings.TrimSpace(os.Getenv("ZUT_SUBAGENT_TRACE_DIR"))107 cfg.TraceDir = strings.TrimSpace(os.Getenv("ZUT_SUBAGENT_TRACE_DIR"))
106 }108 }
107- if mode := strings.TrimSpace(os.Getenv("ZUT_SUBAGENT_TRACE_MODE")); mode == string(subagents.TraceModeDetailed) {109+ if cfg.TraceMode == "" {
108- cfg.TraceMode = subagents.TraceModeDetailed110+ switch strings.ToLower(strings.TrimSpace(os.Getenv("ZUT_SUBAGENT_TRACE_MODE"))) {
111+ case "", string(subagents.TraceModeNormal):
112+ cfg.TraceMode = subagents.TraceModeNormal
113+ case string(subagents.TraceModeDetailed):
114+ cfg.TraceMode = subagents.TraceModeDetailed
115+ default:
116+ fmt.Fprintf(os.Stderr, "zut: ignoring invalid ZUT_SUBAGENT_TRACE_MODE\n")
117+ cfg.TraceMode = subagents.TraceModeNormal
118+ }
119+ }
120+ if cfg.TraceDir == "" {
121+ cfg.TraceMode = subagents.TraceModeNormal
109 }122 }
110 rt := &subagentRuntime{123 rt := &subagentRuntime{
111 args: cfg.Args,124 args: cfg.Args,
@@ -460,10 +473,9 @@ func (rt *subagentRuntime) Close(ctx context.Context) error {
460 if rt.closed {473 if rt.closed {
461 return nil474 return nil
462 }475 }
463- if err := rt.supervisor.StopAllContext(ctx); err != nil {476+ stopErr := rt.supervisor.StopAllContext(ctx)
464- return err477+ closeErr := rt.supervisor.CloseContext(ctx)
465- }478+ if err := errors.Join(stopErr, closeErr); err != nil {
466- if err := rt.supervisor.Close(); err != nil {
467 return err479 return err
468 }480 }
469 rt.closed = true481 rt.closed = true
M packages/agent/subagents/projection.go +28 -28
@@ -2,37 +2,36 @@ package subagents
22
33 import (
44 "sort"
5+ "strings"
56 "time"
67 )
78
8-// Operation is a factual operation inferred from paired trace boundaries. An
9-// empty FinishedAt means that no terminal boundary has been observed yet.
9+// Operation is a currently open factual operation inferred from its start
10+// boundary. Completed operations are deliberately not retained in the public
11+// projection: their terminal trace fact remains available as LastEvent.
1012 type Operation struct {
11- Type string
12- AgentID string
13- TurnID string
14- StartedAt time.Time
15- FinishedAt time.Time
16- Outcome string
13+ Type string
14+ AgentID string
15+ TurnID string
16+ CallID string
17+ StartedAt time.Time
1718 }
1819
19-// Open reports whether the trace contains a start boundary without a matching
20-// terminal boundary.
21-func (o Operation) Open() bool { return o.FinishedAt.IsZero() }
20+// Open is always true because the projection exposes only unmatched starts.
21+func (o Operation) Open() bool { return true }
2222
23-// Duration measures an observed operation. Open operations are measured at
24-// now; callers supply now to make rendering and tests deterministic.
23+// Duration measures this open operation at now; callers supply now to make
24+// rendering and tests deterministic.
2525 func (o Operation) Duration(now time.Time) time.Duration {
26- end := o.FinishedAt
27- if end.IsZero() {
28- end = now
29- }
30- if end.Before(o.StartedAt) {
26+ if now.Before(o.StartedAt) {
3127 return 0
3228 }
33- return end.Sub(o.StartedAt)
29+ return now.Sub(o.StartedAt)
3430 }
3531
32+// Label returns the shared concise label for an open operation.
33+func (o Operation) Label() string { return strings.TrimSuffix(o.Type, ".started") + " open" }
34+
3635 // AgentTraceView is the only user-facing execution state. It intentionally
3736 // contains operations and terminal facts rather than lifecycle guesses such as
3837 // working, idle, or alive.
@@ -60,17 +59,15 @@ func ProjectTrace(events []TraceEvent) map[string]AgentTraceView {
6059 if open[event.AgentID] == nil {
6160 open[event.AgentID] = make(map[string]Operation)
6261 }
63- open[event.AgentID][key] = Operation{Type: event.Type, AgentID: event.AgentID, TurnID: event.TurnID, StartedAt: event.Timestamp}
62+ callID, _ := event.Data["call_id"].(string)
63+ open[event.AgentID][key] = Operation{Type: event.Type, AgentID: event.AgentID, TurnID: event.TurnID, CallID: callID, StartedAt: event.Timestamp}
6464 }
6565 if ends {
66- if operation, ok := open[event.AgentID][key]; ok {
67- operation.FinishedAt = event.Timestamp
68- operation.Outcome = event.Type
69- delete(open[event.AgentID], key)
70- }
66+ delete(open[event.AgentID], key)
7167 }
7268 if terminal != "" {
7369 view.Terminal = terminal
70+ delete(open, event.AgentID)
7471 }
7572 views[event.AgentID] = view
7673 }
@@ -87,7 +84,10 @@ func ProjectTrace(events []TraceEvent) map[string]AgentTraceView {
8784 if left.Type != right.Type {
8885 return left.Type < right.Type
8986 }
90- return left.TurnID < right.TurnID
87+ if left.TurnID != right.TurnID {
88+ return left.TurnID < right.TurnID
89+ }
90+ return left.CallID < right.CallID
9191 })
9292 views[agentID] = view
9393 }
@@ -100,11 +100,11 @@ func traceBoundary(event TraceEvent) (key string, starts, ends bool, terminal st
100100 switch event.Type {
101101 case "turn.started":
102102 return "turn:" + event.TurnID, true, false, ""
103- case "turn.finished", "turn.failed":
103+ case "turn.finished", "turn.failed", "turn.cancelled":
104104 return "turn:" + event.TurnID, false, true, ""
105105 case "tool.started":
106106 return "tool:" + event.TurnID + ":" + callID, true, false, ""
107- case "tool.finished":
107+ case "tool.finished", "tool.failed", "tool.cancelled":
108108 return "tool:" + event.TurnID + ":" + callID, false, true, ""
109109 case "agent.finished":
110110 return key, false, false, "completed"
@@ -2,37 +2,36 @@ package subagents
2 2
3 import (3 import (
4 "sort"4 "sort"
5+ "strings"
5 "time"6 "time"
6 )7 )
7 8
8-// Operation is a factual operation inferred from paired trace boundaries. An9+// Operation is a currently open factual operation inferred from its start
9-// empty FinishedAt means that no terminal boundary has been observed yet.10+// boundary. Completed operations are deliberately not retained in the public
11+// projection: their terminal trace fact remains available as LastEvent.
10 type Operation struct {12 type Operation struct {
11- Type string13+ Type string
12- AgentID string14+ AgentID string
13- TurnID string15+ TurnID string
14- StartedAt time.Time16+ CallID string
15- FinishedAt time.Time17+ StartedAt time.Time
16- Outcome string
17 }18 }
18 19
19-// Open reports whether the trace contains a start boundary without a matching20+// Open is always true because the projection exposes only unmatched starts.
20-// terminal boundary.21+func (o Operation) Open() bool { return true }
21-func (o Operation) Open() bool { return o.FinishedAt.IsZero() }
22 22
23-// Duration measures an observed operation. Open operations are measured at23+// Duration measures this open operation at now; callers supply now to make
24-// now; callers supply now to make rendering and tests deterministic.24+// rendering and tests deterministic.
25 func (o Operation) Duration(now time.Time) time.Duration {25 func (o Operation) Duration(now time.Time) time.Duration {
26- end := o.FinishedAt26+ if now.Before(o.StartedAt) {
27- if end.IsZero() {
28- end = now
29- }
30- if end.Before(o.StartedAt) {
31 return 027 return 0
32 }28 }
33- return end.Sub(o.StartedAt)29+ return now.Sub(o.StartedAt)
34 }30 }
35 31
32+// Label returns the shared concise label for an open operation.
33+func (o Operation) Label() string { return strings.TrimSuffix(o.Type, ".started") + " open" }
34+
36 // AgentTraceView is the only user-facing execution state. It intentionally35 // AgentTraceView is the only user-facing execution state. It intentionally
37 // contains operations and terminal facts rather than lifecycle guesses such as36 // contains operations and terminal facts rather than lifecycle guesses such as
38 // working, idle, or alive.37 // working, idle, or alive.
@@ -60,17 +59,15 @@ func ProjectTrace(events []TraceEvent) map[string]AgentTraceView {
60 if open[event.AgentID] == nil {59 if open[event.AgentID] == nil {
61 open[event.AgentID] = make(map[string]Operation)60 open[event.AgentID] = make(map[string]Operation)
62 }61 }
63- open[event.AgentID][key] = Operation{Type: event.Type, AgentID: event.AgentID, TurnID: event.TurnID, StartedAt: event.Timestamp}62+ callID, _ := event.Data["call_id"].(string)
63+ open[event.AgentID][key] = Operation{Type: event.Type, AgentID: event.AgentID, TurnID: event.TurnID, CallID: callID, StartedAt: event.Timestamp}
64 }64 }
65 if ends {65 if ends {
66- if operation, ok := open[event.AgentID][key]; ok {66+ delete(open[event.AgentID], key)
67- operation.FinishedAt = event.Timestamp
68- operation.Outcome = event.Type
69- delete(open[event.AgentID], key)
70- }
71 }67 }
72 if terminal != "" {68 if terminal != "" {
73 view.Terminal = terminal69 view.Terminal = terminal
70+ delete(open, event.AgentID)
74 }71 }
75 views[event.AgentID] = view72 views[event.AgentID] = view
76 }73 }
@@ -87,7 +84,10 @@ func ProjectTrace(events []TraceEvent) map[string]AgentTraceView {
87 if left.Type != right.Type {84 if left.Type != right.Type {
88 return left.Type < right.Type85 return left.Type < right.Type
89 }86 }
90- return left.TurnID < right.TurnID87+ if left.TurnID != right.TurnID {
88+ return left.TurnID < right.TurnID
89+ }
90+ return left.CallID < right.CallID
91 })91 })
92 views[agentID] = view92 views[agentID] = view
93 }93 }
@@ -100,11 +100,11 @@ func traceBoundary(event TraceEvent) (key string, starts, ends bool, terminal st
100 switch event.Type {100 switch event.Type {
101 case "turn.started":101 case "turn.started":
102 return "turn:" + event.TurnID, true, false, ""102 return "turn:" + event.TurnID, true, false, ""
103- case "turn.finished", "turn.failed":103+ case "turn.finished", "turn.failed", "turn.cancelled":
104 return "turn:" + event.TurnID, false, true, ""104 return "turn:" + event.TurnID, false, true, ""
105 case "tool.started":105 case "tool.started":
106 return "tool:" + event.TurnID + ":" + callID, true, false, ""106 return "tool:" + event.TurnID + ":" + callID, true, false, ""
107- case "tool.finished":107+ case "tool.finished", "tool.failed", "tool.cancelled":
108 return "tool:" + event.TurnID + ":" + callID, false, true, ""108 return "tool:" + event.TurnID + ":" + callID, false, true, ""
109 case "agent.finished":109 case "agent.finished":
110 return key, false, false, "completed"110 return key, false, false, "completed"
M packages/agent/subagents/projection_test.go +24 -2
@@ -36,10 +36,32 @@ func TestProjectTraceClosesPairedOperationsAndKeepsTerminalFact(t *testing.T) {
3636 {Seq: 6, Timestamp: started.Add(5 * time.Second), Type: "agent.finished", AgentID: "agent-1"},
3737 })
3838 view := views["agent-1"]
39- if len(view.OpenOperations) != 1 || view.OpenOperations[0].Type != "tool.started" {
40- t.Fatalf("open operations = %#v, want unmatched tool", view.OpenOperations)
39+ if len(view.OpenOperations) != 0 {
40+ t.Fatalf("open operations = %#v, want none after terminal agent event", view.OpenOperations)
4141 }
4242 if view.Terminal != "completed" {
4343 t.Fatalf("terminal = %q, want completed", view.Terminal)
4444 }
4545 }
46+
47+func TestProjectTraceClosesFailedToolByCallID(t *testing.T) {
48+ started := time.Date(2026, 8, 12, 7, 0, 0, 0, time.UTC)
49+ view := ProjectTrace([]TraceEvent{
50+ {Timestamp: started, Type: "tool.started", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "one"}},
51+ {Timestamp: started, Type: "tool.started", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "two"}},
52+ {Timestamp: started.Add(time.Second), Type: "tool.failed", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "one"}},
53+ })["agent-1"]
54+ if len(view.OpenOperations) != 1 || view.OpenOperations[0].CallID != "two" {
55+ t.Fatalf("open operations = %#v, want only call two", view.OpenOperations)
56+ }
57+}
58+
59+func TestProjectTraceFailedAgentClosesOpenOperations(t *testing.T) {
60+ view := ProjectTrace([]TraceEvent{
61+ {Type: "tool.started", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "one"}},
62+ {Type: "agent.failed", AgentID: "agent-1"},
63+ })["agent-1"]
64+ if len(view.OpenOperations) != 0 || view.Terminal != "failed" {
65+ t.Fatalf("view = %#v, want failed terminal with no open operations", view)
66+ }
67+}
@@ -36,10 +36,32 @@ func TestProjectTraceClosesPairedOperationsAndKeepsTerminalFact(t *testing.T) {
36 {Seq: 6, Timestamp: started.Add(5 * time.Second), Type: "agent.finished", AgentID: "agent-1"},36 {Seq: 6, Timestamp: started.Add(5 * time.Second), Type: "agent.finished", AgentID: "agent-1"},
37 })37 })
38 view := views["agent-1"]38 view := views["agent-1"]
39- if len(view.OpenOperations) != 1 || view.OpenOperations[0].Type != "tool.started" {39+ if len(view.OpenOperations) != 0 {
40- t.Fatalf("open operations = %#v, want unmatched tool", view.OpenOperations)40+ t.Fatalf("open operations = %#v, want none after terminal agent event", view.OpenOperations)
41 }41 }
42 if view.Terminal != "completed" {42 if view.Terminal != "completed" {
43 t.Fatalf("terminal = %q, want completed", view.Terminal)43 t.Fatalf("terminal = %q, want completed", view.Terminal)
44 }44 }
45 }45 }
46+
47+func TestProjectTraceClosesFailedToolByCallID(t *testing.T) {
48+ started := time.Date(2026, 8, 12, 7, 0, 0, 0, time.UTC)
49+ view := ProjectTrace([]TraceEvent{
50+ {Timestamp: started, Type: "tool.started", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "one"}},
51+ {Timestamp: started, Type: "tool.started", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "two"}},
52+ {Timestamp: started.Add(time.Second), Type: "tool.failed", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "one"}},
53+ })["agent-1"]
54+ if len(view.OpenOperations) != 1 || view.OpenOperations[0].CallID != "two" {
55+ t.Fatalf("open operations = %#v, want only call two", view.OpenOperations)
56+ }
57+}
58+
59+func TestProjectTraceFailedAgentClosesOpenOperations(t *testing.T) {
60+ view := ProjectTrace([]TraceEvent{
61+ {Type: "tool.started", AgentID: "agent-1", TurnID: "turn-1", Data: map[string]any{"call_id": "one"}},
62+ {Type: "agent.failed", AgentID: "agent-1"},
63+ })["agent-1"]
64+ if len(view.OpenOperations) != 0 || view.Terminal != "failed" {
65+ t.Fatalf("view = %#v, want failed terminal with no open operations", view)
66+ }
67+}
M packages/agent/subagents/runner_e2e_test.go +1 -1
@@ -214,7 +214,7 @@ func TestRunnerEndToEndWithStubChild(t *testing.T) {
214214 // stdout is no longer duplicated into a per-agent durable event log.
215215 waitFor := func(want string) {
216216 t.Helper()
217- deadline := time.Now().Add(5 * time.Second)
217+ deadline := time.Now().Add(runnerProcessTimeout)
218218 for time.Now().Before(deadline) {
219219 if strings.Contains(strings.Join(a.Transcript(), "\n"), want) {
220220 return
@@ -214,7 +214,7 @@ func TestRunnerEndToEndWithStubChild(t *testing.T) {
214 // stdout is no longer duplicated into a per-agent durable event log.214 // stdout is no longer duplicated into a per-agent durable event log.
215 waitFor := func(want string) {215 waitFor := func(want string) {
216 t.Helper()216 t.Helper()
217- deadline := time.Now().Add(5 * time.Second)217+ deadline := time.Now().Add(runnerProcessTimeout)
218 for time.Now().Before(deadline) {218 for time.Now().Before(deadline) {
219 if strings.Contains(strings.Join(a.Transcript(), "\n"), want) {219 if strings.Contains(strings.Join(a.Transcript(), "\n"), want) {
220 return220 return
M packages/agent/subagents/supervisor.go +5 -2
@@ -648,11 +648,14 @@ func (f *Supervisor) TraceViews() map[string]AgentTraceView {
648648
649649 // Close flushes and closes the optional trace bundle. It does not stop agents;
650650 // their lifetime remains controlled by the supervisor context and Stop.
651-func (f *Supervisor) Close() error {
651+func (f *Supervisor) Close() error { return f.CloseContext(context.Background()) }
652+
653+// CloseContext flushes and closes the optional trace bundle within ctx.
654+func (f *Supervisor) CloseContext(ctx context.Context) error {
652655 if f == nil || f.trace == nil {
653656 return nil
654657 }
655- return f.trace.Close()
658+ return f.trace.CloseContext(ctx)
656659 }
657660
658661 func (f *Supervisor) CancelTurn(id string) error {
@@ -648,11 +648,14 @@ func (f *Supervisor) TraceViews() map[string]AgentTraceView {
648 648
649 // Close flushes and closes the optional trace bundle. It does not stop agents;649 // Close flushes and closes the optional trace bundle. It does not stop agents;
650 // their lifetime remains controlled by the supervisor context and Stop.650 // their lifetime remains controlled by the supervisor context and Stop.
651-func (f *Supervisor) Close() error {651+func (f *Supervisor) Close() error { return f.CloseContext(context.Background()) }
652+
653+// CloseContext flushes and closes the optional trace bundle within ctx.
654+func (f *Supervisor) CloseContext(ctx context.Context) error {
652 if f == nil || f.trace == nil {655 if f == nil || f.trace == nil {
653 return nil656 return nil
654 }657 }
655- return f.trace.Close()658+ return f.trace.CloseContext(ctx)
656 }659 }
657 660
658 func (f *Supervisor) CancelTurn(id string) error {661 func (f *Supervisor) CancelTurn(id string) error {
M packages/agent/subagents/trace.go +112 -37
@@ -7,6 +7,7 @@ import (
77 "fmt"
88 "os"
99 "path/filepath"
10+ "reflect"
1011 "strings"
1112 "sync"
1213 "time"
@@ -89,6 +90,7 @@ type TraceWriter struct {
8990 payloadDir string
9091 memory bool
9192 sequence uint64
93+ dropped uint64
9294 events []TraceEvent
9395
9496 errorMu sync.Mutex
@@ -103,6 +105,7 @@ type traceRecord struct {
103105
104106 const (
105107 traceQueueInitialCapacity = 64
108+ traceQueueLimit = 1024
106109 memoryTraceEventLimit = 4096
107110 )
108111
@@ -144,6 +147,11 @@ func NewTraceWriter(dir string, modes ...TraceMode) (*TraceWriter, error) {
144147 if err := os.Chmod(absolute, 0o700); err != nil {
145148 return nil, fmt.Errorf("trace writer directory permissions: %w", err)
146149 }
150+ if _, err := os.Stat(filepath.Join(absolute, "trace.jsonl")); err == nil {
151+ return nil, fmt.Errorf("trace writer: bundle already exists")
152+ } else if !errors.Is(err, os.ErrNotExist) {
153+ return nil, fmt.Errorf("trace writer trace file: %w", err)
154+ }
147155 payloadDir := filepath.Join(absolute, "payloads")
148156 if err := os.MkdirAll(payloadDir, 0o700); err != nil {
149157 return nil, fmt.Errorf("trace writer payload directory: %w", err)
@@ -244,6 +252,18 @@ func (w *TraceWriter) Record(event TraceEvent) {
244252 }
245253 w.sequence++
246254 record.event.Seq = w.sequence
255+ if len(w.pending) >= traceQueueLimit {
256+ for index, pending := range w.pending {
257+ if pending.barrier == nil {
258+ copy(w.pending[index:], w.pending[index+1:])
259+ w.pending[len(w.pending)-1] = traceRecord{}
260+ w.pending = w.pending[:len(w.pending)-1]
261+ w.dropped++
262+ w.setError(errors.New("trace event queue overflow"))
263+ break
264+ }
265+ }
266+ }
247267 w.pending = append(w.pending, record)
248268 w.cond.Signal()
249269 }
@@ -303,11 +323,7 @@ func (w *TraceWriter) write(record traceRecord) {
303323 event := record.event
304324 if w.memory {
305325 w.mu.Lock()
306- w.events = append(w.events, event)
307- if overflow := len(w.events) - memoryTraceEventLimit; overflow > 0 {
308- copy(w.events, w.events[overflow:])
309- w.events = w.events[:memoryTraceEventLimit]
310- }
326+ w.appendEvent(event)
311327 w.mu.Unlock()
312328 return
313329 }
@@ -347,17 +363,24 @@ func (w *TraceWriter) write(record traceRecord) {
347363 w.setError(fmt.Errorf("trace event sync: %w", err))
348364 }
349365 w.mu.Lock()
350- w.events = append(w.events, event)
366+ w.appendEvent(event)
351367 w.mu.Unlock()
352368 }
353369
354-// Events returns a stable copy of observations durably written by this writer.
355-// It supports the in-process projection used by status and the dashboard.
370+func (w *TraceWriter) appendEvent(event TraceEvent) {
371+ w.events = append(w.events, event)
372+ if overflow := len(w.events) - memoryTraceEventLimit; overflow > 0 {
373+ copy(w.events, w.events[overflow:])
374+ w.events = w.events[:memoryTraceEventLimit]
375+ }
376+}
377+
378+// Events returns a stable copy of events already written by this writer. It
379+// never waits for queued filesystem work, keeping status and UI reads cheap.
356380 func (w *TraceWriter) Events() []TraceEvent {
357381 if w == nil {
358382 return nil
359383 }
360- _ = w.Flush()
361384 w.mu.Lock()
362385 defer w.mu.Unlock()
363386 return append([]TraceEvent(nil), w.events...)
@@ -375,48 +398,66 @@ func (w *TraceWriter) setError(err error) {
375398 }
376399
377400 // Flush waits until all events recorded before the call have been written.
378-// It is useful to readers that need a live, consistent view; agent recording
379-// itself remains asynchronous.
380-func (w *TraceWriter) Flush() error {
401+func (w *TraceWriter) Flush() error { return w.FlushContext(context.Background()) }
402+
403+// FlushContext honors cancellation while waiting for the trace writer.
404+func (w *TraceWriter) FlushContext(ctx context.Context) error {
381405 if w == nil {
382406 return nil
383407 }
408+ if ctx == nil {
409+ ctx = context.Background()
410+ }
384411 w.mu.Lock()
412+ wait := w.finished
385413 if !w.closed {
386414 barrier := make(chan struct{})
387415 w.pending = append(w.pending, traceRecord{barrier: barrier})
388416 w.cond.Signal()
389- w.mu.Unlock()
390- <-barrier
391- } else {
392- w.mu.Unlock()
393- <-w.finished
417+ wait = barrier
418+ }
419+ w.mu.Unlock()
420+ select {
421+ case <-wait:
422+ return w.recordingError()
423+ case <-ctx.Done():
424+ return fmt.Errorf("flush trace: %w", ctx.Err())
394425 }
395- return w.recordingError()
396426 }
397427
398-// Close drains the queue, syncs each event, and closes the bundle. It is safe
399-// to call more than once. Close is the only operation that reports persistence
400-// errors; Record remains best effort.
401-func (w *TraceWriter) Close() error {
428+// Close drains the queue, syncs each event, and closes the bundle.
429+func (w *TraceWriter) Close() error { return w.CloseContext(context.Background()) }
430+
431+// CloseContext starts closing once and honors cancellation while waiting. A
432+// later caller can retry the wait without repeating file-close work.
433+func (w *TraceWriter) CloseContext(ctx context.Context) error {
402434 if w == nil {
403435 return nil
404436 }
437+ if ctx == nil {
438+ ctx = context.Background()
439+ }
405440 w.closeOnce.Do(func() {
406- w.mu.Lock()
407- w.closed = true
408- w.cond.Broadcast()
409- w.mu.Unlock()
410- <-w.finished
411- if !w.memory {
412- if err := w.traceFile.Close(); err != nil {
413- w.setError(fmt.Errorf("trace file close: %w", err))
441+ go func() {
442+ w.mu.Lock()
443+ w.closed = true
444+ w.cond.Broadcast()
445+ w.mu.Unlock()
446+ <-w.finished
447+ if !w.memory {
448+ if err := w.traceFile.Close(); err != nil {
449+ w.setError(fmt.Errorf("trace file close: %w", err))
450+ }
414451 }
415- }
416- close(w.closeDone)
452+ close(w.closeDone)
453+ }()
417454 })
418- <-w.closeDone
419- return w.recordingError()
455+ select {
456+ case <-w.closeDone:
457+ return w.recordingError()
458+ case <-ctx.Done():
459+ return fmt.Errorf("close trace: %w", ctx.Err())
460+ }
420461 }
421462
422463 func (w *TraceWriter) recordingError() error {
@@ -502,7 +543,7 @@ func redactTraceMap(data map[string]any) map[string]any {
502543 }
503544 out := make(map[string]any, len(data))
504545 for key, value := range data {
505- if _, sensitive := sensitiveTraceKeys[normalizeTraceKey(key)]; sensitive {
546+ if sensitiveTraceKey(normalizeTraceKey(key)) {
506547 out[key] = "[REDACTED]"
507548 continue
508549 }
@@ -511,7 +552,19 @@ func redactTraceMap(data map[string]any) map[string]any {
511552 return out
512553 }
513554
555+func sensitiveTraceKey(key string) bool {
556+ for fragment := range sensitiveTraceKeys {
557+ if strings.Contains(key, fragment) {
558+ return true
559+ }
560+ }
561+ return false
562+}
563+
514564 func redactTraceValue(value any) any {
565+ if value == nil {
566+ return nil
567+ }
515568 switch value := value.(type) {
516569 case map[string]any:
517570 return redactTraceMap(value)
@@ -521,9 +574,31 @@ func redactTraceValue(value any) any {
521574 out[i] = redactTraceValue(value[i])
522575 }
523576 return out
524- default:
525- return value
526577 }
578+ v := reflect.ValueOf(value)
579+ if v.Kind() == reflect.Map && v.Type().Key().Kind() == reflect.String {
580+ out := make(map[string]any, v.Len())
581+ iter := v.MapRange()
582+ for iter.Next() {
583+ key := iter.Key().String()
584+ if sensitiveTraceKey(normalizeTraceKey(key)) {
585+ out[key] = "[REDACTED]"
586+ } else {
587+ out[key] = redactTraceValue(iter.Value().Interface())
588+ }
589+ }
590+ return out
591+ }
592+ if v.Kind() == reflect.Struct {
593+ encoded, err := json.Marshal(value)
594+ if err == nil {
595+ var data map[string]any
596+ if json.Unmarshal(encoded, &data) == nil {
597+ return redactTraceMap(data)
598+ }
599+ }
600+ }
601+ return value
527602 }
528603
529604 func cloneTraceMap(data map[string]any) map[string]any {
@@ -7,6 +7,7 @@ import (
7 "fmt"7 "fmt"
8 "os"8 "os"
9 "path/filepath"9 "path/filepath"
10+ "reflect"
10 "strings"11 "strings"
11 "sync"12 "sync"
12 "time"13 "time"
@@ -89,6 +90,7 @@ type TraceWriter struct {
89 payloadDir string90 payloadDir string
90 memory bool91 memory bool
91 sequence uint6492 sequence uint64
93+ dropped uint64
92 events []TraceEvent94 events []TraceEvent
93 95
94 errorMu sync.Mutex96 errorMu sync.Mutex
@@ -103,6 +105,7 @@ type traceRecord struct {
103 105
104 const (106 const (
105 traceQueueInitialCapacity = 64107 traceQueueInitialCapacity = 64
108+ traceQueueLimit = 1024
106 memoryTraceEventLimit = 4096109 memoryTraceEventLimit = 4096
107 )110 )
108 111
@@ -144,6 +147,11 @@ func NewTraceWriter(dir string, modes ...TraceMode) (*TraceWriter, error) {
144 if err := os.Chmod(absolute, 0o700); err != nil {147 if err := os.Chmod(absolute, 0o700); err != nil {
145 return nil, fmt.Errorf("trace writer directory permissions: %w", err)148 return nil, fmt.Errorf("trace writer directory permissions: %w", err)
146 }149 }
150+ if _, err := os.Stat(filepath.Join(absolute, "trace.jsonl")); err == nil {
151+ return nil, fmt.Errorf("trace writer: bundle already exists")
152+ } else if !errors.Is(err, os.ErrNotExist) {
153+ return nil, fmt.Errorf("trace writer trace file: %w", err)
154+ }
147 payloadDir := filepath.Join(absolute, "payloads")155 payloadDir := filepath.Join(absolute, "payloads")
148 if err := os.MkdirAll(payloadDir, 0o700); err != nil {156 if err := os.MkdirAll(payloadDir, 0o700); err != nil {
149 return nil, fmt.Errorf("trace writer payload directory: %w", err)157 return nil, fmt.Errorf("trace writer payload directory: %w", err)
@@ -244,6 +252,18 @@ func (w *TraceWriter) Record(event TraceEvent) {
244 }252 }
245 w.sequence++253 w.sequence++
246 record.event.Seq = w.sequence254 record.event.Seq = w.sequence
255+ if len(w.pending) >= traceQueueLimit {
256+ for index, pending := range w.pending {
257+ if pending.barrier == nil {
258+ copy(w.pending[index:], w.pending[index+1:])
259+ w.pending[len(w.pending)-1] = traceRecord{}
260+ w.pending = w.pending[:len(w.pending)-1]
261+ w.dropped++
262+ w.setError(errors.New("trace event queue overflow"))
263+ break
264+ }
265+ }
266+ }
247 w.pending = append(w.pending, record)267 w.pending = append(w.pending, record)
248 w.cond.Signal()268 w.cond.Signal()
249 }269 }
@@ -303,11 +323,7 @@ func (w *TraceWriter) write(record traceRecord) {
303 event := record.event323 event := record.event
304 if w.memory {324 if w.memory {
305 w.mu.Lock()325 w.mu.Lock()
306- w.events = append(w.events, event)326+ w.appendEvent(event)
307- if overflow := len(w.events) - memoryTraceEventLimit; overflow > 0 {
308- copy(w.events, w.events[overflow:])
309- w.events = w.events[:memoryTraceEventLimit]
310- }
311 w.mu.Unlock()327 w.mu.Unlock()
312 return328 return
313 }329 }
@@ -347,17 +363,24 @@ func (w *TraceWriter) write(record traceRecord) {
347 w.setError(fmt.Errorf("trace event sync: %w", err))363 w.setError(fmt.Errorf("trace event sync: %w", err))
348 }364 }
349 w.mu.Lock()365 w.mu.Lock()
350- w.events = append(w.events, event)366+ w.appendEvent(event)
351 w.mu.Unlock()367 w.mu.Unlock()
352 }368 }
353 369
354-// Events returns a stable copy of observations durably written by this writer.370+func (w *TraceWriter) appendEvent(event TraceEvent) {
355-// It supports the in-process projection used by status and the dashboard.371+ w.events = append(w.events, event)
372+ if overflow := len(w.events) - memoryTraceEventLimit; overflow > 0 {
373+ copy(w.events, w.events[overflow:])
374+ w.events = w.events[:memoryTraceEventLimit]
375+ }
376+}
377+
378+// Events returns a stable copy of events already written by this writer. It
379+// never waits for queued filesystem work, keeping status and UI reads cheap.
356 func (w *TraceWriter) Events() []TraceEvent {380 func (w *TraceWriter) Events() []TraceEvent {
357 if w == nil {381 if w == nil {
358 return nil382 return nil
359 }383 }
360- _ = w.Flush()
361 w.mu.Lock()384 w.mu.Lock()
362 defer w.mu.Unlock()385 defer w.mu.Unlock()
363 return append([]TraceEvent(nil), w.events...)386 return append([]TraceEvent(nil), w.events...)
@@ -375,48 +398,66 @@ func (w *TraceWriter) setError(err error) {
375 }398 }
376 399
377 // Flush waits until all events recorded before the call have been written.400 // Flush waits until all events recorded before the call have been written.
378-// It is useful to readers that need a live, consistent view; agent recording401+func (w *TraceWriter) Flush() error { return w.FlushContext(context.Background()) }
379-// itself remains asynchronous.402+
380-func (w *TraceWriter) Flush() error {403+// FlushContext honors cancellation while waiting for the trace writer.
404+func (w *TraceWriter) FlushContext(ctx context.Context) error {
381 if w == nil {405 if w == nil {
382 return nil406 return nil
383 }407 }
408+ if ctx == nil {
409+ ctx = context.Background()
410+ }
384 w.mu.Lock()411 w.mu.Lock()
412+ wait := w.finished
385 if !w.closed {413 if !w.closed {
386 barrier := make(chan struct{})414 barrier := make(chan struct{})
387 w.pending = append(w.pending, traceRecord{barrier: barrier})415 w.pending = append(w.pending, traceRecord{barrier: barrier})
388 w.cond.Signal()416 w.cond.Signal()
389- w.mu.Unlock()417+ wait = barrier
390- <-barrier418+ }
391- } else {419+ w.mu.Unlock()
392- w.mu.Unlock()420+ select {
393- <-w.finished421+ case <-wait:
422+ return w.recordingError()
423+ case <-ctx.Done():
424+ return fmt.Errorf("flush trace: %w", ctx.Err())
394 }425 }
395- return w.recordingError()
396 }426 }
397 427
398-// Close drains the queue, syncs each event, and closes the bundle. It is safe428+// Close drains the queue, syncs each event, and closes the bundle.
399-// to call more than once. Close is the only operation that reports persistence429+func (w *TraceWriter) Close() error { return w.CloseContext(context.Background()) }
400-// errors; Record remains best effort.430+
401-func (w *TraceWriter) Close() error {431+// CloseContext starts closing once and honors cancellation while waiting. A
432+// later caller can retry the wait without repeating file-close work.
433+func (w *TraceWriter) CloseContext(ctx context.Context) error {
402 if w == nil {434 if w == nil {
403 return nil435 return nil
404 }436 }
437+ if ctx == nil {
438+ ctx = context.Background()
439+ }
405 w.closeOnce.Do(func() {440 w.closeOnce.Do(func() {
406- w.mu.Lock()441+ go func() {
407- w.closed = true442+ w.mu.Lock()
408- w.cond.Broadcast()443+ w.closed = true
409- w.mu.Unlock()444+ w.cond.Broadcast()
410- <-w.finished445+ w.mu.Unlock()
411- if !w.memory {446+ <-w.finished
412- if err := w.traceFile.Close(); err != nil {447+ if !w.memory {
413- w.setError(fmt.Errorf("trace file close: %w", err))448+ if err := w.traceFile.Close(); err != nil {
449+ w.setError(fmt.Errorf("trace file close: %w", err))
450+ }
414 }451 }
415- }452+ close(w.closeDone)
416- close(w.closeDone)453+ }()
417 })454 })
418- <-w.closeDone455+ select {
419- return w.recordingError()456+ case <-w.closeDone:
457+ return w.recordingError()
458+ case <-ctx.Done():
459+ return fmt.Errorf("close trace: %w", ctx.Err())
460+ }
420 }461 }
421 462
422 func (w *TraceWriter) recordingError() error {463 func (w *TraceWriter) recordingError() error {
@@ -502,7 +543,7 @@ func redactTraceMap(data map[string]any) map[string]any {
502 }543 }
503 out := make(map[string]any, len(data))544 out := make(map[string]any, len(data))
504 for key, value := range data {545 for key, value := range data {
505- if _, sensitive := sensitiveTraceKeys[normalizeTraceKey(key)]; sensitive {546+ if sensitiveTraceKey(normalizeTraceKey(key)) {
506 out[key] = "[REDACTED]"547 out[key] = "[REDACTED]"
507 continue548 continue
508 }549 }
@@ -511,7 +552,19 @@ func redactTraceMap(data map[string]any) map[string]any {
511 return out552 return out
512 }553 }
513 554
555+func sensitiveTraceKey(key string) bool {
556+ for fragment := range sensitiveTraceKeys {
557+ if strings.Contains(key, fragment) {
558+ return true
559+ }
560+ }
561+ return false
562+}
563+
514 func redactTraceValue(value any) any {564 func redactTraceValue(value any) any {
565+ if value == nil {
566+ return nil
567+ }
515 switch value := value.(type) {568 switch value := value.(type) {
516 case map[string]any:569 case map[string]any:
517 return redactTraceMap(value)570 return redactTraceMap(value)
@@ -521,9 +574,31 @@ func redactTraceValue(value any) any {
521 out[i] = redactTraceValue(value[i])574 out[i] = redactTraceValue(value[i])
522 }575 }
523 return out576 return out
524- default:
525- return value
526 }577 }
578+ v := reflect.ValueOf(value)
579+ if v.Kind() == reflect.Map && v.Type().Key().Kind() == reflect.String {
580+ out := make(map[string]any, v.Len())
581+ iter := v.MapRange()
582+ for iter.Next() {
583+ key := iter.Key().String()
584+ if sensitiveTraceKey(normalizeTraceKey(key)) {
585+ out[key] = "[REDACTED]"
586+ } else {
587+ out[key] = redactTraceValue(iter.Value().Interface())
588+ }
589+ }
590+ return out
591+ }
592+ if v.Kind() == reflect.Struct {
593+ encoded, err := json.Marshal(value)
594+ if err == nil {
595+ var data map[string]any
596+ if json.Unmarshal(encoded, &data) == nil {
597+ return redactTraceMap(data)
598+ }
599+ }
600+ }
601+ return value
527 }602 }
528 603
529 func cloneTraceMap(data map[string]any) map[string]any {604 func cloneTraceMap(data map[string]any) map[string]any {
M packages/agent/subagents/trace_read.go +59 -4
@@ -2,32 +2,87 @@ package subagents
22
33 import (
44 "bufio"
5+ "context"
56 "encoding/json"
7+ "errors"
68 "fmt"
79 "os"
810 "path/filepath"
11+ "strings"
912 )
1013
11-// ReadTrace reads the stable JSONL event stream from a trace bundle.
14+// ReadTrace reads a trace bundle using a background context.
1215 func ReadTrace(bundle string) ([]TraceEvent, error) {
13- file, err := os.Open(filepath.Join(bundle, "trace.jsonl"))
16+ return ReadTraceContext(context.Background(), bundle)
17+}
18+
19+// ReadTraceContext reads the manifest-selected JSONL trace stream. If a stream
20+// ends with malformed data, events decoded before it are returned with the
21+// error so callers can still diagnose the valid prefix.
22+func ReadTraceContext(ctx context.Context, bundle string) ([]TraceEvent, error) {
23+ if ctx == nil {
24+ ctx = context.Background()
25+ }
26+ manifestPath := filepath.Join(bundle, "manifest.json")
27+ data, err := os.ReadFile(manifestPath)
28+ if err != nil {
29+ return nil, fmt.Errorf("read trace manifest: %w", err)
30+ }
31+ var manifest TraceManifest
32+ if err := json.Unmarshal(data, &manifest); err != nil {
33+ return nil, fmt.Errorf("decode trace manifest: %w", err)
34+ }
35+ if manifest.Version != 1 {
36+ return nil, fmt.Errorf("unsupported trace manifest version %d", manifest.Version)
37+ }
38+ tracePath, err := traceBundlePath(bundle, manifest.TraceFile)
39+ if err != nil {
40+ return nil, err
41+ }
42+ file, err := os.Open(tracePath)
1443 if err != nil {
1544 return nil, fmt.Errorf("open trace: %w", err)
1645 }
1746 defer file.Close()
47+
1848 var events []TraceEvent
1949 scanner := bufio.NewScanner(file)
2050 buffer := make([]byte, 64*1024)
2151 scanner.Buffer(buffer, 4*1024*1024)
2252 for scanner.Scan() {
53+ if err := ctx.Err(); err != nil {
54+ return events, fmt.Errorf("read trace: %w", err)
55+ }
2356 var event TraceEvent
2457 if err := json.Unmarshal(scanner.Bytes(), &event); err != nil {
25- return nil, fmt.Errorf("decode trace event: %w", err)
58+ return events, fmt.Errorf("decode trace event: %w", err)
2659 }
2760 events = append(events, event)
2861 }
2962 if err := scanner.Err(); err != nil {
30- return nil, fmt.Errorf("read trace: %w", err)
63+ return events, fmt.Errorf("read trace: %w", err)
64+ }
65+ if err := ctx.Err(); err != nil {
66+ return events, fmt.Errorf("read trace: %w", err)
3167 }
3268 return events, nil
3369 }
70+
71+func traceBundlePath(bundle, name string) (string, error) {
72+ if strings.TrimSpace(name) == "" || filepath.IsAbs(name) {
73+ return "", errors.New("invalid trace manifest path")
74+ }
75+ root, err := filepath.Abs(bundle)
76+ if err != nil {
77+ return "", fmt.Errorf("resolve trace bundle: %w", err)
78+ }
79+ path, err := filepath.Abs(filepath.Join(root, name))
80+ if err != nil {
81+ return "", fmt.Errorf("resolve trace path: %w", err)
82+ }
83+ rel, err := filepath.Rel(root, path)
84+ if err != nil || rel == ".." || strings.HasPrefix(rel, ".."+string(filepath.Separator)) {
85+ return "", errors.New("trace manifest path escapes bundle")
86+ }
87+ return path, nil
88+}
@@ -2,32 +2,87 @@ package subagents
2 2
3 import (3 import (
4 "bufio"4 "bufio"
5+ "context"
5 "encoding/json"6 "encoding/json"
7+ "errors"
6 "fmt"8 "fmt"
7 "os"9 "os"
8 "path/filepath"10 "path/filepath"
11+ "strings"
9 )12 )
10 13
11-// ReadTrace reads the stable JSONL event stream from a trace bundle.14+// ReadTrace reads a trace bundle using a background context.
12 func ReadTrace(bundle string) ([]TraceEvent, error) {15 func ReadTrace(bundle string) ([]TraceEvent, error) {
13- file, err := os.Open(filepath.Join(bundle, "trace.jsonl"))16+ return ReadTraceContext(context.Background(), bundle)
17+}
18+
19+// ReadTraceContext reads the manifest-selected JSONL trace stream. If a stream
20+// ends with malformed data, events decoded before it are returned with the
21+// error so callers can still diagnose the valid prefix.
22+func ReadTraceContext(ctx context.Context, bundle string) ([]TraceEvent, error) {
23+ if ctx == nil {
24+ ctx = context.Background()
25+ }
26+ manifestPath := filepath.Join(bundle, "manifest.json")
27+ data, err := os.ReadFile(manifestPath)
28+ if err != nil {
29+ return nil, fmt.Errorf("read trace manifest: %w", err)
30+ }
31+ var manifest TraceManifest
32+ if err := json.Unmarshal(data, &manifest); err != nil {
33+ return nil, fmt.Errorf("decode trace manifest: %w", err)
34+ }
35+ if manifest.Version != 1 {
36+ return nil, fmt.Errorf("unsupported trace manifest version %d", manifest.Version)
37+ }
38+ tracePath, err := traceBundlePath(bundle, manifest.TraceFile)
39+ if err != nil {
40+ return nil, err
41+ }
42+ file, err := os.Open(tracePath)
14 if err != nil {43 if err != nil {
15 return nil, fmt.Errorf("open trace: %w", err)44 return nil, fmt.Errorf("open trace: %w", err)
16 }45 }
17 defer file.Close()46 defer file.Close()
47+
18 var events []TraceEvent48 var events []TraceEvent
19 scanner := bufio.NewScanner(file)49 scanner := bufio.NewScanner(file)
20 buffer := make([]byte, 64*1024)50 buffer := make([]byte, 64*1024)
21 scanner.Buffer(buffer, 4*1024*1024)51 scanner.Buffer(buffer, 4*1024*1024)
22 for scanner.Scan() {52 for scanner.Scan() {
53+ if err := ctx.Err(); err != nil {
54+ return events, fmt.Errorf("read trace: %w", err)
55+ }
23 var event TraceEvent56 var event TraceEvent
24 if err := json.Unmarshal(scanner.Bytes(), &event); err != nil {57 if err := json.Unmarshal(scanner.Bytes(), &event); err != nil {
25- return nil, fmt.Errorf("decode trace event: %w", err)58+ return events, fmt.Errorf("decode trace event: %w", err)
26 }59 }
27 events = append(events, event)60 events = append(events, event)
28 }61 }
29 if err := scanner.Err(); err != nil {62 if err := scanner.Err(); err != nil {
30- return nil, fmt.Errorf("read trace: %w", err)63+ return events, fmt.Errorf("read trace: %w", err)
64+ }
65+ if err := ctx.Err(); err != nil {
66+ return events, fmt.Errorf("read trace: %w", err)
31 }67 }
32 return events, nil68 return events, nil
33 }69 }
70+
71+func traceBundlePath(bundle, name string) (string, error) {
72+ if strings.TrimSpace(name) == "" || filepath.IsAbs(name) {
73+ return "", errors.New("invalid trace manifest path")
74+ }
75+ root, err := filepath.Abs(bundle)
76+ if err != nil {
77+ return "", fmt.Errorf("resolve trace bundle: %w", err)
78+ }
79+ path, err := filepath.Abs(filepath.Join(root, name))
80+ if err != nil {
81+ return "", fmt.Errorf("resolve trace path: %w", err)
82+ }
83+ rel, err := filepath.Rel(root, path)
84+ if err != nil || rel == ".." || strings.HasPrefix(rel, ".."+string(filepath.Separator)) {
85+ return "", errors.New("trace manifest path escapes bundle")
86+ }
87+ return path, nil
88+}
M packages/agent/subagents/trace_test.go +13 -6
@@ -54,6 +54,9 @@ func TestTraceWriterCreatesPrivateBundleAndDetailedPayload(t *testing.T) {
5454 if len(lines) != 1 {
5555 t.Fatalf("got %d trace lines, want 1", len(lines))
5656 }
57+ if strings.Contains(lines[0], "sensitive result") {
58+ t.Fatalf("trace event leaked detailed payload: %s", lines[0])
59+ }
5760 var event TraceEvent
5861 if err := json.Unmarshal([]byte(lines[0]), &event); err != nil {
5962 t.Fatal(err)
@@ -71,8 +74,9 @@ func TestTraceWriterCreatesPrivateBundleAndDetailedPayload(t *testing.T) {
7174
7275 func TestMemoryTraceWriterRetainsOrderedEventsWithoutBundle(t *testing.T) {
7376 writer := NewMemoryTraceWriter()
74- writer.Record(TraceEvent{Type: "agent.started"})
75- writer.Record(TraceEvent{Type: "turn.started", TurnID: "turn-1"})
77+ for index := 0; index < memoryTraceEventLimit+2; index++ {
78+ writer.Record(TraceEvent{Type: "event", Data: map[string]any{"index": index}})
79+ }
7680 if err := writer.Flush(); err != nil {
7781 t.Fatal(err)
7882 }
@@ -80,8 +84,8 @@ func TestMemoryTraceWriterRetainsOrderedEventsWithoutBundle(t *testing.T) {
8084 t.Fatalf("memory trace dir = %q, want empty", got)
8185 }
8286 events := writer.Events()
83- if len(events) != 2 || events[0].Seq != 1 || events[1].Seq != 2 {
84- t.Fatalf("memory events = %#v, want contiguous sequence", events)
87+ if len(events) != memoryTraceEventLimit || events[len(events)-1].Seq != memoryTraceEventLimit+2 {
88+ t.Fatalf("memory events retained %d with final seq %d", len(events), events[len(events)-1].Seq)
8589 }
8690 if err := writer.Close(); err != nil {
8791 t.Fatal(err)
@@ -95,8 +99,11 @@ func TestTraceWriterNormalRedactsSensitiveValues(t *testing.T) {
9599 }
96100 secret := "super-secret-value"
97101 writer.Record(TraceEvent{
98- Type: "request",
99- Data: map[string]any{"name": "safe", "authorization": secret, "nested": map[string]any{"token": secret, "count": 2}},
102+ Type: "request",
103+ Data: map[string]any{
104+ "name": "safe", "authorization": secret, "openai_api_key": secret,
105+ "nested": map[string]any{"token": secret, "count": 2, "headers": map[string]string{"x_api_key": secret}},
106+ },
100107 Payload: map[string]any{"content": secret},
101108 })
102109 if err := writer.Close(); err != nil {
@@ -54,6 +54,9 @@ func TestTraceWriterCreatesPrivateBundleAndDetailedPayload(t *testing.T) {
54 if len(lines) != 1 {54 if len(lines) != 1 {
55 t.Fatalf("got %d trace lines, want 1", len(lines))55 t.Fatalf("got %d trace lines, want 1", len(lines))
56 }56 }
57+ if strings.Contains(lines[0], "sensitive result") {
58+ t.Fatalf("trace event leaked detailed payload: %s", lines[0])
59+ }
57 var event TraceEvent60 var event TraceEvent
58 if err := json.Unmarshal([]byte(lines[0]), &event); err != nil {61 if err := json.Unmarshal([]byte(lines[0]), &event); err != nil {
59 t.Fatal(err)62 t.Fatal(err)
@@ -71,8 +74,9 @@ func TestTraceWriterCreatesPrivateBundleAndDetailedPayload(t *testing.T) {
71 74
72 func TestMemoryTraceWriterRetainsOrderedEventsWithoutBundle(t *testing.T) {75 func TestMemoryTraceWriterRetainsOrderedEventsWithoutBundle(t *testing.T) {
73 writer := NewMemoryTraceWriter()76 writer := NewMemoryTraceWriter()
74- writer.Record(TraceEvent{Type: "agent.started"})77+ for index := 0; index < memoryTraceEventLimit+2; index++ {
75- writer.Record(TraceEvent{Type: "turn.started", TurnID: "turn-1"})78+ writer.Record(TraceEvent{Type: "event", Data: map[string]any{"index": index}})
79+ }
76 if err := writer.Flush(); err != nil {80 if err := writer.Flush(); err != nil {
77 t.Fatal(err)81 t.Fatal(err)
78 }82 }
@@ -80,8 +84,8 @@ func TestMemoryTraceWriterRetainsOrderedEventsWithoutBundle(t *testing.T) {
80 t.Fatalf("memory trace dir = %q, want empty", got)84 t.Fatalf("memory trace dir = %q, want empty", got)
81 }85 }
82 events := writer.Events()86 events := writer.Events()
83- if len(events) != 2 || events[0].Seq != 1 || events[1].Seq != 2 {87+ if len(events) != memoryTraceEventLimit || events[len(events)-1].Seq != memoryTraceEventLimit+2 {
84- t.Fatalf("memory events = %#v, want contiguous sequence", events)88+ t.Fatalf("memory events retained %d with final seq %d", len(events), events[len(events)-1].Seq)
85 }89 }
86 if err := writer.Close(); err != nil {90 if err := writer.Close(); err != nil {
87 t.Fatal(err)91 t.Fatal(err)
@@ -95,8 +99,11 @@ func TestTraceWriterNormalRedactsSensitiveValues(t *testing.T) {
95 }99 }
96 secret := "super-secret-value"100 secret := "super-secret-value"
97 writer.Record(TraceEvent{101 writer.Record(TraceEvent{
98- Type: "request",102+ Type: "request",
99- Data: map[string]any{"name": "safe", "authorization": secret, "nested": map[string]any{"token": secret, "count": 2}},103+ Data: map[string]any{
104+ "name": "safe", "authorization": secret, "openai_api_key": secret,
105+ "nested": map[string]any{"token": secret, "count": 2, "headers": map[string]string{"x_api_key": secret}},
106+ },
100 Payload: map[string]any{"content": secret},107 Payload: map[string]any{"content": secret},
101 })108 })
102 if err := writer.Close(); err != nil {109 if err := writer.Close(); err != nil {
M packages/agent/tools/subagent_status_test.go +28 -10
@@ -119,11 +119,7 @@ func TestSubagentStatusListsLiveWorkerWithoutPrivateOutput(t *testing.T) {
119119 }
120120
121121 var got struct {
122- Agents []struct {
123- ID string `json:"agent_id"`
124- StartedAt string `json:"started_at"`
125- TaskSummary string `json:"task_summary"`
126- } `json:"agents"`
122+ Agents []json.RawMessage `json:"agents"`
127123 }
128124 text := textResult(res.Content)
129125 if err := json.Unmarshal([]byte(text), &got); err != nil {
@@ -132,14 +128,36 @@ func TestSubagentStatusListsLiveWorkerWithoutPrivateOutput(t *testing.T) {
132128 if len(got.Agents) != 1 {
133129 t.Fatalf("listed agents = %d, want 1", len(got.Agents))
134130 }
135- if got.Agents[0].ID != agent.ID {
136- t.Fatalf("agent id = %q, want %q", got.Agents[0].ID, agent.ID)
131+ var entry map[string]json.RawMessage
132+ if err := json.Unmarshal(got.Agents[0], &entry); err != nil {
133+ t.Fatal(err)
134+ }
135+ for _, forbiddenField := range []string{"state", "process_state", "turn_state", "updated_at", "finished_at"} {
136+ if _, present := entry[forbiddenField]; present {
137+ t.Fatalf("status response exposes removed lifecycle field %q", forbiddenField)
138+ }
139+ }
140+ var agentID, startedAt, taskSummary string
141+ if err := json.Unmarshal(entry["agent_id"], &agentID); err != nil {
142+ t.Fatal(err)
143+ }
144+ if err := json.Unmarshal(entry["started_at"], &startedAt); err != nil {
145+ t.Fatal(err)
146+ }
147+ if err := json.Unmarshal(entry["task_summary"], &taskSummary); err != nil {
148+ t.Fatal(err)
149+ }
150+ if agentID != agent.ID {
151+ t.Fatalf("agent id = %q, want %q", agentID, agent.ID)
137152 }
138- if got.Agents[0].StartedAt == "" {
153+ if startedAt == "" {
139154 t.Fatal("started_at is empty")
140155 }
141- if got.Agents[0].TaskSummary != "review architecture" {
142- t.Fatalf("task summary = %q, want first task line", got.Agents[0].TaskSummary)
156+ if taskSummary != "review architecture" {
157+ t.Fatalf("task summary = %q, want first task line", taskSummary)
158+ }
159+ if _, present := entry["last_event"]; !present {
160+ t.Fatal("status response omits trace-backed last_event")
143161 }
144162 for _, forbidden := range []string{
145163 "PRIVATE_PROMPT_MARKER",
@@ -119,11 +119,7 @@ func TestSubagentStatusListsLiveWorkerWithoutPrivateOutput(t *testing.T) {
119 }119 }
120 120
121 var got struct {121 var got struct {
122- Agents []struct {122+ Agents []json.RawMessage `json:"agents"`
123- ID string `json:"agent_id"`
124- StartedAt string `json:"started_at"`
125- TaskSummary string `json:"task_summary"`
126- } `json:"agents"`
127 }123 }
128 text := textResult(res.Content)124 text := textResult(res.Content)
129 if err := json.Unmarshal([]byte(text), &got); err != nil {125 if err := json.Unmarshal([]byte(text), &got); err != nil {
@@ -132,14 +128,36 @@ func TestSubagentStatusListsLiveWorkerWithoutPrivateOutput(t *testing.T) {
132 if len(got.Agents) != 1 {128 if len(got.Agents) != 1 {
133 t.Fatalf("listed agents = %d, want 1", len(got.Agents))129 t.Fatalf("listed agents = %d, want 1", len(got.Agents))
134 }130 }
135- if got.Agents[0].ID != agent.ID {131+ var entry map[string]json.RawMessage
136- t.Fatalf("agent id = %q, want %q", got.Agents[0].ID, agent.ID)132+ if err := json.Unmarshal(got.Agents[0], &entry); err != nil {
133+ t.Fatal(err)
134+ }
135+ for _, forbiddenField := range []string{"state", "process_state", "turn_state", "updated_at", "finished_at"} {
136+ if _, present := entry[forbiddenField]; present {
137+ t.Fatalf("status response exposes removed lifecycle field %q", forbiddenField)
138+ }
139+ }
140+ var agentID, startedAt, taskSummary string
141+ if err := json.Unmarshal(entry["agent_id"], &agentID); err != nil {
142+ t.Fatal(err)
143+ }
144+ if err := json.Unmarshal(entry["started_at"], &startedAt); err != nil {
145+ t.Fatal(err)
146+ }
147+ if err := json.Unmarshal(entry["task_summary"], &taskSummary); err != nil {
148+ t.Fatal(err)
149+ }
150+ if agentID != agent.ID {
151+ t.Fatalf("agent id = %q, want %q", agentID, agent.ID)
137 }152 }
138- if got.Agents[0].StartedAt == "" {153+ if startedAt == "" {
139 t.Fatal("started_at is empty")154 t.Fatal("started_at is empty")
140 }155 }
141- if got.Agents[0].TaskSummary != "review architecture" {156+ if taskSummary != "review architecture" {
142- t.Fatalf("task summary = %q, want first task line", got.Agents[0].TaskSummary)157+ t.Fatalf("task summary = %q, want first task line", taskSummary)
158+ }
159+ if _, present := entry["last_event"]; !present {
160+ t.Fatal("status response omits trace-backed last_event")
143 }161 }
144 for _, forbidden := range []string{162 for _, forbidden := range []string{
145 "PRIVATE_PROMPT_MARKER",163 "PRIVATE_PROMPT_MARKER",
M packages/agent/tracecmd.go +3 -3
@@ -17,11 +17,11 @@ func runDebugCommand(rawArgs []string, stdout io.Writer) (bool, error) {
1717 if len(rawArgs) != 3 || rawArgs[1] != "trace" {
1818 return true, fmt.Errorf("usage: zut debug trace <bundle>")
1919 }
20- events, err := subagents.ReadTrace(rawArgs[2])
21- if err != nil {
20+ events, readErr := subagents.ReadTrace(rawArgs[2])
21+ if err := renderTraceInspection(stdout, subagents.ProjectTrace(events), time.Now()); err != nil {
2222 return true, err
2323 }
24- return true, renderTraceInspection(stdout, subagents.ProjectTrace(events), time.Now())
24+ return true, readErr
2525 }
2626
2727 func renderTraceInspection(w io.Writer, views map[string]subagents.AgentTraceView, now time.Time) error {
@@ -17,11 +17,11 @@ func runDebugCommand(rawArgs []string, stdout io.Writer) (bool, error) {
17 if len(rawArgs) != 3 || rawArgs[1] != "trace" {17 if len(rawArgs) != 3 || rawArgs[1] != "trace" {
18 return true, fmt.Errorf("usage: zut debug trace <bundle>")18 return true, fmt.Errorf("usage: zut debug trace <bundle>")
19 }19 }
20- events, err := subagents.ReadTrace(rawArgs[2])20+ events, readErr := subagents.ReadTrace(rawArgs[2])
21- if err != nil {21+ if err := renderTraceInspection(stdout, subagents.ProjectTrace(events), time.Now()); err != nil {
22 return true, err22 return true, err
23 }23 }
24- return true, renderTraceInspection(stdout, subagents.ProjectTrace(events), time.Now())24+ return true, readErr
25 }25 }
26 26
27 func renderTraceInspection(w io.Writer, views map[string]subagents.AgentTraceView, now time.Time) error {27 func renderTraceInspection(w io.Writer, views map[string]subagents.AgentTraceView, now time.Time) error {

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