SENTINEL
AI Code Review System
SENTINEL reads your pull requests like a senior architect — mapping blast radius across services, enforcing your team's architecture patterns, scoring risk across four dimensions, and delivering inline fixes with working code. Not pattern matching. System-level intelligence.
SENTINEL at a Glance
How SENTINEL Works
Submit
Paste a PR URL, code snippet, or receive from FOUNDRY pipeline
Map & Analyze
Maps blast radius, reads architecture context, runs OWASP security + org rules
Score
4-dimension risk scoring — Security, Quality, Architecture, Test Impact — per file and per PR
Review
Interactive code review with inline comments, fix suggestions, Ask AI, and SHIELD handoff
What Only SENTINEL Does
Six capabilities no other code review tool offers — because they require system-level intelligence, not pattern matching.
Blast Radius Analysis
When a PR touches a shared service, SENTINEL maps every downstream consumer — which endpoints, which tests, which other services are affected. Visualized as an interactive dependency graph.
CodeRabbit, Qodo, Copilot — none map cross-service impact
Architectural Drift Detection
SENTINEL models your codebase’s implicit patterns. When new code violates established architecture — raw SQL in handlers when you use repository pattern — it flags the drift, not just the syntax.
Linters catch syntax. SENTINEL catches design violations.
4-Dimension Risk Scoring
Not one number — four: Security exposure, Code quality delta, Architecture compliance, and Test coverage impact. Each scored 0–100 with per-file breakdown. Click any dimension to filter findings.
SonarQube has quality gates. Nobody has multi-dimensional PR risk.
SENTINEL → SHIELD Handoff
When SENTINEL finds a missing validation, it pre-generates a test specification and passes it to SHIELD. ‘Generate a test verifying webhook signature rejects forged payloads.’ One click.
Only possible because GyanMatrix owns both systems.
ARCHITECT Context Injection
If ARCHITECT documents event-driven architecture, SENTINEL enforces it. Code bypassing the message bus for direct HTTP calls gets flagged — not from a YAML rule, from the living architecture model.
No standalone review tool has architecture-level context.
Requirement Traceability
Connect Jira/Linear/GitHub Issues. SENTINEL pulls the linked ticket and verifies: does this PR implement the acceptance criteria? Did the developer miss a requirement? Review against spec, not just code.
Bito has early ticket validation. Nobody else attempts this.
All 22 Capabilities
Across 5 categories — Security, Intelligence, Custom Rules, Output, and Enterprise.
OWASP Top 10
SecurityAutomatic detection of SQL injection, XSS, CSRF, SSRF, broken auth, and all OWASP Top 10 vulnerability classes in every PR.
Secret & Key Scanning
SecurityDetects hardcoded API keys, tokens, passwords, and connection strings before they reach your main branch.
Dependency Vulnerabilities
SecurityScans added or updated dependencies against CVE databases. Flags known vulnerabilities with severity and remediation.
Auth & Access Audit
SecurityVerifies authentication middleware is applied, authorization checks are present, and access control patterns are consistent.
Blast Radius Mapping
SecurityMaps every downstream consumer affected by the change — endpoints, tests, services — visualized as an interactive dependency graph.
Inline Review Comments
IntelligenceContextual comments directly on the diff — not generic warnings, but specific observations tied to the exact line and surrounding logic.
4-Dimension Risk Score
IntelligenceSecurity, Quality, Architecture, and Test Impact — each scored 0–100 with per-file breakdown and interactive filtering.
Architecture Drift Detection
IntelligenceIdentifies when new code violates established codebase patterns — repository pattern bypassed, wrong layer accessed, implicit conventions broken.
Fix Suggestions with Code
IntelligenceNot just “this is wrong” — concrete fix suggestions with working code you can copy, paste, and commit.
Requirement Traceability
IntelligencePulls linked Jira/Linear/GitHub Issues and verifies acceptance criteria are implemented. Flags missing requirements.
Ask AI Follow-up
IntelligenceConversational follow-up on any finding. Ask why a risk was flagged, request alternative fixes, or dive deeper into blast radius.
YAML Org Rules
Custom RulesDefine your team’s standards in YAML — naming conventions, banned patterns, required middleware, import restrictions, and more.
Rule Compliance Report
Custom RulesEvery review includes a compliance summary showing which org rules passed, which failed, and which files are affected.
Pattern Violation Detection
Custom RulesGoes beyond linting — detects violations of architectural patterns, not just syntax rules. Repository pattern, service layer, event-driven conventions.
Interactive HTML Report
OutputFull review output as a navigable HTML report with collapsible sections, severity filters, and jump-to-file navigation.
Markdown & PDF Export
OutputExport any review as Markdown for wikis or PDF for compliance records and audit trails.
GitHub PR Comments
OutputPost review findings directly as GitHub PR comments with inline annotations, threaded discussion, and status checks.
SHIELD Test Handoff
OutputOne-click handoff to SHIELD — SENTINEL pre-generates test specifications for missing validations, edge cases, and security scenarios.
Historical Trends
EnterpriseTrack risk scores, common violations, and security posture over time. See whether review quality is improving across sprints.
Team Analytics
EnterprisePer-developer and per-team review metrics — common issues, fix acceptance rate, risk score trends, and improvement areas.
Complete Audit Trail
EnterpriseEvery review, every finding, every developer response — logged and searchable for compliance, audits, and retrospectives.
Compounding Intelligence
EnterpriseSENTINEL learns your codebase patterns over time. Review #51 is sharper than review #1 — fewer false positives, better context.
Why SENTINEL, Not Generic Review Tools
Other tools match patterns. SENTINEL understands your system.
SENTINEL is System 3 of 7
SENTINEL reviews code from FOUNDRY, triggers SHIELD for tests, and feeds findings to CHRONICLE for documentation.
Ready to review code through SENTINEL?
Paste a PR URL. Get a senior engineer-grade review in seconds.
Try Live Demo →