getdebug

getdebug is an AI code-review and auto-remediation service that continuously watches a Git repository for bugs, regressions, and security vulnerabilities, then generates a validated patch and opens a pull request for human review. Instead of stopping at a SARIF report the way most SAST tools do, getdebug produces a diff, runs the project's own build and test signals against that diff, and only proposes the change once it has evidence the fix compiles and does not obviously regress behavior. The engineer stays in the merge loop — nothing lands on main without approval.

The target user is a shipping product team that already has CI but does not have the bandwidth for a dedicated appsec or code-quality function. Typical workflows include continuous background scanning of one or more tracked repos, on-demand deep scans of a branch before a release, triage of committed secrets (Pro Plus adds git-history provenance to identify the first-seen commit), and cross-file refactor suggestions surfaced as reviewable PRs. Findings persist as stateful suppressions so the same false positive does not reappear every scan.

Pricing is per active seat with the first seat free, so solo developers and evaluators can use the paid tiers without paying, and pending invites are never billed. A free tier covers small teams that bring their own LLM API key; paid tiers include hosted inference. An adjacent public benchmark, codesecbench.org, is published by the same team and is used to calibrate the underlying detectors. getdebug is developer-facing infrastructure rather than a chatbot — the value shows up as merged PRs, not as prompts you have to write.