Changelog

Every update to GridWork HQ, documented. New features, improvements, and fixes shipped to production.

v2.0.0

SaaS LaunchApril 2026

Added

Multi-tenant SaaS architecture — every org gets isolated data, RLS-enforced at the database layer
Better Auth v1.x with organization plugin — invite flow, RBAC (owner/admin/member), seat caps
Stripe billing — Solo $79/mo, Team $249/mo, 14-day free trial, credit wallet with atomic deductions
45 AI pipelines with visual kanban editor, node-level cost tracking, and real-time SSE streaming
Pipeline versioning — side-by-side diff, restore to any version, stable/testing tags
Smart client memory — 3-layer system (facts, rules, learned observations) injected into pipeline prompts
Output viewer — Tiptap rich-text editor, markdown roundtrip, auto-save, diff view, PDF export
16 integrations: Slack, Discord, Resend, Google Sheets/Calendar/Drive, HubSpot, Jira, Calendly, QuickBooks, Twilio, Linear, Zapier, Make, Generic Webhook, OpenAI, Ollama
Onboarding checklist with 5-step guided setup and role selection
Day 0/1/3/7/12/14 email sequence via Resend
Analytics: spend by model/pipeline/period, revenue vs cost margin, CSV export
Audit log with filterable DataTable, expandable rows, CSV export
Real-time activity feed panel (SSE from activity_log)
Settings: billing portal, credit history, team management, integration status, notification preferences
Deployed to Vercel (dashboard) + Fly.io (pipeline server) + Neon (Postgres) + Upstash (Redis)

Security

Row-level security on all 12 tenant tables — org_id isolation enforced at Postgres layer
PyJWKClient JWKS verification on pipeline server with org_id header claim matching
HKDF per-tenant encryption for OAuth tokens
Opaque path_token webhook routing (no org IDs in URLs)
Rate limiting keyed by org_id, Jinja2 SandboxedEnvironment for prompt templates
Sentry error monitoring across all 3 services with alert rules

Removed

Self-hosted distribution — GridWork HQ is SaaS-only from v2.0
Docker bundle — replaced by Vercel + Fly.io managed deployment
SQLite — replaced by Neon Postgres 16 with multi-tenant RLS

v1.3.0

Postgres + Python Pipeline ServerApril 2026

Upgrade note: Run `bunx drizzle-kit migrate` after pulling v1.3.0 to apply the Postgres 16 schema. Update your .env with DATABASE_URL pointing to your Postgres instance. SQLite is no longer supported (superseded by Postgres 16).

Added

Python FastAPI pipeline server replaces the Node.js pipeline server
Better Auth v1.x replaces NextAuth — email/password auth, organization plugin, RBAC (owner/admin/member), invite flow, and tier gating
Cloud deployment for all services (gridwork-hq + pipeline-server)
Ollama integration for local AI inference (qwen2.5-coder:7b, qwen3:8b)

Changed

Migrated from SQLite (superseded by Postgres 16) to Postgres 16 (WAL mode) across all three schemas: pipeline, chat, hq
Unified deployment replaces standalone service instructions — single command brings up the full stack

v1.2.1

Launch PolishMarch 28, 2026

GridWork HQ dashboard with skeleton loaders and design token system

Added

Cmd+K command palette for quick navigation across all pages
Skeleton loaders on 7 dashboard pages for instant perceived performance
Empty states on 5 pages with contextual guidance for new users
Breadcrumb navigation on all nested views
File type icons in Knowledge Vault (20+ extensions recognized)
Trend arrows on stat cards showing week-over-week direction
Resizable panels on Mission Control and Knowledge Vault
Gauge severity colors on Operations monitor health cards
Grouped settings layout with collapsible integration sections
Login page gradient background with animated GridWork HQ wordmark

Improved

226 design tokens migrated from hardcoded hex to gw-* CSS custom properties
Two-tier contrast system: AA for body text, AAA for headings and interactive elements
Inter font loaded via next/font with Latin subset optimization
Table hover states and row selection across Clients, Pipelines, and Cron pages
Kanban drag-and-drop feedback with drop zone highlighting and card shadows
Model name truncation on pipeline cards (prevents layout overflow on long model IDs)

v1.2.0

Production ReadyMarch 10, 2026

Upgrade note: After pulling v1.2 updates, run `bun run migrate` in gridwork-hq to apply SQLite (superseded by Postgres 16) schema changes.

GridWork HQ system monitor showing health checks and pipeline status

Added

SQLite-first storage via Drizzle ORM (superseded by Postgres 16) (Notion becomes optional, not required)
Demo mode with one-click sample data seeding and clean removal
Cloud-hosted deployment with automated provisioning
Setup wizard: 5-phase guided configuration with health checks and placeholder substitution
Vercel data loss warning banner when SQLite (superseded by Postgres 16) is detected on serverless
CI/CD workflow with type checking, linting, and build verification
Distribution packaging script with ZIP exclusion rules
Health check endpoint on pipeline server at /health

Improved

Complete rebrand from internal codename to GridWork HQ across all files
65 API routes now require authentication via getServerSession()
Shell injection eliminated: all execSync calls replaced with execFileSync + arg arrays
52 CSS custom properties (gw-*) replace every hardcoded color value
Zero hardcoded hex values remain outside shadcn primitives

Security

CSP nonce extended to style-src for inline style protection
SSE rate limiting: 20 connections per minute per IP on pipeline server
Path traversal guards on Knowledge Vault file operations
Cron command allowlist prevents arbitrary shell execution

v1.1.0

AI PipelinesMarch 1, 2026

GridWork HQ Mission Control with split-panel layout and AI chat

Added

17 AI-powered pipelines at launch: prospect, outreach, audit, internal-audit, propose, plan, build, brand, seo, content, report, follow-up, spec, kb-librarian, archive-outputs, friday-update, scope-audit (Note: Pipeline count was 17 at time of v1.1 release. Current count: 45.)
Document compiler with 7 output types (proposal, report, plan, audit, spec, brand guide, content brief)
Cron engine with 12 scheduled jobs for automated operations
Operations monitor with 6 real-time health cards (CPU, memory, disk, pipelines, cron, SSE)
Mission Control: split-panel layout with pipeline launcher and AI chat sidebar
Pipeline server with SSE streaming for real-time job progress
Telegram notifications for pipeline completions and cron failures
49 knowledge vault templates with {{AGENCY_*}} placeholder system

v1.0.0

FoundationFebruary 15, 2026

Added

Dashboard with 8 stat cards showing clients, revenue, pipelines, and activity
Client Tracker with filter tabs (active, prospect, archived) and inline editing
Pipeline Kanban with 5-column drag-and-drop (queued, running, review, done, failed)
Knowledge Vault: git-backed file explorer with create, edit, and delete
Finance Panel with revenue charts and invoice tracking
Settings page with 10 integration configurations
Reports page with exportable pipeline and client summaries
Cron Jobs dashboard with status indicators and manual trigger
NextAuth with GitHub OAuth for single-user authentication (superseded by email/password auth)
Notion, Stripe, Vercel, and Google Analytics 4 integrations
Full security headers (HSTS, X-Frame-Options, X-Content-Type-Options, Referrer-Policy)
WebP image optimization across all static assets

Changelog

Every update to GridWork HQ, documented. New features, improvements, and fixes shipped to production.

v2.0.0

SaaS LaunchApril 2026

Added

Multi-tenant SaaS architecture — every org gets isolated data, RLS-enforced at the database layer
Better Auth v1.x with organization plugin — invite flow, RBAC (owner/admin/member), seat caps
Stripe billing — Solo $79/mo, Team $249/mo, 14-day free trial, credit wallet with atomic deductions
45 AI pipelines with visual kanban editor, node-level cost tracking, and real-time SSE streaming
Pipeline versioning — side-by-side diff, restore to any version, stable/testing tags
Smart client memory — 3-layer system (facts, rules, learned observations) injected into pipeline prompts
Output viewer — Tiptap rich-text editor, markdown roundtrip, auto-save, diff view, PDF export
16 integrations: Slack, Discord, Resend, Google Sheets/Calendar/Drive, HubSpot, Jira, Calendly, QuickBooks, Twilio, Linear, Zapier, Make, Generic Webhook, OpenAI, Ollama
Onboarding checklist with 5-step guided setup and role selection
Day 0/1/3/7/12/14 email sequence via Resend
Analytics: spend by model/pipeline/period, revenue vs cost margin, CSV export
Audit log with filterable DataTable, expandable rows, CSV export
Real-time activity feed panel (SSE from activity_log)
Settings: billing portal, credit history, team management, integration status, notification preferences
Deployed to Vercel (dashboard) + Fly.io (pipeline server) + Neon (Postgres) + Upstash (Redis)

Security

Row-level security on all 12 tenant tables — org_id isolation enforced at Postgres layer
PyJWKClient JWKS verification on pipeline server with org_id header claim matching
HKDF per-tenant encryption for OAuth tokens
Opaque path_token webhook routing (no org IDs in URLs)
Rate limiting keyed by org_id, Jinja2 SandboxedEnvironment for prompt templates
Sentry error monitoring across all 3 services with alert rules

Removed

Self-hosted distribution — GridWork HQ is SaaS-only from v2.0
Docker bundle — replaced by Vercel + Fly.io managed deployment
SQLite — replaced by Neon Postgres 16 with multi-tenant RLS

v1.3.0

Postgres + Python Pipeline ServerApril 2026

Added

Python FastAPI pipeline server replaces the Node.js pipeline server
Better Auth v1.x replaces NextAuth — email/password auth, organization plugin, RBAC (owner/admin/member), invite flow, and tier gating
Cloud deployment for all services (gridwork-hq + pipeline-server)
Ollama integration for local AI inference (qwen2.5-coder:7b, qwen3:8b)

Changed

Migrated from SQLite (superseded by Postgres 16) to Postgres 16 (WAL mode) across all three schemas: pipeline, chat, hq
Unified deployment replaces standalone service instructions — single command brings up the full stack

v1.2.1

Launch PolishMarch 28, 2026

Added

Cmd+K command palette for quick navigation across all pages
Skeleton loaders on 7 dashboard pages for instant perceived performance
Empty states on 5 pages with contextual guidance for new users
Breadcrumb navigation on all nested views
File type icons in Knowledge Vault (20+ extensions recognized)
Trend arrows on stat cards showing week-over-week direction
Resizable panels on Mission Control and Knowledge Vault
Gauge severity colors on Operations monitor health cards
Grouped settings layout with collapsible integration sections
Login page gradient background with animated GridWork HQ wordmark

Improved

226 design tokens migrated from hardcoded hex to gw-* CSS custom properties
Two-tier contrast system: AA for body text, AAA for headings and interactive elements
Inter font loaded via next/font with Latin subset optimization
Table hover states and row selection across Clients, Pipelines, and Cron pages
Kanban drag-and-drop feedback with drop zone highlighting and card shadows
Model name truncation on pipeline cards (prevents layout overflow on long model IDs)

v1.2.0

Production ReadyMarch 10, 2026

Upgrade note: After pulling v1.2 updates, run `bun run migrate` in gridwork-hq to apply SQLite (superseded by Postgres 16) schema changes.

Added

SQLite-first storage via Drizzle ORM (superseded by Postgres 16) (Notion becomes optional, not required)
Demo mode with one-click sample data seeding and clean removal
Cloud-hosted deployment with automated provisioning
Setup wizard: 5-phase guided configuration with health checks and placeholder substitution
Vercel data loss warning banner when SQLite (superseded by Postgres 16) is detected on serverless
CI/CD workflow with type checking, linting, and build verification
Distribution packaging script with ZIP exclusion rules
Health check endpoint on pipeline server at /health

Improved

Complete rebrand from internal codename to GridWork HQ across all files
65 API routes now require authentication via getServerSession()
Shell injection eliminated: all execSync calls replaced with execFileSync + arg arrays
52 CSS custom properties (gw-*) replace every hardcoded color value
Zero hardcoded hex values remain outside shadcn primitives

Security

CSP nonce extended to style-src for inline style protection
SSE rate limiting: 20 connections per minute per IP on pipeline server
Path traversal guards on Knowledge Vault file operations
Cron command allowlist prevents arbitrary shell execution

v1.1.0

AI PipelinesMarch 1, 2026

Added

17 AI-powered pipelines at launch: prospect, outreach, audit, internal-audit, propose, plan, build, brand, seo, content, report, follow-up, spec, kb-librarian, archive-outputs, friday-update, scope-audit (Note: Pipeline count was 17 at time of v1.1 release. Current count: 45.)
Document compiler with 7 output types (proposal, report, plan, audit, spec, brand guide, content brief)
Cron engine with 12 scheduled jobs for automated operations
Operations monitor with 6 real-time health cards (CPU, memory, disk, pipelines, cron, SSE)
Mission Control: split-panel layout with pipeline launcher and AI chat sidebar
Pipeline server with SSE streaming for real-time job progress
Telegram notifications for pipeline completions and cron failures
49 knowledge vault templates with {{AGENCY_*}} placeholder system

v1.0.0

FoundationFebruary 15, 2026

Added

Dashboard with 8 stat cards showing clients, revenue, pipelines, and activity
Client Tracker with filter tabs (active, prospect, archived) and inline editing
Pipeline Kanban with 5-column drag-and-drop (queued, running, review, done, failed)
Knowledge Vault: git-backed file explorer with create, edit, and delete
Finance Panel with revenue charts and invoice tracking
Settings page with 10 integration configurations
Reports page with exportable pipeline and client summaries
Cron Jobs dashboard with status indicators and manual trigger
NextAuth with GitHub OAuth for single-user authentication (superseded by email/password auth)
Notion, Stripe, Vercel, and Google Analytics 4 integrations
Full security headers (HSTS, X-Frame-Options, X-Content-Type-Options, Referrer-Policy)
WebP image optimization across all static assets

Changelog

v2.0.0

Added

Security

Removed

v1.3.0

Added

Changed

v1.2.1

Added

Improved

v1.2.0

Added

Improved

Security

v1.1.0

Added

v1.0.0

Added

Ready to run your agency on AI?

Changelog

v2.0.0

Added

Security

Removed

v1.3.0

Added

Changed

v1.2.1

Added

Improved

v1.2.0

Added

Improved

Security

v1.1.0

Added

v1.0.0

Added

Ready to run your agency on AI?