Stop fighting your AI.
Start shipping with it.

You are a senior security engineer. Review the following {{LANGUAGE}} code for:
1. Injection vulnerabilities (SQL, XSS, command injection)
2. Insecure data exposure or hardcoded secrets
3. Missing input validation
4. Unsafe dependency usage

Return findings as: [SEVERITY: LOW|MED|HIGH] — Line X — Issue — Fix.
Only report confirmed issues. Do not suggest style changes.

Code:
{{CODE_BLOCK}}

[HIGH] — Line 34 — SQL query built with string concatenation — use parameterized queries.
[MED] — Line 67 — API key exposed in response payload — strip from output before serialization.
[LOW] — Line 12 — User-controlled redirect path not validated — whitelist allowed paths.

Analyze this error as a senior backend engineer. Identify the root cause, not just symptoms.

Error: {{ERROR_MESSAGE}}
Stack trace: {{STACK_TRACE}}
Context: {{WHAT_YOU_WERE_DOING}}
Environment: {{RUNTIME}} {{VERSION}}, {{OS}}

Respond with:
1. Root cause (1–2 sentences, plain language)
2. Why it happens (technical explanation)
3. Minimal fix (code snippet)
4. How to prevent it in future

Root cause: The async iterator exhausts before the downstream consumer is ready — classic backpressure failure in Node.js streams.
Fix: Add highWaterMark: 0 and implement explicit flow control with pause()/resume()...

Act as a staff engineer at a high-scale startup. Review my proposed architecture:

System: {{SYSTEM_DESCRIPTION}}
Expected load: {{RPS}} req/s, {{DATA_SIZE}} data
Current stack: {{TECH_STACK}}
Constraints: {{BUDGET_TIMELINE_CONSTRAINTS}}

Identify: single points of failure, bottlenecks, cost traps, scaling cliffs.
Suggest the 3 highest-leverage changes only. Ignore premature optimizations.

1. Scaling cliff at ~800 RPS: Your synchronous DB calls will bottleneck before your load balancer. Add a connection pool (PgBouncer) now.
2. Cost trap: NAT gateway charges will exceed EC2 costs at scale — move to VPC endpoints...
3. SPOF: Redis is your only session store with no replica...