Reads the diff
Code Sentinel analyzes changed files, surrounding context, and the pull request description before it responds.
AI pull request intelligence
Code Sentinel
Automated pull request review with inline feedback, change summaries, and best-practice recommendations powered by configurable models through OpenRouter.
Inline comments
Risk summaries
Model routing
42 sec
median first pass
3 modes
summary, inline, advisory
OpenRouter
choose the model per repository
Pull request workflow
Review signal where teams already work.
Ranks the risk
Findings are grouped by severity so reviewers can separate true regressions from style suggestions.
Leaves the comment
Inline notes, concise summaries, and best-practice guidance land directly on the pull request.
Reviewer-grade output
Sharper context before human review starts.
The page stays focused on what maintainers need most: what changed, what might break, and what deserves a second look before merge.
feature/model-router
Review ready
Add repository-level model selection
Configurable intelligence
Route each review to the model that fits.
Fast passes
Use lighter models for summaries, typo catches, and low-risk refactors.
Deep reviews
Send security-sensitive or high-churn changes to stronger reasoning models.
Team policy
Keep review behavior aligned with repository needs and cost targets.
OpenRouter
Switch providers without rewriting the review workflow.
Ready for the next pull request
Give every PR a first responder.
Code Sentinel catches risk early, gives reviewers a clean starting point, and keeps feedback close to the code.
Request Access
Token values are logged when the OpenRouter request fails. Mask the header before returning the error payload.
Add a fallback for repositories without an explicit model configuration.
Extract the prompt template into a named helper before adding another review mode.