Molecule AI
ArchitectureTechnical Reference

Operations, Licensing & Roadmap

Recent features, known gaps, licensing, OSS growth, technical debt, production deployment, summary statistics, and the vision.

Part of the Comprehensive Technical Documentation. Definitive reference based on a non-invasive scan of the molecule-core repository.

22. Recent Feature Highlights

FeatureDescription
A2A streaming responseReal-time task result delivery via SSE (message/sendSubscribe)
Onboarding wizard4-step guided first-run experience in Canvas
Global API keysPlatform-wide secrets with per-workspace override + AES-256 encryption
Coordinator enforcementTeam leads cannot do work, only route and aggregate
Cascade pause/resumePausing a parent cascades to all children; paused children can't be individually resumed
Graceful A2A errors[A2A_ERROR] sentinel + retry with exponential backoff + fallback
Canvas error boundaryReact class component catches render errors, shows retry button
Hydration retryBanner with "Retry" button + PLATFORM_URL hint on WebSocket stale state
Activity log retentionConfigurable cleanup (default 7 days, ACTIVITY_RETENTION_DAYS)
Security hardeningHub double-close race fix (sync.Once), A2A proxy timeout (5min canvas, ∞ workspace), Python JSON decode guards

23. Known Gaps & Backlog

Test Coverage

18 of 26 Go handler files have zero unit tests: a2a_proxy, workspace, templates, registry, discovery, secrets, etc. Current: 278 tests with 25% baseline enforced.

Silent Failures

6+ locations with fire-and-forget ExecContext DB writes need proper error handling (activity log inserts, event broadcasts).

Python Tool Error Handling

Tools call resp.json() without catching JSON decode errors. Should wrap in try/except for malformed responses.

Branch-Level Work

BranchFeatureStatus
feat/nemoclaw-t4-dockerNemoClaw adapter (NVIDIA T4 support)WIP
BacklogFirecracker backend (faster cold starts)Planned
BacklogE2B backend (cloud-hosted code sandbox)Planned
Backlogpgvector semantic memory searchPlanned
BacklogCanvas search, batch operations, keyboard shortcutsPlanned

24. Licensing & Commercialization Path

Open Source (Current)

  • License: MIT
  • Strategy: Maximize adoption, zero friction
  • Model: Follows n8n Community Edition approach

SaaS Path (Future molecule-cloud repo)

FeatureTechnology
AuthenticationClerk or Auth.js
Multi-tenancyorg_id column added to schema
BillingStripe integration
Managed infrastructureECS + Neon + Upstash
White-labelingCustom Canvas branding

Key principle: No changes to core open-source repo. SaaS layer is purely additive.


25. OSS Growth Research

Analysis of 8 OSS agent projects (from oss-agent-growth-research.md):

Winning Launch Formula

[Viral Demo] + [HN Front Page] + [One Major Amplifier] + [Zero-Friction Install]
     ↓              ↓                   ↓                         ↓
  60s video     400+ upvotes      Karpathy / Altman /       docker compose up
  screen rec    top comment       Major YouTuber             3 commands max

Every Tier 1 launch (Open Interpreter, CrewAI) had all four elements.

Documentation Best Practice (Diataxis Model)

TypePurposeExample
TutorialsLearning-oriented"Build your first agent team in 5 minutes"
How-to guidesTask-oriented"How to configure RBAC for production"
ExplanationUnderstanding-oriented"Why memory follows org boundaries"
ReferenceInformation-orientedAPI route tables, config schema

26. Technical Debt & Constraints

Hard Design Constraints

  1. Platform never routes agent messages — A2A is strictly peer-to-peer
  2. Postgres is fact source, Redis is cache — Redis loss is fully recoverable
  3. structure_events is append-only — Never UPDATE, never DELETE
  4. workspace-template has no business logic — Logic lives in workspace-configs-templates/
  5. Bundles never include secrets — API keys forbidden from serialization
  6. Hierarchy = topology — No manual edge wiring; all communication derived from parent_id

27. Production Deployment

Multi-Host Configuration

  • Docker-internal URLs (http://ws-{id}:8000) work directly between containers
  • Nginx on host handles TLS termination
  • For external HTTPS: proxy requests to host-mapped URLs

Volume Management

ModeConfigurationBehavior
DefaultNo WORKSPACE_DIREach workspace gets isolated Docker volume ws-{id}-workspace
SharedWORKSPACE_DIR=/pathAll agents mount same host directory (read/write)

29. Summary Statistics

MetricValue
Python runtime files95
Go platform files94
TypeScript/JS canvas files62
Runtime adapter implementations6
Go handler files26
Postgres migrations11
Core workspace tools14
Platform API endpoints40+
MCP tools20+
Go tests278 (with -race flag)
Canvas Vitest tests188
Python pytest tests148
Total tests614
Activity retention7 days (configurable)
Heartbeat interval30s (default)
Redis liveness TTL60s
Health sweep interval15s (proactive)
Skill hot-reload propagation~3 seconds
Coverage baseline (Go)25% enforced in CI

30. Vision: From Agent Teams to Robot Teams

Molecule AI's workspace abstraction is runtime-agnostic by design. A workspace is a role with an A2A interface — not an LLM with a prompt. The same hierarchy, memory boundaries, approval chains, and governance that organize AI agents in containers today can organize any autonomous system that speaks A2A.

PhaseEraSystemsStatus
NOWSoftware Agent TeamsLLM agents in Docker, 7 runtimes, HMA, Langfuse, A2ALIVE on main
NEXTTerminal + Device AgentsTerminal bots, browser agents, IoT controllers, CI/CD agentsBUILDING
HORIZONEmbodied Robot TeamsWarehouse robots, autonomous vehicles, manufacturing cells, field inspectionHORIZON

The workspace is the role. The protocol is A2A. The boundary between digital and physical disappears — the organizational layer remains.


On this page