Welcome Screen And Chat#

This page covers the Agent Master surface itself — the welcome screen, the main window, how to load data, and how the chat works — so the later pages can focus on workflows, configuration, and outputs.

Welcome Screen#

Agent Master opens on a welcome screen (see installation) with four capability cards — Load EDI, Chat & Plan, AI Inversion, Reports — and a Start Agent Master button that takes you into the main window.

The Main Window#

The Agent Master main window with the empty-state suggestion chips

The main window before the first message: command bar on top, History sidebar on the left, the chat area with suggestion chips in the centre, and the natural-language input at the bottom.#

Command bar (top)

  • Load EDI — open the data loader (below).

  • Save — save the current session.

  • a theme toggle, Help, and Settings (the gear — the LLM configuration, see LLM Configuration).

  • Once data is loaded, a status badge shows the survey size, e.g. 128 EDI · 5 line(s).

History sidebar (left)

  • Chat / Session tabs switch between the message history and saved sessions.

  • New Chat starts a fresh conversation.

  • Pinned keeps prompts you want to reuse.

  • Figures collects every figure the agents generate, as thumbnails, during the session.

Chat area (centre)

Before the first message, the chat area shows suggestion chips — ready-made requests such as Load EDI data, run quality control and clean the data, Run full AI-assisted 1-D neural inversion, Run phase tensor and dimensionality analysis, or Run full pipeline: QC, inversion, and report. Click one to send it, or type your own.

Input bar (bottom)

Type a request in plain language. The + and paperclip let you attach context, and the send button submits; while a workflow runs, the send button becomes a stop control.

Loading EDI Data#

Everything Agent Master does operates on a loaded survey. Open the loader with Load EDI on the command bar.

The Load EDI Data drawer with source and line-assignment controls

The Load EDI Data drawer: browse to a folder or drop files, then choose how stations are grouped into lines.#

  • Select EDI source — Browse folder to a survey directory, or drop EDI files (or a folder) onto the drop zone; the chosen path appears under Selected path.

  • Line assignment mode — group stations into survey lines by Folder names (default), Auto-detect from IDs, or Edit / Rename.

  • Load into session brings the survey in.

When you point at a multi-line survey, the loader reports the lines it found — for example five lines L18PLT (28 EDI), L22PLT, L26PLT, L30PLT, and L34PLT (25 EDI each). After loading, the command-bar badge shows the total (128 EDI · 5 line(s)), and the survey is available to every request you make.

Chatting#

Once data is loaded, describe what you want. Agent Master replies in a chat thread of alternating bubbles — your requests on the right, the assistant’s responses on the left.

Agent Master answering "what can you do?" with its workflow list

Asking what can you do? lists the workflows Agent Master can run on the loaded data — qc, static_shift, denoise, phase_analysis, tipper, rotation, freq_decimation, sensitivity, pre_inversion, and more.#

A good first message is simply what can you do? — the assistant introduces itself as the pyCSAMT assistant and lists the workflows it can run. From there you can ask for any of them in plain language; how those requests are planned and executed is covered in Workflows And Agents.

Sessions And History#

  • New Chat starts over without losing loaded data.

  • Pinned prompts are kept in the sidebar for reuse across chats.

  • Save (command bar) and the Session tab persist and restore conversations; generated figures and traces are covered in Tools, Memory, And Outputs.

Next Steps#