SyntheraOS / project information architecture
From project inputs
to reviewed engineering outputs
A complete logical workflow for one project: sources, connected records, analysis, decisions, deliverables and feedback. The detailed diagrams retain the full 60-module inventory.
Based on the 14 September 2026 assessment. These are information handoffs, not a claim that every service is deployed. Human approval, revision-bound evidence and operational feedback are required controls where the prototype does not yet enforce them.
The shared graph is the connecting record. Every discipline reads project context and produces records or views that other disciplines consume. Intelligence is cross-cutting; human review owns decisions. Revisions and lessons return to the graph. Browser-local state and server authority remain distinct.
- G1Intake and source context
- G2Engineering definition and traceability
- G3Assurance, evidence and readiness
- G4Decisions, alternatives and change feedback
- G5Delivery, procurement and deliverables
- G6Operations and organizational learning
- G7Intelligence and project oversight
- G8Identity, persistence and governance boundaries
01 / G1
Intake and source context
Convert external information into project records with an explicit source and owner.
Current capability. Intake includes deterministic fallbacks. File selection does not prove PDF/Word extraction. The marine meeting example can contain water-plant sample text.
Review requirement. confirm source, units, domain and ownership before relying on generated content.
Next use: Used by engineering definition and project intelligence.
Inspect 4 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P01 Project intake | Name; objective; domain; lifecycle; criticality; standards; budget; currency; duration months; team size; extra context; optional source document; quick natural-language brief | Structured ProjectGraph with project, engineering nodes and relationships; suggested standards; draft requirements, risks, architecture and delivery items | Source implemented with deterministic generation fallback; external generation service unverified PDF/DOC/DOCX file selection is not proof of content extraction; text-like files can supply text. Standards suggestions are presets, not authoritative compliance determinations. Source: lib/intake/schema.ts:21; app/(authed)/projects/new/page.tsx:128,148,180 |
| P02 Project JSON interchange | User-selected JSON project file; schema/graph fields | Validated/imported project graph and errors; settings export downloads project JSON | Source implemented browser file processing Separate from document report format buttons. Imported field completeness and provenance need acceptance tests. Source: lib/ai/import-project.ts; app/(authed)/projects/new/page.tsx:193; app/(authed)/projects/[projectId]/settings/project/page.tsx:81 |
| P41 Meeting intelligence | Pasted transcript; sample transcript; extraction action; selected extracted object for conversion | Proposed requirements/decisions/risks/assumptions/actions/issues; confidence; source text; graph object conversion | Deterministic client extraction; graph conversion implemented Marine project falls through to island-water-plant sample/domain label. No audio recording, transcription or live meeting connection demonstrated. Source: meeting-intelligence/page.tsx:16,33,43; lib/intel/meeting.ts |
| P52 Project settings | Existing Project fields from brief/import; export-master-file click. Project name/system type/industry/location/capacity/budget/timeline/lifecycle/objectives/frameworks are read-only on this page. | Read-only project metadata, objectives and regulatory environment; downloadable portable master project JSON. | Source implemented read-only settings and browser JSON download. Project details must be changed through re-import of an updated master file; this settings page has no edit or reset control. Source: settings/project/page.tsx:10; lib/ai/import-project.ts |
02 / G2
Engineering definition and traceability
Connect what the project must achieve to how the system is structured and verified.
Current capability. Authoring, graph relationships and diagrams exist. Baseline labels do not prove immutable snapshots; a legal link may still connect the wrong engineering concepts.
Review requirement. validate acceptance criteria, semantic links and the revision being released.
Next use: Used by assurance, change analysis, cost and schedule.
Inspect 9 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P10 Ontology catalogue | Selected object class or relationship; ontology metadata | Class definitions; expected fields; allowed relationship semantics; vocabulary and lifecycle context | Source implemented static ontology plus graph references; live bundle confirmed 36 NodeType declarations; some classes have no dedicated create/edit route. Declared fields differ from some canonical ontology field names. Source: components/ontology/ontology-explorer.tsx; lib/types.ts; live bundle 17o10c5d2hu0b.js |
| P12 Requirements register | Search ID/name/owner/text; sort/filter; requirement graph records | Requirement table; quality scores/flags; status, level, owner and method; requirement detail links | Source implemented read/filter interface Displayed qualityScore can originate in seeded/imported data; assess consistency with recomputed analyzer scores. Source: app/(authed)/projects/[projectId]/requirements/page.tsx:44 |
| P13 Requirement authoring and links | Need text; level; requirement type; name/statement; rationale; acceptance criteria; verification methods; owner/status; component/risk/case selection | Draft Requirement; edited/versioned records; allocation/risk/verification links; AI review; updated local graph/audit trail | Source implemented graph mutations; live write persistence not exercised AI drafting and autosuggest need human review. Mutation rules and server enforcement are separate assessment topics. Source: components/requirements/ai-draft-requirement.tsx:32; requirements/[reqId]/page.tsx:34; components/common/inline-field.tsx |
| P15 Requirement traceability and baselines | Requirement graph; selected allocation target; baseline/status/change records | Coverage matrix and gaps; allocation links; baselines listing; requirement change-request cross-links | Source implemented secondary routes; HTTP render not sampled individually Baseline labels and status are not evidence of immutable snapshot management or electronic approval. Source: requirements/traceability/page.tsx:79,118; requirements/baselines/page.tsx:13; requirements/change-requests/page.tsx |
| P16 Architecture | Stakeholders, needs, functions, systems, subsystems, elements, interfaces; selected context/functional/physical/operational view | Four architecture views; physical hierarchy; function allocation; operational context; counts and gaps | Source implemented graph projection; architecture bundle matched local output Operational narrative and system relationships are model descriptions rather than executed simulation. Source: architecture/page.tsx:11; architecture/context/page.tsx; architecture/physical/page.tsx; architecture/operational/page.tsx |
| P17 Architecture authoring | Architecture drafting prompt; function; target physical element; allocation/removal selection | Proposed architecture objects; function-to-element allocation/unallocation; corresponding relationship and success/error messages | Source implemented graph writes; no write exercised UI draft quality, cross-domain defaults and validation require representative project testing. Source: components/architecture/ai-draft-architecture.tsx; architecture/functional/page.tsx:28,44 |
| P18 Interfaces | System A; system B; interface name; kind; technical specification; verification method; owner/status; linked risks/issues | Interface record; endpoints; N-squared matrix; definition gaps; linked risks/issues and verification context | Source implemented UI and graph writes technicalSpec is free text; no demonstrated signal-level payload, units, rates, direction, timing or safe-state schema. Source: interfaces/page.tsx:56,93; interfaces/[interfaceId]/page.tsx:38; lib/types.ts:373 |
| P32 Assumptions | Assumption text; owner/status; confidence low/medium/high; needs-confirmation flag | Assumption register; confidence and confirmation indicators; downstream trace context | Source implemented graph records Confidence is asserted metadata unless a method, evidence and reviewer are supplied. Source: assumptions/page.tsx:18; lib/types.ts:404 |
| P48 Diagram studio | Diagram/notation type; selected root/scope/depth; graph nodes/edges; layout/position changes; export format | Generated diagram; validation diagnostics; layout; downloadable SVG and PNG; browser print-to-PDF | Source implemented diagram generation/export SVG/PNG paths create artifacts, unlike non-PDF document buttons. Not proof of full native SysML/MBSE interchange or executable model semantics. Source: components/diagrams/DiagramStudio.tsx; DiagramExportPanel.tsx:17; lib/services/diagram |
03 / G3
Assurance, evidence and readiness
Separate planned verification, recorded results and accepted evidence for the current requirement.
Current capability. Current logic can count stale passes and label approval as compliance satisfaction. Entering a pass is not execution of the physical test. Signer text is not a verified digital signature.
Review requirement. bind evidence to the exact requirement, test and baseline revisions before acceptance.
Next use: Failures and gaps create issues or change requests; affected requirements need re-verification.
Inspect 11 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P11 Ontology intelligence | Graph, declared class/relationship rules and selected ontology lens | Conformance findings; illegal/dangling relationships; missing required traceability/provenance; ontology statistics and guidance | Source implemented rules; live bundle confirmed Rules are finite programmed checks; do not substitute for semantic engineering validation. Source: components/ontology/ontology-intelligence.tsx; lib/data/ontologyValidationRules.ts; live bundle 17o10c5d2hu0b.js |
| P14 Requirement quality | Requirement statements, criteria, methods and traceability; selected requirement; quality-analysis action | Quality scores; ambiguous/non-testable/missing criteria/duplicate/conflict/traceability flags; proposed improved wording and evidence links | Source implemented heuristic/analysis presentation; external augmentation unverified A rewrite suggestion is an output proposal until explicitly applied to graph. Source: app/(authed)/projects/[projectId]/requirements/quality/page.tsx; components/req-quality/requirement-intelligence-panel.tsx |
| P19 Risks | Search/status/band/sort; cause/consequence; probability/impact; mitigation; residual probability/impact; owner; due date; links to requirement/component/test/interface | Risk register and heatmap; inherent/residual scores; individual risk analysis; linked records; AI finding; proposed change requests | Source implemented CRUD/derived analysis Risk numbers are ordinal 1–5 model inputs; residual score is not a measured risk reduction or validated prediction. Source: risks/page.tsx:49; risks/[riskId]/page.tsx:52,160,285; lib/types.ts:386 |
| P20 Verification cases | Requirement target; case name; method Test/Inspection/Analysis/Demonstration/Simulation; stage; procedure; search and outcome filter | Verification plan/case; VERIFIES link; matrix; coverage; case details; pass/fail/planned counts | Source implemented local-first graph creation A case row proves planned coverage only; it does not prove completed testing. Source: components/verification/new-verification-case.tsx:35; verification/page.tsx |
| P21 Test results | Case ID; pass/fail/conditional outcome; measured value; notes/evidence reference; editable outcome | TestResult node; RESULT_OF edge; case lifecycle update; result display and acceptance eligibility | Source implemented local-first graph mutation Evidence capture is text/reference oriented; no demonstrated raw measurement upload, instrument traceability, calibration or independent witness verification. Source: components/verification/record-result.tsx:39; verification/results/[resultId]/page.tsx:25 |
| P22 Acceptance certificates | Passed test-result ID; signer text; optional certificate number and remarks | AcceptanceCertificate record; generated number if blank; signed date; result acceptance link; audit entry | Source implemented graph record creation; authority/identity service unverified Typing Signed by is not equivalent to cryptographic signature or authenticated approval authority. Source: components/verification/issue-acceptance-cert.tsx:21; lib/types.ts:662 |
| P23 Compliance | Regulatory frameworks; compliance-type requirements; verification methods; evidence kind/reference/note and target; owner/status | Compliance matrix; attached Evidence node and EVIDENCES link; framework coverage and conformance findings; deep-link check route redirects to matrix | Source implemented model-derived checks and evidence records No proof of up-to-date licensed standards ingestion, clause interpretation, accredited review or regulatory acceptance. Source: app/(authed)/projects/[projectId]/compliance/page.tsx:46,59; compliance/[checkId]/page.tsx:8; lib/types.ts:647 |
| P33 Review board | Project graph; run-review selection; review agent roles/rules | Discipline review findings; severity; readiness verdict; chair synthesis; critical/major/confidence counts | Client analysis/simulated-agent presentation plus optional K service unverified Multiple displayed reviewers do not establish independent model calls, independent engineering review or accreditation. Source: review-board/page.tsx:21; lib/intel/review-board.ts; components/review-board |
| P34 Completeness | Project graph; run scan; selected missing item; local owner/status/not-applicable overrides; generate-missing action | Completeness percentage; present/absent/not-applicable checklist; critical gaps; draft missing requirements; optional K output | Client-derived rules with page-local overrides; external K unverified Changing checklist override changes display state, not necessarily the underlying controlled artifact or persistent evidence. Source: completeness/page.tsx:30,38,73; components/completeness |
| P37 Risk prediction | Project graph signals; risk/verification/interface/schedule state; selected forecast; create-risk action | Predicted risk cards; confidence; mitigation suggestions; quantification; emerging-risk record creation; optional K signals | Source implemented heuristic predictions plus optional external K service unverified Prediction labels and confidence percentages lack demonstrated calibration or external outcome validation. Source: risk-prediction/page.tsx:10; components/risk-prediction/risk-prediction-dashboard.tsx |
| P57 Review and readiness gates | Project graph; selected design/compliance/construction/commissioning/operations route; compliance signer name | Gate checklists; pass/block criteria; design/compliance review summary; compliance sign-off record; readiness scores | Client rule checks and some graph write-back; secondary routes source inspected Some gates equate approved-count thresholds with baselined readiness; sensor count with telemetry configuration. These proxies need explicit stronger acceptance rules. Source: reviews/design/page.tsx:8; reviews/compliance/page.tsx:17,68; readiness/*/page.tsx |
04 / G4
Decisions, alternatives and change
Turn an issue or proposed change into a reasoned decision and a controlled feedback loop.
Current capability. Impact estimates are graph heuristics and ignore change-description magnitude. Some costs may be missing. PRISM code paths exist; deployed service enforcement is separately qualified.
Review requirement. a model recommendation must not approve itself. Re-verify after a relevant change.
Next use: Apply accepted changes to engineering records; re-open affected assurance checks.
Inspect 5 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P27 Change control | Change title/reason; affected requirement/component/risk IDs; cost impact; schedule days; approval decision and rationale | ChangeRequest; affected-object grouping; impact analysis; submitted/under review/approved/rejected/baselined states; linked audit | Source implemented write path and lifecycle rules Approval status requires identity/authorization and authoritative persistence assessment; presence of AI impact text is not independent validation. Source: changes/new/page.tsx:13,32; changes/[crId]/page.tsx:28,71,92; lib/types.ts:505 |
| P30 Decisions / PRISM | Decision statement; rationale; alternatives; status quo; linked risks/requirements; falsification criteria; review trigger; PRISM tier/focus; reviewer disposition; outcome status | Decision register; process-quality score; adversarial debate/record; confidence, research and assumptions; approved write-back; review/outcome history | Source implemented local decision workflow; live PRISM/backend capabilities separately unverified Model output and citations are not automatically verified evidence. Human approval and research verification are distinct inputs. Source: decisions/page.tsx:27; components/decisions/prism-decision-panel.tsx; lib/types.ts:410; live bundle 17o10c5d2hu0b.js |
| P31 Issues | Issue name/description; severity; related object ID; owner/status; resolution | Issue register; severity/status presentation; related-object navigation and resolution record | Source implemented stored graph records Issue closure state should be backed by linked verified resolution evidence. Source: issues/page.tsx:21; lib/types.ts:518 |
| P35 Impact simulator | Selected requirement; proposed change text; change type/preset; graph dependency links | Affected objects and propagation graph; blast radius; cost/rework and schedule estimates; approval warning; draft ChangeRequest | Source implemented deterministic graph-impact model plus optional K response unverified Estimates are heuristics over stored links, not CFD/FEA, schedule simulation or validated cost engineering. Source: impact-simulator/page.tsx:27,49,63; components/impact-simulator/change-input-panel.tsx |
| P36 Trade studies | Authored study/options/criteria; Generate action; criterion weight sliders; selected option; PRISM query selection | Weighted comparison matrix; ranking and recommendation; CAPEX/OPEX comparison; downloadable CSV of current matrix | Client authored study with local recomputation; genuine CSV download implemented Generate reveals preset graph-appropriate content. Sliders do not demonstrate option discovery, external research or persistent design decision. Source: trade-studies/page.tsx:17,59,66; lib/intel/trade-studies.ts |
05 / G5
Delivery, procurement and deliverables
Use project records for delivery planning, and distinguish a preview from a produced artifact.
Current capability. Word/Excel/CSV/JSON buttons on document detail do not create a file. Stated page counts are metadata. A print pathway does not prove complete document assembly.
Review requirement. declare success only after the correct, source-bound artifact actually exists.
Next use: Real source-code exports elsewhere: project JSON, diagram SVG/PNG and trade-study CSV.
Inspect 5 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P24 Cost and procurement | CAPEX/OPEX; CBS code; linked component/supplier; estimated/committed/actual values; currency; procurement stage | Cost register; totals and variance; committed-versus-estimated visual; item detail and linked supplier/component | Source implemented graph model and calculations Not a financial ledger or ERP integration; actual cost values are modeled inputs unless independently sourced. Source: cost/page.tsx:50; cost/[costItemId]/page.tsx:33; lib/types.ts:471 |
| P25 Schedule | Activity start/end; milestone flag; predecessors; component/test links; delay-risk band; optional cost impact | Schedule/Gantt-like view; milestones; activity detail; dependencies and linked verification | Source implemented graph-derived schedule No demonstrated enterprise scheduler import, resource leveling, baseline comparison or full CPM calculation. Source: schedule/page.tsx:13; schedule/[activityId]/page.tsx:16; lib/types.ts:484 |
| P26 Suppliers | Supplier identity/organisation; scope; quote amount; procurement state; linked components and owner/status | Supplier register; quote/procurement information and linked supply scope | Source implemented stored records Supplier names/amounts in sample are demonstration data, not independently verified quotations. Source: suppliers/page.tsx:13; lib/types.ts:496 |
| P28 Documents | Existing Document nodes; sample-reference add action; format selection | Document catalogue; descriptions; source-object lists; stated pages and available formats; sample Drawing metadata nodes | Metadata/reference presentation and simulated sample add; PDF print path; non-PDF export success only Add sample reference cycles canned names; no file picker/upload. Word/Excel/CSV/JSON export buttons do not create a file in inspected source. Source: documents/page.tsx:28,47,75; documents/[docId]/page.tsx:68; deployed /proto/_next/static/chunks/2v60drsv6g4ot.js |
| P29 Reports | Document graph nodes excluding Drawing; source-object IDs; project metrics | Report cards and document previews; maturity/risk/requirements/quality/verification KPIs; regenerated-ago label | Graph metadata presentation, not demonstrated report assembly The preview is description plus linked source objects. Stated page counts are data fields, not generated pagination evidence. Source: reports/page.tsx:26; documents/[docId]/page.tsx; deployed /proto/_next/static/chunks/2v60drsv6g4ot.js |
06 / G6
Operations and organizational learning
Bring operating evidence back into engineering decisions, with live ingestion explicitly marked as planned.
Current capability. The ferry twin is not connected to live telemetry. Signal maps, timing, calibration, quality and failure responses are incomplete. Physical actuation through SyntheraOS was not demonstrated.
Review requirement. distinguish a modelled sensor from a measured value and an accepted operating result.
Next use: Feed future project briefs and affected current decisions; this is not autonomous learning.
Inspect 2 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P40 Engineering memory | Decision and assumption records; search; all/decision/assumption/unconfirmed filters | Searchable memory timeline; governance summary; superseded/needs-confirmation counts and linked objects | Source implemented graph-memory view plus optional K unverified Local view of project records is not proof of enterprise cross-project learning or validated retrieval over external documents. Source: memory/page.tsx:17,40,117 |
| P51 Digital twin | Sensor node definitions; protocol; monitored component; metric/unit; optional linked requirement | Sensor catalogue; four ferry telemetry placeholders; awaiting-telemetry state | Modelled only; explicitly not connected to live telemetry No streaming values, historical ingestion or live device connection established; future operations claim is prospective. Source: twin/page.tsx:13; lib/types.ts:572; lib/seed/eferry-coastal.ts:329 |
07 / G7
Intelligence and project oversight
Derive views and advice from the same project context; preserve provenance and human authority.
Current capability. Mix of stored sample recommendations, local rules, model paths and optional backends. The optional enterprise health service was offline; local browser health scores remained available. Understanding workbench was source-only in this review.
Review requirement. show execution mode, source revision, engine/model version and missing-data state.
Next use: Supports every workflow; advice becomes a record only through the appropriate write/review path.
Inspect 17 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P03 Project overview | Project graph; selected project; dashboard panel priority; sample-banner dismissal | Project objectives and attributes; maturity/health; requirements/quality/verification/risk/architecture/cost metrics; priority actions; AI recommendation cards; graph navigation | Derived client presentation over graph; live route HTTP200 Metrics describe stored model records. They do not independently establish vessel safety, real verification or engineering maturity. Source: app/(authed)/projects/[projectId]/page.tsx:129; lib/graph-utils.ts |
| P04 Action center | Requirements, risks, interfaces, changes, issues and derived findings; action navigation | Prioritized action queue; critical/attention/total counts; linked objects and proposed next steps; optional K-engine action results | Derived client rules plus optional external K service unverified One source defect can appear repeatedly across multiple dashboards; counts should not be summed across modules. Source: app/(authed)/projects/[projectId]/action-center/page.tsx:26 |
| P05 Engineering health | Current project graph and scoring rules | Overall health and category scores; category explanations; weaknesses and recommendations; optional server health output | Derived client analysis plus optional K service unverified Health is a modeled completeness/control score, not independent assurance or measured probability of failure. Source: app/(authed)/projects/[projectId]/health/page.tsx:18; lib/graph-utils.ts |
| P06 Lifecycle GPS | Project lifecycle stage; nodes, evidence and gates; selected stage | Lifecycle path; stage completion/maturity; stage-specific required artifacts, blockers, recommended actions and linked records | Source implemented graph-derived stage UI Stage completion relies on presence/status heuristics; formal authority and evidence quality must be verified separately. Source: app/(authed)/projects/[projectId]/gps/page.tsx:18; components/gps/stage-detail.tsx; lib/gps-config.ts |
| P07 Engineering graph | Graph nodes/relationships; filters; selected object; view controls | 2D/3D relationship view; node detail; neighborhoods; navigation and trace context | Source implemented visual projection Edges displayed are stored assertions; link existence does not prove engineering correctness. Source: app/(authed)/projects/[projectId]/graph/page.tsx:19; components/graph/engineering-graph.tsx |
| P08 Views catalogue | Project graph; selected view/template | View catalogue with readiness/status/counts; selected graph-derived engineering projection; missing-data guidance | Source implemented projection catalogue Availability of view type is not proof all notation semantics, editing and exchange capabilities are complete. Source: app/(authed)/projects/[projectId]/views/page.tsx:20; views/[viewId]/page.tsx:24 |
| P09 Traceability thread | Selected root requirement/object; graph links; local filters | Need→requirement→function/physical element→verification/result path; traceability matrix; verified trace coverage; broken links | Source implemented graph traversal Distinct from digital-thread route. Empty or incorrect graph relationships directly affect results. Source: app/(authed)/projects/[projectId]/thread/page.tsx:20; components/thread/digital-thread-view.tsx |
| P38 Recommendations | AIRecommendation graph nodes; category/source/confidence/cited IDs; user selection | Recommendation register; sources, assumptions, confidence and next actions | Stored/generated recommendation presentation Some recommendations are hard-coded seed records with modelUsed/promptVersion labels; these labels do not prove a model ran. Source: recommendations/page.tsx; lib/types.ts:597 |
| P39 Executive brief | Current graph; health/risk/cost/schedule/change/readiness metrics | Executive dashboard and narrative; weakest dimension; graph integrity; linked critical issues and optional K-engine briefing | Derived client summary plus optional external K unverified A synthesized brief inherits every uncertainty and stale/missing source record in the graph. Source: executive/page.tsx:33 |
| P42 AI assistant | Free-text question; Ask/quality/gaps/generate/change/summary mode; project graph/focused object; context and conversation state | Answer with cited objects/source classes/confidence; summaries; quality/gap findings; draft objects/change proposals; errors/fallbacks | Source implemented client+API path; live service outcome separately unverified Prompt, graph data and external model traffic are information outputs to service; exact provider deployment/retention not established by UI. Source: assistant/page.tsx:35,47,70; lib/ai |
| P43 Engineering query / EQL | Natural-language query; suggested example; query history selection; graph objects/relationships | Parsed intent; filtered object list/table; explanation; source trace; confidence/warnings; query history | Local query interpretation/evaluation implemented; relevant live bundle matched local output Query coverage is bounded by supported grammar/intents; unknown requests should produce uncertainty rather than invented results. Source: components/eql/EngineeringQueryDashboard.tsx; components/eql/EQLSearchBar.tsx; lib/services/eql |
| P44 Graph intelligence | Selected object; traversal direction/depth; source and target; graph dependencies | Graph metrics; impact/dependency map; traceability paths; broken-link and source-context panels | Source implemented graph algorithms Computed paths establish graph connectivity only, not physical causality. Source: components/graph-intelligence/GraphIntelligenceDashboard.tsx; lib/services/graph |
| P45 AI context engine | AI task type; target object; depth; mode; optional question; graph objects/relationships | Context package/preview; source list; relevance; warnings; quality and token estimate; AI answer when service succeeds | Client context build implemented; external AI request unverified Context quality score is not groundedness proof; leakage/tenant scope and evidence completeness must be tested separately. Source: components/ai-context/AIContextEngineDashboard.tsx:58,88,121 |
| P46 Digital-thread assessment | Graph objects/relationships; selected trace/path/timeline lens | Thread completeness score; broken paths; audit panel; object timeline; linked lifecycle trace | Source implemented graph-derived dashboard Different route and scoring basis from thread; align definitions to avoid contradictory coverage claims. Source: components/digital-thread/DigitalThreadDashboard.tsx; lib/services/thread |
| P47 Graph health | Graph; health rules; chosen issue/remediation/benchmark/trend lens | Health score/breakdown; issues; remediation; AI-readiness; benchmark/trend presentation | Source implemented health analysis; historical/benchmark provenance requires verification Presence of mock scenario files warrants checking runtime selection; do not treat benchmark/trend values as measured longitudinal data without provenance. Source: components/graph-health/GraphHealthDashboard.tsx; lib/data/mockGraphHealthScenario.ts |
| P50 Agent catalogue | Selected agent metadata/capability; status catalogue; graph context navigation | Agent cards with active/in-design/planned state; role/capability descriptions | Source implemented capability catalogue An Active label alone is not runtime evidence of an independently operating agent or successful model/tool execution. Source: agents/page.tsx:25,33 |
| P58 Understanding workbench | Question; template; assurance level; role; target object IDs; counterfactual ID; baseline-as-of; attestation; comprehension answer | Compiled understanding state; canvas/evidence/impact/time/decision/checks/quality views; engine JSON packet; release request result | Local route/API source present; not included in 45 live route sampling and service unverified Treat as repository capability until deployed route and backend are validated. Source: understanding/page.tsx:83,318,328,554; lib/m-api.ts |
08 / G8
Identity, persistence and governance
Keep browser state, external services and authoritative project controls as separate boundaries.
Current capability. Browser storage, Supabase paths and a separate Python backend coexist. General cloud graph sync is disabled. Invitations and connection toggles are simulations; local audit is not immutable.
Review requirement. fail closed, reject stale writes, enforce project access and retain evidence/revision history.
Next use: Applies across every project workflow. No single deployed authoritative storage path was established.
Inspect 7 mapped modules and their exact inputs / outputs
| Module | Information inputs | Information outputs | Implementation and limits |
|---|---|---|---|
| P49 Access log | Current authenticated/local actor; page/action events; local ring buffer | Activity/event table for access/actions | Local browser storage ring buffer; guest visibility restricted in source Not equivalent to immutable central security audit, server access log or cross-device forensic trail. Source: access-log/page.tsx:4; lib/activityLog.ts |
| P53 Members | Stakeholder records; invite email; cancel-invite action | People list; in-memory Invitation pending display; validation messages | Simulated local invite state; no email/network call in page Send invite does not demonstrate invitation delivery or access provisioning; people list is derived from Stakeholder nodes. Source: settings/members/page.tsx:26,37,54 |
| P54 Roles | Role catalogue / current persona context | Role descriptions and permissions presentation | Source implemented role catalogue; authorization enforcement separately assessed Sixteen roles in schema. Displayed role selection is not a substitute for server-enforced permissions. Source: settings/roles/page.tsx:33; lib/types.ts Role |
| P55 Integrations | Connect/disconnect clicks for Slack and Jira/Azure DevOps; catalogue entries for Postgres/Neo4j/Vector DB/Redis | Simulated Connecting then Connected labels; status cards | Explicitly simulated in preview; no OAuth credentials exchanged Do not count this page as working Slack/Jira/DevOps integration or proof every listed core datastore is deployed. Source: settings/integrations/page.tsx:8,32 |
| P56 Audit history | Node auditTrail data; actor/action/field/from/to/date; filter | Combined chronological audit feed with kind markers | Source implemented graph-derived audit UI Audit array provenance/integrity is distinct from append-only server audit and signed change evidence. Source: settings/audit/page.tsx:42,105; lib/types.ts:173 |
| P59 Platform identity and navigation | Username and password; password visibility toggle; existing session; current role/persona; search/command palette; theme; selected project | Session/auth state; error/redirect; role-gated views; navigation results; persistent local UI preferences | Client/auth source present; no credentials submitted in assessment Authentication and role authorization are separate; assessment did not create an account or perform privileged mutations. Source: app/(auth)/login/page.tsx:46,63; lib/gate.ts; components/layout/top-bar.tsx; components/layout/command-palette.tsx |
| P60 Shared graph and external service boundary | Bundled seed; local graph override; imported JSON; optional cloud-loaded graph; typed node/edge changes; AI requests | Browser store/local persistence; optional sync payloads; service requests; graph JSON; audit entries; recalculated views | Live bundle confirms loadOverride/saveOverride, Supabase graph loader, mutation guards and API pathways; service success not established by presence Separate model information leaving browser for AI/cloud services from UI outputs. Exact configured backend, tenancy, retention, backups and network behavior require runtime evidence. Source: live 17o10c5d2hu0b.js; lib/store/graph-store.ts; lib/sync; lib/data/load-graph.ts |