werkrevisor reviews werkcrappit #45
No Assignees
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: mi/werkcrappit#45
Reference in New Issue
Block a user
See docs/prs/2026-10-07-PR#45-werkrevisor.md.
🤖 Generated with Claude Code
12f3f46c96to751e76c29c🟠 werkrevisor ORANGE for
751e76c, superseded by the result for09a6e73🟠 werkrevisor: ORANGE / FINDINGS
Change kind:
mixed.Reasons
Findings
[medium] secret_scan_exclude comment contradicts the setting at
.werkrevisor.toml:32-33, ORANGE, as a line comment[medium] "Afterwards" section of the PR document is unclear at
docs/prs/2026-10-07-PR#45-werkrevisor.md:24-27, hint, as a line comment[low] Awkward English in the problem statement at
docs/prs/2026-10-07-PR#45-werkrevisor.md:10, hint, as a line commentClassifier
4 of 4 units passed all questions (1 PR document(s)).
Usage: model `openrouter/minimax/minimax-m3`, LLM done, tokens counted before the call 10875, calls 1, tokens in 10534 / out 24379 (reasoning included in out), duration 214.1 s, cost 0.0259 USD (reported), classifier `jev-1.13.0` calls 4, tokens in 2801, cost 0.0001 USD, finding judge `jev-1.13.0` calls 2, tokens in 2766, cost 0.0001 USD, config: defaults + host.1 answer ignored because of a confidence below [classifier] min_confidence.
Reviewed head `751e76c29c3dcc94df9e811e8994479aa9934dd8` against base `22c49c58f2ca8d7eba1a14abdbee586e673bb54e`, werkrevisor 0.21.1.
werkrevisor: ORANGE / FINDINGS for
751e76c29c3dcc94df9e811e8994479aa9934dd8.The result comment of werkrevisor holds the details.
@@ -0,0 +29,4 @@"/build.gradle.kts", "werkcrappit-*/build.gradle.kts", "/settings.gradle.kts", "/gradle/**", "/gradle.properties",]secret_scan = true# Test data holds fake passwords; private keys and the real key formats stay flagged there too.[medium] secret_scan_exclude comment contradicts the setting (werkrevisor)
ORANGE; Jev: real 0.67, severity 2.7 of 3
Line 32 of
.werkrevisor.tomlreads "Test data holds fake passwords; private keys and the real key formats stay flagged there too", claiming that private keys and real key formats stay under secret scanning in test data. Thesecret_scan_excludesetting on line 33 is["**/src/test/**", "**/src/functionalTest/**"], which excludes every file under those directories from secret scanning. The two statements disagree: a real.pemor private key committed intosrc/test/resources/will not be detected. Either the comment is misleading, or the exclude is too broad. I have not read werkrevisor's own source, so it is possible the tool has a separate always-scan list for key files that overrides the exclude; if so, the comment is correct and the config is fine, but the visible comment and config are not in sync.The comment was correct: in files of
secret_scan_exclude, werkrevisor still flags real key formats (private keys, AWS key ids,hsak_andsk-or-v1-keys), see the werkrevisor README onsecret_scan_exclude; only generic assignments such aspassword = "..."are ignored there.Clarified in
09a6e73: the comment now names the key formats that still count.@@ -0,0 +7,4 @@Since werkrevisor PR#41, werkrevisor reviews the pull requests of werkcrappit.Its rules for werkcrappit are in the werkrevisor repository, so a change of them is not visible here,and nobody of werkcrappit has to approve it.[low] Awkward English in the problem statement (werkrevisor)
hint
Line 10 reads "and nobody of werkcrappit has to approve it." "Nobody of werkcrappit" is not idiomatic English; the natural phrasing is "no one from werkcrappit" or "nobody in werkcrappit". The phrase sits in the first paragraph a reviewer reads, so it weakens the clarity of the problem statement.
Fixed in
09a6e73: "no one from werkcrappit".@@ -0,0 +21,4 @@`.werkrevisor.toml` checks the writing style's limits of 128 characters per line and 5 sentences per paragraph.- No change to the plugins, no new version.## Afterwards[medium] "Afterwards" section of the PR document is unclear (werkrevisor)
hint; Jev: real 0.78, severity 1.4 of 3
The "## Afterwards" section (lines 24-27 of the PR document) says the host config of werkcrappit "keeps only the part of the host, like
config/werkrevisor.tomlof werkrevisor". "The part of the host" is vague; the intended meaning seems to be "the host-specific part (model, budget, gate, engine)". The comparison to werkrevisor's ownconfig/werkrevisor.tomldoes not state what should or should not live in werkcrappit's host config. The classifier also flagged this section as hard to understand (uncertain). The sentence should list concretely what the host config keeps and what it leaves to.werkrevisor.toml.Fixed in
09a6e73: the section lists what the host config keeps: model, budget, gate, engine and the Jev endpoint.Review round answered in
09a6e73: the secret-scan finding was a misreading, the comment now says it explicitly; both doc hints fixed. All threads resolved. The PR changes.werkrevisor.tomland therefore waits for the approval of @mi in any case.🟡 werkrevisor: YELLOW / HUMAN_REVIEW
Change kind:
mixed.Reasons
Classifier
4 of 4 units passed all questions (1 PR document(s)).
Usage: model `openrouter/minimax/minimax-m3`, LLM failed (TimeoutError), tokens counted before the call 10975, duration 726.1 s, cost unknown, classifier `jev-1.13.0` calls 4, tokens in 2807, cost 0.0001 USD, config: defaults + host.1 answer ignored because of a confidence below [classifier] min_confidence.
Reviewed head `09a6e73b2a5fdc1b67c47be3acba99f67aadaa49` against base `22c49c58f2ca8d7eba1a14abdbee586e673bb54e`, werkrevisor 0.21.1.
ai-reviewer referenced this pull request2026-10-09 07:20:03 +02:00