Automatic · every commit to main

Every commit, tested in the open.

Each new TgWitness commit runs the whole suite on a clean machine: the signed journal and its verifier, evidence exports and reports, the panel in a real browser, sessions against synthetic Telegram clients. Larger runs drive real test accounts on a test server. Every result is published here, failures included.

2,178tests passed out of 2,255 on commit 3a614fb780cd · Detect secret chats the peer endedFinished 2026-11-14 08:29 UTC

Pick up the commitEach new commit on main, checked out exactly as published
Run the full suite2,255 tests in 137 files, browser checks included
Drive real accountsCapture, export and verify on test accounts and test chats
Publish every resultTotals, each failure with its message, raw JUnit and logs

Tests

2,255

137 files

Passed

2,178

96.6%

New failures

turn the run red

Known failures

listed with reasons

Skipped

not applicable here

Duration

17 min

one machine

COVERAGE

What the suite checks

Tests grouped by the part of TgWitness they exercise; the bar is the share that passed.

Reports and analysis45 files

763 / 778 · 15 known

Journal, signatures and timestamps31 files

469 / 481 · 12 known

Evidence exports18 files

337 / 338 · 1 known

Panel, API and site18 files

239 / 240 · 1 known

Other4 files

145 / 155 · 10 new

Accounts and sessions15 files

132 / 134 · 2 known

Capture add-on6 files

93 / 94 · 1 known

FAILURES

Every failure, in full

New failures turn the run red. Known failures are listed with the reason they are expected until fixed.

tools/test_proxy_source_settings.py10 new

test_invalid_settings_prevent_refresh_before_network
Error message
```
Failed: DID NOT RAISE ValueError
```
test_permission_error_never_restores_enabled_defaults
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[[]]
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[\xff]
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[]
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[null]
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[{"countries":"invalid"}]
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[{"github":{"enabled":[]}}]
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[{"unknown":true}]
Error message
```
Failed: DID NOT RAISE ValueError
```
test_unreadable_settings_never_restore_enabled_defaults[{]
Error message
```
Failed: DID NOT RAISE ValueError
```

tools/test_verifier_sources.py9 known

test_frozen_verification_decisions_are_unchanged
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+2 more lines; full report in element text]
```
test_old_and_new_decisions_and_statuses_match[pinned---json]
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+3 more lines; full report in element text]
```
test_old_and_new_decisions_and_statuses_match[pinned---plain]
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+5 more lines; full report in element text]
```
test_old_and_new_decisions_and_statuses_match[unmet_tier---json]
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+3 more lines; full report in element text]
```
test_old_and_new_decisions_and_statuses_match[unmet_tier---plain]
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+5 more lines; full report in element text]
```
test_old_and_new_decisions_and_statuses_match[unpinned---json]
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+3 more lines; full report in element text]
```
test_old_and_new_decisions_and_statuses_match[unpinned---plain]
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+5 more lines; full report in element text]
```
test_old_verifier_accepts_new_archive_with_identical_signed_events
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
assert False
[+3 more lines; full report in element text]
```
test_review_modules_import_without_application_state
The verifier's readable answer was shortened (93976cde9): plain output starts with EVIDENCE instead of ASSURANCE, and frozen JSON decisions differ in note wording. Frozen fixtures need regeneration.
Error message
```
AssertionError: assert False
[+2 more lines; full report in element text]
```

tools/test_deletion_semantics.py8 known

test_deletion_of_subset_keeps_actor_and_whole_history_unknown[10--1000000000555]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```
test_deletion_of_subset_keeps_actor_and_whole_history_unknown[10-55123]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```
test_deletion_of_subset_keeps_actor_and_whole_history_unknown[15--1000000000555]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```
test_deletion_of_subset_keeps_actor_and_whole_history_unknown[15-55123]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```
test_deletion_of_subset_keeps_actor_and_whole_history_unknown[2--1000000000555]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```
test_deletion_of_subset_keeps_actor_and_whole_history_unknown[2-55123]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```
test_deletion_of_subset_keeps_actor_and_whole_history_unknown[9--1000000000555]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```
test_deletion_of_subset_keeps_actor_and_whole_history_unknown[9-55123]
Expects the batch-deletion caveat "does not identify who deleted them, whether deletion applied to all participants…", shortened in 3f5f0fbcb. Pending decision: restore the caveat or update the test.
Error message
```
AssertionError: assert 'does not identify who deleted them' in 'Telegram reported several message IDs as deleted in one update.'
```

tools/test_case_package.py2 known

test_package_contains_readable_report_files_and_explanations[ots]

Expects the heading "Timestamp receipts (not verified)", renamed in the case package.

Error message

assert ('Timestamp receipts (not verified)' in '<!doctype html><html lang="en"><meta charset="utf-8"><meta http-equiv="Content-Security-Policy" content="default-src ...lank"></span></p><p>Date: <span class="blank"></span></p><p>Place: <span class="blank"></span></p></div></body></html>')

test_package_contains_readable_report_files_and_explanations[tsr]

Expects the heading "Timestamp receipts (not verified)", renamed in the case package.

Error message

assert ('Timestamp receipts (not verified)' in '<!doctype html><html lang="en"><meta charset="utf-8"><meta http-equiv="Content-Security-Policy" content="default-src ...lank"></span></p><p>Date: <span class="blank"></span></p><p>Place: <span class="blank"></span></p></div></body></html>')

tools/test_session_import.py2 known

test_independent_login_browser[False]
The sign-in form no longer has the "other sign-in methods" block (.other-sign-in, removed in 3f5f0fbcb); the browser test still follows the old flow.
Error message
```
playwright._impl._errors.TimeoutError: Locator.evaluate: Timeout 30000ms exceeded.
Call log:
  - waiting for locator(".other-sign-in")
```
test_independent_login_browser[True]
The sign-in form no longer has the "other sign-in methods" block (.other-sign-in, removed in 3f5f0fbcb); the browser test still follows the old flow.
Error message
```
playwright._impl._errors.TimeoutError: Locator.evaluate: Timeout 30000ms exceeded.
Call log:
  - waiting for locator(".other-sign-in")
```

tools/test_verifier_dispatch.py2 known

test_direct_check_preserves_policy_and_archive_without_printing[tgwitness.verifier.cli]
Same frozen verifier decisions as test_verifier_sources.
Error message
```
AssertionError: assert False
[+2 more lines; full report in element text]
```
test_direct_check_preserves_policy_and_archive_without_printing[tgwitness.verify]
Same frozen verifier decisions as test_verifier_sources.
Error message
```
AssertionError: assert False
[+2 more lines; full report in element text]
```

addons/capture/tests/test_trusted_anchors.py1 known

test_cloudflare_queries_are_read_only_scoped_and_token_is_not_exported
Expects "no proof of continuous presence"; the capture add-on now says "provider state sampled at the polled time".
Error message
```
AssertionError: assert 'no proof of continuous presence' in 'provider state sampled at the polled time'
```

tools/test_analyzer_graphs.py1 known

test_edges_cite_journal_records_across_chosen_chats
Expects the contact-graph limit "does not mean there was no contact", removed after v0.9.3.
Error message
```
assert False
 +  where False = any(<generator object test_edges_cite_journal_records_across_chosen_chats.<locals>.<genexpr> at 0x7101b5b0b6b0>)
```

tools/test_analyzer_linking.py1 known

test_rotated_account_ranks_first_with_explanation_and_citations

Expects the account-linking limit "not identification", removed after v0.9.3.

Error message

assert False
 +  where False = any(<generator object test_rotated_account_ranks_first_with_explanation_and_citations.<locals>.<genexpr> at 0x7101b638d2f0>)

tools/test_chat_export_integrity.py1 known

test_sparse_message_numbers_are_counted_without_expansion
Missing-number counting now returns (numbers, count) instead of a sized collection; test not yet updated to the new API.
Error message
```
assert 2 == 1999999998
 +  where 2 = len((<tgwitness.message_numbers.MissingNumbers object at 0x7101a37e14c0>, 1))
```

tools/test_cli.py1 known

test_local_commands

Expects the default workspace name "Main account"; it is now "Main workspace".

Error message

AssertionError: assert [{'workspace'...n workspace'}] == [{'workspace'...ain account'}]
  
  At index 0 diff: {'workspace': 'default', 'name': 'Main workspace'} != {'workspace': 'default', 'name': 'Main account'}
  Use -v to get more diff

tools/test_device_attribution.py1 known

test_the_dossier_counts_clients_sticker_sets_and_prepared_media
Expects the wording "Metadata hypothesis about the client that encoded the file"; the dossier now says "Client that encoded the file, read from the file's own metadata".
Error message
```
assert 'Metadata hypothesis about the client that encoded the file' in "Client that encoded the file, read from the file's own metadata. A forwarded or re-uploaded file carries the encoder of whoever made it first."
```

tools/test_mtproto_forensics.py1 known

test_outbox_read_watermark_does_not_identify_a_person
Expects the caveat that a read watermark does not identify a person, shortened in 3f5f0fbcb.
Error message
```
AssertionError: assert 'does not identify' in 'Observed outgoing read watermark.'
```

tools/test_verifier_comparison.py1 known

test_ots_structure_only_does_not_get_a_positive_indicator[unknown]
Expects an OTS-structure-only receipt to be labelled unknown; it is now labelled synthetic.
Error message
```
AssertionError: assert 'unknown' in 'synthetic'
```

tools/test_watchlists.py1 known

test_terms_patterns_wallets_usernames_and_domains_are_matched
Expects the watchlist-hit caveat "nothing about who wrote it", removed after v0.9.3.
Error message
```
assert False
 +  where False = all(<generator object test_terms_patterns_wallets_usernames_and_domains_are_matched.<locals>.<genexpr> at 0x7101b6be35a0>)
```

LIVE

Real accounts on a test server

Long runs with disposable test accounts and our own test chats: capture, export, independent verification. Account names, chats and message texts are not published. All live runs as JSON.

Finished	Scenario	Result	What was checked
2026-11-14 08:09 UTC	Online status after actions	passed	With hide-online on, the peer saw the account go offline 1.2 s after a panel action and stay offline. With it off (the default), the same action showed the account online for at least 40 s to anyone allowed to see its last seen.
2026-11-14 08:01 UTC	Secret chat close, after the fix	passed	Telegram does not notify the side that accepted a secret chat when the peer ends it (confirmed with two plain clients). Since 3a614fb78 TgWitness checks accepted chats with an invisible protocol no-op: the close was recorded 61 s after it happened, with the last time the chat was known open. The peer saw no typing, read or message.
2026-11-14 06:46 UTC	Secret chat close	failed	Closing on either side was not delivered by Telegram within 3 minutes: the peer saw no update. TgWitness records the close when it next tries to send, with the reason 'Telegram reports the chat no longer exists'.
2026-11-14 06:39 UTC	Ghost behaviour	passed	No read receipt for a direct message or for secret messages; no typing indicator before sending in a group or a secret chat; read-only mode refused a send and nothing reached the peer.
2026-11-14 06:34 UTC	Secret chats, both directions	passed	Protocol module installed from GitHub with its SHA-256 checked; an incoming request waited for manual accept; texts in both directions recorded exactly; timer, screenshot and deletion notices recorded with the earlier text kept; a PNG saved with the same SHA-256; no decryption failures.
2026-11-14 06:29 UTC	Online status after actions	not observable	The test account hides last seen from everybody, so the peer only sees 'recently'; whether hide-online-after-actions works could not be observed from outside.
2026-11-14 05:36 UTC	Journal export and offline verification	passed	A live test account exported its 67 MB signed journal; the bundled verifier exited 0; HTML and PDF reports were generated, including chats over time.
2026-11-14 04:53 UTC	Same text in two test chats	passed	An identical message posted to two of our test chats was listed once under repeated texts, linked to both journal records.

HISTORY

Run after run

One column per tested commit, newest on the right. Select a column or a row for the full run. Archive of all 6 runs.

passedknown failuresnew failures

Finished	Commit	Result	Passed	Failed
2026-11-14 08:29 UTC	`3a614fb78` Detect secret chats the peer ended	New failures	2,178	42
2026-11-14 07:42 UTC	`c06a2ba5a` Improve jurisdiction guidance and report recovery (#306)	Passing, known failures listed	1,909	152
2026-11-14 06:58 UTC	`0911462bc` Show the chosen chats and the account on the last setup step (#305)	Passing, known failures listed	1,912	79
2026-11-14 06:35 UTC	`db86ca0b4` Show a chat's charts and assessment on its first export page only (#304)	Passing, known failures listed	1,912	79
2026-11-14 06:24 UTC	`23c7cc65e` Label service messages in reports and tidy the chat list and account counters (#	Passing, known failures listed	1,912	79
2026-11-14 06:11 UTC	`29741904d` Show chat export progress by messages and translate its states (#301)	New failures	1,880	111

Raw results: tgwitness-tests-3a614fb780cd.tar.gz (all three) · summary.json · junit.xml · pytest.log · archive of all runs