Skip to content

OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security

fix(history): use chronological cutoffs for explicit comparisons (#1774)

Michael D'Angelo0bfdf7229399
diff --git a/plugins/codex-security/scripts/workbench_scan_history.py b/plugins/codex-security/scripts/workbench_scan_history.py
index e0654d6a7..0f5eacc33 100644
--- a/plugins/codex-security/scripts/workbench_scan_history.py
+++ b/plugins/codex-security/scripts/workbench_scan_history.py
@@ -667,14 +667,14 @@ def compare_scans(
for pair in related
]
if include_matching_inputs:
+ after_key = (timestamp_key(after["started_at"]), after["id"])
known_scan_ids = {
scan["id"]
for scan in connection.execute(
- "SELECT * FROM scans WHERE status = 'complete' "
- "AND (started_at < ? OR (started_at = ? AND id <= ?))",
- (after["started_at"], after["started_at"], after["id"]),
+ "SELECT id, started_at, target_id, target_path FROM scans WHERE status = 'complete'"
)
- if _same_repository(scan, after)
+ if (timestamp_key(scan["started_at"]), scan["id"]) <= after_key
+ and _same_repository(scan, after)
}
excluded_pairs = {(before["id"], after["id"]), (after["id"], before["id"])}
known_groups = _known_finding_groups(
diff --git a/plugins/codex-security/tests/test_workbench_scan_history.py b/plugins/codex-security/tests/test_workbench_scan_history.py
index c3b6ed26b..b2cd6eb6c 100644
--- a/plugins/codex-security/tests/test_workbench_scan_history.py
+++ b/plugins/codex-security/tests/test_workbench_scan_history.py
@@ -7,6 +7,7 @@ import sqlite3
import subprocess
import sys
import uuid
+from argparse import Namespace
from pathlib import Path
from typing import Any
@@ -18,6 +19,7 @@ from workbench_test_support import (
fail_scan,
get_scan,
initialize_git_repository,
+ load_script,
mark_deep_coordinator_succeeded,
resume_deep_scan,
run_workbench,
@@ -1098,6 +1100,97 @@ def test_semantic_scan_comparison_caches_matches_and_exposes_related_findings(
assert latest_finding["knownSince"] == known_since
+ "timestamps",
+ [
+ (
+ "2026-08-01T03:00:00+02:00",
+ "2026-08-01T01:15:00Z",
+ "2026-08-01T01:30:00Z",
+ "2026-08-01T00:45:00-01:00",
+ ),
+ (
+ "2026-08-01T01:00:00Z",
+ "2026-08-01T01:00:00.000001Z",
+ "2026-08-01T01:00:00.000002Z",
+ "2026-08-01T01:00:00.000003Z",
+ ),
+ (
+ "2026-08-01t03:00:00.000000+02:00",
+ "2026-08-01T01:00:00Z",
+ "2026-08-01T01:00:00.0Z",
+ "2026-08-01T01:00:00.000Z",
+ ),
+ ],
+ ids=["offsets", "microseconds", "equal-instants"],
+)
+def test_explicit_comparison_uses_chronological_matching_cutoff(
+ tmp_path: Path, monkeypatch: pytest.MonkeyPatch, timestamps: tuple[str, ...]
+) -> None:
+ state_dir = tmp_path / "state"
+ repository = tmp_path / "repository"
+ repository.mkdir()
+ scans = sorted(
+ (
+ create_cli_scan(
+ state_dir, tmp_path / "scans", repository, identity_anchor=f"synthetic-{index}"
+ )
+ for index in range(4)
+ ),
+ key=lambda scan: scan["scanId"],
+ )
+ middle, after, future = scans[1:]
+ findings = [get_scan(state_dir, scan["scanId"])["scan"]["findings"][0] for scan in scans]
+ for left, right in ((0, 1), (1, 3)):
+ save_scan_matches(
+ state_dir,
+ scans[left],
+ scans[right],
+ confirmed_match(findings[left]["occurrenceId"], findings[right]["occurrenceId"]),
+ )
+ with sqlite3.connect(state_dir / "workbench.sqlite3") as connection:
+ connection.executemany(
+ "UPDATE scans SET started_at = ? WHERE id = ?",
+ [(timestamp, scan["scanId"]) for timestamp, scan in zip(timestamps, scans)],
+ )
+ original = list(connection.iterdump())
+ expected = [sorted(finding["findingId"] for finding in findings[:2])]
+
+ for before_scan, after_scan in ((middle, after), (after, middle)):
+ compared = compare_scan_pair(
+ state_dir, before_scan, after_scan, "--include-matching-inputs"
+ )
+ assert (compared["beforeScanId"], compared["afterScanId"]) == (
+ before_scan["scanId"],
+ after_scan["scanId"],
+ )
+ assert compared["matchingInputs"]["knownFindingGroups"] == expected
+
+ history = load_script("workbench_scan_history")
+
+ def unexpected_git_lookup(*_args: Any) -> str:
+ raise AssertionError("Future scans must not trigger repository discovery")
+
+ monkeypatch.setattr(history, "git_output", unexpected_git_lookup)
+ with sqlite3.connect(state_dir / "workbench.sqlite3") as connection:
+ assert list(connection.iterdump()) == original
+ connection.row_factory = sqlite3.Row
+ connection.execute(
+ "UPDATE scans SET target_id = NULL, target_path = ? WHERE id = ?",
+ (str(tmp_path / "future-repository"), future["scanId"]),
+ )
+ compared = history.compare_scans(
+ connection,
+ Namespace(before_scan_id=middle["scanId"], after_scan_id=after["scanId"]),
+ require_scan=lambda db, scan_id: db.execute(
+ "SELECT * FROM scans WHERE id = ?", (scan_id,)
+ ).fetchone(),
+ read_coverage=lambda _: {},
+ include_matching_inputs=True,
+ )
+ assert compared["matchingInputs"]["knownFindingGroups"] == expected
+
+
def test_semantic_scan_matching_backfills_legacy_finding_details(tmp_path: Path) -> None:
state_dir = tmp_path / "state"
repository = tmp_path / "repository"
diff --git a/sdk/typescript/tests-ts/workbench-scan-history.test.ts b/sdk/typescript/tests-ts/workbench-scan-history.test.ts
index 57a4a335f..169556bd5 100644
--- a/sdk/typescript/tests-ts/workbench-scan-history.test.ts
+++ b/sdk/typescript/tests-ts/workbench-scan-history.test.ts
@@ -106,7 +106,7 @@ for index, (scan, repository) in enumerate([
('before', 'selected'), ('after', 'selected')
]):
connection.execute('INSERT INTO scans VALUES (?, NULL, ?, ?, ?)',
- (scan, str(Path(sys.argv[2]) / repository), 'complete', str(index)))
+ (scan, str(Path(sys.argv[2]) / repository), 'complete', f'2026-08-01T00:00:0{index}Z'))
connection.executemany('INSERT INTO finding_occurrences VALUES (?, ?, ?)', [
('unrelated-first', 'unrelated-identity-a', 'unrelated-before'),
('unrelated-second', 'unrelated-identity-b', 'unrelated-after')