From 5be023a6c6bede073f8eb3ea8ea4771acf217e7b Mon Sep 17 00:00:00 2001
From: Daniel <rootdevdb@gmail.com>
Date: Fri, 11 Sep 2026 14:52:26 +0000
Subject: [PATCH 1/3] fix(parser): dedupe Codex inferences with empty IDs


From 5657df9882509a10ab5ac62466b8fa0562d5477e Mon Sep 17 00:00:00 2001
From: Daniel <rootdevdb@gmail.com>
Date: Fri, 11 Sep 2026 14:52:27 +0000
Subject: [PATCH 2/3] test(parser): Codex empty-ID inference dedup

---
 internal/parser/codex_dedup_test.go | 83 +++++++++++++++++++++++++++++
 1 file changed, 83 insertions(+)
 create mode 100644 internal/parser/codex_dedup_test.go

diff --git a/internal/parser/codex_dedup_test.go b/internal/parser/codex_dedup_test.go
new file mode 100644
index 0000000..a5b3204
--- /dev/null
+++ b/internal/parser/codex_dedup_test.go
@@ -0,0 +1,83 @@
+package parser
+
+import (
+	"encoding/json"
+	"testing"
+)
+
+func TestBuildTraceFromCodex_DedupsEmptyInferenceIDs(t *testing.T) {
+	inf, err := json.Marshal(codexInference{
+		Model:        "gpt-5",
+		InputTokens:  100,
+		OutputTokens: 50,
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	entries := []codexEntry{
+		{
+			Type:      "tool_call_started",
+			ID:        "tool-1",
+			SessionID: "s1",
+			Timestamp: "2026-07-18T10:00:00.000Z",
+			Data:      json.RawMessage(`{"name":"Edit","input":"{\"file_path\":\"a.go\"}"}`),
+		},
+		{
+			Type:      "inference_completed",
+			ID:        "",
+			SessionID: "s1",
+			Timestamp: "2026-07-18T10:00:01.000Z",
+			Data:      inf,
+		},
+		{
+			// Duplicate empty-ID inference with identical payload/timestamp.
+			Type:      "inference_completed",
+			ID:        "",
+			SessionID: "s1",
+			Timestamp: "2026-07-18T10:00:01.000Z",
+			Data:      inf,
+		},
+	}
+
+	tr, err := buildTraceFromCodex(entries, "synthetic.codex")
+	if err != nil {
+		t.Fatalf("buildTraceFromCodex: %v", err)
+	}
+	it := tr.Get(0)
+	if it == nil {
+		t.Fatal("expected iteration")
+	}
+	if it.Tokens.InputTokens != 100 {
+		t.Fatalf("InputTokens = %d, want 100 (deduped once)", it.Tokens.InputTokens)
+	}
+	if it.Tokens.OutputTokens != 50 {
+		t.Fatalf("OutputTokens = %d, want 50 (deduped once)", it.Tokens.OutputTokens)
+	}
+}
+
+func TestBuildTraceFromCodex_CountsDistinctEmptyIDInferences(t *testing.T) {
+	inf1, _ := json.Marshal(codexInference{Model: "gpt-5", InputTokens: 10, OutputTokens: 5})
+	inf2, _ := json.Marshal(codexInference{Model: "gpt-5", InputTokens: 20, OutputTokens: 7})
+	entries := []codexEntry{
+		{
+			Type:      "tool_call_started",
+			SessionID: "s1",
+			Timestamp: "2026-07-18T10:00:00.000Z",
+			Data:      json.RawMessage(`{"name":"Edit","input":"{\"file_path\":\"a.go\"}"}`),
+		},
+		{Type: "inference_completed", Timestamp: "2026-07-18T10:00:01.000Z", Data: inf1},
+		{Type: "inference_completed", Timestamp: "2026-07-18T10:00:02.000Z", Data: inf2},
+	}
+	tr, err := buildTraceFromCodex(entries, "synthetic.codex")
+	if err != nil {
+		t.Fatal(err)
+	}
+	it := tr.Get(0)
+	if it.Tokens.InputTokens != 30 {
+		t.Fatalf("InputTokens = %d, want 30", it.Tokens.InputTokens)
+	}
+	if it.Tokens.OutputTokens != 12 {
+		t.Fatalf("OutputTokens = %d, want 12", it.Tokens.OutputTokens)
+	}
+}

From 51532ac2d5cc71368f9f982a0e936e7fbb972295 Mon Sep 17 00:00:00 2001
From: Daniel <rootdevdb@gmail.com>
Date: Fri, 11 Sep 2026 14:52:54 +0000
Subject: [PATCH 3/3] fix(parser): dedupe Codex inferences with empty IDs

---
 internal/parser/codex.go | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/internal/parser/codex.go b/internal/parser/codex.go
index c2300fc..8e3e9db 100644
--- a/internal/parser/codex.go
+++ b/internal/parser/codex.go
@@ -141,12 +141,16 @@ func buildTraceFromCodex(entries []codexEntry, sourcePath string) (*trace.Trace,
 				model = inf.Model
 			}
 
-			if entry.ID != "" && seenInference[entry.ID] {
-				continue
+			dedupKey := entry.ID
+			if dedupKey == "" {
+				// Empty IDs used to bypass dedup and double-count tokens/cost.
+				dedupKey = fmt.Sprintf("%s|%d|%d|%d|%s",
+					inf.Model, inf.InputTokens, inf.OutputTokens, inf.ReasoningOut, entry.Timestamp)
 			}
-			if entry.ID != "" {
-				seenInference[entry.ID] = true
+			if seenInference[dedupKey] {
+				continue
 			}
+			seenInference[dedupKey] = true
 
 			if currentIter != nil {
 				currentIter.Tokens.InputTokens += inf.InputTokens