cyberk-dev/Ultrareview

AI-powered code review: spec-aware Intent injection + GitNexus graph context + multi-step agent loop. Posts line-level comments on PRs.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
ai-api-keyAPI key for AI proxy (cliproxy)yes
ai-base-urlAI proxy base URLnohttps://api.openai.com
analysis-modelModel for deep analysisnogpt-5.4
judge-modelModel for bug judging (should differ from analysis model)nogpt-5.2
classify-modelModel for bug classificationnogpt-5.4-mini
min-confidenceMinimum confidence threshold (0-1)no0.7
min-judge-scoreMinimum judge score threshold (0-1)no0.6
timeout-minutesPipeline timeout in minutes (default 10, set 30 for thorough analysis)no10
feedbackEnable self-improving feedback loop (saves review data for prompt optimization)notrue
gitnexus-enabledEnable GitNexus code graph context (requires npx gitnexus in PATH)notrue
gitnexus-secret-scanScan .gitnexus/ index for leaked secrets after indexing (fails with warning, never blocks PR)notrue
intent-enabledEnable Intent injection: inject PR title/body and classified spec files (OpenSpec, CK-Plan, docs) at the top of the analyzer promptnotrue
flow-diagram-enabledEmbed a Mermaid flowchart of the bot's understanding at the top of the PR review summary (one extra cheap LLM call per PR)notrue

no outputs