LIT OS

Operational model

Functional agents, not vendor names

The UI exposes LIT capabilities. Under the hood Hermes can use multiple ChatGPT credentials and local Ollama fallback, but the operator sees task lanes and safety states.

LIT OS ONLINE

LIT OS ↔ Hermes Bridge

Control-room message

Bridge v0.3 UX

Worker

unknown

Pending

n/a

Processing

n/a

Answered

n/a

Errors

n/a

Last heartbeat

n/a

Last message

n/a

Last reply

n/a

Send text with images/audio from the OS. Telegram remains secondary; this queue is controlled and does not expose a shell/tool endpoint.

How to obtain the token without exposing it here

LIT Core

online

General orchestration, task routing and morning briefing

GPT pool + local fallback

Hermes Bridge

local_active

Direct LIT OS message queue with text, image and audio uploads

v0.1 queue active

Zoho Agent

researching

CRM / WorkDrive / Projects connection plan

API verified only

Client Portal Agent

designing

Client_Portal_Package and customer delivery workflow

manual review required

Thermography Agent

planned

Radiometric package review; preview is never measurement

temp.npy/raw.npy first

Radar Agent

planned

Stack-up/signal workflow; Python math required

no LLM thickness math

Endoscopy Agent

local_active

Image/video annotation and evidence extraction

V0.2 player + future full-frame extractor

NAS Organizer

blocked_for_approval

Archive organization with reversible move plans

no destructive overnight work

Report Agent

planned

Report drafting and package assembly

evidence-driven language