Starter
Try AI PR review.
- 2 reviewers
- Code Review
- Security Review
- 3 seats included
- 15,000 credits to start, then 5,000/mo
- About 70 reviews to start, then about 25 a month
- Unlimited repos
Flags admin actions with no audit log, disabled CI gates and exposed secrets.
HighKnown issue in the SOC 2 Review benchmark · caught by DeepSeek V4.1 FlashSee the graded case
Real IonWarp comments on open-source pull requests, graded against the benchmark's answer key.
Missed byGPT 5.6 Luna Pro 2DeepSeek V4.1 Flash 0DeepSeek V4 Pro 2Muse Spark 1.3 3
2 of 4 · See the benchmark
P1[C1] Password-reset bearer token logged in plaintext at debug level · SOC 2 Review
When resolvePasswordResetToken is called with an expired, invalid or userless token, the raw reset token is written to debug logs ({ token }). The reset token is a bearer credential that lets its holder change the account password; anyone with log access can replay it within its 20m TTL to take over the account.
HighKnown issue in the SOC 2 Review benchmark · caught by GLM 5.3 FlashSee the graded case
P1[C1] Elasticsearch credentials in REPO_INDEXER_CONN_STR are written to logs · SOC 2 Review
When an operator configures REPO_INDEXER_TYPE=elasticsearch with credentials in the connstr (the documented example is http://elastic:changeme@...), the trace log (elastic_search.go:62, enabled at DEBUG/TRACE) and the init-failure log.Fatal in indexer.go:134 print the raw connection string including the password to production logs.
HighKnown issue in the SOC 2 Review benchmark · caught by GLM 5.3 FlashSee the graded case
GLM 5.3 Flash is its best model on this benchmark; SOC 2 Review runs DeepSeek V4.1 Flash unless you pick another.
IonWarp's models and the review apps, graded on the same open-source pull requests.
| PRs | |||||
|---|---|---|---|---|---|
| IonWarp SOC 2 · GLM 5.3 Flash | |||||
| IonWarp SOC 2 · GPT 5.6 Luna Pro | |||||
| IonWarp SOC 2 · DeepSeek V4.1 Flash | |||||
| IonWarp SOC 2 · DeepSeek V4 Pro | |||||
| IonWarp SOC 2 · Muse Spark 1.3 | |||||
| CodeRabbit | |||||
| Cursor Bugbot | |||||
| Greptile v4.1 | |||||
| Qodo v2 |
Why IonWarp runs DeepSeek V4.1 Flash: in a trial of our specialist reviewers on production pull requests on 23 September 2026 it filed about as many findings per review as GLM 5.3 Flash at high effort (0.93 against 1.02) for a third of the cost ($0.039 against $0.125), and 3 of its 64 findings were refuted.
One workspace price with seats included. Credits pool across your team. Unlimited repos on every plan.
SOC 2 Review is on Max: $149 a month with 10 seats and a 7-day trial.
Try AI PR review.
Get AI PR review right from day 1.
Scale AI PR review.
Custom reviewers + credits · GitHub Enterprise · SSO/SAML + audit logs · Custom security review + SLA
P0[C1] Raw submitted Basic-auth public key written to production logs unmasked · SOC 2 Review
When the secret-hash lookup resolves a key whose publicKey differs from the submitted Basic-auth username, the new warning interpolates the raw `publicKey` (the caller-supplied username from the Authorization header) directly into the log message. The username is never validated to be a `pk-lf-...` public key, so a misconfigured or partially-rotated client that puts its secret key in the public-key position (or otherwise places a credential there) has that value written to production logs in cleartext, where it is retained and readable by anyone with log access. The repo's own security checklist requires secrets to be omitted from logs, and the codebase already ships a masking idiom for this purpose. The added test at api-auth.servertest.ts:361 even asserts the raw submitted value is present in the log, locking in the unmasked behavior.