SecEng Workbench · Map
SecEng Surface Scanner
Find every AI vendor, SDK, runtime, and shadow model before it becomes an attack surface.
Surface Scanner runs across your browser, repos, and IDE — detecting AI model providers, agent frameworks, vector stores, widgets, and inference endpoints from signals alone. Every hit resolves against a versioned vendor catalog and flows into a structured AI asset register.
Platforms
Chrome extension
In-page side panel, one-click scan, 120+ vendor catalog
GitHub scanning
Repo import, dependency graph, SDK fingerprinting
VS Code extension
Inline flags, real-time catalog resolution, workspace scan
CLI / API
Scripted surface scans, CI/CD integration, JSON output
Capabilities
What Surface Scanner does.
Browser AI fingerprinting
The Chrome extension captures DOM selectors, script URLs, globals, cookies, local storage, network requests, and headers — resolving every AI signal against a versioned vendor catalog without touching internal APIs.
GitHub repository scanning
Scans repos for AI SDK imports, framework usage, config patterns, and model references. Identifies LangChain, LlamaIndex, OpenAI client, HuggingFace, AutoGen, and 60+ other libraries from code structure alone.
VS Code workspace analysis
Flags AI dependencies, risky patterns, and unreviewed SDK additions inline as you code. Integrates with the shared savvy-stacks catalog for real-time vendor resolution inside the editor.
AI vendor registry resolution
Every detected signal resolves against a versioned catalog of 120+ AI vendors, model families, agent frameworks, vector stores, guardrails, eval harnesses, and inference runtimes — one canonical source.
Shadow AI detection
Surfaces undocumented AI endpoints, embedded widgets, runtime globals, and SDK bundles that don't appear in engineering docs or supply-chain manifests — found by behavioral signal analysis.
Attack surface register export
Generates a structured AI asset inventory with vendor, family, confidence, trust boundary, owner, and risk tags — ready for governance, audit, or red-team handoff to the Attack phase.
Live demo
See a surface discovery run — rendered from a real repo fixture.
The demo walks through a GitHub repo scan of an AI platform: 112 signals detected, vendor catalog resolution, RAG boundary analysis, and a structured risk output — the same format used in red-team handoffs.
Output format
Other Map instruments