How logan23info/ai-agentic-rcm works
AI Agentic RCM is a multi-tenant assurance workbench for audit teams assessing Microsoft and Google AI agents. Its main workflow moves from engagement scoping and agent inventory through applicability, evidence collection, TOD/TOI/TOE testing, signed conclusions, findings, QA, and an issued report. Separate product branches support agent design/readiness, client self-assessment, and monitoring control-library changes. The graph combines README-documented workflows with relationships visible in the source index and excerpts; several route and subsystem implementations were not sampled.
Main entry points: assurance.ts Test conclusions · workflow.ts Control workpapers · service.ts Self-assessment · controls.ts Pinned control library
Components 24
Assurance engagements
- Engagement workspace
apps/web/app/app/e/[id]/page.tsx - Scoping
apps/web/app/app/e/[id]/scoping/page.tsx - Agent inventory
apps/web/app/app/e/[id]/agents/page.tsx - Control workpapers
apps/web/lib/workflow.ts - Evidence
apps/web/lib/evidence.ts - Test conclusions
apps/web/lib/assurance.ts - Findings
apps/web/lib/findings.ts - QA and report
apps/web/lib/export/report.ts
Agent governance
- Design and readiness
apps/web/lib/usecase/service.ts - Self-assessment
apps/web/lib/selfassess/service.ts - Client agent register
apps/web/lib/register.ts - Client 360
apps/web/lib/client360.ts - Optional AI drafting
apps/web/lib/ai
Shared services
- Pinned control library
apps/web/lib/controls.ts - Licence profiles
apps/web/lib/licence-profile.ts - Role and access rules
apps/web/lib/rbac.ts - Application services
apps/web/lib/services.ts - Tenant database
apps/web/lib/db/index.ts
Library monitoring
- Library watch
apps/web/app/app/library/watch/page.tsx - Feed matching and proposals
apps/web/lib/library-watch/service.ts - Scheduled watch
apps/web/app/api/cron/library-watch/route.ts
Other components
- Audit team
- Client team
- Client users
Connections 27
- Audit team → Engagement workspace: works in
- Engagement workspace → Scoping: sets scope
- Scoping → Agent inventory: informs inventory
- Agent inventory → Control workpapers: locks applicability
- Control workpapers → Evidence: requests evidence
- Control workpapers → Test conclusions: records tests
apps/web/lib/workflow.ts - Evidence → Test conclusions: supports testing
- Test conclusions → Findings: informs findings
- Findings → QA and report: feeds QA and issue
- Control workpapers → Test conclusions: uses rules
apps/web/lib/workflow.ts - Control workpapers → Pinned control library: loads controls
apps/web/lib/workflow.ts - Design and readiness → Client agent register: links live agent
apps/web/lib/usecase/service.ts - Self-assessment → Client agent register: reconciles agents
apps/web/lib/register.ts - Self-assessment → Test conclusions: computes risk
apps/web/lib/selfassess/service.ts - Self-assessment → Client team: supports assessment
- Audit team → Client 360: reviews client
- Client 360 → Licence profiles: validates profile
apps/web/lib/client360.ts - Pinned control library → Licence profiles: defines coverage
apps/web/lib/licence-profile.ts - Optional AI drafting → Evidence: drafts from linked text
- Optional AI drafting → Test conclusions: suggests test results
- Application services → Role and access rules: authorizes actions
apps/web/lib/services.ts - Application services → Tenant database: queries data
apps/web/lib/services.ts - Tenant database → Pinned control library: imports controls
apps/web/lib/db/index.ts - Library watch → Feed matching and proposals: manages watch
apps/web/lib/library-watch/service.ts - Scheduled watch → Feed matching and proposals: runs scheduled scan
- Feed matching and proposals → Pinned control library: matches controls
apps/web/lib/library-watch/service.ts - Client users → Self-assessment: self-assesses