Status
Every subsystem, with what we can actually verify about it. Runtime checks as of 27 October 2026; space labels as of 11 November 2026.
One table, no marketing adjectives. Running means it is deployed and someone ran it reproducibly and read the result. Deployed means it runs on our own server but we have not published a reproducible run. Partial means some of it runs, or it runs only on a workstation. Untested means it exists in code but is not running or not confirmed. Research means it runs, but we are not offering it as a product yet. Not running means exactly that.
| Subsystem | State | Note |
|---|---|---|
| Routing (Brain) | Partial | Regex routing, execution and validation verified against the live database on 27 Oct. Request service scaled to zero; semantic fallback broken in code; no space filter |
| Job chain (Brain → OpenFang) | Running | Brain code → job queue → OpenFang agent → check → delivery, verified on 27 Oct |
| Execution (OpenFang) | Running | v0.6.9, 43 agents, 37 as Claude Code subprocesses. WASM sandbox in code, unused |
| MCP tool layer | Running | 50 in-house servers, 798 tools counted in code; registry and runtime configuration have drifted |
| Memory — routing graph | Running | 13 collections; decisions collection holds 171 entries |
| Memory — code index | Partial | Last confirmed 66,979 chunks on 2 Aug; updater failing since 8 Sep |
| Sales station | Running | Deployed on our server; tool layer verified over MCP on 27 Oct. Not routed through the Brain |
| Learning | Running | Deployed on our server, readiness verified on 27 Oct. Not yet open to the public, so the spaces section labels it in progress (since 11 Nov). Not routed through the Brain |
| MiroFish (forecasting) | Deployed | Runs on our server; no published end-to-end run |
| Knowledge vault (Rowboat) | Deployed | Runs on our server, with a workstation part; no published end-to-end run |
| Captain Cook | Research | Runs on our server; treated as research, not offered as a product yet. Our n8n workflows move here. Not routed through the Brain |
| Project intake | Partial | Being built; runs on our server; not open source yet. Telegram assistant, question tree, project brief as PDF. Not routed through the Brain |
| Video | Deployed | API on our server; GPU models on a workstation |
| Marketing | Deployed | API and data on our server; generation on a workstation. Not routed through the Brain |
| Automation (n8n) | Research | Folded into Captain Cook |
| Ideas | Partial | Workstation only, data on our server; bubble operations verified with the Brain’s code on 27 Oct |
| Desktop | Partial | Workstation only; not startable from the main checkout |
| Channels | Partial | Telegram via OpenFang to an operations agent, round trip verified on 27 Oct; WhatsApp via OpenClaw, no published run |
| Voice | Untested | Backend fails at start; dedicated agent removed |
| Coding | Untested | Routing in the registry, engine not running anywhere, preview disabled |
| Scheduling | Untested | Disabled by default; not running |
| Security suite | Untested | Proof-of-concept tool servers; public with the umbrella repository since 1 September 2026 |
| Infrastructure | Partial | Container stacks on our server and workstation; parts of the observability stack degraded |
| Research, wellness (Personal Space), forum bridge, music, requirements wizard, agent farm (part of Captain Cook) | Not running | Code exists; no running service |
| Accessibility | Planned | Early and opt-in |