📦

ntm

Name: ntm
Author: Dicklesworthstone

Content revision r1 Critical ⚙️ External commands🌐 Network access📁 Filesystem access

Coordinate AI Agents in Tmux

Managing several AI coding agents across terminals is hard to track. NTM provides tmux-based sessions, dashboards, routing, and automation for Claude Code, Codex, and related tools.

Supports: Claude Codex Code(CC)

⚠️ 38 Poor

Do not auto-install this skill.

The canonical policy requires operator review before any installation action.

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "ntm" from https://skillstore.io/skills/dicklesworthstone-ntm.md and its manifest at https://skillstore.io/api/skills/dicklesworthstone-ntm/manifest. Verify the artifact. Do not auto-install. Inspect the skill and report your findings, then wait for an operator or manual installation decision.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/dicklesworthstone-ntm.md Signed manifest GET /api/skills/dicklesworthstone-ntm/manifest Signed lockfile GET /api/skills/dicklesworthstone-ntm/lockfile

Test it

Using "ntm". Show me the current agent session status.

Expected outcome:

A session summary with pane names, agent types, activity states, and context indicators.
A list of stalled or unhealthy panes that need attention.

Using "ntm". Distribute this feature across agents.

Expected outcome:

A suggested split between architecture, implementation, tests, and review.
Commands or steps for sending each prompt to the right agent group.

Using "ntm". Capture all Claude pane output for review.

Expected outcome:

A saved or copied transcript summary focused on the selected Claude panes.

Security Audit

Critical

v9 • 7/9/2026 Open versioned report

Most static command findings are false positives caused by Markdown examples and inline command names. The audit confirms critical curl-to-bash installation instructions, persistent shell initialization, hidden prompt logging, outbound webhook support, and high-risk examples that disable agent safety controls.

Files scanned

578

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (4)

Critical

SKILL.md:25

Pipe to shell pattern

curl -fsSL https://raw.githubusercontent.com/Dicklesworthstone/ntm/main/install.sh | bash

The quick start pipes a remote install script directly into bash. This executes network-fetched code without inspection, pinning, or integrity verification.

Critical

SKILL.md:538

Pipe to shell pattern

curl -fsSL https://raw.githubusercontent.com/Dicklesworthstone/ntm/main/install.sh | bash

The installation section recommends curl piped directly to bash. This is a high-risk installation pattern because remote code executes immediately.

High

SKILL.md:455-458

Unsafe Agent Permission Bypass Configuration

The example configuration sets Claude, Codex, and Gemini commands with flags that bypass permissions, approvals, sandboxing, or safety prompts. This can allow spawned agents to run high-impact actions without normal review.

The documented agent command strings explicitly include dangerously-skip-permissions, bypass approvals and sandbox, and yolo flags.

High

SKILL.md:379-395

Configurable Shell Hooks Execute on Agent Events

The skill documents command hooks that run shell commands during events such as pre-send and post-spawn. This is powerful automation, but unsafe hook configuration can execute arbitrary commands or persist sensitive prompt content.

The documented hooks.toml format contains command fields and event triggers, including a pre-send example that logs NTM_MESSAGE.

Capability review items (12)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High

SKILL.md:28

Hidden file in home directory

echo 'eval "$(ntm init zsh)"' >> ~/.zshrc && source ~/.zshrc

The quick start appends executable shell initialization to ~/.zshrc. Modifying a hidden startup file creates persistent behavior in future shells.

High

SKILL.md:390

Hidden file in home directory

command = "echo \"$(date): $NTM_MESSAGE\" >> ~/.ntm-send.log"

The hook example writes agent messages to ~/.ntm-send.log. This may persist prompt or task content in a hidden home-directory log.

Medium

SKILL.md:28

Shell command substitution

echo 'eval "$(ntm init zsh)"' >> ~/.zshrc && source ~/.zshrc

The command appends eval of ntm init output into ~/.zshrc. This creates persistent shell startup execution from generated tool output.

Medium

SKILL.md:390

Shell command substitution

command = "echo \"$(date): $NTM_MESSAGE\" >> ~/.ntm-send.log"

The hook example runs a shell command on a pre-send event and interpolates runtime message data. The example is simple logging, but the hook mechanism executes user-configured shell commands.

Medium

SKILL.md:522

Shell command substitution

After `eval "$(ntm init zsh)"`:

The documentation normalizes eval of ntm init output for shell aliases. This is a persistent shell integration pattern that should be treated carefully.

Medium

SKILL.md:522

Template literal with command substitution

After `eval "$(ntm init zsh)"`:

The inline command contains eval with command substitution. Even as documentation, it encourages a risky persistent shell initialization pattern.

Medium

SKILL.md:355

Slack webhook

url = "https://hooks.slack.com/..."

The notification configuration supports Slack webhook delivery. This is legitimate alerting, but webhook output may expose event metadata outside the local system.

Medium

SKILL.md:28

Hidden file access

echo 'eval "$(ntm init zsh)"' >> ~/.zshrc && source ~/.zshrc

The command writes to ~/.zshrc and sources it immediately. That is legitimate shell integration but still a sensitive hidden-file modification.

Medium

SKILL.md:390

Hidden file access

command = "echo \"$(date): $NTM_MESSAGE\" >> ~/.ntm-send.log"

The hook example writes runtime message content to a hidden home-directory file. This can create unexpected local retention of sensitive prompts.

Low

SKILL.md:25

Hardcoded URL

curl -fsSL https://raw.githubusercontent.com/Dicklesworthstone/ntm/main/install.sh | bash

The hardcoded URL is part of a curl-to-bash installer. The URL itself is expected for installation, but it contributes to unauthenticated remote code execution risk.

Low

SKILL.md:355

Hardcoded URL

url = "https://hooks.slack.com/..."

The Slack webhook URL is a placeholder rather than a real secret. It still documents an outbound webhook capability that can send event data externally.

Low

SKILL.md:538

Hardcoded URL

curl -fsSL https://raw.githubusercontent.com/Dicklesworthstone/ntm/main/install.sh | bash

The hardcoded URL is part of the recommended installer command. It fetches remote code for execution without integrity verification.

Risk Factors

⚙️ External commands (50)

🌐 Network access (4)

SKILL.md:25 SKILL.md:355 SKILL.md:538 SKILL.md:355

📁 Filesystem access (8)

SKILL.md:28 SKILL.md:382 SKILL.md:390 SKILL.md:445 SKILL.md:28 SKILL.md:382 SKILL.md:390 SKILL.md:445

Detected Patterns

Pipe to shell pattern×2

Audited by: claude View Audit History →

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/dicklesworthstone-ntm/audits/9?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/dicklesworthstone-ntm/security.svg)](https://skillstore.io/skills/dicklesworthstone-ntm?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/dicklesworthstone-ntm?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/dicklesworthstone-ntm/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/dicklesworthstone-ntm.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

Dicklesworthstone. (2026). ntm security audit report (audit version 9) [Author version unspecified]. Skillstore. https://skillstore.io/skills/dicklesworthstone-ntm/audits/9

BibTeX citation

@techreport{dicklesworthstone-dicklesworthstone-ntm-2026,
  author = {Dicklesworthstone},
  title = {ntm security audit report (audit version 9)},
  institution = {Skillstore},
  year = {2026},
  number = {9},
  url = {https://skillstore.io/skills/dicklesworthstone-ntm/audits/9},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "ntm security audit report (audit version 9)"
version: "unspecified"
type: report
authors:
  - name: "Dicklesworthstone"
date-released: "2026-07-09"
url: "https://skillstore.io/skills/dicklesworthstone-ntm/audits/9"
identifiers:
  - type: other
    value: "skillstore:dicklesworthstone-ntm:audit:9"
    description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: Medium

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Run parallel implementation tracks

Create one tmux session with separate agents for architecture, implementation, testing, and review work.

Monitor long-running agent sessions

Use dashboards, activity views, health checks, and output capture to supervise several coding agents from one terminal.

Automate agent coordination

Use robot-mode commands to inspect sessions, assign work, replay prompts, and integrate with task automation.

Try These Prompts

Create a basic session

Use NTM to create a tmux session for my project with two Claude Code agents and one Codex agent.

Send work to agents

Use NTM to send an implementation task to Claude agents and a test-writing task to Codex agents in the same session.

Inspect session health

Use NTM robot mode to summarize session health, context usage, recent output, and stalled panes.

Design a multi-agent workflow

Plan an NTM workflow that assigns design, implementation, review, testing, and documentation across multiple agents with checkpoints.

Best Practices

Review install scripts before execution and prefer package-manager installs when possible.
Keep agent permission and sandbox protections enabled unless you have a controlled environment.
Use clear session names and checkpoint before large multi-agent changes.

Avoid

Do not pipe remote installers directly to a shell in sensitive environments.
Do not configure agents with unrestricted bypass flags for normal development work.
Do not log prompts or pane output to persistent files unless the data is safe to retain.

Frequently Asked Questions

What does NTM manage?

It manages named tmux sessions that contain multiple AI coding agents and related workflow panes.

Which tools does it support?

The documentation describes support for Claude Code, Codex, Gemini, and Claude-compatible workflows.

Can it send prompts to multiple agents?

Yes. It documents commands for sending prompts to Claude, Codex, Gemini, or all agents in a session.

Does it include automation APIs?

Yes. Robot mode provides machine-readable commands for status, context, spawning, assignment, and replay.

Is it safe to install with the one-line installer?

The documented curl-to-bash installer is risky. Review the script or use a safer package installation method.

What should I configure carefully?

Review shell hooks, webhook notifications, startup-file changes, and any agent command that disables approvals or sandboxing.

Developer Details

Author

Dicklesworthstone

License

MIT

Skillstore revision

Version notice

The author did not declare a version.

Repository

https://github.com/Dicklesworthstone/agent_flywheel_clawdbot_skills_and_integrations/tree/main/skills/ntm

Ref

1ffa7643651792ccb4bd3b15d924d2c97edff755

Maintenance freshness

7/18/2026

Usage

5 downloads · 205 views

File structure

📄 SKILL.md