More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Tmux Safe Send — Exact-Target Guard for Multi-Pane Agents

    1

    What you get: - tmux_target_check.py asks tmux where the target really resolves and compares it with what you requested.

    $12

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Tmux Safe Send — Exact-Target Guard for Multi-Pane Agents

    Tmux Safe Send — Exact-Target Guard for Multi-Pane Agents

    Example session with this skill installed

    Before I wake the worker in team:0.9, check where tmux will actually send it.

    • Read your context and instructions
    • Compiled the tmux safe send

    $ python3 scripts/tmux_target_check.py team:0.1
    team:0.1 %1
    exit 0

    $ python3 scripts/tmux_target_check.py team:0.9
    ROUNDED: requested team:0.9 -> tmux resolved team:0.1 %1
    exit 5

    team:0.9 does not exist, and raw send-keys would have typed into team:0.1. I did not send anything. Which pane did you mean?

    Connects securely to your tools. The creator never sees your data.

    About this skill

    If you run several Claude Code (or other CLI agent) sessions in tmux panes and wake them with tmux send-keys, you have probably relied on a target like team:0.3 being exactly what it says. It is not guaranteed. On tmux 3.6a, team:0.9 (no such pane) lands on the last real pane, team:7.0 (no such window) lands on a real window, and tea:0.0 (no such session) lands on session team by prefix. send-keys returns 0 each time, so your message goes to the wrong agent and nothing tells you.

    What you get:

    • tmux_target_check.py asks tmux where the target really resolves and compares it with what you requested. Exit 0 = exact, exit 5 = rounded.
    • safe_send.py refuses on rounding, optionally verifies the pane is running the command you expect, types into the resolved pane id (%N), presses Enter in a separate call, then looks for your text in the pane. It never re-types, because a duplicated prompt is worse than a reported failure.
    • A SKILL file that teaches Claude Code to use the guard instead of raw send-keys.
    • 11 tests that run against a private tmux server, so your own sessions are never touched.

    Requirements and safety: Python 3.8+ and tmux. Standard library only. No network calls, no installers, no background services, no credentials. The checker is read-only, and --dry-run shows the destination without typing.

    Requires python3 and specific scripts. Limited to one-line messages. Does not support multi-line pasting to avoid early execution.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 8 days ago

    Requires

    Python

    Permissions

    Terminal / Shell
    Read Files
    Listed8 days ago
    Updated5 days ago

    What's inside

    Frequently Asked Questions

    Unlike standard tmux commands that might send text to the 'nearest match' if an ID is wrong, this skill enforces exact target resolution. It acts as a safety layer that aborts the operation if the pane, window, or session doesn't match your specific criteria, preventing costly mistakes in production environments.

    Yes, you can use the Command Guarding feature to define a regex pattern that matches the process running in the target pane. This ensures that if you intend to send a snippet to a Python REPL, the skill will block the execution if it detects a Bash shell or another process instead.

    This skill requires a Linux or macOS environment with `python3` and `tmux` installed. It is designed to be integrated into any AI agent framework that supports local tool execution and CLI access.

    You will receive the core logic scripts, the target verification engine, and a set of implementation guidelines to integrate these guards into your agent's workflow. It includes both the execution wrappers and the dry-run simulation tools.

    To ensure maximum safety, the skill currently flattens multi-line inputs into a single line. This prevents a common failure mode where a newline character triggers a command execution before the agent has finished sending the full instruction or verified the environment.

    The skill includes a 'Dry Run Mode' which allows you to simulate the routing logic. This lets you test your target identifiers and regex guards to see exactly where a message would land without actually sending any keystrokes to your active terminal.

    New in Workflow Automation