Skip to content

Latest commit

 

History

History
58 lines (46 loc) · 2.39 KB

File metadata and controls

58 lines (46 loc) · 2.39 KB

Roadmap

DeployProof is an open-source preview-parity toolkit. Its job is narrow: compare the observable HTTP contract of a preview deployment with production before a change is merged.

Product principles

  1. Compare, do not guess. The same bounded request is sent to preview and production. Findings describe observed drift rather than speculative risk.
  2. Private by construction. Reports never include response bodies, cookie values, custom request-header values, or raw query values.
  3. Useful without an account. The CLI, library, and GitHub Action work locally or in a user's own CI environment.
  4. Low-noise defaults. Behaviorally equivalent values are normalized before comparison.
  5. Provider-neutral core. Deployment providers supply URLs; the comparison engine remains portable.

Current focus

  • Make the GitHub Action easy to adopt after any preview deployment.
  • Maintain tested recipes for GitHub deployments, Vercel, Cloudflare, and Netlify.
  • Expand deterministic normalization only when a real equivalent-behavior case is demonstrated by a fixture.
  • Improve structured output for CI systems and library consumers.
  • Publish threat-model, benchmarking, and integration documentation.

Next

  • Reusable workflow examples for common monorepo and multi-site layouts.
  • Optional comparison policies that map stable rule IDs to severity without hiding findings.
  • Machine-readable JSON Schema for configuration and results.
  • A benchmark suite for bounded reads, redirect chains, and large route sets.
  • More end-to-end fixtures covering provider-specific redirect and header behavior.

Later, if adoption supports it

  • Additional CI providers and deployment platforms.
  • A plugin interface for new deterministic checks.
  • Historical result comparison through user-owned artifacts.
  • A separately evaluated hosted offering only if users demonstrate a need that cannot be met safely in their own CI.

Non-goals

  • Uptime or synthetic monitoring
  • Browser automation and visual regression testing
  • Mutation of preview or production targets
  • Collection of response bodies or credentials
  • AI-generated findings without deterministic evidence
  • A hosted account, billing, or telemetry requirement for core functionality

Roadmap items are directional, not release commitments. Open a feature request with a concrete use case before investing in a large implementation.