The prototype is deployed. The 25-task FMP Batch B increment is complete in alpha-44340458084-1, following Batch A in alpha-44132016951-1. Batch B has accepted evidence for eight event and document API operations and five representative local widget journeys. Publication verification covers the public artifacts, a network-isolated API check and isolated internal-network readiness and login. Windows host-loopback transport remains unverified. Within the 24-operation expansion plan, Batches A and B are complete (Batch B tasks). The five Batch C DCF and as-reported operations remain follow-ons. The separate modeling foundation retains its approved 18-task plan; repository coordinates and selected-input readiness remain open. This release does not approve those implementations or new owner-server deployment.
Input verification for the first research story is complete: six ETFs and eight FRED series are verified at the data API (checkpoint). The next step is one thin modeling slice for "Commodities vs tech: which regime are we in?". No factor, story or widget exists yet.
Scope: existing Full Workspace and installed data API. Further provider coverage, renderer families, theme changes, commands, AI, Excel and custom sources are not prerequisites. Preserve working functionality and disclose supported capabilities and gaps.
| Slice | Deliverable | Acceptance |
|---|---|---|
| P1: freeze candidate | Reconcile pending fixes, fork pins and supported-capability manifest | No unrelated edits included; source identities and known gaps recorded |
| P2: CI/CD | Portable checks/publication, thin GitHub runner and Dagger image builds | Checks fail closed; PR jobs exclude publication/provider secrets; immutable image identities |
| P3: assembled prototype | Real service images in executable Compose | Existing isolated E2E proves login, supported data/widget use and diagnostics against candidate images |
| P4: install | Pull-only instructions and release manifest | Installation rehearsal; owner-run server outcome recorded separately |
Owner clarification: saved research-context persistence acceptance moves to analyst-workspace. Infrastructure backup/restore and rollback verification remain deferred delivery work. Neither blocks the prototype; the bind mount and existing tests remain intact.
Target: HTTP on 0.0.0.0:23003, qopenbb.example.com through owner-managed cloudflared, relative bind-mounted data. The owner deploys locally with Compose; no SSH, tunnel service or Jupyter dependency.
Latest accepted chart milestone: alpha-40652638534-1 includes the 36-chart historical FMP gallery in Apps. Local acceptance covers the changed combination assignment, native save, fresh authenticated reopening and retained-bind Compose recreation. Publication evidence verifies the public image pair separately; browser acceptance has not been repeated on the CI image bytes. This historical chart record is separate from the current owner-confirmed prototype deployment. See the current chart checklist for individual chart milestones.
The completed prototype batch repairs four observed integration gaps: an absent first-login questionnaire returns HTTP 200/null, the data image packages and serves its existing widget/app catalog, optional external app previews use Workspace's existing fallback, and the data-only backend advertises an empty agent list. alpha-40467871625-1 passes CI, release-asset checks, six anonymous pulls and the fresh-account assembled Compose journey. P3 and the local P4 installation rehearsal are accepted for onboarding, later login, live annual AAPL income values, refresh and clean diagnostics. The generic-widget required-symbol guard, broader provider acceptance, analyst persistence and backup/restore remain open. Current owner-reported startup is recorded below. AI remains a separate spec. See the application review, application result and delivery status.
Latest publication: alpha-43604526965-1 adds the bounded Company Research App. Both CI jobs, three public assets and six anonymous image pulls are verified in the publication receipt. All 13 bounded tasks are accepted. Local financial/runtime acceptance remains separate from the released CI image bytes; full IA-B, house modeling, CI-image browser and owner-server deployment are not inferred.
The preceding alpha-42188370167-1 adds accepted company-name/Symbol selection to the published initial financial-input batch. On 2026-11-06 the owner confirmed that release's Linux pull/update, healthy data API, retained dashboards and working autocomplete. This owner report is separate from local runtime and publication evidence; no independent server inspection or recovery check is claimed.
The 11-endpoint expansion design and 26-task plan are complete. All eleven API models and four representative widgets have actual local evidence under the unchanged 43/48 FMP request ledger; seven widgets and future AI remain unknown. Published source is root 32d1140, API a7c6ae2, Workspace bf180ef; later acceptance metadata is separate. The internal-network CI pair checks do not repeat the local financial browser journey. The 24-operation plan retains B8/C5 as follow-ons.
The bounded local increment is the Company Research dashboard: linked company selection, statements, price context, editable candidate peers and saved research configuration. Readiness reconciliation distinguishes closed acquisition defects from unresolved financial comparability. Requirements are owner-approved. The bounded design is owner-approved. The 13-task plan passed independent review, including the owner-amendment re-review after one repair pass; task/implementation is approved, and foundation/shared contracts are accepted. It reuses existing Apps, groups and persistence, adds inspectable evidence and independently editable peers, and leaves house calculations in the modeling workstream.
- Reconcile the completed periodic-query and statement-period work with the dated IA-A assessment.
- Independently review bounded Company Research requirements, including App-template export/import.
- Approve requirements and independently review the bounded design.
- Approve design and independently review the 13-task implementation plan.
- Approve tasks and accept the isolated foundation and shared contracts.
- Accept the bounded local Company Research journey, template portability and retained-data recreation. Runtime evidence.
- Publish the accepted sources as alpha-43604526965-1; verify image/asset identities and anonymous pullability (receipt).
The market-data requirements are owner-approved. The design is also owner-approved. The independently reviewed first task batch is owner-approved and implementation has started: 26 bounded tasks covering shared contracts, FMP financial inputs, FRED provenance, consumer integration and milestone delivery. Later DCF, narrative and wider provider families remain explicit follow-on work. Research confirms reuse of the owned OpenBB framework and retained legacy provider code. Broad exploratory provider coverage supplies the analyst dashboards; it is not limited to one example company.
The investment-analysis contract defines the next analyst workflow: quarterly performance and cash conversion, defensible peer valuation, FMP DCF with a cost-of-capital/macro bridge, then linked expectations, news, transcripts and thesis context. Reuse OpenBB widgets and linking; no embedded browser is required. CFA-level analytical discipline is a quality target, not a certification claim.
- Verify actual FMP/FRED data coverage and existing implementation reuse for the first financial workflow.
- Deliver quarterly financial analysis and peer valuation, with consistent periods, metric definitions and saved research context.
- Deliver house DCF/WACC and verified sensitivities through the separate modeling layer; retain FMP DCF as an optional comparison.
- Complete linked expectations, inline news/transcripts and thesis tracking.
These are planned financial capabilities, not completed chart-rendering features. Detailed scope and acceptance are IA1-IA10 in the contract.
First-spec closure records nine completed delivery milestones and transferred/deferred obligations.
The initial provider batch and company selector are complete. The completed IA-A assessment, guided by the input requirements, covered a representative nonfinancial company and at least two candidate peers, quarterly statement compatibility, dated price/share/EV inputs, FMP DCF feasibility and FRED Treasury context. Source findings identify the remaining gaps without repeating completed checks. House calculations stay in the separate modeling workstream.
- Reconcile completed initial-provider and company-selector milestones, including the owner-reported Linux update.
- Independently review the bounded IA-A requirements: six requirements, 24 criteria, one repair pass.
- Obtain owner requirements approval (2026-11-06).
- Approve the bounded verification design (2026-11-06).
- Approve the six-task plan and begin implementation.
- Collect the bounded real-provider evidence: 46 native FMP requests and three direct DCF probes; five live journeys passed.
- Independently accept the input assessment and IA-B readiness decision.
- Deliver quarterly analysis and peer valuation through the separately approved dashboard/model contracts. A raw-data App alone does not complete IA-B.
No deployed application change or image rebuild is part of the IA-A verification batch. IA-A is complete as a reviewed gap assessment. IA-B input and workflow readiness remain open; the periodic-query provenance design is owner-approved. Its approved four-task plan has accepted source/native integration and four real FMP cases. The separately published repair passes image, catalog and offline packaged checks; independent delivery review passed on 2026-11-10. It repairs one existing metadata omission; broader financial compatibility remains open.
- Independently review the bounded statement-period provenance contract.
- Obtain statement-period requirements approval and independently review the bounded design.
- Obtain design approval and independently review the seven-task plan, with provider and consumer lanes after the shared contract.
- Approve the seven-task implementation plan.
- Prepare isolated sources and the native provider baseline (evidence).
- Accept the typed statement core contract (evidence).
- Accept the statement consumer contract (evidence).
- Accept FMP statement provenance source integration (evidence).
- Accept the installed statement/source join and 18-case real FMP matrix (evidence).
- Repair optional native statement limits (evidence).
- Accept the packaged Income/Balance/Cash browser journey (evidence).
- Deliver Income, Balance and Cash period provenance (publication); broader financial compatibility remains open.
| Spec | Responsibility | Sequence |
|---|---|---|
| openbb-self-service | Foundation and prototype delivery | Prototype delivered; closure and remaining work |
| market-data-integrations | Broad primary FMP coverage, yfinance price gaps, FRED economic series and future providers; consistent UX with labeled provenance | After prototype; installed adapters stay in baseline |
| financial-modeling | Separate-repository house methods, financial assertions, model API contracts and reproducible calculations | Stub only now; implementation follows verified provider inputs and Kiro approvals; later three historical / five forecast years |
| analyst-workspace | Saved research context, widgets, commands, linked instrument/time views, Bloomberg-inspired theme and remaining renderer/table functionality | After prototype; build on existing work |
| workspace-ai | Anthropic, grounding, streaming and recovery | Independent of prototype |
| openbb-excel | Add-in, generated functions, authentication and Office acceptance | Retained dependency: Anthropic accepted first |
| custom-data-source | Own datasets and later PostgreSQL | Deferred |
Build the analyst experience on existing widget, theme and ECharts work. FMP, yfinance and FRED should use interchangeable interaction patterns where their data contracts permit, with provider provenance visible. Shared controls must preserve instrument identity, time ranges, currency, units and frequency rather than imply that different datasets are equivalent.
The capability reuse inventory anchors the next analyst slice in existing OpenBB source and installed catalog evidence. After the current delivery, uv/Python packaging and Valkey work, specify a Company Research App with one shared name/symbol selection across price, profile, statements, news, earnings calls and analyst views. Reuse the existing App JSON, widget controls, linked groups, tables and chart renderers; extend missing installed catalog/provider paths in small increments. Each increment needs source, installed configuration, real-provider and save/reopen acceptance before broader provider expansion. Existing mechanisms are not proof that the proposed App already works; the successor requirements remain scope stubs until their bounded contract is approved.
Renderer responsibilities belong to analyst-workspace:
| Renderer | Existing scope and remaining work |
|---|---|
| ECharts | Integrated table charts for selected families; remaining controls, families and saved-state behavior need their own acceptance |
| Perspective 5.6 | Packages, runtime host and isolated tests exist; production dashboard binding and analytical workflow acceptance remain open |
| Plotly | Retain useful supported widgets and their financial semantics |
| Highcharts | Dependencies/imports removed; legacy ChartHighcharts still maps to UnsupportedRenderer. Full former-caller parity remains open |
These renderer obligations do not block the prototype. Full original visual parity is not a prototype release gate. The Bloomberg-inspired theme should extend the existing work and keep widget controls, commands and linked views consistent.
Quant plugins remain later work. Successor specs are initialized scope descriptions, not approved detailed requirements or tasks. Do not generate all their implementation plans before shipping the prototype.
Spec ownership records transferred obligations. The existing breadth map retains historical criteria, actual tests and evidence. New capability records need one owning spec/criterion, implementation state, actual test paths, evidence identity and release relevance. Planned checks never count as passing tests.
The provider map describes horizontal scope. Track provider availability separately from adapter, widget, consistency and E2E acceptance. Shared semantics cover readable identities, units, time, provenance and linking; provider origin should not change the analyst's interaction pattern.
Keep isolated Playwright/CDP, bounded xdist, financial semantics and existing persistence and failure/recovery regressions. Run affected checks per change and the integrated prototype journey at the release gate. Saved research-context acceptance belongs to analyst-workspace; infrastructure backup/restore and rollback verification remain deferred. Publish concise reviewed evidence; do not turn each check into another planning project.
The agile delivery amendment supersedes older sequencing and whole-spec gates. Retain independent review, complexity caps and per-task commits. Transfer unfinished work without marking it complete.
The owner authorized starting the next family while the accepted predecessor finishes publication on 2026-10-24. Each release remains pinned to its accepted source; the installed Compose environment is unchanged by development.
| Order | Work | State | Dependency |
|---|---|---|---|
| 1 | Nightingale | Accepted and published | Tagged assets and public image pulls verified |
| 2 | Radial Column and Radial Bar | Accepted and published; owner approved both appearances | RC5 assets and anonymous pulls verified |
| 3 | Treemap and Sunburst | Accepted and published for onboarded users; both appearances approved | H5 assets and anonymous pulls verified |
| 4 | Heatmap | Accepted and published for onboarded users | HM5 assets and anonymous pulls verified |
| 5 | Funnel, Cone Funnel and Pyramid | Accepted and published for onboarded users | F5 tagged assets and public image pulls verified |
| 6 | Stacked Line and 100% Stacked Line | Accepted and published | Continue with the next bounded roadmap item |
| 7 | Sankey (additional chart type) | Accepted and published; light/dark appearances approved | Tagged assets and public image pulls verified |
| 8 | Theme Settings preview | Accepted and published; Padding works and both appearances approved | TP4 tagged assets and public image pulls verified |
| Later | Remaining caller/table obligations | Open | Follow the capability matrix and actual original reachability |
Renderer and control lanes may run concurrently after their shared contract is accepted. Shared application integration, local heavy checks and publication gates remain serialized. This queue does not mark planned tests or unsupported original paths complete.
- Close the first-login onboarding, catalog, preview and agent-discovery batch with the independently accepted published-image Compose journey and local installation rehearsal.
- Record owner-confirmed Linux update, retained dashboards and working company autocomplete (2026-11-06).
- Complete the generic-widget empty-required-symbol guard; broader provider and analyst workflows retain separate acceptance.
- Complete shared advanced image, gradient and pattern palettes with renderer, persistence and export acceptance. Owner decision on 2026-10-26: track this separately while Funnel proceeds with stock/plain CSS colors.
The current gallery milestone packages fixed FMP observations around January 2025 into reusable dashboards. It adds no chart renderer. The analyst guide explains the questions, dimensions and transformations for all 36 chart types.
- Verify native saved states and deterministic historical projections.
- Implement packaged snapshot lookup and discovery without replacing the existing catalogs.
- Verify all rendered examples and retained-state recreation in the assembled application.
- Capture and review the README overview and all 36 per-chart screenshots.
- Publish the gallery in the prebuilt images.
First data batch progress (tasks):
Company name autocomplete with adjacent Symbol is accepted: 10/10 tasks passed on 2026-11-06, including real FMP data, compatible groups, independent comparison/FRED and retained-state reopening. Task plan and acceptance. The larger analyst-dashboard sequence remains separate.
-
Accept packaged company selection and retained-state reopening.
-
Publish and verify the company-selector milestone alpha: alpha-42188370167-1.
-
Restore and characterize the existing provider verification baseline.
-
Implement typed financial metadata through the native result envelope.
-
Implement safe financial failures and separate query/result validation.
-
Harden FMP transport while preserving baseline consumers.
-
Preserve FRED acquisition times through versioned disk, memory and coalesced requests.
-
Add financial metadata to existing FMP profile and income without baseline contract changes.
-
Preserve generic FRED series semantics and observation provenance.
-
Restore the initial FMP datasets, wire shared service contracts and verify the bounded real-provider queries; 26/26 first-batch tasks are accepted. Wider families remain open.
- Restore bounded balance-sheet and cash-flow adapters with signed/null values, period/date semantics and acquisition provenance. Ten of26initial data tasks are accepted. Reviewed evidence.
- Register the initial restored datasets and verify installed contracts and scoped live access in tasks 5.1-6.3, recorded below. Historical source-only checkpoints retain their original scope.
At that historical checkpoint these adapters were source-only; the initial batch is now published through task 6.3 below. The separate modeling workstream remains queued after verified provider inputs.
- Source adapters for named/symbol discovery, provider peer candidates and bounded daily prices accepted. Twelve of 26 initial data tasks are accepted. Evidence.
- Complete initial periodic metrics, ratios and enterprise-value inputs, then register consumers and verify the scoped real-provider queries through tasks 5.1-6.3.
No image was published at that adapter-only checkpoint; task 6.3 below records subsequent publication.
- Source adapters for periodic key metrics and ratios, with explicit units, provider-calculation labels and unsupported TTM modes. Fourteen of 26 initial data tasks accepted. Evidence.
- Add initial enterprise-value inputs and complete schema/runtime integration, scoped live-provider verification and the packaged Income journey through tasks 5.1-6.3. Full analyst workflows remain open.
No image was published at that source-only checkpoint; task 6.3 below records subsequent publication.
- Typed enterprise-value input dataset and isolated native command accepted. Fifteen of 26 initial tasks are source-accepted. Evidence.
- Task 4.1: authoritative financial schema projection. All eleven repair tasks are accepted, with real native registry/route agreement, 389 affected tests and final source/size review. Acceptance evidence. Initial batch: 26/26 accepted and published. Broader market-data families remain open follow-on work.
- Task 4.2: descriptor-driven explicit provider guard; baseline defaults preserved, omitted selections rejected before execution. Acceptance. Installed service activation is accepted in task 5.2.
- Task 4.3: scoped safe API failure component, native mounted-route/canary checks and legacy status compatibility. Acceptance. Installed service activation is accepted in task 5.2.
- Task 4.4: preserve and validate financial catalog descriptors and detect consumer drift. Acceptance.
- Task 4.5: schema-driven discovery and a versioned coverage index for all thirteen FMP families, with separate source, access and consumer evidence. Acceptance.
- Task 5.1: register/mount the accepted first-batch models in an exact-source installed candidate. Acceptance.
- Task 5.2: activate discovery, provider policy and safe errors in the exact-source installed service. Acceptance.
- Task 5.3: eleven reusable native widgets and coordinated source/catalog pin promotion. Acceptance.
- Task 6.1: real FMP financial inputs and FRED Treasury context, source/API equality and scoped access evidence. Acceptance.
- Task 6.2: packaged native Income exploration across companies/periods, visible real values, error recovery and save/reopen. Acceptance. Eleven definitions are discovered; this is one complete UI journey.
- Task 6.3: publish and verify both images, the tagged prerelease, three installation assets, six anonymous pulls and installed native catalog. Acceptance. Initial batch complete; the wider specification remains open.
The repair keeps static financial declarations in one owner, with dynamic FRED/result semantics resolved at runtime. Existing adapter values and installed baseline behavior remain intact.
- Approve the native Windows workflow requirements: Podman/Compose runtime, native Dagger and Playwright migration.
- Approve the independently reviewed migration design.
- Approve the reviewed 15-task implementation plan.
- Qualify the native toolchain and command/environment entry: task 2 evidence.
- Verify Windows private-file permissions and checked-handle access: task 3.1 evidence.
- Accept private-file lifecycle and Windows interruption recovery: task 3.2 evidence. Actual POSIX execution remains in the Linux acceptance task.
- Freeze validated runtime descriptors and pipe-only private IPC: task 4 evidence. Compose, browser and build lanes can now proceed independently within the worker cap.
- Accept native synthetic browser infrastructure: task 7 evidence. Financial consumers and journeys remain separate.
- Verify actual native Compose startup, bind mapping and login: task 5 evidence. Recreation and cleanup are accepted in task 6.
- Accept native build-command adapters: task 9 evidence. Actual foundation export/load is accepted in task 10.
- Accept the checked browser-consumer bridge: task 8 evidence. Fork pin promotion and application journeys remain in the integration step.
- Verify state-preserving recreation and owned cleanup: task 6 evidence. POSIX explicit directory deletion fails closed; final Linux acceptance is recorded below.
- Qualify pinned native Dagger foundation export/load: task 10 evidence. Group 5-10 is complete; application-image builds and financial journeys follow the command/source join.
- Join native commands and reviewed source pins; synthetic verification remains distinct from financial acceptance.
- Qualify both native-built application images, generated-file permissions and retained-state recreation: task 11.2 evidence. Financial browser acceptance and final Linux checks are accepted below.
- Correct container DNS selection and verify packaged widget discovery on Podman. Acceptance. The AAP guard correction and all selected financial journeys are accepted. Financial evidence.
- Complete private session storage, Compose, Dagger and Playwright migration, including Linux compatibility acceptance. Closeout. Task graph. Sol workers own separate implementation lanes; Astra reviews difficult integration and the controller joins evidence. Linux CI and server deployment stay supported.