Term Cli logo

Term Cli

Community
EliasOenal
term-cli

Controls interactive terminal sessions for running long-lived processes, servers, REPLs, debuggers, and TUI programs without blocking. Use when you need to run dev servers (npm run dev), debuggers (pdb, gdb), REPLs (python, node), databases (psql, mysql), SSH sessions, or editors (vim, nano) — any interactive or blocking program.

Overview

PublisherEliasOenal
Repositoryterm-cli
Skill nameterm-cli
Stars
104
Forks
10
Bundled files
Instructions only
LicenseBSD-3-Clause-Clear
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by EliasOenal on GitHub. Read the source before you install it.

Installation

Install the Term Cli AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/EliasOenal/term-cli/tree/main/skills/term-cli
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/EliasOenal/term-cli.git /tmp/term-cli
mkdir -p .claude/skills
cp -r /tmp/term-cli/skills/term-cli .claude/skills/term-cli
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Term Cli in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Term Cli on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Term Cli is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Terminal Session Control

You can now run any interactive application using term-cli — dev servers, debuggers, REPLs, databases, SSH sessions, editors, and TUIs — without blocking.

Debuggers encouraged: Prefer interactive debuggers (pdb, gdb) via term-cli whenever appropriate for the task. This complements (and is often superior to) print-based debugging: set breakpoints, step, inspect state, then capture to reason.

Think concurrently: term-cli runs long tasks in the background. Start tests/builds/servers in one session while you write docs or investigate code in parallel. Once all other tasks are done, check back with wait/capture. Tip: occasionally run date +"%H:%M:%S" in a session to see when things happened, estimate future execution and to calibrate your own loop speed.

Self-documenting: Run term-cli --help or term-cli <command> --help for complete usage details.

Quick Start

bash
term-cli start --session dev && term-cli run --session dev "make test"
# ... do other work ...
term-cli wait --session dev && term-cli capture --session dev
term-cli kill --session dev

Commands

Session Management

bash
term-cli start --session NAME --cwd /path
term-cli kill --session NAME
term-cli list
term-cli status --session NAME

Running Commands

bash
term-cli run --session NAME "make test" --wait --timeout 60   # default: 10s

Sending Input

bash
term-cli send-text --session NAME ":wq" --enter && term-cli wait --session NAME
term-cli send-key --session NAME C-c
term-cli send-stdin --session NAME < file.txt
term-cli send-mouse --session NAME --text "Item B"
term-cli send-mouse --session NAME --text "Item B" --scroll-down 3

Keys: C-c C-d C-z C-u C-l (ctrl), Enter Escape Tab Space BSpace, Up Down Left Right, Home End PPage NPage, F1-F12

send-mouse works only in alternate screen mode (active TUI). Prefer --text targets over coordinates when possible.

Capturing Output

bash
# Visible screen only (default: physical rows, trimmed, no ANSI)
# Prefer this — it's almost always enough
term-cli capture --session NAME

# Last N physical rows from the bottom of the visible screen
term-cli capture --session NAME --tail 10

# Last N logical lines from scrollback+visible history (joins wrapped lines)
# Use only when you need output that scrolled off-screen
# In alternate screen (TUIs / nested tmux), this is blocked by default
term-cli capture --session NAME --scrollback 20

# Override alternate-screen protection (may return stale/misleading history)
term-cli capture --session NAME --scrollback 20 --force

# Include ANSI escape codes (colors) — works with any mode
term-cli capture --session NAME --scrollback 10 --raw

# Annotated capture — pane metadata + highlight annotations
# Shows screen mode, bell, cursor position, and highlighted TUI elements
# Use this when operating TUIs where you need to identify selected items
term-cli capture --session NAME --annotate

# Default capture auto-enables annotations for active alternate-screen TUIs.
# Force plain output when needed:
term-cli capture --session NAME --no-annotate

# Optional line numbers (1-based) for visible-screen captures (not scrollback)
term-cli capture --session NAME --annotate --line-numbers

When to use --annotate vs --raw vs plain capture:

  • Plain capture — default for shells, REPLs, command output. Fast, clean text.
  • --annotate — for TUIs (menus, file managers, settings panels) where you need to know what's happening. Returns visible content plus an Annotations: section with screen mode (normal/alternate), bell alerts, cursor position (Cursor: row,col, 1-based), optional mouse mode (Mouse: ... when enabled), and highlighted rows (1-based). Bell alerts are one-shot: shown once then cleared.
  • Auto behavior — plain capture auto-enables annotations only when an active alternate-screen TUI is detected; idle shell prompts stay plain. Use --annotate / --no-annotate to override.
  • --raw — when you need exact ANSI escape codes (e.g., parsing color semantics yourself). Higher token cost.

Waiting

Prefer wait — it detects shell prompts ($, %, #), REPL prompts (>>>, (Pdb), >), and is fastest.

bash
term-cli wait --session NAME

Use wait-idle for TUIs (vim, htop, less) that don't have a detectable prompt — waits for screen to settle:

bash
term-cli wait-idle --session NAME                                      # defaults: 2s idle, 10s timeout

Use wait-for sparingly — waits for pattern in screen output:

bash
term-cli wait-for --session NAME "error" "success" --ignore-case --print-match
term-cli wait-for --session NAME "error" --print-match-context 3             # match + 3 lines above/below

⚠️ Warning: wait-for matches the entire screen including your echoed command:

bash
# WRONG — pattern appears in echoed command, triggers immediately - WRONG
term-cli send-text --session NAME "./build.sh && echo DONE" --enter
term-cli wait-for --session NAME "DONE"
# ↑ Problem: The screen now shows "./build.sh && echo DONE" so "DONE" matches instantly

# RIGHT — wait for program output (not your own marker)
term-cli run --session NAME "npm run dev"
term-cli wait-for --session NAME "Listening on port"

# RIGHT — assemble pattern during print so it doesn't appear in command
term-cli send-text --session NAME "./build.sh && printf 'DON' && printf 'E\n'" --enter
term-cli wait-for --session NAME "DONE"

# BEST — use wait or run --wait when possible
term-cli run --session NAME "./build.sh" --wait

Human Assistance

When you need human help (passwords, CAPTCHAs, manual intervention):

bash
term-cli request --session NAME --message "Please enter SSH password"
term-cli request-wait --session NAME                                   # default timeout: 300s (5 min)
term-cli request-status --session NAME
term-cli request-cancel --session NAME

How it works: Your human runs term-assist list to see pending requests, then term-assist attach --session NAME to join. A status bar shows your message. When finished, they press Ctrl+B Enter to complete — they can optionally type a response message for you. request-wait then returns and you continue.

If they detach with Ctrl+B d while a request is still pending, request-wait fails with exit code 4.

File Transfer (over SSH / any shell)

Transfer files in-band through the terminal session — works over SSH, in containers, or any shell. No scp/sftp needed.

bash
# Upload a local file to the remote session
term-cli upload --session NAME local_file.txt remote_file.txt

# Download a remote file to local
term-cli download --session NAME remote_file.txt local_file.txt

# Overwrite existing files (both commands refuse by default)
term-cli upload --session NAME file.txt file.txt --force --timeout 30

# Pipe support: upload from stdin (REMOTE_PATH required)
cat config.json | term-cli upload --session NAME - /remote/config.json

# Pipe support: download to stdout (status message goes to stderr)
term-cli download --session NAME /remote/data.csv - | jq .

Key behaviors:

  • Hash verification — SHA-256 integrity check
  • Gzip compression — transfers are always gzip-compressed
  • Pipe support — use -, status/verbose output goes to stderr
  • Requires Python 3 on the remote — transfer deploys a Python helper; fails if python3 is not available
  • Prompt required — session must be at a clean shell prompt; transfers fail (exit 2) if a command is running, a TUI is active, or there is partial input on the line
  • Overwrite protection — refuses without --force
  • Fast Uploads
  • Optimized Downloads — starts with pipe-pane only in normal screen mode; in alternate screen mode (nested tmux/TUI), it goes directly to chunked mode for reliability.
  • Default timeout: 120s — set with --timeout

Other Commands

bash
term-cli resize --session NAME --cols 120 --rows 40
term-cli scroll --session NAME -50
term-cli pipe-log --session NAME /tmp/out.log
term-cli unpipe --session NAME

Example: Concurrency (Run long tasks while you work)

bash
term-cli start --session tests && term-cli run --session tests 'date +"%H:%M:%S"; pytest -q'
# ... do other work in parallel ...
term-cli wait --session tests --timeout 1800; term-cli capture --session tests
term-cli kill --session tests

Example: Dev Server

bash
term-cli start --session server && term-cli run --session server "npm run dev"
term-cli wait-idle --session server --timeout 15 && term-cli capture --session server
# ... later ...
term-cli send-key --session server C-c && term-cli wait --session server
term-cli kill --session server

Example: Python Debugger (pdb)

bash
term-cli start --session debug && term-cli run --session debug "python3 -m pdb script.py" && term-cli wait --session debug
term-cli send-text -s debug "b 42" --enter && term-cli wait -s debug      # breakpoint
term-cli send-text -s debug "c" --enter && term-cli wait -s debug         # continue
term-cli send-text -s debug "p some_var" --enter && term-cli wait -s debug && term-cli capture -s debug --tail 5
term-cli send-text -s debug "q" --enter; term-cli wait -s debug           # quit (may need 'y' to confirm)
term-cli kill --session debug

Example: SSH with Password (Human Helps)

bash
term-cli start --session remote && term-cli run --session remote "ssh user@host"
term-cli wait --session remote && term-cli capture --session remote
# If password prompt shown, request human help; if shell prompt, key auth succeeded
term-cli request --session remote --message "Please enter SSH password"
term-cli request-wait --session remote && term-cli capture --session remote

Example: File Transfer over SSH

bash
# Assumes passwordless login
term-cli start --session remote && term-cli run --session remote "ssh user@host" --wait
# Upload a config file
term-cli upload --session remote ./nginx.conf /etc/nginx/nginx.conf
# Download a log file
term-cli download --session remote /var/log/app.log ./app.log

Tips

  • Defaults are sane. Plain capture (visible screen), wait (10s timeout), and start (80x24) work for most cases. Use --tail to focus on just the last few rows. Only add --timeout or --scrollback when the default isn't enough.
  • Use long-form flags. --session, --timeout, --scrollback over -s, -t, -n. Short forms save almost no tokens and hurt readability.
  • Chain commands: Fewer tool calls = less overhead, lower token usage, faster wall clock time. Use && for dependent operations (send-text --enter && wait), ; to always run the next command (wait --timeout 5; capture — see output even on timeout).
  • Remember: you can run interactive apps through term-cli (debuggers, TUIs, SSH, installers).
  • Lost context after memory compaction? Plain capture first (visible screen is usually enough). Use --scrollback 30 if relevant output scrolled off in normal screen mode. In alternate screen mode, prefer --tail/--annotate; use --force only when you explicitly accept potentially stale history.
  • Prefer debuggers (pdb/gdb) via term-cli for bug analysis; step + inspect beats print-debugging for many issues.
  • Think concurrently: start long tasks in one session and keep working; check progress with status, wait, and capture.
  • Use --help on any command for details: term-cli run --help
  • Default timeout is 10s for most wait commands — increase for builds, tests, or slow operations
  • run --wait already waits for prompt — don't add another wait after it
  • Capture before sending keys to verify screen state
  • Unsure if a TUI is running? term-cli status --session NAME shows Screen: alternate when a full-screen app (vim, htop, less) is active. Use wait-idle for alternate screen, wait for normal.
  • Navigating TUIs? Use capture --annotate to see screen mode, bell alerts, cursor position (Cursor: row,col, 1-based), and which items are highlighted. Plain capture loses all styling, and --raw costs many tokens.
  • Default request-wait timeout is 5 minutes — usually no need to override
  • Locked sessions (exit code 5): agent can only capture, status, wait-*, request*, list, scroll, pipe-log, unpipe. Commands run, send-*, resize, kill, upload, download are blocked

Exit Codes

CodeMeaning
0Success
1Runtime error
2Invalid input
3Timeout
4Human detached while request pending
5Session locked (agent read-only)
127tmux not found

Feedback

As an agent, you're well-positioned to spot improvements. Found a bug or have an idea? Email the maintainer: Elias Oenal termcli@eliasoenal.com

Frequently asked questions

What does the Term Cli AI skill do?

Controls interactive terminal sessions for running long-lived processes, servers, REPLs, debuggers, and TUI programs without blocking. Use when you need to run dev servers (npm run dev), debuggers (pdb, gdb), REPLs (python, node), databases (psql, mysql), SSH sessions, or editors (vim, nano) — any interactive or blocking program.

Why use Term Cli on TypingMind?

Because you install it once and use it with any model. Term Cli is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Term Cli in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/EliasOenal/term-cli/tree/main/skills/term-cli. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Term Cli?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Term Cli?

As many as you like. As long as a model supports skills, you can use Term Cli with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Term Cli AI skill free?

Yes. It is published on GitHub by EliasOenal under the BSD-3-Clause-Clear license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇