Skip to content
New — orchestrate a team of Mel agents from your terminal. Learn more about orchestrating Mel agents
Mel

Mel – the agentic terminal
The AI terminal for teams.

Prompt & workflow gallery

Do things with Mel

Do things with Mel is a gallery of real prompts and workflows the Mel team uses day to day — browse, copy one into the agent input, and make it yours. Every entry names the shipped feature it runs on.

Always review what a prompt will do — the agent follows the approvals you set in Settings > Agent mode.

Review my open changes

promptAgent Mode

Have the agent review uncommitted and committed changes on your branch — logic bugs, edge cases, and project-convention drift — before you open the PR.

Create a PR from open changes

promptAgent Mode

Commit messages, a sensible PR description, push, and the open-in-browser step — one prompt end to end using git and the GitHub CLI.

Fetch PR comments into the terminal

promptAgent Mode

Pull every review comment for the current branch's PR with the GitHub CLI, grouped by file and thread, so you can address feedback without leaving Mel.

Set up an unfamiliar project

promptAgent Mode

Point the agent at a fresh clone: it reads the README and lockfiles, detects the stack and package manager, installs, and starts the project.

Investigate a failing test

promptAgent Mode

The agent reproduces the failure first, traces the cause through the code, and proposes a fix for approval before editing anything.

Plan before building

promptwrite_plan

Ask for a plan document first: Mel's write_plan tool opens the plan in a side panel next to the conversation before any code is written.

Ship a PDF report of the work

promptcreate_pdf

Turn an investigation or audit into a typeset PDF — headings, tables, syntax-highlighted code — with the agent's create_pdf tool.

Drive a team of agents

workflow/orchestrate

The lead agent splits your task, opens a team of agents in their own panes, gives each a piece, and checks on them until the work is put together.

Melt an SSH session

workflowSSH + Agent Mode

ssh into a box, then press Ctrl+Enter: the agent's file and command tools now run on the remote host — no remote install needed.

One worktree per task

workflowGit worktrees

The + tab menu creates a git worktree on a fresh branch for any tracked repo — parallel tasks get parallel checkouts, not stashes.

Launch a saved multi-pane workbench

workflowTab configs

A .toml tab config describes panes, working directories, and startup commands — one click opens your whole layout, parameters included.

Summarize what changed since main

promptAgent Mode

A readable narrative of your branch: what changed, why it holds together, and what a reviewer should look at first.