Code triage
Sentinel follows data through your dependencies, auth flows and business logic, so findings reflect real exploit paths instead of regex matches.
Powered by Anthropic Claude 3.5 Sonnet
Yoru Sentinel Security audits that ship the fix
Continuous code security and remediation powered by deep static analysis. Sentinel reads your repos and DNS, flags what is actually exploitable, and opens a pull request with the patch.
Early access for engineering teams. No spam.
$ yoru scan --repo github.com/acme/payments-api --depth deep
⟳ Connecting to Claude 3.5 Sonnet…
✓ Engine ready · model=claude-3-5-sonnet · latency=312ms
⟳ Indexing 847 files, building AST graph…
✗ HIGH src/routes/checkout.ts:142 SQL injection via user_id
user_id is concatenated into a raw query string
- db.query(`SELECT * FROM orders WHERE user_id = ${req.body.id}`)
+ db.query('SELECT * FROM orders WHERE user_id = $1', [req.body.id])
⚠ MED dns: mail.acme.io is missing SPF and DMARC records
✓ Patch drafted · report/summary.pdf generated
$ yoru remediate --apply checkout-sqli --open-pr
✓ Pull request #284 opened, awaiting review
findings: 2high: 1medium: 1patched: 1
Core capabilities
Built to find real bugs, then fix them.
Static analysis, live infrastructure checks and Claude's reasoning in one pipeline. Made for small teams that ship fast and can't afford a security backlog.
Patch generation
Every fix arrives as a pull request with the diff, a short risk note and the checks it ran. Review it like any other PR.
Infrastructure and DNS recon
Tracks open ports, HTTP security headers, TLS configuration and DNS records. You get an alert when something drifts.
Compliance reports
Each scan produces a short summary for leadership and evidence mapped to SOC 2 and ISO 27001. No spreadsheets.
How it works
From first scan to merged fix
Three steps. Most teams get their first findings within an hour.
-
01
Connect a repo or domain
Link GitHub or GitLab, or point us at a public domain. Nothing to install.
-
02
Sentinel analyzes it
Code, dependencies and infrastructure are checked together. Findings are ranked by how exploitable they are.
-
03
Review and merge the fix
You get an alert in Slack or through a webhook. Open the patch, review it, merge.
Where Claude fits in
Ingestion workers normalize repos and recon data. The reasoning layer sends Claude structured prompts with AST snippets and earlier findings. Results feed the patch generator, the report builder and your webhooks.
- → Repo ingest · semantic chunking · secret redaction
- → Triage · severity scoring · exploit context
- → Diff synthesis · PR open · compliance PDF
// pipeline.ts async function audit(target) { const graph = await ingest(target); const findings = await claude.analyze({ model: "claude-3-5-sonnet", context: graph, vectors: ["code", "dns", "headers"] }); return remediate(findings); }
Pricing
Simple pricing while we're early
Founding rates for early teams. Lock in Pro pricing before general availability.
Developer
$0forever
For solo builders checking an early product.
- 3 repository scans per month
- Standard severity models
- Basic summary report (PDF)
- Community support on Discord
Recommended
Team
$149per month
Continuous auditing for startups with production traffic.
- Unlimited audit reports
- Continuous repo and domain monitoring
- Patch pull requests generated for you
- Webhooks, Slack and CI gates
- Onboarding call with our team