User guide

One terminal for the shell, your APIs, and MCP.

Termorq keeps the call, the secret, and the check together. termorq-cli runs that same collection in a pipeline.

Termorq: one window for the shell, HTTP APIs, MCP tools, and a model. Secrets stay in the OS keychain.

Killer features

  1. 1

    One prompt, four kinds of work

    A shell line, an HTTP call, an MCP tool, or a model, from the same place. The scrollback records which one ran and how long it took.

  2. 2

    Agent, in the prompt New

    Turn it on and the line is chat. A line that starts with @ still runs. The system prompt is in Settings.

    Ask the model
  3. 3

    A collection is a runbook

    Favourites run first. A capture fills the next line. A failed check stops the run.

  4. 4

    termorq-cli, the headless runner New

    The same collection, with no window. Checks, captures, and secrets included. JUnit for GitHub, GitLab, and Azure DevOps.

    Run a collection
  5. 5

    Secrets stay out of the diff

    The file keeps a name or a vault reference. The value stays in the OS keychain, Azure Key Vault, or 1Password.

  6. 6

    A workspace the team can review

    @workspace save writes pretty JSON under .ai-terminal/. You commit the diff yourself.

What you can do

Shell, tabs, and a real terminal

Bash, zsh, PowerShell, split panes, and an interactive PTY when a program needs a keyboard.

Read it

HTTP APIs with checks

Save a connection once, call endpoints with flags, and stop a runbook when a response check fails.

Read it

MCP tools

Connect HTTP, Azure, or stdio servers. Inspect tools, resources, and prompts before you call them.

Read it

Collections as runbooks

Favourite lines run first. A capture such as method = $.method feeds the next line as {{method}}.

Read it

Termorq CLI

List, show, check, call, and run a collection with no window.

Read it

GitHub, GitLab, Azure DevOps

The same run writes JUnit. Secrets come from the pipeline, not from the repo.

Read it

Why termorq-cli is in this guide

Same engine

termorq-cli calls through the app library. A check that fails in the terminal fails in CI.

Secrets stay out of the file

The workspace stores a name and a vault reference. The value arrives as AIT_SECRET_… or from Azure Key Vault or 1Password.

A result your forge already understands

Text for a person. JSON or YAML for a script. JUnit for GitHub, GitLab, and Azure DevOps.

Run a collection

1

Build termorq-cli

From a checkout of this repo. The binary does not open a window.

cargo build --release --bin termorq-cli --manifest-path src-tauri/Cargo.toml
2

Point at the workspace

The path is absolute, or it starts with ~. A relative . is rejected.

termorq-cli list --workspace ~/src/acme-apis
3

Check before you call

Missing secrets and unknown endpoints are blocking. Nothing is called.

termorq-cli check Smoke --workspace ~/src/acme-apis
4

Run

Exit 0 is a pass, skips included. Exit 1 is a failed step. Exit 2 is configuration.

termorq-cli run Smoke --workspace ~/src/acme-apis --env Staging --format junit

Five things worth learning first

1

One prompt for four kinds of work

Type a shell line, @endpoint, @tool, or @chat. The scrollback marks which one ran, how long it took, and the HTTP status.

2

A collection is a runbook

Favourites run first. Captures fill {{name}} on later lines. A failed check, a non-zero exit, or Ctrl+C stops the run in the app.

3

Secrets are not in the git diff

Keys go to the OS keychain. A team folder keeps akv:// and op:// references. Save refuses a file that still contains a token.

4

termorq-cli is that runbook, headless

Every subcommand, the secret rules, and a pipeline for GitHub, GitLab, and Azure DevOps.

Run a collection

5

A workspace your team can review

@workspace save writes pretty JSON under .ai-terminal/. You commit the diff yourself. The app does not push.

Open, save, status, close

The guide is ready to read.

Install, the window, HTTP, MCP, models, the team folder, secrets, and termorq-cli in a pipeline.