Roadmap¶
Current Status¶
SynthOrg is in active development at v0.6.x. The core subsystems are built, tested (13,000+ unit tests, 80%+ coverage), and integrated through a REST + WebSocket API, React 19 dashboard, and Go CLI.
What works today:
- Agent engine with ReAct, Plan-and-Execute, Hybrid execution loops, crash recovery, and task decomposition
- Budget & cost management with per-agent limits, auto-downgrade, spending reports, and anomaly detection
- Security with fail-closed rule engine, 4 autonomy tiers, progressive trust, output scanning, and audit logging
- Memory with hybrid retrieval (dense + BM25 sparse), tool-based injection, procedural memory auto-generation from failures, and consolidation
- Communication with message bus, delegation, conflict resolution, and meeting protocols
- Workflow engine with Kanban, Agile sprints, ceremony scheduling (8 strategies), visual workflow editor, and workflow execution from graph definitions
- Web dashboard (React 19 + shadcn/ui) with org chart, task board, agent detail, budget tracking, provider management, workflow editor, ceremony policy settings, and setup wizard
- CLI (Go) with init, start, stop, doctor, config, wipe, cleanup, and cosign/SLSA verification
- Docker deployment with Chainguard distroless images, Trivy + Grype scanning, and cosign signatures
- Multi-user access with JWT auth, session management, and concurrent access handling
- Local model management for Ollama and LM Studio (browse, pull, delete, configure launch parameters)
What's not there yet:
- End-to-end production runs -- subsystems are integrated but the full autonomous loop (agents receiving work, executing, producing artifacts, iterating) has not been validated as a cohesive product
- PostgreSQL persistence -- SQLite only (planned for v0.8)
- Distributed backends -- message bus, task queue, and persistence are local-process only (planned for v0.8)
- Tool ecosystem -- built-in tools cover file system, git, sandbox, and MCP bridge; web, database, terminal, and other tool categories are planned for v0.7
What's Next¶
All work is tracked on the GitHub issue tracker with version labels (v0.6, v0.6.2, etc.).
v0.6 (current) -- Workflows, Memory, Providers¶
- Capability-aware prompt profiles for model tier adaptation
- Workflow execution lifecycle (COMPLETED/FAILED transitions)
- Workflow editor improvements (list page, YAML import, conditionals, minimap, copy/paste)
- Procedural memory auto-generation from agent failures
- Quality scoring Layers 2+3 (LLM judge + human override)
- Fine-tuning pipeline for embedding models
- Memory consolidation upgrades (LLM Merge, Search-and-Ask)
- Coordination metrics pipeline
- Automated reporting system
v0.7 -- Tools, Security, Research¶
- Sandbox security improvements (auth proxy, gVisor, Chainguard packages)
- Core tool categories (web, database, terminal/shell)
- Advanced memory research (GraphRAG, consistency protocols, RL consolidation)
- Safety classifier for approval gates
- Hallucination detection
- Plugin system, benchmarking suite, A2A protocol compatibility
v0.8 -- Production Hardening¶
- PostgreSQL persistence backend
- Distributed message bus and task queue
- Dynamic company scaling
- Multi-project support
See Future Vision for the longer-term direction and Open Questions for unresolved design decisions.