Blueprint · Everything the Software division runs

App routes52

Indexed links59

Divisions11

Link audit200 · 2026-08-26

The full app — every surface, nothing hidden workspace, engineering, the Hermes console and system — intact, now linked

Workspace · 5 routes

BlueprintOpen Operating PageOpen Agent TeamOpen Work ItemsOpen Division SettingsOpen

the rail's own destinations — all editable

Engineering Ops · 8 routes

PipelineOpen EngineOpen WizardOpen AutomationOpen MaintenanceOpen FilesOpen DivisionsOpen StackOpen

the build and ship surfaces this division owns

Hermes Console · 11 surfaces

ChatOpen SessionsOpen AgentsOpen ModelsOpen SkillsOpen LogsOpen CronOpen PluginsOpen ConfigOpen HermesOpen WorkflowsOpen

served by hermes-light-proxy-v2 on this host

System · 7 routes

Security & RAMOpen Workflow BuilderOpen MemoryOpen Hive MindOpen OpenCLIOpen OpenCLI AdminOpen Hermes ActivityOpen

host health, memory and the CLI console

Every link above answered HTTP 200 on the 2026-08-26 audit, probed against the live host. Nothing here was invented: the Hermes Console column indexes eleven surfaces that Caddy already serves on this host but that had no inbound link anywhere in the app. Surfaces that did NOT answer 200 are listed in the register below with their real status instead of being rendered as dead links. Out of scope for this index — standalone apps that are not this division's codebase: HoopTrack, ShotIQ, Postiz, SEO Engine, Coach AI, the shared opendesign.* instance (ContentHub tooling).

Surfaces — what did not answer 200 the honest register: a broken surface goes here, not into a dead link

SurfaceWhat is actually happeningStatus

/ (site root)Caddy `handle /` serves the static kevinclaw-platform-vnext SPA from disk, so this app's own root page never renders. The Blueprint therefore lives at /blueprint, which does reach Next.200 — shadowed by Caddy

/dashboard/hive-mindSame shadow: Caddy serves the vNext SPA for this exact path, so src/app/dashboard/hive-mind/page.tsx does not render in production. Restyled in-app anyway so it is correct the moment the shadow is lifted.200 — shadowed by Caddy

/chat (bare)Caddy 302s bare /chat to the OLD KevinClaw hub (kevinclaw.194-146-12-139.sslip.io/hermes/dev/chat). Never linked. The 200 target on this host is /chat/ with the trailing slash, which is what the rail and the Hermes Console column now use.302 to old hub

/operatingNo route in this app. The Software operating surface is /service, which is what the rail links.404

/apps307 to /?section=products, a legacy dashboard section param whose target is the Caddy-shadowed root. No 200 surface to link.307

/hermes-extras307 to /hermes-extras/status. The 200 target is linked in the CLI reference cluster.307

/api/kc/rail-tallyCaddy's kc_search_api snippet claims ALL of /api/kc/* for kc-search-service on :4099, so this app's own route was unreachable and the rail's three tally cells always read n/a. Endpoint moved to /api/software/rail-tally, which does reach Next.404 — path claimed by Caddy

cognee-softwareCaddy vhost cognee-software.194-146-12-139.sslip.io is configured but the upstream is down.502

Voice Loop (voicehub)Caddy vhost voicehub.194-146-12-139.sslip.io is configured but the upstream is down.502

The division map every sibling division's operating page — this replaces the retired division strip

All 11 answered 200 on 2026-08-26. Each points at that division's current operating surface — never a legacy landing route, and never the old KevinClaw hub.

Cross-division work for Software who is doing what for this division — read from the shared work-graph record

Cross-division record empty — backfill pending.

The store is live (13,884 handoff rows fleet-wide, newest Aug 31, 2026), but no handoff has been recorded for or by Software yet. This is an empty read, not a failed one.

Agent network — proof of work real messages from the Buzz relay; Jane's verdict is her own read

Jane: unavailableHer verdict could not be read box-internally — The operation was aborted due to timeout. The verdict queue itself is still reachable.Jane's verdict queue →

software · 77 messages

AgentMessagesLast active

Jarvis38Sep 23, 14:41 UTC

Dev — Software Hermes23Sep 23, 14:34 UTC

Dev — Software Buzz15Sep 23, 14:34 UTC

baller701Aug 22, 20:52 UTC

Newest · Jarvis at Sep 23, 14:41 UTC: “@Dev — Software Hermes two alerts are yours to investigate as shared host/software reliability, no Kevin escalation yet. Verified evidence: ”

Read live from buzz-prod postgres (docker exec, local socket) · generated Sep 23, 14:47 UTC (60s cache). The editable roster with retasking is on Agent Team.

Hermes CLI reference 28 pages that had no inbound link anywhere in the app before this index existed

/hermes-extras · 29 routes, 28 answering 200

status acp backup checkpoints claw completion computer-use curator debug doctor dump gateway hooks import kanban lsp mcp memory pairing postinstall proxy send setup tools uninstall update version webhook

29 page files, 28 links: the bare /hermes-extras index 307s to /hermes-extras/status, so the redirect is not linked and its 200 target is listed as status above.

On-box tooling — no web surface checked against the filesystem on every load

ToolingStatus

kc-* CLI tooling172 kc-* executables in /usr/local/bin

kc-alert frameworkpresent at /opt/kc-alert

kc-guardpresent at /opt/kc-guard

Off-box CI runnerpresent at /opt/actions-runner

Division apps — separate codebases standalone apps this division owns, off this Next app

Owned apps · all answering 200

App Factorypm2 app-factory :3138 OpenDesigndocker :7458 vNext Previewstatic build Buzz Relayagent message bus