Tower logo

Tower

Community
koolamusic
tower

Run the orchestrator's war room on visible Herdr panes so the user can watch every child agent work. Use when the user says 'tower on', 'orchestrate this in herdr', wants routed agents they can see, or when the orchestrator is on inside a Herdr-managed pane.

Overview

Publisherkoolamusic
Repositoryclaudefiles
Skill nametower
Stars
121
Forks
14
Bundled files
Instructions only
LicenseMIT
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 koolamusic on GitHub. Read the source before you install it.

Installation

Install the Tower 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/koolamusic/claudefiles/tree/main/skills/tower
  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/koolamusic/claudefiles.git /tmp/claudefiles
mkdir -p .claude/skills
cp -r /tmp/claudefiles/skills/tower .claude/skills/tower
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Tower 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 Tower 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 Tower 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.

Tower

The orchestrator skill decides who does what. This skill decides where they run: one Herdr pane per child, visible to the user, instead of invisible background subagents.

Gate first:

bash
test "${HERDR_ENV:-}" = 1

If the check fails, say the session is not inside Herdr and fall back to the orchestrator's own substrate. Do not run any herdr command.

Commands

  • tower on — orchestrator mode with this substrate. Loads orchestrator and herdr if they are not already in context.
  • tower off — same as orchestrator off.
  • tower status — the orchestrator status check, with pane states from Herdr.

What stays the same

Everything the orchestrator skill defines: the three slots, personas, routing rules, the delegation brief, the war room table, the stall cap, and the safety rules. Read them there. This file replaces only the "Child Agent Substrate" section. The herdr skill remains the authority for CLI syntax and safety.

Substrate mapping

Orchestrator stepTower
Spawn a childpane split (or tab create for an executor), then agent start <name> --kind claude --pane <id>
Continue a childagent prompt <name> "<text>" --wait --timeout <ms>
Status checkagent list; per child agent get <name>
Read a closeoutagent read <name> --source recent-unwrapped --lines 200
Worktree isolationworktree create --branch <name> for the executor's workspace
Stallagent_status unchanged across two status checks

Naming

The agent name is the war room key: <branch-or-key>-<slot>, lowercase, hyphens, under 32 chars. auth-exec, auth-audit, search-survey. The name follows the pane occupant, so the same child answers follow-up prompts on that branch, and agent list reads as the roster.

Placement

  • Surveyors and auditors are short-lived: split beside the calling pane with --no-focus, right when wide, down when tall, and close the pane after the closeout is recorded.
  • Executors are long-lived and own a branch: give each its own tab, labelled with the branch, so its transcript is not fighting for space. Use a worktree workspace when two executors are mutating at once.
  • Never create a workspace, tab, or pane the user did not ask for beyond these rules, and never close one this session did not create.

Dispatch

bash
herdr pane split --current --direction right --cwd "$PWD" --no-focus
herdr agent start auth-exec --kind claude --pane <returned-pane-id>
herdr agent prompt auth-exec "<delegation brief>" --wait --timeout 600000

Wait for agent start to return ready before prompting. The delegation brief is the orchestrator's, verbatim, plus one line: "Finish your closeout as a short Markdown block so it can be read from the terminal."

Waiting

agent prompt --wait blocks this session on one child. With several children active, put each wait in a background Bash call, or prompt without --wait and use agent wait <name> --timeout <ms> at the next status check. A timeout does not prove the prompt was lost; read the pane before re-sending.

Blocked children

A child in blocked has a permission dialog or a question on screen. Do not answer it. Report the row as blocked, name the pane so the user can look, and route the decision to the user. This is the visibility the substrate exists for.

Reading results

Read the closeout with agent read. If the transcript is longer than the read returns, ask the child to write the closeout to a Markdown file in a temporary directory and reply with the path, then read that file. Findings from surveyors and auditors still come back to this session, and this session routes any fix to the owning executor.

War room table

Same four columns. Put the agent name in the Next cell on the first row for a branch so it survives summarization:

md
| auth | executor (backend) | working | pane auth-exec: migration in progress |

Success criteria

  • tower on inside Herdr yields a war room where every child is a visible pane.
  • Outside Herdr it degrades to the orchestrator's own substrate with one line of notice.
  • Same-branch follow-ups go to the same named agent.
  • A blocked child reaches the user, not an automatic answer.
  • Nothing in orchestrator or herdr is duplicated here beyond the table above.

Frequently asked questions

What does the Tower AI skill do?

Run the orchestrator's war room on visible Herdr panes so the user can watch every child agent work. Use when the user says 'tower on', 'orchestrate this in herdr', wants routed agents they can see, or when the orchestrator is on inside a Herdr-managed pane.

Why use Tower on TypingMind?

Because you install it once and use it with any model. Tower 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 Tower in TypingMind?

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

Which AI models can use Tower?

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 Tower?

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

Is the Tower AI skill free?

Yes. It is published on GitHub by koolamusic under the MIT 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 👇