pymupdf 2.0

Source: https://github.com/pymupdf/pymupdf
Tag scanned: 2.0
Compared with: 1.28.2 (71 signals introduced since)
Files: 227 scanned, 745 skipped (binary, oversized, or unrecognised)
Signals: 0 critical · 0 warn · 95 info
Scan time: 63.0 s
LLM review: Reviewed by claude-haiku-5-5 (Anthropic): 87 of 87 warn and critical signals; 87 demoted to info.
Disclude version: 4.0.0
Scanned (UTC): 2026-10-20 06:11

Signals

Ordered by severity, then signals introduced in this release. Hidden characters are shown as ⟨U+…⟩.

info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:109:45 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                            ['2023', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
dismissed LLM, confidence 0.90 Test fixture stores UTF-8 Chinese text as a bytes literal.
Reasoning
The bytes literal is UTF-8 encoded Chinese text used as test data for table content. The escape density reflects non-ASCII text written as bytes, not a binary payload.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:109:47 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)
```
                            ['2023', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
dismissed LLM, confidence 0.90 Raw-pass flag on a run of \xNN escapes that encode Chinese UTF-8 test text.
Reasoning
The raw escape run is the same UTF-8 Chinese test text encoded as bytes. Raw-pass escape flags on legitimate multibyte text are a known false-positive pattern.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:109:232 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa
```
dismissed LLM, confidence 0.90 Raw-pass flag on a further segment of encoded Chinese test text.
Reasoning
This is another segment of the same UTF-8 Chinese test string, with normal multibyte sequences and newlines. It is ordinary encoded text, not obfuscation.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:109:317 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\x
```
dismissed LLM, confidence 0.90 Raw-pass flag on encoded Chinese test text within the same fixture.
Reasoning
The escape run is part of the same Chinese UTF-8 test data, which contains ordinary multibyte characters. The entropy figure reflects natural UTF-8, not an encrypted blob.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:109:415 new in this release

bytes literal contains 78 `\xNN` escapes (97% of content) — binary payload shape (ast pass, confidence 0.85)
```
\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xb
```
dismissed LLM, confidence 0.90 Test fixture bytes literal holding Chinese text for table tests.
Reasoning
The bytes literal is the same Chinese UTF-8 test content, with newlines and punctuation encoded as escapes. It is legitimate test data.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:109:417 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\
```
dismissed LLM, confidence 0.90 Raw-pass flag on escape-encoded Chinese text in a test fixture.
Reasoning
This raw escape run is a part of the Chinese UTF-8 test fixture. Multibyte natural-language encoding triggers the escape heuristic without any malicious signal.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:109:491 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa
```
dismissed LLM, confidence 0.90 Raw-pass flag on escape-encoded Chinese text in a test fixture.
Reasoning
The escape run encodes more Chinese UTF-8 test text, including punctuation and line breaks. This is expected test data.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:109:654 new in this release

18 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.65)
```
e8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
dismissed LLM, confidence 0.90 Raw-pass flag on the end of an escape-encoded Chinese test fixture.
Reasoning
The short escape run is the tail of the same Chinese UTF-8 test fixture. Nothing suggests a payload or hidden logic.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:110:45 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                            ['2022', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
likely benign LLM, confidence 0.90 Parallel test fixture bytes literal holding Chinese text.
Reasoning
This is a repeat of the same Chinese UTF-8 test fixture in a parallel test case, encoded as bytes for expected table content. It is benign test data.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:110:47 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)
```
                            ['2022', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
dismissed LLM, confidence 0.95 Test fixture holding UTF-8 encoded Chinese rating-description text in a bytes literal.
Reasoning
The raw pass flagged a run of \xNN escapes, but the bytes are UTF-8 encoding of Chinese financial-rating text (for example 偿还能力很强, 受不利经济, 环境的影响不大), decoded with .decode() in a test fixture. This is legitimate internationalized test data, not obfuscation.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:110:232 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa
```
dismissed LLM, confidence 0.95 Consecutive UTF-8 byte escapes spelling Chinese words inside a decoded test string.
Reasoning
Same test fixture: the escapes encode readable Chinese text (for example 违约风险很低 and 中诚信国际信用评级), and the bytes are immediately decoded as UTF-8. The raw-pass escape-run heuristic fires on ordinary multibyte text.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:110:317 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\x
```
dismissed LLM, confidence 0.95 UTF-8 escape run of Chinese text in a test data literal, decoded to a string.
Reasoning
The escape run is a fragment of the same UTF-8 encoded Chinese rating text, decoded in a test. It is a normal way to write non-ASCII test strings and has no execution or hidden-payload behavior.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:110:415 new in this release

bytes literal contains 78 `\xNN` escapes (97% of content) — binary payload shape (ast pass, confidence 0.85)
```
\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xb
```
dismissed LLM, confidence 0.90 Large bytes literal of UTF-8 Chinese text in a test fixture that is decoded to a string.
Reasoning
The AST pass labels this a binary payload shape because most of the bytes literal is \xNN escapes. However, the content is UTF-8 Chinese prose that is decoded to str in the test, and the literal is not executed, decoded into code, or written to disk. The shape is a multibyte-text false positive.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:110:417 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\
```
dismissed LLM, confidence 0.95 Escape run of UTF-8 Chinese text within a decoded test fixture.
Reasoning
Another fragment of the same UTF-8 Chinese text in a test fixture, decoded right after. Its entropy reflects the multibyte encoding of natural-language text, not a hidden payload.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:110:491 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa
```
dismissed LLM, confidence 0.95 UTF-8 byte escapes for Chinese rating text in a test data tuple.
Reasoning
The run encodes Chinese rating-description text (for example 联合资信评估) and is decoded with .decode() in a test. There is no execution sink or obfuscation intent.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:110:654 new in this release

18 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.65)
```
e8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
dismissed LLM, confidence 0.95 Chinese-language UTF-8 byte escapes in a test fixture, decoded to strings.
Reasoning
The run encodes Chinese text (for example 稳定 and 有限公司) and is decoded in a test. Its entropy is typical of UTF-8 multibyte text.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:111:45 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                            ['2021', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
dismissed LLM, confidence 0.90 Bytes literal of UTF-8 Chinese rating text in a test table, decoded later.
Reasoning
This AST finding is the same test-data pattern on the adjacent line. The bytes literal is UTF-8 Chinese text that is decoded to str, with no dynamic execution or dropper behavior.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:111:47 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)
```
                            ['2021', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
dismissed LLM, confidence 0.95 Run of UTF-8 escapes for Chinese text in a test fixture row.
Reasoning
The escape-soup run is UTF-8 encoded Chinese text in the same test data table, and it is decoded in the test. The heuristic reacts to multibyte encoding, not to malicious intent.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:111:232 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x8
```
dismissed LLM, confidence 0.95 UTF-8 Chinese byte escapes in a test table row that is decoded to strings.
Reasoning
This escape run is UTF-8 Chinese text (for example 联合资信评估 and 违约风险) inside a decoded test fixture. No execution or hidden-logic indicators are present.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:111:317 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode()
```
dismissed LLM, confidence 0.95 UTF-8 escape run of Chinese rating text in a decoded test fixture.
Reasoning
This is another piece of the same UTF-8 Chinese test fixture, with explicit .decode() calls. It is a false positive from the multibyte-escape heuristic.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:111:417 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
likely benign LLM, confidence 0.97 Test fixture holding UTF-8 Chinese strings written as byte escapes and decoded.
Reasoning
The escapes are UTF-8 bytes for Chinese text (e.g. 联合资信评估, 有限公司, 稳定) decoded with .decode(), inside a test fixture of Chinese financial table data. Legitimate internationalization test data, not obfuscation.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:112:45 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                            ['2020', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
likely benign LLM, confidence 0.96 Bytes literal of UTF-8 Chinese text used as test table content.
Reasoning
The bytes literal is UTF-8 encoded Chinese prose (e.g. 偿还债务的能力很强, 受不利经济的影响) in a test file of table data. The binary-payload shape is an artifact of non-ASCII test text, not a dropper.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:112:47 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)
```
                            ['2020', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8
```
likely benign LLM, confidence 0.96 Escaped UTF-8 Chinese text within a test fixture.
Reasoning
Same fixture as the other findings: UTF-8 Chinese text escaped as bytes in a test file. High escape density is expected for non-ASCII content.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:112:232 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x8
```
likely benign LLM, confidence 0.96 Chinese text escaped as UTF-8 bytes in a test table.
Reasoning
Continuation of the same Chinese-language test fixture with UTF-8 byte escapes decoded to text. No execution or hidden payload.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:112:317 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode()
```
likely benign LLM, confidence 0.96 UTF-8 Chinese strings encoded as escapes in a test.
Reasoning
Same Chinese text fixture decoded from UTF-8 bytes. Consistent with i18n test data.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:112:417 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
likely benign LLM, confidence 0.96 Chinese financial-rating text stored as byte escapes in tests.
Reasoning
Chinese financial-report wording (联合资信评估有限公司, 稳定) encoded as UTF-8 escapes in a test fixture. Benign localization data.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:115:70 new in this release

18 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
         [b'\xe5\xba\x8f\xe5\x8f\xb7'.decode(), b'\xe9\x87\x91\xe8\x9e\x8d\xe6\x9c\xba\xe6\x9e\x84\xe5\x90\x8d\xe7\xa7\xb0'.decode(), b'\xe6\x8e\x88\xe4\xbf\xa1\xe6\x80\xbb\xe9\xa2\x9d'.decode(), b'\xe5\xb7\xb2\xe4\xbd\xbf\xe7\x94\xa8\xe6\x83\x85\xe5\x86\xb5'.decode(), b'\xe5\x89\xa9\xe4\xbd\x99\xe9\xa2\x9d\xe5\xba\xa6
```
likely benign LLM, confidence 0.96 Chinese table column headers written as byte escapes and decoded.
Reasoning
Decoded UTF-8 Chinese column headers (序号, 金融机构名称, 授信总额) in a test table. Benign.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:117:35 new in this release

bytes literal contains 36 `\xNN` escapes (100% of content) — binary payload shape (ast pass, confidence 0.85)
```
                            ['2', b'\xe5\xbe\xbd\xe5\x95\x86\xe9\x93\xb6\xe8\xa1\x8c\xe5\x8d\x97\xe4\xba\xac\xe4\xb8\xad\xe5\xb1\xb1\xe5\x8c\x97\xe8\xb7\xaf\xe6\x94\xaf\xe8\xa1\x8c'.decode(), '33043530\n, .', '25673530\n, .', '7370000\n, .', ],
```
likely benign LLM, confidence 0.96 UTF-8 Chinese bank branch name stored as bytes in test data.
Reasoning
Bytes literal of UTF-8 Chinese bank name (徽商银行南京中山北路支行) decoded in a test table. Binary-shape flag is a false positive from non-ASCII text.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:117:37 new in this release

36 consecutive `\xNN` escapes (entropy 4.3 bits/byte) (raw pass, confidence 0.80)
```
                            ['2', b'\xe5\xbe\xbd\xe5\x95\x86\xe9\x93\xb6\xe8\xa1\x8c\xe5\x8d\x97\xe4\xba\xac\xe4\xb8\xad\xe5\xb1\xb1\xe5\x8c\x97\xe8\xb7\xaf\xe6\x94\xaf\xe8\xa1\x8c'.decode(), '33043530\n, .', '25673530\n, .', '7370000\n, .', ],
```
likely benign LLM, confidence 0.96 Escaped UTF-8 CJK bank name in a test fixture.
Reasoning
Same Chinese bank-name fixture; escape soup is just UTF-8 encoding of CJK text in a test.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:118:37 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                            ['3', b'\xe5\x85\xb4\xe4\xb8\x9a\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '30260000\n, .', '11060000\n, .', '19200000\n, .', ],
```
likely benign LLM, confidence 0.96 Escaped UTF-8 Chinese bank name in test table.
Reasoning
Chinese bank branch name (兴业银行盐城分行) as UTF-8 escapes in test data. Benign.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:119:37 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                            ['4', b'\xe6\xb0\x91\xe7\x94\x9f\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '26200000\n, .', '8850000\n, .', '17350000\n, .', ],
```
likely benign LLM, confidence 0.96 Escaped UTF-8 Chinese bank name in test table.
Reasoning
Chinese bank branch name (民生银行盐城分行) as UTF-8 escapes in a test fixture. Benign.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:120:37 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                            ['5', b'\xe4\xb8\xad\xe4\xbf\xa1\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '20508800\n, .', '20508800\n, .', '000\n.', ],
```
likely benign LLM, confidence 0.96 Escaped UTF-8 Chinese bank name in test table.
Reasoning
Chinese bank branch name (中信银行盐城分行) as UTF-8 escapes in test data. Benign.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:121:37 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                            ['6', b'\xe5\x8d\x97\xe4\xba\xac\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '17517000\n, .', '17517000\n, .', '000\n.', ],
```
likely benign LLM, confidence 0.90 Test fixture row holding UTF-8 encoded Chinese bank name bytes.
Reasoning
The escapes are UTF-8 bytes for Chinese bank and branch names (e.g. 南京银行, 南京分行) inside a test fixture table. This is ordinary non-ASCII test data written as byte escapes, not obfuscation.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:125:38 new in this release

30 consecutive `\xNN` escapes (entropy 4.3 bits/byte) (raw pass, confidence 0.80)
```
                            ['10', b'\xe6\x81\x92\xe4\xb8\xb0\xe9\x93\xb6\xe8\xa1\x8c\xe5\x8d\x97\xe4\xba\xac\xe6\xb1\x9f\xe5\x8c\x97\xe6\x94\xaf\xe8\xa1\x8c'.decode(), '9390000\n, .', '4350000\n, .', '5040000\n, .', ],
```
likely benign LLM, confidence 0.90 Test fixture entry with UTF-8 bytes for a Chinese bank name.
Reasoning
The bytes decode to Chinese bank branch names (e.g. 恒丰银行, 南京江北支行). The file is a test table of such names, so the escape-heavy encoding is a byte-literal representation of normal Unicode text.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:126:38 new in this release

24 consecutive `\xNN` escapes (entropy 4.3 bits/byte) (raw pass, confidence 0.80)
```
                            ['11', b'\xe4\xb8\x8a\xe6\xb5\xb7\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '4626500\n, .', '4626500\n, .', '000\n.', ],
                        ],
                    ]
```
likely benign LLM, confidence 0.90 Test fixture entry with UTF-8 bytes for a Chinese bank name.
Reasoning
The bytes decode to Chinese bank branch text (上海银行监管分行-style names), consistent with a financial-institution lookup table used in tests. No executable or decoding-to-run behaviour is present.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:262:41 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                        ['2023', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf
```
likely benign LLM, confidence 0.88 Test fixture bytes literal containing UTF-8 encoded Chinese disclosure text.
Reasoning
The AST pass flagged a bytes literal, but its content is UTF-8 Chinese prose (credit and risk disclosure text) used as expected test data. Nothing executes or decodes it into code, so the 'binary payload' shape is a false positive from non-ASCII text.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:262:43 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)
```
                        ['2023', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf
```
likely benign LLM, confidence 0.88 Escape run in a test bytes literal encoding Chinese text.
Reasoning
The escape run is UTF-8 for Chinese disclosure text inside a bytes literal in a test fixture. It is a text encoding artifact, not an encoded dropper or hidden payload.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:262:228 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa
```
likely benign LLM, confidence 0.86 Continuation of a UTF-8 Chinese text bytes literal in a test fixture.
Reasoning
This is a continuation of the same UTF-8 Chinese text in a test fixture. The paired .decode() call shows the bytes are being turned into ordinary strings for comparison.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:262:313 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\x
```
likely benign LLM, confidence 0.86 Escape run encoding Chinese text decoded in a test fixture.
Reasoning
The escapes encode Chinese risk and disclosure wording, and the bytes are decoded with .decode() in a test table. This is normal internationalised test data.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:262:411 new in this release

bytes literal contains 78 `\xNN` escapes (97% of content) — binary payload shape (ast pass, confidence 0.85)
```
\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xb
```
likely benign LLM, confidence 0.88 Test fixture bytes literal with UTF-8 Chinese text.
Reasoning
The AST pass flagged another bytes literal, but its content is UTF-8 Chinese text paired with .decode() in a test. No code is built from it and nothing is executed, so this is a false positive.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:262:413 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\
```
likely benign LLM, confidence 0.86 Escape run in a test fixture encoding Chinese text.
Reasoning
This is more UTF-8 Chinese text in the same test fixture, with the bytes decoded to strings for comparison. The escape density comes from the non-ASCII content, not from concealment.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:262:487 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa
```
likely benign LLM, confidence 0.86 Escape run in a test fixture encoding Chinese text.
Reasoning
The run encodes Chinese text for a test table (credit and rating wording) and is decoded in place. Nothing suggests obfuscated execution or malicious intent.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:262:650 new in this release

18 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.65)
```
e8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
likely benign LLM, confidence 0.86 Short UTF-8 bytes literal for a Chinese word in a test fixture.
Reasoning
The bytes decode to the Chinese word 稳定 ('stable') in a test table. This is a short non-ASCII literal written as escapes, which is a normal coding pattern in a test file.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:263:41 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                        ['2022', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf
```
likely benign LLM, confidence 0.93 A test fixture holding UTF-8 encoded Chinese financial text as a bytes literal.
Reasoning
The bytes literal is UTF-8 encoding of Chinese text (financial rating and risk wording) in a test fixture, decoded with .decode(). The escape-heavy shape is the normal result of writing non-ASCII test data as bytes, not a binary payload. No execution, network, or dropper behaviour is present.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:263:43 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)
```
                        ['2022', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf
```
likely benign LLM, confidence 0.92 Consecutive UTF-8 byte escapes for Chinese text inside a test data row.
Reasoning
The raw pass flags a run of \xNN escapes, but they are the UTF-8 bytes of Chinese characters inside a test data table. The raw pass has a high false-positive rate on multibyte text, and the context shows an ordinary test fixture.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:263:228 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa
```
likely benign LLM, confidence 0.92 A run of UTF-8 escapes encoding a Chinese sentence in a test fixture.
Reasoning
Same fixture: the escape run encodes Chinese words and punctuation (a '\n' and full-width comma sit among them). This is legitimate multilingual test data with no decoding into code or execution.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:263:313 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\x
```
likely benign LLM, confidence 0.92 UTF-8 escape sequence for Chinese text in a test data tuple.
Reasoning
The escapes are UTF-8 for Chinese text in a test table, decoded to str. Entropy is typical of multibyte CJK text, not of an encrypted or packed blob.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:263:411 new in this release

bytes literal contains 78 `\xNN` escapes (97% of content) — binary payload shape (ast pass, confidence 0.85)
```
\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xb
```
likely benign LLM, confidence 0.92 Another bytes literal of UTF-8 Chinese text in a test fixture.
Reasoning
The bytes literal is a second Chinese text value in the same test fixture, decoded via .decode(). The escape ratio follows from the script being non-ASCII, not from hidden code.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:263:413 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe4\xb8\xad\xe8\xaf\x9a\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\
```
likely benign LLM, confidence 0.92 UTF-8 byte escapes for Chinese text within a test data row.
Reasoning
Escapes encode Chinese text in a test fixture tuple. The entropy is in the normal range for CJK text in UTF-8, and nothing is executed.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:263:487 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
\xe4\xbf\xa1\xe5\x9b\xbd\xe9\x99\x85\xe4\xbf\xa1\n\xe7\x94\xa8\xe8\xaf\x84\xe7\xba\xa7\xe6\x9c\x89\xe9\x99\x90\xe8\xb4\xa3\n\xe4\xbb\xbb\xe5\x85\xac\xe5\x8f\xb8 \xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa
```
likely benign LLM, confidence 0.92 A run of UTF-8 escapes for Chinese financial text in a test fixture.
Reasoning
The run encodes more Chinese financial wording in a decoded test value. It is test data, not an encoded dropper or staged payload.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:263:650 new in this release

18 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.65)
```
e8\x81\x94\xe5\x90\x88\xe8\xb5\x84\n\xe3\x80\x81\n\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\n\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
likely benign LLM, confidence 0.92 A short UTF-8 Chinese word stored as a bytes literal in a test fixture.
Reasoning
A short bytes literal of Chinese text (稳定) decoded in a test tuple. Too short and too plain to be a payload, and it sits in ordinary test code.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:264:41 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                        ['2021', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf
```
likely benign LLM, confidence 0.93 A UTF-8 encoded Chinese text bytes literal in a second test fixture row.
Reasoning
Same test fixture pattern as line 263: a UTF-8 bytes literal of Chinese text with a decode step, in a test row for a different year. This is test data, not a binary payload.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:264:43 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)
```
                        ['2021', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf
```
likely benign LLM, confidence 0.92 Consecutive UTF-8 byte escapes for Chinese text in a test row.
Reasoning
The escape run is UTF-8 for Chinese text in the test fixture. The raw-pass signal is a false positive on multibyte non-ASCII data.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:264:228 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x8
```
likely benign LLM, confidence 0.92 A UTF-8 escape run encoding Chinese text in a test fixture.
Reasoning
Chinese text escapes in the same test fixture, decoded to str. This is multilingual test content with no execution or hidden logic.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:264:313 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode()
```
likely benign LLM, confidence 0.95 Chinese UTF-8 text stored as byte-escape literals in a test table fixture.
Reasoning
The escapes are UTF-8 bytes for Chinese text (e.g. 影响不大, 联合资信评估, 稳定) decoded immediately with .decode(). This is a test fixture for localized table data, not a payload.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:264:413 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
likely benign LLM, confidence 0.94 Chinese financial-rating text encoded as UTF-8 byte literals in test data.
Reasoning
Same pattern: UTF-8 Chinese strings written as escaped bytes and decoded in a test data table. Benign localization fixture.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:265:41 new in this release

bytes literal contains 87 `\xNN` escapes (98% of content) — binary payload shape (ast pass, confidence 0.85)
```
                        ['2020', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf
```
likely benign LLM, confidence 0.94 A bytes literal holding Chinese prose in a test table row, flagged as a binary payload.
Reasoning
The bytes literal is UTF-8 Chinese prose (e.g. 偿还债务的能力很强) inside a test table row. The 'binary payload' heuristic misfires on non-ASCII text; there is no execution or decoding to a hidden stage.

info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:265:43 new in this release

27 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.80)

                        ['2020', 'AA+', b'\xe5\x81\xbf\xe8\xbf\x98\xe5\x80\xba\xe5\x8a\xa1\xe7\x9a\x84\xe8\x83\xbd\xe5\x8a\x9b\xe5\xbe\x88\xe5\xbc\xba \xe5\x8f\x97\xe4\xb8\x8d\xe5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf

likely benign LLM, confidence 0.94 UTF-8 Chinese text escape run in a test table.

Reasoning

Raw escape run is the same UTF-8 Chinese text in a test fixture. Benign.

info ~~warn~~ → info by review encoding-hex tests/test_tables.py:265:228 new in this release

21 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)

e5\x88\xa9\xe7\xbb\x8f\xe6\xb5\x8e\n\xef\xbc\x8c\n\xe7\x8e\xaf\xe5\xa2\x83\xe7\x9a\x84\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x8

likely benign LLM, confidence 0.94 UTF-8 Chinese text escape run in a test table.

Reasoning

Continuation of the same Chinese UTF-8 test data. Benign.

info ~~warn~~ → info by review encoding-hex tests/test_tables.py:265:313 new in this release

18 consecutive `\xNN` escapes (entropy 3.9 bits/byte) (raw pass, confidence 0.65)
```
4\xe5\xbd\xb1\xe5\x93\x8d\xe4\xb8\x8d\xe5\xa4\xa7 \xe8\xbf\x9d\xe7\xba\xa6\xe9\xa3\x8e\xe9\x99\xa9\xe5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode()
```
likely benign LLM, confidence 0.94 UTF-8 Chinese text as byte literals decoded in test data.
Reasoning
Same Chinese UTF-8 byte literals decoded in a test fixture. Benign.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:265:413 new in this release

18 consecutive `\xNN` escapes (entropy 3.7 bits/byte) (raw pass, confidence 0.65)
```
e5\xbe\x88\xe4\xbd\x8e\n\xef\xbc\x8c'.decode(), b'\xe8\x81\x94\xe5\x90\x88\xe8\xb5\x84\xe4\xbf\xa1\xe8\xaf\x84\xe4\xbc\xb0\n\xe6\x9c\x89\xe9\x99\x90\xe5\x85\xac\xe5\x8f\xb8'.decode(), b'\xe7\xa8\xb3\xe5\xae\x9a'.decode(), ],
```
likely benign LLM, confidence 0.94 UTF-8 Chinese text as byte literals decoded in test data.
Reasoning
Same Chinese UTF-8 byte literals decoded in a test fixture. Benign.
info ~~warn~~ → info by review encoding-hex tests/test_tables.py:268:66 new in this release

18 consecutive `\xNN` escapes (entropy 3.8 bits/byte) (raw pass, confidence 0.65)
```
         [b'\xe5\xba\x8f\xe5\x8f\xb7'.decode(), b'\xe9\x87\x91\xe8\x9e\x8d\xe6\x9c\xba\xe6\x9e\x84\xe5\x90\x8d\xe7\xa7\xb0'.decode(), b'\xe6\x8e\x88\xe4\xbf\xa1\xe6\x80\xbb\xe9\xa2\x9d'.decode(), b'\xe5\xb7\xb2\xe4\xbd\xbf\xe7\x94\xa8\xe6\x83\x85\xe5\x86\xb5'.decode(), b'\xe5\x89\xa9\xe4\xbd\x99\xe9\xa2\x9d\xe5\xba\xa6
```
likely benign LLM, confidence 0.94 Chinese table column headers written as UTF-8 byte literals in a test.
Reasoning
Column headers such as 序号, 金融机构名称, 授信总额 are UTF-8 bytes decoded in a test table. Benign.
info ~~warn~~ → info by review payload-bytes-literal tests/test_tables.py:270:31 new in this release

bytes literal contains 36 `\xNN` escapes (100% of content) — binary payload shape (ast pass, confidence 0.85)
```
                        ['2', b'\xe5\xbe\xbd\xe5\x95\x86\xe9\x93\xb6\xe8\xa1\x8c\xe5\x8d\x97\xe4\xba\xac\xe4\xb8\xad\xe5\xb1\xb1\xe5\x8c\x97\xe8\xb7\xaf\xe6\x94\xaf\xe8\xa1\x8c'.decode(), '33043530\n, .', '25673530\n, .', '7370000\n, .', ],
```
likely benign LLM, confidence 0.93 Chinese bank name stored as a bytes literal in a test table row.
Reasoning
The bytes literal is Chinese bank-name text decoded immediately, next to ordinary numeric strings in a test table. No execution or hidden payload.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:270:33 new in this release

36 consecutive `\xNN` escapes (entropy 4.3 bits/byte) (raw pass, confidence 0.80)
```
                        ['2', b'\xe5\xbe\xbd\xe5\x95\x86\xe9\x93\xb6\xe8\xa1\x8c\xe5\x8d\x97\xe4\xba\xac\xe4\xb8\xad\xe5\xb1\xb1\xe5\x8c\x97\xe8\xb7\xaf\xe6\x94\xaf\xe8\xa1\x8c'.decode(), '33043530\n, .', '25673530\n, .', '7370000\n, .', ],
```
likely benign LLM, confidence 0.94 Chinese bank name as UTF-8 escapes in a test table.
Reasoning
Same Chinese bank-name UTF-8 bytes in a test fixture. Benign.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:271:33 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                        ['3', b'\xe5\x85\xb4\xe4\xb8\x9a\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '30260000\n, .', '11060000\n, .', '19200000\n, .', ],
```
likely benign LLM, confidence 0.94 Chinese bank branch name as UTF-8 escapes in a test table.
Reasoning
Chinese branch-name UTF-8 bytes decoded in a test table alongside numeric strings. Benign.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:272:33 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                        ['4', b'\xe6\xb0\x91\xe7\x94\x9f\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '26200000\n, .', '8850000\n, .', '17350000\n, .', ],
```
likely benign LLM, confidence 0.94 Chinese bank branch name as UTF-8 escapes in a test table.
Reasoning
Chinese bank branch name as UTF-8 escapes decoded in a test table. Benign.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:273:33 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                        ['5', b'\xe4\xb8\xad\xe4\xbf\xa1\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '20508800\n, .', '20508800\n, .', '000\n.', ],
```
likely benign LLM, confidence 0.93 Test fixture holding UTF-8-encoded Chinese bank names as byte escapes.
Reasoning
The escapes are UTF-8 bytes for Chinese bank-branch names (中信银行, 盐城分行 style text), decoded immediately with .decode(). This is test fixture data for table extraction, not obfuscation. The raw-pass entropy signal is a false positive on legitimate multibyte text.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:274:33 new in this release

24 consecutive `\xNN` escapes (entropy 4.1 bits/byte) (raw pass, confidence 0.80)
```
                        ['6', b'\xe5\x8d\x97\xe4\xba\xac\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '17517000\n, .', '17517000\n, .', '000\n.', ],
```
likely benign LLM, confidence 0.93 Test fixture with UTF-8 byte-escaped Chinese text, decoded to a string.
Reasoning
Same pattern: UTF-8 bytes for Chinese text decoded in a test table fixture. Repeated escapes are just how non-ASCII text is written in a bytes literal, and nothing executes or hides behaviour.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:278:34 new in this release

30 consecutive `\xNN` escapes (entropy 4.3 bits/byte) (raw pass, confidence 0.80)
```
                        ['10', b'\xe6\x81\x92\xe4\xb8\xb0\xe9\x93\xb6\xe8\xa1\x8c\xe5\x8d\x97\xe4\xba\xac\xe6\xb1\x9f\xe5\x8c\x97\xe6\x94\xaf\xe8\xa1\x8c'.decode(), '9390000\n, .', '4350000\n, .', '5040000\n, .', ],
```
likely benign LLM, confidence 0.93 Test data row with UTF-8 byte-escaped Chinese bank name.
Reasoning
A decoded UTF-8 Chinese string in a test data row alongside plain numeric strings. The high escape count reflects non-ASCII content, not a payload. The raw-pass signal is a false positive.
info ~~warn~~ → info by review encoding-escape-soup tests/test_tables.py:279:34 new in this release

24 consecutive `\xNN` escapes (entropy 4.3 bits/byte) (raw pass, confidence 0.80)
```
                        ['11', b'\xe4\xb8\x8a\xe6\xb5\xb7\xe9\x93\xb6\xe8\xa1\x8c\xe7\x9b\x90\xe5\x9f\x8e\xe5\x88\x86\xe8\xa1\x8c'.decode(), '4626500\n, .', '4626500\n, .', '000\n.', ],
                    ],
                ]
```
likely benign LLM, confidence 0.93 Test fixture row with UTF-8 byte-escaped Chinese text.
Reasoning
Same benign pattern: UTF-8 Chinese text as bytes, decoded in a table-extraction test fixture with ordinary numeric strings.

info high-complexity tests/test_tables.py:1117:19 new in this release

compression ratio 0.573 is >2σ above file mean 0.436 (raw pass, confidence 0.50)

        text = """|Column 1|Column 2|Column 3|\n|---|---|---|\n|Zelle (0,0)|**Bold (0,1)**|Zelle (0,2)|\n|~~Strikeout (1,0), Zeile 1~~<br>~~Hier kommt Zeile 2.~~|Zelle (1,1)|~~Strikeout (1,2)~~|\n|**`Bold-monospaced`**<br>**`(2,0)`**|_Italic (2,1)_|**_Bold-italic_**<br>**_(2,2)_**|\n|Zelle (3,0)|~~**Bold-strikeout**~~<

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:3:1

line length 7422 characters (raw pass, confidence 0.65)
```
var _self="undefined"!=typeof window?window:"undefined"!=typeof WorkerGlobalScope&&self instanceof WorkerGlobalScope?self:{},Prism=function(e){var n=/(?:^|\s)lang(?:uage)?-([\w-]+)(?=\s|$)/i,t=0,r={},a={manual:e.Prism&&e.Prism.manual,disableWorkerMessageHandler:e.Prism&&e.Prism.disableWorkerMessageHandler,util:{encode:
```
likely benign LLM, confidence 0.97 Minified vendored Prism.js syntax highlighting library.
Reasoning
Minified Prism syntax highlighter bundle with long lines is standard for minified vendored JS. The long-line signal is a formatting property, not obfuscation of malicious intent.
info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:4:1

line length 2850 characters (raw pass, confidence 0.65)
```
Prism.languages.markup={comment:{pattern://,greedy:!0},prolog:{pattern:/<\?[\s\S]+?\?>/,greedy:!0},doctype:{pattern:/<!DOCTYPE(?:[^>"'[\]]|"[^"]*"|'[^']*')+(?:\[(?:[^<"'\]]|"[^"]*"|'[^']*'|<(?!!--)|)*\]\s*)?>/i,greedy:!0,inside:{"internal-subset":{pattern:/(^[^\[]*\[)[
```
likely benign LLM, confidence 0.97 Prism markup language grammar definition in minified vendored library.
Reasoning
Prism language grammar definitions (markup) with regex patterns; long lines are from minification of a known library.

info high-complexity docs/_static/prism/prism.js:4:2390

compression ratio 0.891 is >2σ above file mean 0.656 (outside string/comment) (raw pass, confidence 0.25)

\s\S]+/,inside:{value:{pattern:/(^=\s*(["']|(?!["'])))\S[\s\S]*(?=\2$)/,lookbehind:!0,alias:[e,"language-"+e],inside:Prism.languages[e]},punctuation:[{pattern:/^=/,alias:"attr-equals"},/"|'/]}}}})}}),Prism.languages.html=Prism.languages.markup,Prism.languages.mathml=Prism.languages.markup,Prism.languages.svg=Prism.lang

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:7:1

line length 4611 characters (raw pass, confidence 0.65)
```
Prism.languages.javascript=Prism.languages.extend("clike",{"class-name":[Prism.languages.clike["class-name"],{pattern:/(^|[^$\w\xA0-\uFFFF])(?!\s)[_$A-Z\xA0-\uFFFF](?:(?!\s)[$\w\xA0-\uFFFF])*(?=\.(?:constructor|prototype))/,lookbehind:!0}],keyword:[{pattern:/((?:^|\})\s*)catch\b/,lookbehind:!0},{pattern:/(^|[^.]|\.\.\.
```
likely benign LLM, confidence 0.97 Prism JavaScript grammar definition in minified vendored library.
Reasoning
Prism javascript grammar definition; standard minified library content with no execution or encoded payload.

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:8:1

line length 6143 characters (raw pass, confidence 0.65)

!function(e){var t="\\b(?:BASH|BASHOPTS|BASH_ALIASES|BASH_ARGC|BASH_ARGV|BASH_CMDS|BASH_COMPLETION_COMPAT_DIR|BASH_LINENO|BASH_REMATCH|BASH_SOURCE|BASH_VERSINFO|BASH_VERSION|COLORTERM|COLUMNS|COMP_WORDBREAKS|DBUS_SESSION_BUS_ADDRESS|DEFAULTS_PATH|DESKTOP_SESSION|DIRSTACK|DISPLAY|EUID|GDMSESSION|GDM_LANG|GNOME_KEYRING_C

likely benign LLM, confidence 0.97 Prism bash/shell grammar definition in minified vendored library.

Reasoning

Prism bash grammar with keyword lists; long line from minification, no malicious behavior.

info long-line docs/_static/prism/prism.js:10:1

line length 6179 characters (mostly string/comment) (raw pass, confidence 0.39)

!function(e){function n(e,n){return e.replace(/<<(\d+)>>/g,(function(e,s){return"(?:"+n[+s]+")"}))}function s(e,s,a){return RegExp(n(e,s),a||"")}function a(e,n){for(var s=0;s<n;s++)e=e.replace(/<<self>>/g,(function(){return"(?:"+e+")"}));return e.replace(/<<self>>/g,"[^\\s\\S]")}var t="bool byte char decimal double dyn

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:11:1

line length 2596 characters (raw pass, confidence 0.65)

!function(e){var t=/\b(?:alignas|alignof|asm|auto|bool|break|case|catch|char|char16_t|char32_t|char8_t|class|co_await|co_return|co_yield|compl|concept|const|const_cast|consteval|constexpr|constinit|continue|decltype|default|delete|do|double|dynamic_cast|else|enum|explicit|export|extern|final|float|for|friend|goto|if|im

likely benign LLM, confidence 0.97 Prism C-like/C++ grammar definition in minified vendored library.

Reasoning

Prism C++ keyword grammar; minified vendored code with no suspicious operations.

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:12:1

line length 2770 characters (raw pass, confidence 0.65)

!function(e){var n=/\b(?:abstract|assert|boolean|break|byte|case|catch|char|class|const|continue|default|do|double|else|enum|exports|extends|final|finally|float|for|goto|if|implements|import|instanceof|int|interface|long|module|native|new|non-sealed|null|open|opens|package|permits|private|protected|provides|public|reco

likely benign LLM, confidence 0.97 Prism Java grammar definition in minified vendored library.

Reasoning

Prism Java keyword grammar; minified vendored code, benign.

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:15:1

line length 2113 characters (raw pass, confidence 0.65)

Prism.languages.python={comment:{pattern:/(^|[^\\])#.*/,lookbehind:!0,greedy:!0},"string-interpolation":{pattern:/(?:f|fr|rf)(?:("""|''')[\s\S]*?\1|("|')(?:\\.|(?!\2)[^\\\r\n])*\2)/i,greedy:!0,inside:{interpolation:{pattern:/((?:^|[^{])(?:\{\{)*)\{(?!\{)(?:[^{}]|\{(?!\{)(?:[^{}]|\{(?!\{)(?:[^{}])+\})+\})+\}/,lookbehind

likely benign LLM, confidence 0.97 Prism Python grammar definition in minified vendored library.

Reasoning

Prism Python grammar with regex patterns; standard minified library content.

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:17:1

line length 2939 characters (raw pass, confidence 0.65)

Prism.languages.swift={comment:{pattern:/(^|[^\\:])(?:\/\/.*|\/\*(?:[^/*]|\/(?!\*)|\*(?!\/)|\/\*(?:[^*]|\*(?!\/))*\*\/)*\*\/)/,lookbehind:!0,greedy:!0},"string-literal":[{pattern:RegExp('(^|[^"#])(?:"(?:\\\\(?:\\((?:[^()]|\\([^()]*\\))*\\)|\r\n|[^(])|[^\\\\\r\n"])*"|"""(?:\\\\(?:\\((?:[^()]|\\([^()]*\\))*\\)|[^(])|[^\\

likely benign LLM, confidence 0.97 Prism Swift grammar definition in minified vendored library.

Reasoning

Prism Swift grammar with regex patterns; standard minified library content.

info ~~warn~~ → info by review long-line docs/_static/prism/prism.js:18:1

line length 2896 characters (raw pass, confidence 0.65)
```
!function(){if("undefined"!=typeof Prism){var e=Object.assign||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n]);return e},t={"remove-trailing":"boolean","remove-indent":"boolean","left-trim":"boolean","right-trim":"boolean","break-lines":"number",indent:"number","remove-initial-line-feed":"boolean","tabs-t
```
likely benign LLM, confidence 0.96 Prism plugin configuration code in minified vendored library.
Reasoning
Prism plugin code (toolbar/normalize-whitespace style config) in a documentation static asset; no network, exec, or encoded payload visible.
info ~~critical~~ → info by review dynamic-execution docs/conf.py:73:29

`eval` at module scope (runs on import) on a non-literal expression (ast pass, confidence 0.80)
```
    python_versions_minor = m.group(1)
    python_versions_minor = eval(m.group(1))
else:
    # PyMuPDF version is set in setup.py, so we import it here.
```
likely benign LLM, confidence 0.80 eval applied to a regex capture group while parsing a version string in Sphinx config.
Reasoning
Sphinx conf.py evaluating a regex group from a version string at import time is a crude but common pattern for reading version info from a source file; the input is the project's own version text, not attacker-supplied. Context suggests a PyMuPDF version parse, so the risk is low, though eval on non-literal input is poor practice.

info long-line docs/page.rst:456:1

line length 564 characters (raw pass, confidence 0.50)

   .. method:: find_tables(clip=None, strategy=None, vertical_strategy=None, horizontal_strategy=None, vertical_lines=None, horizontal_lines=None, snap_tolerance=None, snap_x_tolerance=None, snap_y_tolerance=None, join_tolerance=None, join_x_tolerance=None, join_y_tolerance=None, edge_min_length=3, min_words_vertical=3

info unicode-mixed-script docs/samples/national-capitals.py:218:7

identifier `Nukuʻalofa` mixes scripts: Latin + Other (in string/comment) (raw pass, confidence 0.40)

Marshall Islands;Majuro;27797;66.1%;2011
Tonga;Nukuʻalofa;27600;26.0%;2022
Seychelles;Victoria;26450;24.8%;2010
French Polynesia (France);Papeete;26926;8.9%;2017

info ~~warn~~ → info by review dynamic-execution docs/samples/story-write.py:18:6

`open(__file__)` — file reads its own source, common pattern for extracting payloads hidden in comments (ast pass, confidence 0.85)
```
#
with open(__file__) as f:
    text = f.read()
text = html.escape(text)
```
likely benign LLM, confidence 0.85 Sample script reading its own source file to HTML-escape it for documentation output.
Reasoning
Sample script reads its own source to escape it into HTML for display; this is a documentation sample, not payload extraction. The message's claim about hidden payloads does not match the code, which only html-escapes the text.
info ~~critical~~ → info by review dynamic-execution setup.py:1046:9

`subprocess.run` with `shell=True` and a non-literal command — runtime-determined shell invocation (ast pass, confidence 0.85)
```
        log( f'Building MuPDF by running: {command}')
        subprocess.run( command, shell=True, check=True)
        log( f'Finished building mupdf.')
    
```
likely benign LLM, confidence 0.80 Build script invoking MuPDF build command through the shell with a constructed command string.
Reasoning
Build script runs a command string with shell=True to build MuPDF; this is expected in a native build setup script where the command is constructed locally from build configuration. Non-literal command warrants a look, but the context is a legitimate build step with logging of the command.
info ~~critical~~ → info by review dynamic-execution setup.py:1218:9

`subprocess.run` with `shell=True` and a non-literal command — runtime-determined shell invocation (ast pass, confidence 0.85)
```
        log( f'Building MuPDF by running: {command}')
        subprocess.run( command, shell=True, check=True)
        log( f'Finished building mupdf.')
    
```
likely benign LLM, confidence 0.85 setup.py runs a locally constructed build command through a shell to compile MuPDF.
Reasoning
The shell=True call runs a command string built by the MuPDF build script, which is a known build step in PyMuPDF's setup.py. The command is locally constructed and logged, with no sign of fetching remote content or hiding intent. The non-literal command is expected for a build driver, so this is a legitimate pattern, though it deserves a look in a supply-chain context.
info ~~warn~~ → info by review frame-introspection src/__init__.py:248:17

`inspect.stack` — call-stack frame introspection (ast pass, confidence 0.70)
```
    try:
        stack = inspect.stack(context=0)
    except StopIteration:
        pass
```
likely benign LLM, confidence 0.80 Library code calls inspect.stack to inspect the current call stack.
Reasoning
inspect.stack is used for caller-frame introspection, commonly for logging or locating the calling source location in a Python library. Nothing in the snippet indicates evasion or hidden execution.

info long-line tests/resources/simple_toc.txt:1:1

line length 1651 characters (raw pass, confidence 0.50)

[1, 'HAUPTÜBERSICHT', -1][1, 'Januar 01/2006', -1][2, 'SPEKTROGRAMM', -1][3, 'Urzeit-Godzilla', -1][3, 'Frühchristliche Mosaike im Knast', -1][3, 'Evolution auf Eis', -1][3, 'Entwarnung bei Kondensstreifen', -1][3, 'Spermatausch beim Schnecken-Sex', -1][3, 'Mehr Monde für Pluto', -1][3, 'Endlich ein Malaria-Impfstof

info long-line tests/test_geometry.py:342:1

line length 558 characters (raw pass, confidence 0.50)

    b = {'number': 0, 'type': 0, 'bbox': (403.3577880859375, 330.8871765136719, 541.2731323242188, 349.5766296386719), 'lines': [{'spans': [{'size': 14.0, 'flags': 4, 'font': 'SFHello-Medium', 'color': 1907995, 'ascender': 1.07373046875, 'descender': -0.26123046875, 'text': 'Inclusion and diversity', 'origin': (403.357

info ~~warn~~ → info by review dynamic-execution tests/test_story.py:120:22

`eval` called on a non-literal expression (ast pass, confidence 0.70)
```
        '''
        fit_result = eval(call)
        
        print(f'test_fit_springer(): {call=} => {fit_result=}.')
```
likely benign LLM, confidence 0.80 Test code uses eval on a test-generated call expression and prints the result.
Reasoning
eval evaluates a call string inside a test that prints the result. This is test scaffolding over strings the test controls, not a payload or obfuscated execution path.

info high-complexity tests/test_textextract.py:893:25

compression ratio 0.747 is >2σ above file mean 0.490 (raw pass, confidence 0.50)

    expected_1_23_5 = b'JOB No.: \nShipper (complete name and address) \xe5\x8f\x91\xe8\xb4\xa7\xe4\xba\xba(\xe5\x90\x8d\xe7\xa7\xb0\xe5\x8f\x8a\xe5\x9c\xb0\n\xe5\x9d\x80) \nSINORICH TRANSPORT LIMITED\nADD:7C,WEST BLDG.,ZHONGQU\nMANSION,211 ZHONGSHAN\nRD. SHANTOU,515041 CN\nTEL:0754-88570001 FAX:0754-88572709\nS/O No.

info ~~warn~~ → info by review payload-bytes-literal tests/test_textextract.py:896:29

bytes literal contains 57 `\xNN` escapes (55% of content) — binary payload shape (ast pass, confidence 0.85)
```
    expected_mupdf_1_26_1 = b'JOB No.: Shipper (complete name and address) \xe5\x8f\x91\xe8\xb4\xa7\xe4\xba\xba(\xe5\x90\x8d\xe7\xa7\xb0\xe5\x8f\x8a\xe5\x9c\xb0\xe5\x9d\x80)  Tel:                                  Fax: \n \nS/O No. \xe6\x89\x98\xe8\xbf\x90\xe5\x8d\x95\xe5\x8f\xb7\xe7\xa0\x81     \nSINORICH TRANSPORT LIM
```
likely benign LLM, confidence 0.90 Expected-output bytes literal of extracted shipping-document text in a textextract test.
Reasoning
This is an expected-output bytes literal for a PDF text-extraction test (MuPDF version comparison). It contains readable English and Chinese shipping-order text encoded as UTF-8. The AST payload-shape heuristic fires on the escape density, but the content is extraction ground truth, not an executable or binary payload.
info ~~warn~~ → info by review payload-bytes-literal tests/test_textextract.py:899:29

bytes literal contains 51 `\xNN` escapes (51% of content) — binary payload shape (ast pass, confidence 0.85)
```
    expected_mupdf_1_27_0 = b'JOB No.: \n \nS/O No. \xe6\x89\x98\xe8\xbf\x90\xe5\x8d\x95\xe5\x8f\xb7\xe7\xa0\x81   \nSINORICH TRANSPORT LIMITED \nSHIPPING ORDER \n\xe6\x89\x98\xe8\xbf\x90\xe5\x8d\x95 \n \xe5\xb8\x82\xe5\x9c\xba\xe9\x83\xa8: \n88570009 \n88577019 \n88572702 \n \xe6\x93\x8d\xe4\xbd\x9c\xe9\x83\xa8: \n885
```
likely benign LLM, confidence 0.90 Expected-output bytes literal for a text-extraction test with shipping-order text.
Reasoning
Another expected-output bytes literal for a text-extraction test, containing mixed English and Chinese document text. It is test data for comparing extractor output, and nothing in the snippet decodes, executes, or drops anything.

Reproduce This Scan

Run disclude locally to verify. Add --format sarif for SARIF output. Add --llm (with an API key in the environment) to include the LLM review.

cargo install disclude --version 4.0.0
git clone https://github.com/pymupdf/pymupdf pymupdf && cd pymupdf
git checkout 2.0
disclude scan --severity info --diff 1.28.2 .