Change log

See below for what’s new since our last release.

• Strategy OS: Experiment outcome and learning note

See below for what’s new since our last release.

• MCP: Create library assets with typed frameworkCanvases so strategy-rail canvas slots turn green
• Workspace: Mission and vision strategy-rail rows link to library assets and open in preview when content exists

See below for what’s new since our last release.

• Integrations: Project hub, org OAuth apps, and connector setup guides
• Workspace: Merge MCP tab into Integrations and fix project create navigation
• Ci: Fixed use preinstalled nvm Node before network install
• Agentic delivery: MCP write tools for agent pipelines
• Workspace: Fixed improve Quality index badge contrast on light sidebar
• Localisation: Fixed localize project integrations hub and asset picker (#155)
• Workspace: Project integrations hub UI (#155)
• Strategy Os: Fixed resolve connectors via project bindings (#155)
• Connectors: Tenant-only org OAuth connect split (#155)
• Api: Project integrations bindings REST routes (#155)
• Db: Added ProjectIntegrationBinding and tenant account refs (#155)
• Handoff: Added GitHub Issues as one-way execution handoff target
• Handoff: Added GitLab Issues as one-way execution handoff target
• Workspace: Added Project Integrations hub under sidebar PROJECT
• Workspace: Show version count instead of document count in strategy gallery
• Okr: Show Lamorte-style KR bands and commitment on OKRs
• Workspace: Unity v9 chrome, standalone landing pages, and OKR scoreboard
• Agentic delivery: Run live strategy workshops with timed phases in Sculptor
• Connectors: Fixed keep HubSpot evidence loaders client/server split
• Connectors: HubSpot pipeline grounding reaches agentic turns
• Delivery: Fixed pin DeliveryRun org/project from Asset gate
• Delivery: Persist DeliveryRun phase shell for refresh durability
• Connectors: Renew Reddit and LinkedIn OAuth access via refresh hub
• Connectors: Fixed keep OAuth pending tokens out of redirect URLs
• Agentic delivery: i18n-message-parity
• Agentic delivery: Adapt Matt Pocock skills into Sculptor ADLC
• Agentic delivery: Extract agent-platform-kit and pin Sculptor to v0.1.2
• Agentic delivery: Adapt Superpowers ADLC patterns into Sculptor gates
• Delivery: Block Start until strategy readiness passes
• Agentic delivery: Fixed prune abandoned issue branches when all MRs are closed
• Connectors: Fixed wire Bitbucket OAuth and fail-closed authorize families
• Agentic delivery: Fixed auto-bootstrap MR review checklist and high-risk approval in CI
• Strategy agents: Avoid glab --related-issue wrong MR source branch
• Delivery: Human-approved sandbox deploy preview and org hosting settings
• Delivery: Improved reliability for job pop dead-letter and retry a11y
• Delivery: Run Analysis→Document phases with artefacts and retry
• Delivery: Improved reliability for sandbox trust boundary and a11y
• Delivery: Build solutions from strategy into hosted sandbox
• Agentic delivery: Split programme parents into linked child work items
• Agentic delivery: Fixed close empty MR when issue branch already matches main
• Agentic delivery: Fixed persist next vertical slice for /agent continue
• Agentic delivery: WSFF planning depth in Agent Work Items and MR gates
• Account & privacy: Microsoft 365 sign-in on login when AUTH_MICROSOFT_ENTRA_ID_* is set
• Account & privacy: Google sign-in on login when AUTH_GOOGLE_* is set
• Agentic delivery: Fixed map practices-auditor paths to dedicated test script

See below for what’s new since our last release.

• Polish mobile chrome, navigation persistence, and org LLM autofill
• Restored Ollama Cloud org chat when production has no tsx MCP
• Fixed improve workspace dictation on Edge with Auto cloud fallback

See below for what’s new since our last release.

• Fixed align localized marketing HTML with hreflang metadata
• Added integrations, trust, and solutions pages with GTM proof gallery
• Complete go-to-market step integration
• Require explicit license choice and add cs/sk hreflang
• Close register-to-checkout funnel and SEO gaps
• Serve Strategy Context over scoped MCP API keys
• Added conditional score pills to CEP map tables
• Limit connector evidence toolbar to active org connectors
• Added ping/pong verify for shared API keys
• Fixed localize sidebar accordion groups and CEP phase labels

See below for what’s new since our last release.

• CEP map Load connector evidence now covers all 24 org connectors (backlog, paid search, behavior, documents, social)
• Expand catalog with MCP bridge, storage, and API-key flows
• Added GitLab OAuth connector alongside GitHub
• Fixed resolve lint and build errors for connectors and CEP client bundle
• Added setup instructions to plugin/integration pages
• Expose connector OAuth env vars in platform admin console
• Added company connectors tab, GA4/GitHub OAuth, and verification docs
• Added external_connections migration and GSC OAuth flow
• Fixed raise production nginx LLM timeouts to avoid chat/agent 504s
• Phase C GSC capture-gap connectors foundation
• Phase B SerpAPI grounding service and billing quota gate
• Enforce 8-row agent quality gate and validation pack tests
• Fixed CEPMAP row ids and clamp scores for small models
• Fixed allow CEP-* row ids in cep-map specialist validation
• Phase A — assumption-mode CEP canvas and /cep command
• Restored offline prisma patches and clear lint a11y warnings
• Added integrated outline view with hierarchy and asset drafts
• Checkpoint comments/history/titles + add integrated-layout wireframes
• Polish public surveys and round comparison chart
• Added scorecard, shareable surveys, and execution tracking
• Document local Mac OOM workarounds and enable webpack memory opts
• Fixed correct requireProjectEditorRouteGate signature in strategy execution routes

See below for what’s new since our last release.

• Shareable strategy confidence surveys on the Scorecard: public or password-protected links, expiry, QR + copy/email share, response charts and round comparison
• Hypotheses in Strategy execution now use a drag-and-drop Kanban board (Unvalidated → Validating → Validated / Invalidated)
• Fixed correct requireProjectEditorRouteGate signature in strategy execution routes
• Fixed resolve 16 code review issues across agentic and chat modules
• Add comprehensive code review task list
• Comprehensive review of remaining 17 issues
• Fixed allow finance playground href builder from server routes
• Add final comprehensive code review report
• Added error logging to additional user management routes
• Added error handling and cleanup to multi-step repository operations
• Added error logging to auth routes
• Removed unused llmCredentialModeUserChosen blocking Docker build
• Add comprehensive code review session 2 completion report
• Fixed resolve phase template and trajectory TypeScript build errors
• Fixed resolve React Query dependency array ESLint warnings
• Update memory tracking progress on code review fixes
• Fixed wrap notification preference updates in transaction to prevent lost writes
• Fixed wrap database read-modify-write operations in transactions to prevent lost writes
• Fixed resolve 3 critical issues causing provider errors and data corruption
• Fixed resolve phase template efficiency and validation issues
• Removed unused PhaseTemplateUiStage import blocking build
• Fixed resolve stale cache showing as error on mobile bfcache resume
• Fixed handle mobile bfcache resume for LLM credential context
• Fixed resolve architectural issues in phase execution context and children integration
• Fixed resolve critical bugs found in code review
• Added trajectory adjustment, calculators, and orchestration gating
• Phase template UI orchestration
• Added 6 core strategy phase templates for complete coverage
• Integrate phase dependency indicators into strategy rail
• Completed-phases hooks and strategy-rail integration guide
• Integration examples and copy-paste-ready instructions
• Complete integration guide for phase templates system
• Full phase template UI integration with dependency checking and validation
• Integrate phase templates into coach system prompt, dependency checking, and SQI grading
• Phase templates with output structure & checklists
• Per-run execution mode config + timeout fallback
• Server-side zombie recovery via job leases
• Fixed critical issues in server-side execution default
• Server-side execution default for agentic runs
• LLM credential resolution & org cloud workspace integration
• Fixed trajectory & success bar duplication + missing parser field
• Live trajectory preview in project details (mid-journey pivot)
• Recalibrate SQI grading bands per success bar (experience-adjusted)
• Wire recommended trajectory into system prompt for phase-aware coaching

See below for what’s new since our last release.

• Fixed allow finance playground href builder from server routes
• Add final comprehensive code review report
• Added error logging to additional user management routes
• Added error handling and cleanup to multi-step repository operations
• Added error logging to auth routes
• Removed unused llmCredentialModeUserChosen blocking Docker build
• Add comprehensive code review session 2 completion report
• Fixed resolve phase template and trajectory TypeScript build errors
• Fixed resolve React Query dependency array ESLint warnings
• Update memory tracking progress on code review fixes
• Fixed wrap notification preference updates in transaction to prevent lost writes
• Fixed wrap database read-modify-write operations in transactions to prevent lost writes
• Fixed resolve 3 critical issues causing provider errors and data corruption
• Fixed resolve phase template efficiency and validation issues
• Removed unused PhaseTemplateUiStage import blocking build
• Fixed resolve stale cache showing as error on mobile bfcache resume
• Fixed handle mobile bfcache resume for LLM credential context
• Fixed resolve architectural issues in phase execution context and children integration
• Fixed resolve critical bugs found in code review
• Added trajectory adjustment, calculators, and orchestration gating
• Phase template UI orchestration
• Added 6 core strategy phase templates for complete coverage
• Integrate phase dependency indicators into strategy rail
• Completed-phases hooks and strategy-rail integration guide
• Integration examples and copy-paste-ready instructions
• Complete integration guide for phase templates system
• Full phase template UI integration with dependency checking and validation
• Integrate phase templates into coach system prompt, dependency checking, and SQI grading
• Phase templates with output structure & checklists
• Per-run execution mode config + timeout fallback
• Server-side zombie recovery via job leases
• Fixed critical issues in server-side execution default
• Server-side execution default for agentic runs
• LLM credential resolution & org cloud workspace integration
• Fixed trajectory & success bar duplication + missing parser field
• Live trajectory preview in project details (mid-journey pivot)
• Recalibrate SQI grading bands per success bar (experience-adjusted)
• Wire recommended trajectory into system prompt for phase-aware coaching

See below for what’s new since our last release.

• Add Strategy Operating System PRD and release 2.6.54 notes
• Added business reality tool and compact WS_CTX summary
• Upgrade business reality fields to full TipTap toolbar
• Added project business reality fields and BASELINE-001 intake

Business reality in Project brief, lighter coach context, and baseline-aware strategy phases.

• Project brief — new Business reality section with six optional as-is fields: operating model, team and skills, finances, pipeline, constraints and risks, and strengths
• Rich text on Business reality — full editor with formatting toolbar for lists, emphasis, and notes on each field
• Intake (`/intake`) outputs BASELINE-001 alongside SCOPE-001 so strategy runs start from documented facts
• Coach uses a compact business-reality summary and pulls full detail on demand — less noise in chat, sharper follow-ups
• Lead notes on the project brief are optional; when provided, they improve strategy suggestions
• Org, finance, sales, and risk phases honor your Business reality baseline instead of re-asking stated facts

See below for what’s new since our last release.

• Fixed clear unused imports and hook deps blocking production build
• Project brief UX, chat polish, dictation, and framework fixes
• Added fork conversation action in thread menu

See below for what’s new since our last release.

• Fixed invite signup contrast and masked email registration
• Finance hub i18n, locale overlay fix, and LEARNING.md
• Show 3D relation map edges between nodes

Financial scenarios, project currency, and a finance playground you can save to the workspace.

• Financial scenarios — branch what-if assumptions from /finance assets; compare CAC, LTV, payback, and revenue side by side
• Finance playground — try SaaS, marketplace, and other templates at /tools/finance-scenarios; save scenarios to a workspace finance phase when ready
• Project currency — set ISO currency in Project details; finance outputs and the playground format amounts consistently
• Finance guide on Guides hub and landing — unit economics, scenarios, and abbreviations (CAC, LTV, burn, runway)

See below for what’s new since our last release.

• Added full change log page and polish sidebar UX

See below for what’s new since our last release.

• Improved reliability for automatic batch conversation edge cases
• Fixed pin in-flight batch thread and harden rail smoke tests
• Fixed resolve conversation for run-selected automatic batch

See below for what’s new since our last release.

• Restore automatic batch progress popup after app restart
• Strategy phases rail marks completed outputs from loaded bundle bodies
• Block starting a duplicate automatic batch while one is still running

See below for what’s new since our last release.

• Harden automatic batch runs and improve progress popup UX

See below for what’s new since our last release.

• Org cloud: Run selected automatically now runs on the server with a background worker — progress survives tab sleep and is visible to project teammates
• Floating progress popup shows Queued, Running, and terminal states with per-item status in the strategy rail
• Team in-app notifications when automatic generation starts, completes, fails, or is cancelled
• Fixed guard invalid coach run ids and LLM URL validation on agent route
• Added context-derived typed attributes for strategy personas
• Fixed stabilize RTL smokes, SVG preview SSR, and i18n token parity
• Fixed fallback PMS portfolio status labels when i18n missing
• Fixed clear unused vars and hook dependency warning
• Playbook palette export, quality badges, and critique UX polish
• Distribute public-try endpoint rate limiters
• Route remaining rate limiters through the distributed limiter
• Distributed rate limiting for auth brute-force limits (#5)
• Improved reliability for rate-limit IP, webhook idempotency, body size, local LLM proxies
• Fixed sanitize HTML/SVG dangerouslySetInnerHTML sinks
• Added includeSolarch to run config callback deps
• Added workspace theme palettes to playbook documents
• Fixed block generate on INI/EPC summary rail rows
• Fixed align proactive coach strip button layout
• Fixed resolve DocumentChrome module load error
• Mark Sprint C shipped in UI/UX backlog (adb43c7b)
• Fixed map localized pricing tier names to license slugs
• Fixed move what's new entry below feedback support
• Ship Sprint A onboarding, mobile nav, and rail polish
• Resolve agentic post-run handoff to in-panel embed (#4C)
• Validate UI redesign backlog against codebase
• Lock sidebar IA — strategy document and in-flow agentic
• Lock UI redesign backlog to landing-first brand constraint
• Ship solution architecture phase and strengthen LLM SEO discovery
• Added SolutionArchitectureAgent and refine workspace chrome
• Removed resume run row and chat subtitle from project accordion

See below for what’s new since our last release.

• Added includeSolarch to run config callback deps
• Added workspace theme palettes to playbook documents
• Fixed block generate on INI/EPC summary rail rows
• Fixed align proactive coach strip button layout
• Fixed resolve DocumentChrome module load error
• Mark Sprint C shipped in UI/UX backlog (adb43c7b)
• Fixed map localized pricing tier names to license slugs
• Fixed move what's new entry below feedback support
• Ship Sprint A onboarding, mobile nav, and rail polish
• Resolve agentic post-run handoff to in-panel embed (#4C)
• Validate UI redesign backlog against codebase
• Lock sidebar IA — strategy document and in-flow agentic
• Lock UI redesign backlog to landing-first brand constraint
• Ship solution architecture phase and strengthen LLM SEO discovery
• Added SolutionArchitectureAgent and refine workspace chrome
• Removed resume run row and chat subtitle from project accordion

See below for what’s new since our last release.

• Added includeSolarch to run config callback deps
• Added workspace theme palettes to playbook documents
• Fixed block generate on INI/EPC summary rail rows
• Fixed align proactive coach strip button layout
• Fixed resolve DocumentChrome module load error
• Mark Sprint C shipped in UI/UX backlog (adb43c7b)
• Fixed map localized pricing tier names to license slugs
• Fixed move what's new entry below feedback support
• Ship Sprint A onboarding, mobile nav, and rail polish
• Resolve agentic post-run handoff to in-panel embed (#4C)
• Validate UI redesign backlog against codebase
• Lock sidebar IA — strategy document and in-flow agentic
• Lock UI redesign backlog to landing-first brand constraint
• Ship solution architecture phase and strengthen LLM SEO discovery
• Added SolutionArchitectureAgent and refine workspace chrome
• Removed resume run row and chat subtitle from project accordion

See below for what’s new since our last release.

• Strategy and chat rails load more smoothly, with clearer progress while content is loading.
• Run selected strategy phases and reprioritize backlog from rail

See below for what’s new since our last release.

• Run selected strategy phases and reprioritize backlog from rail
• Strategy and chat rails load more smoothly, with clearer progress while content is loading.

See below for what’s new since our last release.

• Deliverable guides, depth-first SEO/GEO, and metrics hub
• Free framework library section and gated prompt downloads
• Playbook covers, Wardley strategy rails, and workspace polish
• Polish playbook templates, export, and workspace chrome
• C4 — sculptor-* primitives round-trip through HTML/DOCX export

See below for what's new since our last release.

• Workspace sidebar uses a project combobox with search — selected project panel replaces the long desktop project list
• Document editor toolbar adds a Lists & indent submenu for bullet, numbered, and indent controls
• Conversation more menu consolidates new-chat actions and removes per-thread Match settings
• Copy and label refinements (Company heading, Project stage, Generate missing chapters)

See below for what’s new since our last release.

• Fixed unblock org Ollama Cloud turns before LLM fetch
• Chat streaming no longer duplicates thinking-model prose or shows a stray live bubble above the thread

See below for what’s new since our last release.

• Org Ollama Cloud composer: fix silent send failures (prepare-phase crash, missing baseUrl) and add assistant-turn debug with ?debugComposerLlm=1

See below for what’s new since our last release.

• Chat with organisation Ollama Cloud (e.g. gemma4:31b) shows assistant replies again — reasoning-only stream fallback and direct-answer mode on composer proxy

See below for what’s new since our last release.

• Composer org Ollama Cloud health probe calls org models API directly (fixes false unreachable banner)

See below for what’s new since our last release.

• Improved composer Ollama Cloud probe logging and pre-deploy verify workflow
• Fixed align composer Ollama Cloud health probe with org credentials

See below for what’s new since our last release.

• Fixed org Ollama Cloud composer health probe to match chat credential rules
• Fixed clear deploy warnings and composer LLM debug type error
• Added opt-in composer LLM connectivity diagnostics
• Fixed inject org keys when chat uses organisation credential preset

See below for what’s new since our last release.

• Fixed clear deploy warnings and composer LLM debug type error
• Added opt-in composer LLM connectivity diagnostics
• Fixed inject org keys when chat uses organisation credential preset

See below for what’s new since our last release.

• B3 — per-phase signature deliverable requirements
• A4 section-set for go-to-market and discovery-memo
• A4 section-set for investor-view, growth-opps, QBR + outcome headings
• Label playbook generate actions and prefer library on fill
• Fixed stop raw slot keys leaking into section copy
• Drop redundant "About this document" intro block
• Restored mission and vision to executive summary
• Fixed align executive-summary description with tightened sections
• Tighten executive-summary section set (A4 pilot)
• Outcome-named section headings (A3)
• Added decision callout and stat-highlight primitives
• Apply substance bar to specialist slot generation
• Fixed clear lint, typecheck, and build errors across document and library modules.
• Update.cursor README for Sculptor instead of FlowOS
• Fixed sidebar navigation from doc route and chat rail layout
• Framework preview highlights and comment restore on load
• Playbook comments, chat rail UX, and editor polish
• Fixed dark mode gallery, chat rail layout, and localized insert labels
• Embed workspace chat, insert picker, and strategy rail on doc route
• Docuemnt Tempaltes
• Fixed stop LLM health probes from blocking document navigation
• Fixed use browser history for document header back button
• Consultant-grade project brief and section guidance copy
• Visual palette previews in picker
• Detailed-strategy includes full strategy phase tree
• Fixed resolve phase include provider on document page
• Fixed resolve ProjectsListProvider on document page
• Fixed break circular import causing Turbopack HMR panic
• Fixed close playbook 5/5 gaps
• Per-section guidance, chapter subtitles, shared editor shell
• Unify Combined Results pane with shared palettes
• Unify playbook surfaces and guidance leftovers
• Wire playbook slot generate and chapter readiness
• Editable playbook editor with palettes and export
• Phase 1 playbook guidance for template slots
• Strategy pack prefill, reader view, and sidebar link
• Strategy-pack seed gallery with filters and library prefill
• Document block hardening, org delete, project details UX
• Assistant rail with streaming Claude chat
• Inline title editing, document PATCH route, Esc on insert menu
• Doc list on gallery, lazy context form, slash insert command
• Phase 1 — dep engine, subgraph resolve, staleness, TipTap editor, conflict dialogs
• Implement Document-First Workspace Phase 0

See below for what’s new since our last release.

• Update.cursor README for Sculptor instead of FlowOS
• Fixed sidebar navigation from doc route and chat rail layout
• Framework preview highlights and comment restore on load
• Playbook comments, chat rail UX, and editor polish
• Fixed dark mode gallery, chat rail layout, and localized insert labels
• Embed workspace chat, insert picker, and strategy rail on doc route
• Docuemnt Tempaltes
• Fixed stop LLM health probes from blocking document navigation
• Fixed use browser history for document header back button
• Consultant-grade project brief and section guidance copy
• Visual palette previews in picker
• Detailed-strategy includes full strategy phase tree
• Fixed resolve phase include provider on document page
• Fixed resolve ProjectsListProvider on document page
• Fixed break circular import causing Turbopack HMR panic
• Fixed close playbook 5/5 gaps
• Per-section guidance, chapter subtitles, shared editor shell
• Unify Combined Results pane with shared palettes
• Unify playbook surfaces and guidance leftovers
• Wire playbook slot generate and chapter readiness
• Editable playbook editor with palettes and export
• Phase 1 playbook guidance for template slots
• Strategy pack prefill, reader view, and sidebar link
• Strategy-pack seed gallery with filters and library prefill
• Document block hardening, org delete, project details UX
• Assistant rail with streaming Claude chat
• Inline title editing, document PATCH route, Esc on insert menu
• Doc list on gallery, lazy context form, slash insert command
• Phase 1 — dep engine, subgraph resolve, staleness, TipTap editor, conflict dialogs
• Implement Document-First Workspace Phase 0
• Fixed localize settings and sidebar labels; refresh release notes
• Fixed align composer LLM health probe with pill and org credentials
• Fixed resolve hook exhaustive-deps warnings and harden production builds
• Added Fix immediately handoff to agentic revision runs
• LLM framework fence regenerate; fix dashboard step overflow
• Mermaid previews, inline generate UX, and missing-ref actions
• Use phase-tree labels for framework asset titles
• Fixed portal strategy-rail asset preview into chat panel row
• Strategy and chat rails load more smoothly, with clearer progress while content is loading.
• Added Fix asset action for broken framework previews
• Test coverage
• Improve asset relation map navigation and focus highlighting
• Cache strategy digests for long org and project briefs
• Default Ollama keep_alive to 3h without env setup
• Append SCULPT_DEBUG_LLM_IO traces to log file

See below for what’s new since our last release.

• Library relation map — optional 3D WebGL view with bubble grouping, four layout presets, and the same focus filters as the 2D SVG map.
• Slovak and Czech thread-intake welcome on new conversations — known facts, trajectory preview, and review sections no longer fall back to English.
• Fixed satisfy prefer-const in Office document extractors
• Fixed validate org company PATCH and raise description limit
• Fixed use project-scoped library previews for cloud workspaces
• Fixed sync org picker with organisation name edits
• Removed unused var blocking Docker build
• Add intake context and trajectory router plan
• Localize combined document coaching guidance and templates
• Project intake context — offering, segment, goals, and stage — now steers Chat and Agentic runs, adds a Context fit score on the Quality index, and adapts the new-thread intake wizard to your situation.
• New Quality tab on the project workspace — review strategy quality scores, trends, and context fit in one place.
• Strategy pack templates are grouped by journey stage so you can pick the right starting point faster.
• Strategy pack template picker — clearer descriptions and quick-scan pills for each template.
• Combined document and library surfaces respect your light or dark theme consistently.
• Combined strategy document background follows your workspace theme again.
• Wardley map framework output includes richer expanded descriptions for each component.
• Combined strategy documents read more naturally — section openers and closers guide readers through your narrative.
• Combined strategy documents open much faster — especially when loading a scoped slice of your library.
• Faster combined document loading and a unified strategy packs list in the workspace.
• Improved reliability for combined-doc fence repair and strategy pack archive
• Combined strategy documents render charts and markdown more reliably after export and merge.
• Source documents tab now lists only files you uploaded — system-generated assets stay out of the way.
• Security hardening — tighter protection for API keys, invitations, embedded content, and LLM proxy routes.
• Strategy rail phase status is more accurate — incomplete steps show a clear amber indicator.
• Improved parsing for typed bundle envelopes and salvage quality-failed blocks
• Feedback analysis agents gather source context before running — more grounded insights from your inputs.
• Fixed route asset rendering from bundle metadata, not markdown shape
• Fixed chart parse salvage, funding preview, and finance/funding split
• Improved reliability for finance prose parse, quality, and markdown display
• Journey mapping and customer experience are labeled distinctly so frameworks stay unambiguous in your pack.
• Improved parsing for Gemma-style pseudo tool_calls in specialist JSON mode
• Dismissed asset quality review flags stay dismissed after you refresh the workspace.
• Added naming-pillars agent package to unblock chat agent build
• Naming framework output renders as pillar cards and scored tables — easier to scan and compare options.
• Re-bucket optional depth phases and tighten GTM prose output
• Compliance output no longer includes emojis in compliance phase output
• RACI and operating-model views — role cards, org-aligned structure, and plain-language labels in the strategy rail.
• Added role cards and use plain-language framework labels in strategy rail
• Strategy rail shows backlog epics and hypotheses in a clearer tree — plus refreshed in-app typography.
• Landing page copy now highlights Sculptor’s curated, proven strategy frameworks.
• OKRs are grouped in the strategy rail so objectives and key results are easier to browse.
• OKR framework output renders as readable tables instead of raw backlog JSON.
• Backlog hierarchy opens on click and sanitizes AI-generated descriptions for safer display.
• Backlog hierarchy previews — validation, professional tables, and click-to-expand structure.
• Strategy quality scores move into a compact header pill — more room for charts and your workspace content.
• Backlog previews prioritize items using your existing strategy artefacts — not generic placeholders.
• Fixed require customer journey stage matrix depth and harden canvas repair

See below for what’s new since our last release.

• feat: workspace start dashboard, sidebar UX, and library embed improvements
• feat: unified chat display, faster conversation open, and strategy phase rail
• feat: chat-to-agentic handoff and faster workspace project sidebar
• feat: roadmap-grouped strategy progress, coach suggestions, and asset export
• feat: library relation map grouping, revision history, and Mermaid hardening
• feat: mark wizard required fields and tune agentic phases by venture stage
• feat: welcome wizard v2 creates new org with Sculptor AI and agentic polish
• fix: project details autosave loop, cap notes editor, Lucide outline icons
• fix: render framework canvases in chat sidecar assets and polish asset preview.
• chore: remove unused socket.io dependency
• fix: break circular dependency between conversation and message repositories
• Fix chat sidecar asset save and tighten competitor-matrix validation.
• Add Ollama Cloud web search tools and harden framework canvas specialists.
• Add library asset relation map and reorganize framework canvas JSON.
• Harden typed specialist frameworks, gallery previews, and LLM proxy streaming.
• PESTLE ready: typed specialists, gallery previews, and schema hardening.
• Render framework canvases in combined strategy doc and harden agentic output.
• Add platform admin console, support chat, and operator tooling.
• Speed up chat and library loading with paginated display messages and batched asset fetches.
• Open composer settings when clicking the LLM health dot.
• Harden framework canvas output: complete JSON prompts, streaming placeholders, and fence repair.
• Enable LLM prompt caching for chat, agentic, and MCP agent paths.
• Fix mobile sidebar menus above drawer and add landing agent focus cards.
• feat(framework): extend canvas kinds, harden JSON pipeline, and capture LLM backlog
• Visualizers for assets
• Add Wardley map graph and harden extended framework canvas payloads.
• feat(chat): add ten extended framework canvas kinds
• feat(chat): framework canvases, streaming UX, and coach JSON hardening
• Fix conversation titles leaking asset sidecar fragments.
• Align ERRC and block framework canvases with VPC section styling.

Automated from 1 commit(s) since tag deploy/production-20260506T204111Z. Edit this file for customer-facing wording when needed.

• fix(chat): show assistant star on mobile; workspace shell and header tweaks

Automated from 3 commit(s) since tag deploy/production-20260506T192246Z. Edit this file for customer-facing wording when needed.

• feat(api): consolidate first-party LLM chat under /api/first-party/chat
• feat(chat): show sender avatar on org user messages
• feat(chat): composer gear menu for model and voice settings