Apigee as a IDE security plugin

An IDE security plugin analyzes code as you type, highlighting patterns that may lead to runtime issues. Its scope is limited to the files in your workspace and the rules bundled with the plugin. middleBrick operates as a black-box API scanner that evaluates live endpoints, not source code. It does not require access to your repository or build artifacts, and findings are based on what an external observer can observe through the network interface.

middleBrick maps findings to OWASP API Top 10 (2023), covering common risks observable without authentication and those requiring authenticated sessions. The scanner detects input validation issues such as CORS wildcard usage and dangerous HTTP methods, as well as data exposure indicators like API key formats and PII patterns. Where configuration permits, authenticated scans exercise methods such as GET and HEAD, and targeted POST for LLM probes, to surface issues like missing security headers and misconfigured JWT validation.

The platform parses OpenAPI 3.0, 3.1, and Swagger 2.0 documents, resolving recursive $ref references to build a coherent view of the API surface. It cross-references spec definitions against runtime observations, highlighting undefined security schemes, sensitive fields exposed in responses, deprecated operations, and missing pagination. This helps reconcile design intent with deployed behavior without requiring access to source repositories.

Authenticated scanning, available from Starter tier and above, supports Bearer, API key, Basic auth, and Cookie credentials. Domain verification is enforced through DNS TXT records or an HTTP well-known file to ensure only domain owners can scan with credentials. The scanner sends read-only methods, blocks private and metadata endpoints, and forwards a strict allowlist of headers. No remediation or code modification is performed; guidance is provided alongside each finding.

The scanner includes LLM / AI Security testing via 18 adversarial probes across three scan tiers: Quick, Standard, and Deep. These probes assess risks such as system prompt extraction, instruction override attempts, DAN and roleplay jailbreaks, data exfiltration paths, token smuggling, and multi-turn manipulation. Results highlight surface-level model and configuration weaknesses rather than implying a comprehensive evaluation of model behavior.

Findings are delivered through the Web Dashboard, CLI, and MCP Server, with support for JSON and text output formats. The platform integrates into CI/CD via GitHub Actions and provides scheduled rescans and diff detection in Pro tier. middleBrick does not fix, patch, block, or remediate issues, nor does it perform intrusive injection tests that require active exploitation. It surfaces findings relevant to audit evidence and helps you prepare for reviews, but it is not an auditor and cannot certify compliance.