[meta] Hero Sandbox Review #275

New issue

Open

opened 2026-06-09 12:52:53 +00:00 by mik-tf · 47 comments

mik-tf commented 2026-06-09 12:52:53 +00:00

Owner

Copy link

Rolling review of the Hero OS sandbox by the whole team. Post findings as comments, ideally one finding per comment with a screenshot.

How to read this issue

• Current status of every finding in this thread lives in the status recap comment (what is done, what is partial, what is open, what needs a retest). It gets refreshed as things land.
• The original planning items from the first capture are implemented through home#278 (expanded base bundle, hero_office, tester sizing, separating user apps from admin dashboards).
• If your finding is marked UNCLEAR in the recap, a one line restatement or a retest on a fresh machine is the fastest way to get it moving.

Original capture (2026-06-09, kept for history)

Cockpit/Deployer

• add hero_code to it, this has features to download repo's and such
• make sure we can see the admin interface from proc & router
  would be good then we have the terminal available
• Put hero_office
• should give max 4 GB to a zero-os
• have an admin tab where we have the admin apps
  • separated from user apps
• Add more apps to the base bundle
  • e.g.

NOTE: For now we do 1 slice = 4gb and 1 tester vm =2xslice, otherwise we have 1 tester vm = 1vcpu, 4gb, would be too slow

Rolling review of the Hero OS sandbox by the whole team. Post findings as comments, ideally one finding per comment with a screenshot. **How to read this issue** - Current status of every finding in this thread lives in the [status recap comment](https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40699) (what is done, what is partial, what is open, what needs a retest). It gets refreshed as things land. - The original planning items from the first capture are implemented through [home#278](https://forge.ourworld.tf/lhumina_code/home/issues/278) (expanded base bundle, hero_office, tester sizing, separating user apps from admin dashboards). - If your finding is marked UNCLEAR in the recap, a one line restatement or a retest on a fresh machine is the fastest way to get it moving. <details> <summary>Original capture (2026-06-09, kept for history)</summary> # Cockpit/Deployer - add hero_code to it, this has features to download repo's and such - make sure we can see the admin interface from proc & router would be good then we have the terminal available - Put hero_office - should give max 4 GB to a zero-os - have an admin tab where we have the admin apps - separated from user apps - Add more apps to the base bundle - e.g.  NOTE: For now we do 1 slice = 4gb and 1 tester vm =2xslice, otherwise we have 1 tester vm = 1vcpu, 4gb, would be too slow </details>

image.png

257 KiB

mik-tf commented 2026-06-09 16:57:44 +00:00

Author

Owner

Copy link

Cockpit/Deployer

• Make sure we have latest from lab
• Update button
  • Need it from the admin for individual vms and all vms
    • e.g. you can click mass update hero_proc for all sandboxes, or for user 1,2,3, etc.
  • If its hero_proc, it restarts, if not hero_proc, change the binary then hero_proc restarts, etc.
• Can download the main binaries or the development/integration binaries, etc. so we can test not just main
  • By default, all main

# Cockpit/Deployer - Make sure we have latest from lab - Update button - Need it from the admin for individual vms and all vms - e.g. you can click mass update hero_proc for all sandboxes, or for user 1,2,3, etc. - If its hero_proc, it restarts, if not hero_proc, change the binary then hero_proc restarts, etc. - Can download the main binaries or the development/integration binaries, etc. so we can test not just main - By default, all main

mik-tf commented 2026-06-09 19:36:09 +00:00

Author

Owner

Copy link

Admin page

• 
• the active page needs to be highlighted btw
• When deploy on admin to provision + install
  • if it breaks half-way, we need
    • way to re-install, re-provision, and also self-heal

# Admin page -  - the active page needs to be highlighted btw - When deploy on admin to provision + install - if it breaks half-way, we need - way to re-install, re-provision, and also self-heal

image.png

196 KiB

mik-tf commented 2026-06-09 19:36:36 +00:00

Author

Owner

Copy link

Deployer/Cockpit

we need spaces on the control buttons or more room

# Deployer/Cockpit  we need spaces on the control buttons or more room

image.png

449 KiB

mik-tf commented 2026-06-09 19:38:38 +00:00

Author

Owner

Copy link

Cockpit/Deployer

these need to be a toggle button

# Cockpit/Deployer  these need to be a toggle button

image.png

64 KiB

mik-tf commented 2026-06-09 19:38:42 +00:00

Author

Owner

Copy link

Cockpit/Deployer

also need an option to upgrade all services

• note: maybe already done, chekc it is possible for tester vms individually, and deployer/admin for all and individual testers vm (but called by admin)

# Cockpit/Deployer also need an option to upgrade all services - note: maybe already done, chekc it is possible for tester vms individually, and deployer/admin for all and individual testers vm (but called by admin)

mik-tf commented 2026-06-09 19:41:02 +00:00

Author

Owner

Copy link

Kimi Agent

even after an upgrade it does not work
but sometimes kimi agent works after a while. need to investigate
hint: check if its an issue with kimi itself or when it is in the deployer

# Kimi Agent  even after an upgrade it does not work but sometimes kimi agent works after a while. need to investigate hint: check if its an issue with kimi itself or when it is in the deployer

image.png

120 KiB

mik-tf commented 2026-06-09 19:55:06 +00:00

Author

Owner

Copy link

Deployer

• in admin dashboard we can't see the vm admin itself mycelium address, as we can with users vms
• DONE

# Deployer - in admin dashboard we can't see the vm admin itself mycelium address, as we can with users vms - DONE

mik-tf commented 2026-06-09 20:03:09 +00:00

Author

Owner

Copy link

Deployer/Cockpit

also on the whiteboard + the rest of the services all need to have voice integration

# Deployer/Cockpit  also on the whiteboard + the rest of the services all need to have voice integration

image.png

303 KiB

mik-tf commented 2026-06-09 20:03:21 +00:00

Author

Owner

Copy link

Hero_voice

the hero voice quality seems to be not that great?

# Hero_voice the hero voice quality seems to be not that great? 

image.png

83 KiB

mik-tf commented 2026-06-09 20:03:44 +00:00

Author

Owner

Copy link

hero_planner

doesn't get filtered

actually even selecting any of these pins doesn't do an filtering too

Fixed

# hero_planner  ~~doesn't get filtered~~ ~~actually even selecting any of these pins doesn't do an filtering too~~ Fixed

image.png

133 KiB

mik-tf commented 2026-06-09 21:16:55 +00:00

Author

Owner

Copy link

Deployer

Tried deleting one of my VMs since I deployed two. Hit this

---

And this if I hit "install"

Hint: maybe it was due to restarting the admin deployer at the same time. Seemed transient as hinted. considered fixed.

# Deployer  Tried deleting one of my VMs since I deployed two. Hit this ---  And this if I hit "install" Hint: maybe it was due to restarting the admin deployer at the same time. Seemed transient as hinted. considered fixed.

image.png

103 KiB

image.png

69 KiB

mik-tf referenced this issue 2026-06-09 22:16:56 +00:00

[cockpit] Hero sandbox: expanded tester app bundle and cockpit Apps/Services UI #278

mik-tf commented 2026-06-10 00:41:27 +00:00

Author

Owner

Copy link

Hero cockpit/deployer

• here it says qa,test,main but its just the mnemonic set as such, the nodes are just on qa and main
  • it should show qa and main then.
    FIXED

# Hero cockpit/deployer  - here it says qa,test,main but its just the mnemonic set as such, the nodes are just on qa and main - it should show qa and main then. FIXED

image.png

111 KiB

thabeta commented 2026-06-10 07:39:59 +00:00

Owner

Copy link

hero_db_admin

hero_db_admin 

image.png

19 KiB

thabeta commented 2026-06-10 07:40:23 +00:00

Owner

Copy link

upgrading cockpit

upgrading cockpit  

image.png

73 KiB

image.png

292 KiB

thabeta commented 2026-06-10 07:41:10 +00:00

Owner

Copy link

herobiz

wrong theme

herobiz wrong theme 

image.png

280 KiB

👍 1

thabeta commented 2026-06-10 07:41:27 +00:00

Owner

Copy link

a vm install

a vm install 

image.png

116 KiB

thabeta changed title from Hero Sandbox Features to Add to Hero Sandbox Review 2026-06-10 08:20:43 +00:00

thabeta commented 2026-06-10 08:21:18 +00:00

Owner

Copy link

accessing /admin/users

accessing /admin/users 

image.png

62 KiB

thabeta commented 2026-06-10 08:58:59 +00:00

Owner

Copy link

orchestrator launching kimi fleet

orchestrator launching kimi fleet 

image.png

134 KiB

thabeta commented 2026-06-10 09:00:24 +00:00

Owner

Copy link

orchestrator by default not auth, no way to setup auth

also this is the admin page, but it still highlights agents tab

orchestrator by default not auth, no way to setup auth  also this is the admin page, but it still highlights agents tab

image.png

166 KiB

thabeta commented 2026-06-10 09:03:26 +00:00

Owner

Copy link

orchestrator fleet info don't exist (e.g what was the prompt, what were the repos,.. etc)

orchestrator fleet info don't exist (e.g what was the prompt, what were the repos,.. etc) 

image.png

116 KiB

thabeta commented 2026-06-10 09:05:14 +00:00

Owner

Copy link

Apps descriptions (kimi, orchestrator)

Apps descriptions (kimi, orchestrator) 

image.png

44 KiB

thabeta commented 2026-06-10 09:18:56 +00:00

Owner

Copy link

slides claims to create dir if doesn't exist

Also, the buttons look disabled

slides claims to create dir if doesn't exist  Also, the buttons look disabled 

image.png

11 KiB

image.png

94 KiB

AhmedHanafy725 commented 2026-06-10 10:10:57 +00:00

Owner

Copy link

The voice bar doesn't do anything

Also, these options go nowhere

The voice bar doesn't do anything  Also, these options go nowhere 

image.png

5.7 KiB

image.png

24 KiB

AhmedHanafy725 commented 2026-06-10 10:49:46 +00:00

Owner

Copy link

buttons look disabled

pdf view

buttons look disabled  pdf view 

image.png

41 KiB

image.png

160 KiB

AhmedHanafy725 commented 2026-06-10 10:57:40 +00:00

Owner

Copy link

Try to record a question in books

Try to record a question in books 

image.png

116 KiB

AhmedHanafy725 commented 2026-06-10 11:10:20 +00:00

Owner

Copy link

can't install hero agent

can't install hero agent 

image.png

22 KiB

AhmedHanafy725 commented 2026-06-10 11:14:46 +00:00

Owner

Copy link

Click "Huddle" at the top right

Click "Huddle" at the top right 

image.png

44 KiB

AhmedHanafy725 commented 2026-06-10 12:24:52 +00:00

Owner

Copy link

can't install Office

can't install Office 

image.png

22 KiB

mik-tf commented 2026-06-10 12:28:35 +00:00

Author

Owner

Copy link

@AhmedHanafy725 wrote in #275 (comment):

can't install Office

On it

@AhmedHanafy725 wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40106: > can't install Office > > [](/lhumina_code/home/attachments/93d0e352-da7d-4f3c-b2e2-720b8bbfeeec) On it

mik-tf commented 2026-06-10 12:29:32 +00:00

Author

Owner

Copy link

@AhmedHanafy725 wrote in #275 (comment):

can't install hero agent

I am removing hero_agent from the Sandbox for now, since hero_kimi is very good, and hero_orchestrator too. We don't need hero_agent for now and maybe never going forward, especialyl since Thabet is doing a lot of work on hero_shrimp and it's going well. So we can consider this done.

@AhmedHanafy725 wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40086: > can't install hero agent > > [](/lhumina_code/home/attachments/b2e19acd-4733-428d-9fc6-ffba569b1f3b) I am removing hero_agent from the Sandbox for now, since hero_kimi is very good, and hero_orchestrator too. We don't need hero_agent for now and maybe never going forward, especialyl since Thabet is doing a lot of work on hero_shrimp and it's going well. So we can consider this done.

mik-tf commented 2026-06-10 12:30:57 +00:00

Author

Owner

Copy link

@thabeta wrote in #275 (comment):

Apps descriptions (kimi, orchestrator)

Don't know by who, but this seems done. Good.

@thabeta wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40053: > Apps descriptions (kimi, orchestrator) > > [](/lhumina_code/home/attachments/67587655-5ac4-4b65-b569-2ec8d5b6e2db) Don't know by who, but this seems done. Good.

mik-tf commented 2026-06-10 12:33:58 +00:00

Author

Owner

Copy link

@AhmedHanafy725 wrote in #275 (comment):

Try to record a question in books

That's a tricky one. Hero books needs aibroker, but I am not sure if we have aibroker working these days. @thabeta Any insights?

@AhmedHanafy725 wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40085: > Try to record a question in books > > [](/lhumina_code/home/attachments/1feb4267-424d-46ca-b31f-9c5066f66271) That's a tricky one. Hero books needs aibroker, but I am not sure if we have aibroker working these days. @thabeta Any insights?

AhmedHanafy725 commented 2026-06-10 13:14:33 +00:00

Owner

Copy link

Hero Memory's UI is slow. Also, check the scroll bar

Hero Memory's UI is slow. Also, check the scroll bar

Screencast from 2026-06-10 15-54-14.webm

1.1 MiB

timur commented 2026-06-11 12:47:26 +00:00

Owner

Copy link

hero_biz feedback:

• not bugs but stylistic: fonts are too large, some of the pages don't have empty list placeholder (like no contracts instead of completely blank view would be nice, looks like something is broken otherwise.)

• is loading between each tab necessary, makes it feel choppy and slow. can we lazy load instead? or perhaps fix underlying issue, api shouldnt be as slow.

• when creating company (or any object) i get this error: Failed to save company: Internal error: Failed to save company: Network error: builder error

hero_biz feedback: - not bugs but stylistic: fonts are too large, some of the pages don't have empty list placeholder (like no contracts instead of completely blank view would be nice, looks like something is broken otherwise.)  - is loading between each tab necessary, makes it feel choppy and slow. can we lazy load instead? or perhaps fix underlying issue, api shouldnt be as slow. - when creating company (or any object) i get this error: Failed to save company: Internal error: Failed to save company: Network error: builder error 

image.png

311 KiB

image.png

457 KiB

scott commented 2026-06-12 00:05:59 +00:00

Owner

Copy link

@thabeta wrote in #275 (comment):

orchestrator fleet info don't exist (e.g what was the prompt, what were the repos,.. etc)

Resolved in the current orchestrator build (integration). The fleet detail page now shows the prompt, the backend / model / effort / tools badges, the concurrency and repo count, and a per-repo card for every repo in the fleet. Minor polish still on the list (a collapsed full repo-path list + the created-at timestamp), but the core info is there.

@thabeta wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40051: > orchestrator fleet info don't exist (e.g what was the prompt, what were the repos,.. etc) Resolved in the current orchestrator build (integration). The fleet detail page now shows the **prompt**, the **backend / model / effort / tools** badges, the **concurrency** and **repo count**, and a **per-repo card** for every repo in the fleet. Minor polish still on the list (a collapsed full repo-path list + the created-at timestamp), but the core info is there.

scott commented 2026-06-12 00:06:00 +00:00

Owner

Copy link

@thabeta wrote in #275 (comment):

also this is the admin page, but it still highlights agents tab

The tab-highlight half of this is resolved in the current build (verified live): the Admin page now correctly highlights the Admin tab, not Agents.

The other half — "orchestrator by default not auth, no way to setup auth" — is still open and tracked separately.

@thabeta wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40050: > also this is the admin page, but it still highlights agents tab The tab-highlight half of this is resolved in the current build (verified live): the Admin page now correctly highlights the **Admin** tab, not Agents. The other half — *"orchestrator by default not auth, no way to setup auth"* — is still open and tracked separately.

mik-tf commented 2026-06-12 21:36:15 +00:00

Author

Owner

Copy link

Status recap, 2026-06-12

Full pass over every finding in this thread, with current status. Legend: DONE shipped and verified, PARTIAL part of it shipped, OPEN not started or still failing, UNCLEAR needs a restatement or a retest to judge.

Big picture first: both onboarding paths were proven end to end today on a fully refreshed latest-integration channel. A brand new account and an existing account each went create, provision, install, browser SSO login, welcome email, working cockpit. RuyMaldonado, joriss, thabeta and timur all have fresh machines again (new web addresses, sent to each of you by welcome email). All release assets on the latest-integration channel were rebuilt and fresh installs were verified byte identical to the published binaries.

Platform: deployer and cockpit

Item	By	Status	Notes
hero_code in the base bundle	mik-tf	DONE	In the standard install set, running on fresh testers
hero_office in the base bundle	mik-tf	DONE	Running on fresh testers; also closes "can't install Office" below
Tester sizing (2 slices, 2 vCPU, 8 GB)	mik-tf	DONE
Admin apps separated from user apps	mik-tf	DONE	Testers see 17 user apps; the admin machine additionally sees the admin bundles
Expanded base bundle	mik-tf	DONE	21 components installed per tester
Releases always fresh ("latest from lab")	mik-tf	DONE	Uploads now replace stale release assets by checksum
Update services on one VM from admin	mik-tf	DONE
Mass update across all tester VMs in one click	mik-tf	OPEN	Not built yet; today it is one VM at a time
Channel selection (main / integration)	mik-tf	DONE	Selectable at install and update
Active page highlighted in admin nav	mik-tf	DONE	Orchestrator admin tab fixed too (see scott's reply)
Recovery when install breaks halfway	mik-tf	PARTIAL	Re-install and delete plus re-provision both work and were exercised; fully automatic self healing is not there
Control buttons spacing	mik-tf	DONE	Control page re-tiled to match the Apps page
"These need to be a toggle button"	mik-tf	UNCLEAR	Needs a one line restatement of which control was meant
Upgrade all services	mik-tf	PARTIAL	The tester cockpit has Update All; an admin triggered update of every tester at once does not exist
Admin VM's own mycelium address visible	mik-tf	DONE	Marked done in thread
Transient delete / install errors	mik-tf	DONE	Upgrades and restarts now run detached, so a dropped browser connection can no longer strand services. Remaining known case is the cockpit restarting itself: hero_proc#149
Networks listed without any nodes	mik-tf	DONE	The node registry now lists only networks with real nodes
Upgrading cockpit errors	thabeta	PARTIAL	Same root cause as the row above; only cockpit self update remains, tracked in hero_proc#149
hero_db_admin screenshot	thabeta	UNCLEAR	Needs a retest on a fresh tester or a one line restatement
"a vm install" screenshot	thabeta	UNCLEAR	Same
/admin/users error	thabeta	UNCLEAR	The users page is in daily use and looks healthy; please retest
Mixed case usernames could not provision	found while testing	DONE	New finding from today's walk: the VM name was derived from the username unchanged and the daemon refuses capitals. Fixed in 9158307

AI assistants

Item	By	Status	Notes
Kimi agent not working / intermittent	mik-tf	DONE	Works on fresh testers now
Kimi: sending from the empty state does nothing	found while testing	OPEN	You must click New session first. Filed as hero_kimi_rust#8
Can't install hero_agent	AhmedHanafy725	DONE	Resolved by decision: hero_agent removed from the sandbox bundle
Orchestrator has no auth by default	thabeta	OPEN	Tracked separately per scott's reply
Orchestrator fleet info missing	thabeta	DONE	Resolved by scott, minor polish open
Orchestrator launching kimi fleet screenshot	thabeta	UNCLEAR	Not sure if this was a bug report or a showcase
App descriptions for kimi and orchestrator	thabeta	DONE	Confirmed in thread

Voice

Item	By	Status	Notes
Voice integration on whiteboard and all apps	mik-tf	PARTIAL	Shared voice and agent widgets just landed in the shared UI layer; per app wiring is rolling out
Voice bar does nothing, options go nowhere	AhmedHanafy725	PARTIAL	Same strand as the row above
Voice quality	mik-tf	OPEN	With the voice owner, see hero_voice#37
Record a question in Books	AhmedHanafy725	OPEN	Depends on the voice wiring above plus the Books search fix that is built but not yet merged: hero_books#148

App polish

Item	By	Status	Notes
Planner pin filtering	sameh-farouk	DONE	Marked fixed in thread
Slides claims to create a dir, buttons look disabled	thabeta	OPEN	No fix recorded yet
Books buttons look disabled, PDF view	AhmedHanafy725	OPEN	No fix recorded yet
Memory UI slow, scroll bar	AhmedHanafy725	PARTIAL	A memory UI fix shipped since; the slowness needs a retest
Biz wrong theme	thabeta	OPEN	Biz UI work is active, retest after it lands
Biz fonts, empty list placeholders, per tab loading	timur	OPEN	Same, active UI work in 74f56c4 and follow ups
Biz create company gives "builder error"	timur	PARTIAL	Very likely fixed by the base url change in 74f56c4; needs one retest on a fresh tester

Infrastructure note

The rented mainnet node turned out to be a newer light generation machine that only accepts the light workload family, while our stack deploys the classic workload types, so it rejects every deployment. Details and a placement guard proposal in hero_os_tfgrid_deployer#25. Until that is resolved all testers run on the QA node, which is nearly full (room for one more tester). A capacity decision is needed before the next wave.

If a row of yours says UNCLEAR, a one line restatement (or a quick retest on your fresh machine) is the fastest way to get it moving.

# Status recap, 2026-06-12 Full pass over every finding in this thread, with current status. Legend: DONE shipped and verified, PARTIAL part of it shipped, OPEN not started or still failing, UNCLEAR needs a restatement or a retest to judge. Big picture first: both onboarding paths were proven end to end today on a fully refreshed latest-integration channel. A brand new account and an existing account each went create, provision, install, browser SSO login, welcome email, working cockpit. RuyMaldonado, joriss, thabeta and timur all have fresh machines again (new web addresses, sent to each of you by welcome email). All release assets on the latest-integration channel were rebuilt and fresh installs were verified byte identical to the published binaries. ## Platform: deployer and cockpit | Item | By | Status | Notes | |---|---|---|---| | hero_code in the base bundle | mik-tf | DONE | In the standard install set, running on fresh testers | | hero_office in the base bundle | mik-tf | DONE | Running on fresh testers; also closes "can't install Office" below | | Tester sizing (2 slices, 2 vCPU, 8 GB) | mik-tf | DONE | | | Admin apps separated from user apps | mik-tf | DONE | Testers see 17 user apps; the admin machine additionally sees the admin bundles | | Expanded base bundle | mik-tf | DONE | 21 components installed per tester | | Releases always fresh ("latest from lab") | mik-tf | DONE | Uploads now replace stale release assets by checksum | | Update services on one VM from admin | mik-tf | DONE | | | Mass update across all tester VMs in one click | mik-tf | OPEN | Not built yet; today it is one VM at a time | | Channel selection (main / integration) | mik-tf | DONE | Selectable at install and update | | Active page highlighted in admin nav | mik-tf | DONE | Orchestrator admin tab fixed too (see scott's reply) | | Recovery when install breaks halfway | mik-tf | PARTIAL | Re-install and delete plus re-provision both work and were exercised; fully automatic self healing is not there | | Control buttons spacing | mik-tf | DONE | Control page re-tiled to match the Apps page | | "These need to be a toggle button" | mik-tf | UNCLEAR | Needs a one line restatement of which control was meant | | Upgrade all services | mik-tf | PARTIAL | The tester cockpit has Update All; an admin triggered update of every tester at once does not exist | | Admin VM's own mycelium address visible | mik-tf | DONE | Marked done in thread | | Transient delete / install errors | mik-tf | DONE | Upgrades and restarts now run detached, so a dropped browser connection can no longer strand services. Remaining known case is the cockpit restarting itself: [hero_proc#149](https://forge.ourworld.tf/lhumina_code/hero_proc/issues/149) | | Networks listed without any nodes | mik-tf | DONE | The node registry now lists only networks with real nodes | | Upgrading cockpit errors | thabeta | PARTIAL | Same root cause as the row above; only cockpit self update remains, tracked in [hero_proc#149](https://forge.ourworld.tf/lhumina_code/hero_proc/issues/149) | | hero_db_admin screenshot | thabeta | UNCLEAR | Needs a retest on a fresh tester or a one line restatement | | "a vm install" screenshot | thabeta | UNCLEAR | Same | | /admin/users error | thabeta | UNCLEAR | The users page is in daily use and looks healthy; please retest | | Mixed case usernames could not provision | found while testing | DONE | New finding from today's walk: the VM name was derived from the username unchanged and the daemon refuses capitals. Fixed in [9158307](https://forge.ourworld.tf/lhumina_code/hero_os_tfgrid_deployer/commit/91583079ba348c4c08813a3a80bd6fab098eb453) | ## AI assistants | Item | By | Status | Notes | |---|---|---|---| | Kimi agent not working / intermittent | mik-tf | DONE | Works on fresh testers now | | Kimi: sending from the empty state does nothing | found while testing | OPEN | You must click New session first. Filed as [hero_kimi_rust#8](https://forge.ourworld.tf/lhumina_code/hero_kimi_rust/issues/8) | | Can't install hero_agent | AhmedHanafy725 | DONE | Resolved by decision: hero_agent removed from the sandbox bundle | | Orchestrator has no auth by default | thabeta | OPEN | Tracked separately per scott's reply | | Orchestrator fleet info missing | thabeta | DONE | Resolved by scott, minor polish open | | Orchestrator launching kimi fleet screenshot | thabeta | UNCLEAR | Not sure if this was a bug report or a showcase | | App descriptions for kimi and orchestrator | thabeta | DONE | Confirmed in thread | ## Voice | Item | By | Status | Notes | |---|---|---|---| | Voice integration on whiteboard and all apps | mik-tf | PARTIAL | Shared voice and agent widgets just landed in the shared UI layer; per app wiring is rolling out | | Voice bar does nothing, options go nowhere | AhmedHanafy725 | PARTIAL | Same strand as the row above | | Voice quality | mik-tf | OPEN | With the voice owner, see [hero_voice#37](https://forge.ourworld.tf/lhumina_code/hero_voice/issues/37) | | Record a question in Books | AhmedHanafy725 | OPEN | Depends on the voice wiring above plus the Books search fix that is built but not yet merged: [hero_books#148](https://forge.ourworld.tf/lhumina_code/hero_books/issues/148) | ## App polish | Item | By | Status | Notes | |---|---|---|---| | Planner pin filtering | sameh-farouk | DONE | Marked fixed in thread | | Slides claims to create a dir, buttons look disabled | thabeta | OPEN | No fix recorded yet | | Books buttons look disabled, PDF view | AhmedHanafy725 | OPEN | No fix recorded yet | | Memory UI slow, scroll bar | AhmedHanafy725 | PARTIAL | A memory UI fix shipped since; the slowness needs a retest | | Biz wrong theme | thabeta | OPEN | Biz UI work is active, retest after it lands | | Biz fonts, empty list placeholders, per tab loading | timur | OPEN | Same, active UI work in [74f56c4](https://forge.ourworld.tf/lhumina_code/hero_biz/commit/74f56c406231e6d66b45ec06b283c2ba4f7b0894) and follow ups | | Biz create company gives "builder error" | timur | PARTIAL | Very likely fixed by the base url change in [74f56c4](https://forge.ourworld.tf/lhumina_code/hero_biz/commit/74f56c406231e6d66b45ec06b283c2ba4f7b0894); needs one retest on a fresh tester | ## Infrastructure note The rented mainnet node turned out to be a newer light generation machine that only accepts the light workload family, while our stack deploys the classic workload types, so it rejects every deployment. Details and a placement guard proposal in [hero_os_tfgrid_deployer#25](https://forge.ourworld.tf/lhumina_code/hero_os_tfgrid_deployer/issues/25). Until that is resolved all testers run on the QA node, which is nearly full (room for one more tester). A capacity decision is needed before the next wave. If a row of yours says UNCLEAR, a one line restatement (or a quick retest on your fresh machine) is the fastest way to get it moving.

mik-tf referenced this issue 2026-06-12 21:55:23 +00:00

[META] Hero OS sandbox demo, functional readiness: onboarding pipeline + per-app verification #239

mik-tf referenced this issue 2026-06-14 03:22:03 +00:00

[platform] Roadmap audit + issue triage: apply the platform/service taxonomy, close stale, write the phased roadmap #283

mik-tf changed title from Hero Sandbox Review to [meta] Hero Sandbox Review 2026-06-14 04:32:15 +00:00

mik-tf added the

platform

label 2026-06-14 04:32:16 +00:00

mik-tf self-assigned this 2026-06-14 04:32:16 +00:00

thabeta commented 2026-06-14 08:43:09 +00:00

Owner

Copy link

Tried to access the collab app on the running sandbox and i faced this error:

https://thabeta005k.gent01.qa.grid.tf/hero_collab/web/

Cannot connect
Make sure hero_collab_server is running.
Unexpected token 'B', "Backend un"... is not valid JSON

Tried to access the collab app on the running sandbox and i faced this error: https://thabeta005k.gent01.qa.grid.tf/hero_collab/web/ ``` Cannot connect Make sure hero_collab_server is running. Unexpected token 'B', "Backend un"... is not valid JSON ```

Screenshot 2026-06-14 at 11.43.03 AM.png

194 KiB

thabeta commented 2026-06-14 08:49:52 +00:00

Owner

Copy link

Hero AIBroker keep loading models

 Hero AIBroker keep loading models

image.png

117 KiB

thabeta commented 2026-06-14 08:50:41 +00:00

Owner

Copy link

Hero Foundry backend not found

 Hero Foundry backend not found

image.png

117 KiB

ashraf commented 2026-06-14 09:50:03 +00:00

Member

Copy link

hero_biz network err
new contact -> save redirects to edit contact with this following err
same err with adding new employee, new project ... almost adding any new thing

hero_biz network err new contact -> save redirects to edit contact with this following err same err with adding new employee, new project ... almost adding any new thing

Screenshot from 2026-06-14 12-47-10.png

105 KiB

ashraf commented 2026-06-14 10:15:44 +00:00

Member

Copy link

hero_docs download pdf produces err

hero_docs download pdf produces err 

image.png

33 KiB

ashraf commented 2026-06-14 10:18:31 +00:00

Member

Copy link

hero_books read button not working

hero_books read button not working 

image.png

76 KiB

ashraf commented 2026-06-14 10:38:37 +00:00

Member

Copy link

hero slides wizard failure

hero slides wizard failure 

image.png

46 KiB

sameh-farouk commented 2026-06-14 11:02:15 +00:00

Member

Copy link

@AhmedHanafy725 wrote in #275 (comment):

Click "Huddle" at the top right

This is not a collab bug, it is a deployment issue (livekit is not configured or run on the server)

@AhmedHanafy725 wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-40087: > Click "Huddle" at the top right [](/lhumina_code/home/attachments/45eac774-520a-4955-b4a1-21edcf94f094) This is not a collab bug, it is a deployment issue (livekit is not configured or run on the server)

ashraf commented 2026-06-14 11:14:23 +00:00

Member

Copy link

hero voice

hero voice 

image.png

69 KiB

sameh-farouk commented 2026-06-14 11:33:20 +00:00

Member

Copy link

@thabeta wrote in #275 (comment):

Tried to access the collab app on the running sandbox and i faced this error:

https://thabeta005k.gent01.qa.grid.tf/hero_collab/web/

Cannot connect
Make sure hero_collab_server is running.
Unexpected token 'B', "Backend un"... is not valid JSON

Backend down ? Can't see the logs from router dashboard

@thabeta wrote in https://forge.ourworld.tf/lhumina_code/home/issues/275#issuecomment-41043: > Tried to access the collab app on the running sandbox and i faced this error: > > https://thabeta005k.gent01.qa.grid.tf/hero_collab/web/ > > ```text > Cannot connect > Make sure hero_collab_server is running. > Unexpected token 'B', "Backend un"... is not valid JSON > ``` Backend down ? Can't see the logs from router dashboard

Sign in to join this conversation.

No Branch/Tag specified

Branches Tags

development

main

No results found.

Labels

Clear labels   

meeting-notes

Meeting notes and follow-ups

  

meeting-sensitive

Sensitive meeting; no shared archive or AI processing

  

meeting-transcript

Raw meeting transcript; AI drafts the report

  

platform

Hero platform work: the integrated stack + deployer/onboarding/gateway/admin/lab (the whole-product layer)

  

service

Hero service work: a feature/bug inside one app (slides, books, voice, biz, ...)

No labels

meeting-notes

meeting-sensitive

meeting-transcript

platform

service

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

mik-tf

7 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

lhumina_code/home#275

No description provided.