tmux

Remote-control tmux sessions for interactive CLIs by sending keystrokes and scraping pane output.

Install with Tessl CLI

npx tessl i github:HKUDS/nanobot --skill tmux

What are skills?

1.37x

Quality

77%

Does it follow best practices?

Impact

80%

1.37x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./nanobot/skills/tmux/SKILL.md

SKILL.md

Review

Evals

Evaluation results

100%

51%

Automate a Python Data Processing Session

Python REPL session setup

Criteria

Without context

With context

Socket dir env var

100%

Socket named nanobot.sock

100%

PYTHON_BASIC_REPL=1

100%

Monitor commands printed

50%

100%

Literal send flag

100%

capture-pane flags

100%

Private socket flag

100%

Session name no spaces

100%

Session cleanup

100%

Non-interactive preference

100%

Without context: $0.1766 · 52s · 15 turns · 21 in / 2,491 out tokens

With context: $0.6999 · 2m 50s · 34 turns · 38 in / 7,939 out tokens

84%

-2%

Build a tmux Session Health-Check Tool

Session discovery and wait-for-text

Criteria

Without context

With context

find-sessions --all flag

100%

wait-for-text -T flag

100%

wait-for-text -t target

100%

wait-for-text -p pattern

100%

Socket dir env var

100%

Socket named nanobot.sock

Private socket flag

25%

Session name no spaces

100%

HEALTHY/UNHEALTHY output

100%

Session cleanup in demo

100%

wait-for-text not reimplemented

100%

Without context: $0.6285 · 3m 12s · 24 turns · 29 in / 11,428 out tokens

With context: $0.8282 · 2m 55s · 33 turns · 71 in / 10,811 out tokens

57%

17%

Orchestrate Parallel Automated Fixes Across a Repository

Parallel agent orchestration with worktrees

Criteria

Without context

With context

Git worktrees used

Prompt detection character

Socket dir env var

100%

Socket named nanobot.sock

Private socket flag

100%

Bulk session cleanup

100%

Parallel launch

100%

capture-pane flags

100%

Session names no spaces

100%

design_notes branch conflicts

Worktree cleanup

100%

Without context: $0.5656 · 2m 32s · 31 turns · 35 in / 7,931 out tokens

With context: $0.8765 · 3m 31s · 33 turns · 35 in / 12,515 out tokens

Evaluated: 3 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Automate a Python Data Processing Session Build a tmux Session Health-Check Tool Orchestrate Parallel Automated Fixes Across a Repository

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.