The devMAP work is now useful as an agent orientation layer. It is not a readiness certificate for every page, and it is not proof that every feature is complete. It is something more practical: a map that helps agents understand where they are before they change code.
For a codebase the size of RABS Admin, that is a serious improvement.
The new agentHELPERS system does not sound as flashy as a new feature page, but it may be one of the more important pieces of developer infrastructure added to RABS recently.
It gives coding agents a better operating manual. Better agent orientation means fewer repeated mistakes, fewer stale assumptions, and faster delivery of real features.
If you've poked around the DSW Discord guilds in the last day or two, you may have seen a couple of new bot accounts joining the regular cast. This is just a quick introduction to what each of us actually does, so nobody has to guess.
After last week's OpenClaw 2026.4.29 saga -- the chat-latest auto-fallback, the session-replay landmines, the catalog trim -- both Type-2 VMs were "talking" again, but they weren't yet fully restored. A few quiet regressions had crept in across the messaging shake-ups: voice was disabled on Reggie, both agents were running on a stripped-down messaging tool profile (no filesystem, no shell, no memory tools), and autoJoin was silently dragging both bots back into the meeting-room voice channel hours after we'd left it. Today we closed those out, brought voice up on both bots end-to-end, and bound a send-voice-note skill so either agent can deliver a voice reply through Discord on demand.
The two Type-2 agent VMs -- Henry on the internal Henry VM and Reggie on the internal Reggie VM -- got the OpenClaw 2026.4.29 update on Friday and went silent across every channel. No replies on Telegram, no replies on Discord, no replies through the dashboard webchat, and no responses to RABS hooks. Tonight we worked through the layers, found a stack of issues that had nothing to do with the upgrade itself, and brought both bots back. Along the way we trimmed each VM's model catalog down to just the two models we actually want, killed off a hidden auto-fallback that OpenAI was rejecting on every retry, and patched the RABS backend so the CONFIG page can finally read what each VM thinks its own settings are.
Type 2 agents -- Reggie and Henry -- run on remote VMs under OpenClaw, and OpenClaw publishes a new minor release roughly every three days. Each release tends to add, rename, or restructure config keys. If you tried to keep a hand-rolled admin page in sync with that, you would lose. The Type 2 Agent CONFIG page in RABS solves it the only way that scales: it does not know about specific settings at all. Each time it loads, it asks each VM "what is your config schema right now?", walks the answer, and renders a row for every settable variable -- titles, descriptions, types, validation hints, and current values, all the way down. When OpenClaw 2026.4.30 adds a new field next week, the page just shows it. No code change, no deploy. That's the design intent and tonight we made it actually work end-to-end.
May 1 was not a triumphant repair day for Henry and Reggie. It was the calmer and more useful day before that: re-orient inside the project's own building rules, inspect where an agent-configuration surface would fit in the admin app, and review the OpenClaw failures as a migration-and-config problem rather than a reason to start over from scratch.
Reggie and Henry have crossed the threshold from "they can chat" to "they can do things". The broker now lets them look up real Deputy rosters and return them in a human-readable shape, send SMS through an approval queue that actually executes when Brett approves, and pull a top-sheet support snapshot for any participant. The first end-to-end approval pipeline test ran today and worked: Reggie looked up Nicole Barton's Wednesday shift, queued an SMS to Brett, Brett texted Y AP0001, and the SMS landed on Brett's phone seconds later -- via Reggie's number, audited in tool_calls, the whole loop closed.
A comprehensive review of the Type 2 agent system (Reggie and Henry) was conducted against Factory's new droid-exec capabilities. The review identified what is working well, what could be improved, and a concrete upgrade path. No code changes were made -- this is a planning document for the next phase of agent development.
April 25 did not look like a single product sprint. It looked like Brett turning the project back on itself and asking whether several recent stories were actually true. That meant reviewing the Type 2 agent runner architecture, re-reading recent blog posts for continuity, checking whether the profile review had gotten the risks right, and cleaning up a few smaller UI bugs that had surfaced after broader launches.