Skip to content
BugFence

Security Whitepaper & Privacy Architecture

Honest About What We Protect. And What We Don't.

BugFence is architected so sensitive data is neutralized before it ever leaves the reporter's browser. Below is our exact security model, cryptography, and transparent technical boundaries.

Data Ingestion Pipeline

How Data Flows From Browser to Your Private VPC

Unlike cloud recording services, raw unmasked DOM snapshots are never transmitted over the wire.

01

DOM Masking

Passwords, credit card forms, and elements with data-bugfence-mask are masked prior to canvas rasterization.

02

Pattern Scrubbing

Regex engine scrubs JWTs, Bearer tokens, API secret keys, emails, SSNs, and Luhn-valid card numbers from URLs and headers.

03

Reporter Consent

The user reviews a live Privacy Check showing replaced values and can toggle console, network, video, and replay on or off.

04

VPC Ingestion

Encrypted transfer directly to your self-hosted API. Replay payloads are re-scrubbed server-side and stored in your database.

Telemetry & Context Collected

Payload elements assembled by the widget

  • Title, description and optional voice transcript
  • Optional screenshot, video and DOM replay
  • Console and error entries, failed network requests, breadcrumbs
  • Browser, OS and viewport
  • Reporter id, name and email (email handling is configurable: keep, hash or drop)
  • Release metadata and correlation ids

Technical Controls & Hardening

Defense in depth across browser and server

  • Password and card inputs are masked by default; data-bugfence-mask and data-bugfence-block fence anything else
  • Tokens, emails, card numbers, SSNs and phone numbers are scrubbed from text, URLs and headers in the browser, then re-scrubbed on the server
  • Replay is re-scrubbed server-side and rendered in a script-less sandboxed iframe
  • Integration tokens are encrypted with AES-256-GCM; secret keys are stored as hashes; passwords use scrypt
  • Org-scoped auth, CORS allowlist, rate limiting and SSRF guards; the server refuses to start in production with weak secrets
  • AI triage is optional; prompts are a minimal sanitised text subset and provider keys stay on the server

Documented Limitations & Boundaries

We believe in transparent engineering: hear our limitations directly from us.

Detection is pattern-based and best-effort. Names, addresses and free-form text can slip through.
Screen video recordings include in-studio region redaction tools, privacy trimming, and pre-submission reporter review.
Reports, including media, are stored without application-level encryption. Use disk or volume encryption.
Tenant isolation is enforced in application logic, not separate storage.
There is no full audit log yet, and SSO availability should be confirmed for your plan.
We do not claim SOC 2, ISO 27001 or similar certifications.

Where Your Data Lives

BugFence is deployed entirely within your infrastructure (Docker, Kubernetes, or Node.js). Bug reports, screenshots, and logs remain in your PostgreSQL database or file store.

Zero central servers. Zero third-party telemetry. We do not operate multi-tenant subprocessors for your bug report data.

Coordinated Vulnerability Disclosure

We take software security seriously. If you discover a vulnerability in BugFence, please report it responsibly so our team can patch it swiftly.

Read our RFC 9116 security policy at /.well-known/security.txt

Need a Security Questionnaire or Architecture Review?

Our team regularly assists enterprise security, compliance, and DevSecOps teams with VPC deployment architecture.