Loading Forky
LOADING FORKE...
Changelog

Everything we've actually shipped.

Hand-curated, not auto-generated from commits — an auto-derived log fills up with dependency bumps and tells you nothing. One entry here is one user-visible shipping moment, often spanning several repos.

September 2, 2026

1 change
Update

High-Performance Go Backend Engine & Oracle Cloud VM Deployment

Completely re-architected the Forke backend infrastructure by transitioning from monolithic Next.js server actions to a dedicated high-performance Go microservices backend (forke-backend at api.forke.space). Deployed natively on an Oracle Cloud Infrastructure (OCI) Linux VM instance, the Go engine delivers sub-5ms API response times, handles high-concurrency task claims via goroutines, and features automated CI/CD deployment pipelines on every git push.
High-Performance Go Backend Engine & Oracle Cloud VM Deployment
ENLARGE
Improvements
  • Re-engineered core task routing, claim validation, and profile aggregation from Next.js server actions into a high-concurrency Go REST API.
  • Configured automated GitHub Actions CI/CD workflows executing binary builds and zero-downtime SSH deployments to Oracle Cloud Infrastructure VM.
  • Added automated environment toggles for Swagger/OpenAPI documentation (api.forke.space/swagger) with strict production gating.

August 30, 2026

1 change
Update

Cross-Subdomain Session Management & First-Party Attribution Engine

Rolled out unified cross-subdomain cookie architecture across forke.space, dashboard.forke.space, and admin.forke.space. Users can now log in once and seamlessly navigate across the entire platform ecosystem without repeated authentication challenges. Alongside this, our privacy-preserving attribution middleware now captures visitor campaign sources, organic search engines, and referral paths using SHA-256 IP hashing without setting third-party tracking cookies.
Improvements
  • Configured cross-subdomain session cookies (.forke.space) to unify authentication across marketing, dashboard, and administrative surfaces.
  • Implemented dynamic attribution middleware mapping search engine referrers (Google, Bing, DuckDuckGo, Ecosia) and direct UTM campaign tags.
  • Added automated IP salt rotation with 24-hour cryptographic hashing to ensure strict visitor privacy compliance.

August 28, 2026

1 change
Fix

Persistent Device Attribution Deduplication & Edge Caching Fixes

Addressed edge cases in visitor telemetry where rapid consecutive page navigations or multi-tab browsing triggered redundant visit logs. The tracking engine now uses client-side localStorage device fingerprint tokens combined with an in-memory 5-minute deduplication buffer to ensure accurate unique visitor metrics without inflating analytics counts.
Fixes
  • Fixed an issue where opening multiple tabs simultaneously caused duplicate session entries in the conversion funnel.
  • Corrected Edge Middleware URL rewrite loops when visitors accessed nested dashboard routes directly from external campaign links.
  • Resolved a hydration mismatch error in the navigation bar CTA button state during server-to-client rendering handoffs.

August 24, 2026

1 change
Feature

Sandbox Code Playground & Live In-Browser Execution

Shipped a full-fledged in-browser development environment built directly into the developer dashboard. Powered by WebContainer technology, contributors can now test micro-task solutions, execute test suites, and debug code with zero local installation overhead. The sandbox provides instant terminal output, Monaco editor integration, and real-time syntax checking for TypeScript, Python, and Go.
Improvements
  • Integrated isolated WebContainer runtimes for TypeScript, Python, and Go with real-time process monitoring.
  • Built a live streaming terminal component with ANSI color support, custom scrollbars, and instant stderr/stdout piping.
  • Added one-click task workspace branch cloning directly into Sandbox for frictionless prototyping before committing.
Fixes
  • Resolved terminal buffer overflow during long test output runs and improved stream flushing on memory-constrained devices.

August 20, 2026

1 change
Polish

Custom High-Contrast Scrollbars & Platform Typography Refinements

Polished the global design system across the marketing site, developer dashboard, and administrative console. Replaced default browser scrollbars with custom macOS-inspired sleek trackbars featuring theme-aware accent color highlights and smooth opacity transitions on hover. Typography across all headings and cards has been re-tuned for enhanced contrast and readability on OLED displays.
Improvements
  • Designed ultra-thin custom scrollbars that adapt dynamically between dark and light workspace themes.
  • Standardized monospaced font sizing and line-heights across code blocks, ledger lines, and terminal components.
  • Enhanced border contrast ratios across dark mode bento cards to improve accessibility and visual hierarchy.

August 15, 2026

1 change
Feature

Automated Task Claim Locking & Mirror Branch Provisioning

Introduced an anti-sniping task claim mechanism to protect developer focus. When a contributor claims a bounty, the system places it into a 20-minute soft reservation window. Once the developer begins coding, an isolated mirror branch (dev-{username}/task-{id}) is automatically provisioned on the Forke GitHub mirror repository, ensuring client source repos remain untouched until PR review approval.
Improvements
  • 20-minute claim lock with active countdown timer and heartbeat checks in the developer navigation bar.
  • Automatic mirror branch provisioning on GitHub with isolated permissions and zero write access to client master branches.
  • Added warm queue support allowing up to 3 developers to waitlist on high-demand bounties if the primary claimant lapses.

August 10, 2026

1 change
Feature

Automated AI Pull Request Review & Plain-English Verdict Engine

Pull request verification is now completely automated through a multi-stage review pipeline. When a developer submits a PR, our review engine automatically executes baseline test assertions, lint checks, and file scope verification before triggering a Claude-powered code analysis. Repository owners receive a concise, plain-English verdict card summarizing the diff, security considerations, and quality score for one-click merge approval.
Automated AI Pull Request Review & Plain-English Verdict Engine
ENLARGE
Improvements
  • Automated FORKE_SUBMISSION.md verification checking that all acceptance criteria and file scope boundaries are respected.
  • Generates structured plain-English verdict cards with security flags, architectural recommendations, and code quality scoring.
  • Sandboxed test execution validating build compilation and test pass rates before notifying the repo owner.
Fixes
  • Fixed race conditions during simultaneous webhook deliveries for multi-commit pull request updates.

August 5, 2026

1 change

July 28, 2026

1 change
Feature

25-Tier Developer Level Progression & Gamified XP Ranking

Rolled out the platform experience ranking system spanning 25 distinct level tiers, from Newcomer to Commit Warlord. Completing bounties awards XP based on complexity, turnaround velocity, and code cleanliness. Developers unlock higher-tier bounties, exclusive platform titles, and auto-generated OpenGraph social preview cards with embedded verification QR codes.
Improvements
  • 25-tier level progression with custom titles (LVL 1 Newcomer to LVL 25 Commit Warlord).
  • Velocity multipliers providing bonus XP for sub-hour PR turnarounds and zero-defect submissions.
  • Unlocks high-value bounty tiers and priority task claim windows for veteran platform contributors.

July 22, 2026

1 change
Fix

Postgres Connection Pooling & Database Query Latency Fixes

Optimized database connection management across serverless Edge and Node.js runtimes. Implemented transaction-level connection pooling with PgBouncer, tuned idle timeout thresholds, and converted heavy analytical aggregation queries to utilize pre-computed materializations, reducing database round-trip latency by 70%.
Fixes
  • Resolved intermittent "too many clients already" errors during traffic spikes on the marketing waitlist.
  • Fixed query timeouts on large subscriber export operations in the admin console.
  • Implemented automatic retry with exponential backoff on transient database disconnection events.
01 / 03