Roadmap
Ploston is actively developed. This page reflects the current state of the project.
OSS Core — Available Now ✅
Everything in this section is shipped and working.
Setup & Deployment
| Feature |
Description |
ploston bootstrap |
One-command Docker Compose deployment — pulls images, starts CP, Redis, native-tools |
ploston init --import |
Scans Claude Desktop / Cursor configs, interactive import, starts local runner |
| Local Runner |
Daemon that connects your local MCP servers to the Control Plane over WebSocket |
| Docker Compose |
Production-ready compose file with health checks and restart policies |
| Kubernetes |
Helm charts + manifests for K8s deployment (ploston bootstrap --target k8s) |
Execution Engine
| Feature |
Description |
| Workflow Engine |
Sequential step execution with full dependency resolution |
| MCP Frontend |
MCP server over stdio and HTTP+SSE — Claude Desktop, Cursor, any MCP client |
| Tool Invoker |
Calls any MCP tool from a workflow step |
| Python Sandbox |
Secure sandboxed code execution with 7-layer security model |
| Jinja2 Templates |
Parameter templating across tool steps, conditions, and outputs |
| Retry Logic |
Per-step configurable retry with exponential backoff |
| Error Handling |
on_error: fail / skip / retry per step |
Control Plane
| Feature |
Description |
| Tool Registry |
Discovers and manages all MCP tools from connected runners |
| Workflow Registry |
Loads, validates, and hot-reloads workflow YAML files |
| REST API |
Full HTTP API for workflows, tools, executions, runners, config |
| HTTP Transport |
MCP over HTTP+SSE with TLS support |
| Config Tools |
MCP-native configuration management (config_get, config_set, config_done) |
| Runner Management |
Multi-runner support with WebSocket connections and persistent registration |
Observability
| Feature |
Description |
| Prometheus Metrics |
Execution metrics, tool call counts, token savings estimates |
| OpenTelemetry Traces |
Distributed tracing for workflow and step execution |
| Structured Logging |
JSON-formatted logs via structlog |
| Telemetry Store |
SQLite-backed execution history with configurable retention |
| Grafana Dashboards |
Pre-built dashboards for execution metrics and system health |
Developer Experience
| Feature |
Description |
| CLI |
ploston run, ploston validate, ploston workflows list/show, ploston tools list |
| Plugin Framework |
PlostonPlugin base class with pre/post execution hooks |
| Plugin Registry |
Auto-discovery and loading of plugins |
| Error Registry |
Consistent error codes across all failure modes |
| Native Tools |
Built-in MCP server: filesystem, HTTP, data transforms, Kafka, Firecrawl, ML |
Security
| Feature |
Description |
| API Key Auth |
Token-based authentication for the REST API and MCP frontend |
| 7-Layer Sandbox |
Import restrictions, builtin restrictions, tool whitelist, rate limiting, parameter validation, timeout, recursion prevention |
Phase 4: Enterprise Foundation — Planned Q2 2026
Security & Governance
| Feature |
Description |
| RBAC / ABAC |
Role and attribute-based access control for workflows and tools |
| Policy Engine |
Cedar/Rego-based policy enforcement at the execution layer |
| SSO Integration |
SAML and OIDC identity provider support |
| Immutable Audit Logs |
Tamper-proof, compliance-ready execution audit trail |
Advanced Orchestration
| Feature |
Description |
| Parallel Execution |
Run independent steps concurrently |
| Compensation Steps |
Automatic rollback on workflow failure |
| Human Approval |
Pause workflow execution for human decision |
| Async Execution |
Fire-and-forget workflows with webhook callbacks |
| Execution Cancellation |
Cancel in-flight workflow executions |
Operations
| Feature |
Description |
| Multi-Tenancy |
Fully isolated execution environments per tenant |
| Multi-Node Runtime |
Distributed execution across multiple runner nodes |
| Kubernetes Operator |
Native K8s CRD-based workflow management |
Phase 5: Enterprise Intelligence — H2 2026
Learning & Optimization
| Feature |
Description |
| Pattern Mining |
Detect repeated tool sequences from agent session telemetry |
| Workflow Synthesis |
Auto-generate YAML workflows from observed patterns |
| Token Cost Accounting |
Per-workflow token savings tracking and ROI reporting |
| Tool Selection Engine |
Suggest the right tool or workflow for a given task |
Enterprise Console
| Feature |
Description |
| Visual Workflow Builder |
Drag-and-drop YAML workflow creation |
| Workflow Library |
Browse, search, and clone workflows across your organization |
| Tool Catalog |
Explore all registered tools with schemas and usage stats |
| Execution Explorer |
Inspect, replay, and debug any execution |
| Pattern Review |
Review LLM-synthesized workflows before publishing |
OSS vs Enterprise at a glance
| Capability |
OSS |
Enterprise |
| Workflow Engine |
✅ |
✅ |
| MCP Integration |
✅ |
✅ |
| Python Sandbox |
✅ |
✅ |
| CLI + REST API |
✅ |
✅ |
| Docker + K8s deploy |
✅ |
✅ |
| Prometheus + Grafana |
✅ |
✅ |
| Plugin Framework |
✅ |
✅ |
| RBAC / ABAC |
— |
Phase 4 |
| Policy Engine |
— |
Phase 4 |
| Parallel Execution |
— |
Phase 4 |
| Human Approval |
— |
Phase 4 |
| Multi-Tenancy |
— |
Phase 4 |
| Pattern Mining |
— |
Phase 5 |
| Workflow Synthesis |
— |
Phase 5 |
| Visual Builder |
— |
Phase 5 |
Design Partners
We're looking for design partners to shape Enterprise features.
Ideal partners: teams running AI agents in production who need deterministic execution guarantees, compliance/governance requirements, or want to reduce agent token costs at scale.
What you get: early access to Enterprise features, direct input on roadmap priorities, dedicated support channel, discounted pricing at GA.
Contact us or open a GitHub discussion.
Release Philosophy
- OSS-first — core execution is always open source
- Enterprise adds, never subtracts — OSS features stay OSS
- Stability over speed — ship when it's ready
- Feedback-driven — roadmap adapts to what users actually need
Stay Updated