Know what you are shipping. Review local changes, catch bugs and security risks, and bring a clear, tested change to your next pull request.
50 free reviews per month for individual developers. Plan details
Claude Code
GitHub
GitLab
Bitbucket
Payment retry review
def retry_charge(order):
return gateway.charge(
amount=order.total,
customer=order.customer_id
# no stable retry key
)The gateway creates the charge, but its response times out. Retrying without the original key could create a second charge.
PAY-184: retry a payment request without creating a duplicate charge.
The gateway call has no stable idempotency key tying the retry to the original payment attempt.
Pass the original attempt's key on every retry. Use a new key only for a new payment attempt.
Test response timeouts and duplicate requests against the gateway's actual idempotency contract.
Use review feedback throughout the task, then give your teammate the evidence they need to make a decision.
Bring the ticket or acceptance criteria into the coding session. Identify the expected behavior, affected code, and cases that must keep working.
Use Inspect for a focused check of common bugs and essential security issues. Choose Review for a broader pass. Findings include a location, severity, explanation, and suggested fix to help you choose a correction and verify it with tests.
Link the requirement, explain the important tradeoff, and include test results and remaining risks. Request a PR review and involve a maintainer when domain judgment is needed.
Get specific feedback on what deserves another pass. You decide what to change and verify the result.
Compare the change with the requested behavior. A task-completion check makes missing or partial requirements visible before you call the work done. Linked tickets require connected issue access; an inaccessible ticket is not a successful validation.
Look for unsafe inputs, edge cases, and broken failure handling. Generate unit tests and supporting documentation for the change. Inspect and run the tests, including regression cases: generated tests are not evidence of a passing result.
Review secrets, access boundaries, and unsafe data flows across applications and AI tools. Follow findings back to the code and validate the proposed remediation.
Challenge duplicated logic, unused abstractions, and changes that conflict with established patterns. Check callers, dependencies, and abstraction boundaries for architectural impact. Use AI slop findings as reasons to investigate, not a verdict on their own.
Review should reflect the repository you are working in, not just generic coding advice.
Apply explicit instructions such as tenant-scoped queries or required error handling. Related code and tests help explain why the rule matters here.
Review and save rules proposed from selected past PRs. Developer feedback helps refine the review rather than repeating advice the team has already resolved.
Your team's selected checklist defines quality, security, and test expectations. Warning-mode failures report issues; Error-mode failures can block merges when the host's required checks are configured.
Choose your editor or repository workflow. Local review helps before commit; PR review keeps feedback with the shared change.
Claude Code
Install the CloudAEye plugin and complete its connection setup. Inspect or review your pending diff, then ask the agent to address the findings you select.
Plugin setupConfigure CloudAEye's MCP server in a supported coding client. Follow that client's setup and authentication steps; available commands depend on the integration.
MCP setupConnect your GitHub, GitLab, or Bitbucket repository. Request a review in the PR comments or enable automatic reviews. Add issue references when checking task alignment.
Repository setupStart with your next change. 50 free reviews per month.
Start free
Claude Code
cloudaeye:inspect
cloudaeye:review
/check_task [#42]
@cloudaeye /inspect
@cloudaeye /review
@cloudaeye /review [#42]