Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →CLI utility to communicate with other CLI Agents or Scripts in other tmux panes; use it only when user asks you to communicate with other CLI Agents or Scripts in other tmux panes.
.claude/skills/pchalasani-tmux-cli/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -38% | 0% |
Use the tmux-cli command to communicate with other CLI Agents or Scripts in other tmux panes. Do tmux-cli --help to see how to use it!
This command depends on installing the claude-code-tools. If you get an error indicating that the command is not available, ask the user to install it using: uv tool install claude-code-tools.
Use tmux-cli execute when you need to know if a shell command succeeded or failed:
bashtmux-cli execute "make test" --pane=2 # Returns JSON: {"output": "...", "exit_code": 0} tmux-cli execute "npm install" --pane=ops:1.3 --timeout=60 # Returns exit_code=0 on success, non-zero on failure, -1 on timeout
This is useful for:
Note: execute is for shell commands only, not for agent-to-agent chat. For communicating with another Claude Code instance, use send + wait_idle + capture instead.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 8,823 | 3,417 | -61% | 1 | 1 | 0% | 1,107 | 431 | -61% | 0 | 0 | — |
case-06 | fail→pass | 9,559 | 1,875 | -80% | 1 | 1 | 0% | 1,808 | 583 | -68% | 0 | 0 | — |
case-02 | fail→fail | 9,348 | 3,365 | -64% | 1 | 1 | 0% | 1,740 | 428 | -75% | 0 | 0 | — |
case-03 | fail→fail | 8,985 | 3,575 | -60% | 1 | 1 | 0% | 1,759 | 526 | -70% | 0 | 0 | — |
case-04 | fail→pass | 8,498 | 2,533 | -70% | 1 | 1 | 0% | 1,715 | 819 | -52% | 0 | 0 | — |
case-05 | fail→pass | 6,518 | 1,408 | -78% | 1 | 1 | 0% | 1,177 | 509 | -57% | 0 | 0 | — |
case-07 | pass→pass | 11,534 | 3,161 | -73% | 1 | 1 | 0% | 1,941 | 745 | -62% | 0 | 0 | — |
case-08 | fail→pass | 8,716 | 3,739 | -57% | 1 | 1 | 0% | 1,607 | 1,059 | -34% | 0 | 0 | — |
case-09 | fail→pass | 4,152 | 1,110 | -73% | 1 | 1 | 0% | 747 | 466 | -38% | 0 | 0 | — |
case-10 | pass→pass | 9,379 | 3,066 | -67% | 1 | 1 | 0% | 2,087 | 839 | -60% | 0 | 0 | — |
case-15 | fail→pass | 7,682 | 2,773 | -64% | 1 | 1 | 0% | 1,438 | 602 | -58% | 0 | 0 | — |
case-11 | fail→pass | 4,452 | 2,265 | -49% | 1 | 1 | 0% | 878 | 669 | -24% | 0 | 0 | — |
case-12 | pass→pass | 7,648 | 4,632 | -39% | 1 | 1 | 0% | 1,365 | 1,151 | -16% | 0 | 0 | — |
case-13 | pass→pass | 7,188 | 4,881 | -32% | 1 | 1 | 0% | 1,511 | 1,246 | -18% | 0 | 0 | — |
case-14 | fail→fail | 8,056 | 6,560 | -19% | 1 | 1 | 0% | 1,368 | 1,415 | +3% | 0 | 0 | — |
case-16 | fail→pass | 5,847 | 1,472 | -75% | 1 | 1 | 0% | 1,112 | 566 | -49% | 0 | 0 | — |
case-17 | fail→fail | 7,052 | 3,258 | -54% | 1 | 1 | 0% | 490 | 439 | -10% | 0 | 0 | — |
case-18 | fail→fail | 9,280 | 3,624 | -61% | 1 | 1 | 0% | 1,765 | 491 | -72% | 0 | 0 | — |
case-19 | fail→pass | 11,984 | 4,003 | -67% | 1 | 1 | 0% | 2,344 | 1,135 | -52% | 0 | 0 | — |
case-20 | fail→pass | 4,629 | 1,382 | -70% | 1 | 1 | 0% | 931 | 585 | -37% | 0 | 0 | — |
case-21 | fail→pass | 7,876 | 2,134 | -73% | 1 | 1 | 0% | 1,468 | 673 | -54% | 0 | 0 | — |
case-22 | pass→pass | 13,927 | 3,268 | -77% | 1 | 1 | 0% | 2,624 | 873 | -67% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 17 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +50 percentage points is the difference between those two pass rates over the 17 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.