pullrequests Search Results · language:Dune language:Python language:Java language:HTML language:HTML language:Java language:Java
Filter by
114M results
Adds a testing skill documenting how to run and test the Battleship game end-to-end, including key test flows for
gameplay, bomb feature, AudioContext error handling, and hard AI difficulty.
Devin Session: ...
devin-ai-integration[bot]
🤖 AI Security Remediation
🟢 Auto-merge eligible — will merge after CI passes.
📊 Analysis
- Severity: MEDIUM
- Confidence: 85%
- Auto-fix allowed: True
📝 Summary
The application uses the subprocess ...
ai-remediation
severity-medium
gangiiopsdev
- Records being prepared by the system (prefill, file/context generation) previously had to sit in blocked, but
check-files auto-unblocks blocked as soon as input files appear — any legitimate check-files ...
radionest
🤖 AI Security Remediation
🟢 Auto-merge eligible — will merge after CI passes.
📊 Analysis
- Severity: MEDIUM
- Confidence: 85%
- Auto-fix allowed: True
📝 Summary
The application uses the subprocess ...
ai-remediation
severity-medium
Summary
P1 harness bug surfaced by the live run-4 dogfood. scripts/_reviewer_wrapper.py _reviewer_cmd built the claude -p
reviewer argv in space-form:
claude -p --allowedTools X --disallowedTools ...
lyanpark2019
Summary
- initialAmount was permanently immutable after account creation — the UPDATE SQL excluded it, so clients couldn t
correct a wrong opening balance without deleting and recreating the account. ...
mnovelodou
picture source media= (prefers-color-scheme: dark) srcset= https://static.trunk.io/assets/merge_test_pr_banner_dark.svg
source media= (prefers-color-scheme: light) srcset= https://static.trunk.io/assets/merge_test_pr_banner_light.svg ...
trunk-staging-io[bot]
Automated test pull request
aws-aemilia-pdx
决策与证据
Operator 决策（2026-06-12）：gross_profit 采用 Yahoo 标准化值。证据：在唯一可三方对照的公司上（01810 小米，by-function 列报有真实毛利行），AKShare = Yahoo = PDF
完全一致（76,560,194 千 RMB）；而 by-nature 公司（00001）PDF 根本没有毛利行——旧 pdf_required 闸门要求的是不可能完成的验证（M5 ...
xinghuolk
Judges previously inherited DEFAULT_MODEL, so a provider experiment (e.g. ADAPTIVE_LEARNING_MODEL=deepseek-reasoner)
would also swap the graders, making scores incomparable across runs. Judges now default ...
abhijeetgupta23