Skip to main content

Roadmap

Mango is focused on a reliable, self-hosted implementation of the core Claude Managed Agents session API. Priorities are ordered by user-visible capability, not by reproducing Anthropic's internal implementation.

Core compatibility

These are the release-blocking capabilities for the managed-agent harness:

  1. autonomous recovery for interrupted deletion plus sandbox orphan reconciliation;
  2. repeatable black-box conformance for the supported PostgreSQL/Temporal and external-sandbox paths;
  3. bounded context management and compaction over server-owned history.

Durable cross-process interrupt, deterministic finish-vs-interrupt ordering, Sandbox identity, worker-restart reattachment, and deletion cleanup are durable across the local, Docker, E2B, CubeSandbox, OpenSandbox, and Daytona provider adapters. Remote adapters remain Preview until repeatable service-level validation promotes them.

Runtime integrations

After the core loop is complete:

  • MCP tool discovery and execution;
  • supported server tools such as web search and fetch;
  • token usage and model/tool execution spans;
  • additional preview event types;
  • promotion of preview sandbox adapters after recorded live conformance.

Production hardening

Production deployments additionally need:

  • authentication, authorization, and tenant isolation;
  • Worker Versioning and rolling-upgrade coverage;
  • metrics, traces, audit logs, and operational runbooks;
  • object storage for large histories and artifacts;
  • quotas, resource policy, and orphan reconciliation;
  • versioned deployment manifests and migration procedures.

Files, skills, memory, vaults, multi-agent orchestration, schedules, and webhooks remain outside the core harness scope. Current behavior is tracked in the compatibility matrix; architectural guarantees are documented in the architecture overview.