Przejdź do treści
Biblioteka skilli
AI i agenci · Automatyzacja

pack

/pack

Przyjmuje wtyczki i skille Claude Code jako pakiety Regency, włącza je per projekt i uruchamia ich playbooki.

Repozytorium
regency ↗
Licencja
MIT
Ostatnia zmiana repozytorium
06.11.2026

Ocena i bezpieczeństwo

100/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść35 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
1
Skrypty
0
Komentarze HTML
0
Adresy URL w SKILL.md
0
Przestarzałość

Brak sygnałów. Repozytorium zmienione 06.11.2026, bez odwołań do wycofanych modeli i API.

Licencja

MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.

Co robi i kiedy

Przyjmuje wtyczki i skille Claude Code jako pakiety Regency, włącza je per projekt i uruchamia ich playbooki.

Wyzwalacze z opisu

adopt this pluginwhich packs are active

Argumenty

/pack <adopt | run | list | check | activate | ratify | show | suggest | deactivate | mcp> ...
Oryginalny opis (EN)

Regency packs: what regency becomes in one domain. Adopt any Claude Code plugin or skill as it is, switch it on per project, ratify its hooks, run its playbooks. Use for /regency:pack[:adopt|run], 'adopt this plugin', 'which packs are active'.

Jak działa

  1. Commands
  2. adopt
  3. activate · ratify
  4. run
  5. What a pack may and may not do
  6. Report
Sekcje z nagłówków SKILL.md, w oryginale · 81 linii, 729 słów
Podgląd SKILL.md
---
name: pack
description: "Regency packs: what regency becomes in one domain. Adopt any Claude Code plugin or skill as it is, switch it on per project, ratify its hooks, run its playbooks. Use for /regency:pack[:adopt|run], 'adopt this plugin', 'which packs are active'."
argument-hint: "<adopt | run | list | check | activate | ratify | show | suggest | deactivate | mcp> ..."
---

# pack — regency, specialised

> **Root.** This file is `skills/pack/SKILL.md`; the regency root is two directories above it. `scripts/…` and
> `library/…` paths are relative to that root. Doctrine: `library/packs.md`. Read `library/conventions.md` once per
> session.

A pack is what regency becomes in one domain: a folder with `pack.json` and its files, adopted from any Claude Code
plugin or skill as it is, or written native. It lives in one of two layers:

| Layer | Folder | Travels with | Rule |
|---|---|---|---|
| public | `<regency>/packs/<id>/` | regency's repository | an open-source license regency may redistribute |
| project | `<guidelines>/.packs/<id>/` | the project's repository | any license; a project pack shadows a public one of the same id |

Claude Code loads plugins per install, never per project, so regency reads packs itself: this playbook opens a pack's
files, and `scripts/hook-packs.py` runs a pack's hooks only where the pack is active and the Regent ratified them.

## Commands

```bash
python <regency-root>/scripts/packs.py list
python <regency-root>/scripts/packs.py adopt <folder | git url> --id <id> --layer public|project [--when "*.ext,..."]
python <regency-root>/scripts/packs.py adopt ... --yes        # writes; without --yes it only prints the plan
python <regency-root>/scripts/packs.py check
python <regency-root>/scripts/packs.py show <id> [<item key>]
python <regency-root>/scripts/packs.py suggest
python <regency-root>/scripts/packs.py activate <id> [--yes]
python <regency-root>/scripts/packs.py deactivate <id> [--yes]
python <regency-root>/scripts/packs.py ratify <id> <item key | hooks | all>        # the review; records nothing
python <regency-root>/scripts/packs.py mcp <id> [--write --yes]
```

Add `--repo DIR` when the current directory is not inside the project.

## adopt

1. Run the plan without `--yes` and show it to the Regent whole: the source, its license, every item by kind, every
   part **not adopted** with its reason, the requirements (node, python, bash…) and where the pack lands.
2. The layer is the Regent's choice. Public publishes the pack with regency; `adopt` refuses a source with no
   open-source license there. Material the Regent keeps private, or that another project owns (Glyph's, by R2.2),
   goes to the project layer.
3. On the Regent's yes, run it with `--yes`. Nothing runs yet: every item is a proposal.
4. `--when` takes the file globs that suggest the pack for a project (`packs.py suggest` reads them).

## activate · ratify

- `activate` switches a pack on in this project (`<guidelines>/.packs/active.json`). Its playbooks become runnable.
  Its hooks do not run.
- `ratify` without `--yes` is the **review**: each hook's event, matcher and exact command. Show it to the Regent.
- `ratify ... --yes` is the Regent's consent. The hazard hook asks the Regent on that call: approve only when the
  Regent said so for this pack after reading the review. **Never ratify on your own judgement** (T5: no role accepts
  its own work, and the agent that adopted a pack is the last one to clear its code).
- Any later change to any file of the pack makes its ratified hooks STALE: they stop until ratified again.

## run

`/regency:pack run <id> <item> [arguments]` (or `/regency:pack:run`):

1. `packs.py show <id> <item>`. Refuse when the pack is not active here; offer `activate`.
2. A `playbook:` or `command:` item: read the file it names and follow it with the given arguments, as if the
   Regent had invoked it. Its relative paths resolve against its own folder; replace `${CLAUDE_PLUGIN_ROOT}` with
   the root `show` prints. Its requirements (`show` names them) must be on PATH; say which is missing instead of
   improvising.
3. An `agent:` item: dispatch a subagent whose prompt is the agent file's body, then the task.
4. Say `proposal` when the item is not ratified: the Regent's invocation is consent to run it this once.

## What a pack may and may not do

- A pack adds; it never overrides regency's guards. Its hooks run after regency's own, through the dispatcher, which
  passes on deny and ask, and drops allow and a rewritten tool input.
- Precedence: the Regent's message → the project's own guidelines → active packs → regency's conventions → a
  playbook's default.
- Parts a pack cannot bring in this version (LSP servers, output styles, themes, monitors, settings, `bin/` on PATH,
  prompt and http hooks, Stop hooks) are listed in its `pack.json` under `unsupported`, never dropped silently.

## Report

One line per pack touched: `pack <id> · <layer> · <n> item(s) · active | inactive · hooks running <k>/<m>`, then the
next act (`activate`, `ratify`, `run`) the Regent may choose.

Wersje

RepozytoriumWynikLicencja
Ta strona · Trevojan/regency100MIT
lilly-logic-labs/lll-skills100MIT
kouko/monkey-skills100MIT
himanshu-mamgain/offerstar95MIT

Migawka bazy z 13.11.2026 · wersja paczki 5ed630d877be (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

  • 100regency
    Dostarcza harness Regency do pracy zespołu agentów z decyzjami Regenta, regułami, skryptami, trybami i kontrolą zdrowia projektu.
  • 100health
    Wykonuje przegląd zdrowia Regency: jedna linia na wynik z podaniem metody, w tym sprawdzenia skrótów, wersji, hooków i kolejek.
  • 95public-plugin-builder
    Pomaga budować wtyczki, skille i agenty Claude, określać strukturę wtyczki Claude Code i publikować ją w marketplace.
  • 100claude-code-plugin-development
    Prowadzi przez tworzenie pluginów Claude Code, które pakują agentów, hooki, komendy, skille lub serwery MCP w dystrybuowalną całość.
  • 100pack-review
    Porównuje pakiet agentów z wersją Claude Code, sprawdza otwarte punkty z poprzedniego przeglądu i zapisuje raport.
  • 100validate-all
    Uruchamia pełną walidację claudelint na plikach projektu Claude Code: CLAUDE.md, skille, hooki i serwery MCP.

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.