Przejdź do treści
Biblioteka skilli
AI i agenci · Przegląd i audyt

ccretro

/ccretro

Przygotowuje przegląd użycia Claude Code i zamienia sugestie z raportu w zmiany w skillach, CLAUDE.md i hookach.

Repozytorium
ccretro ↗
Licencja
MIT
Ostatnia zmiana repozytorium
08.11.2026

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 08.11.2026, bez odwołań do wycofanych modeli i API.

Licencja

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

Co robi i kiedy

Przygotowuje przegląd użycia Claude Code i zamienia sugestie z raportu w zmiany w skillach, CLAUDE.md i hookach.

Wyzwalacze z opisu

s Claude Code usage, open the report, then turn its振り返りretroccretrolook back at how I used Claude Code

Argumenty

/ccretro [--days N] [--exclude-project name ...] [--redact]
Oryginalny opis (EN)

Run a ccretro retrospective of the user's Claude Code usage, open the report, then turn its "what to do next" suggestions into real changes — new or updated skills, CLAUDE.md lines, hooks, permission rules — after the user picks which ones. Use for "振り返り", "retro", "ccretro", "look back at how I used Claude Code", or a monthly review.

Jak działa

  1. 1. Generate
  2. 2. Read the result
  3. 3. Check each suggestion against the real config
  4. 4. Let the user choose
  5. 5. Apply
  6. 6. Report
Sekcje z nagłówków SKILL.md, w oryginale · 59 linii, 599 słów
Podgląd SKILL.md
---
name: ccretro
description: Run a ccretro retrospective of the user's Claude Code usage, open the report, then turn its "what to do next" suggestions into real changes — new or updated skills, CLAUDE.md lines, hooks, permission rules — after the user picks which ones. Use for "振り返り", "retro", "ccretro", "look back at how I used Claude Code", or a monthly review.
argument-hint: "[--days N] [--exclude-project name ...] [--redact]"
---

# ccretro: look back, then act on it

The report alone changes nothing. This skill closes the loop: generate the retrospective, then apply the suggestions the user agrees with, preferring to update what already exists over adding more.

## 1. Generate

Run the copy of ccretro that ships with this plugin, passing the user's arguments through (default period: 30 days). It prints the HTML path on stdout; the JSON with the same basename sits next to it.

```bash
node "${CLAUDE_SKILL_DIR}/../../bin/ccretro.mjs" $ARGUMENTS
```

- If the user gave a bare number ("/ccretro 7"), turn it into `--days 7`; the CLI takes no positional arguments.
- If that path doesn't exist (the skill was copied without the rest of the repo), fall back to `npx --yes --allow-git=root github:0-draft/ccretro` (npm 12 refuses git packages without `--allow-git`).
- It calls `claude -p` for the narrative. If that fails (no network in a sandbox, not logged in), it falls back to rule-based output and says so on stderr. Tell the user which one they got.
- If the user has private projects they don't want summarised, pass `--exclude-project <name>` or point them at `~/.config/ccretro/config.json` (`{"excludeProjects": [...]}`).

## 2. Read the result

Read the `.json` file, not the HTML. Use:

- `insight.nextActions`: the suggestions (`kind`, `title`, `why`, `how`, optional `snippet`)
- `repeated`: prompts typed again and again, with counts
- `inventory`: skills, subagents, hooks, and per-project CLAUDE.md the user already has
- `insight.fluency` / `insight.patterns`: context for why a suggestion matters

Give the user a short summary in their language: the headline, and the suggestions as a numbered list with the evidence count behind each.

## 3. Check each suggestion against the real config

Before proposing a change, look at what exists. The JSON inventory is a hint; the files are the truth.

- Skills: list `~/.claude/skills/*` and the project's `.claude/skills/*`. Resolve symlinks — if a skill dir links into a dotfiles repo, edit the file in that repo.
- Read any skill whose name or description overlaps the suggestion. **If one already covers most of it, propose an edit to that skill instead of a new one.**
- CLAUDE.md: read `~/.claude/CLAUDE.md` (follow its symlink) and the project's. Don't add a line that restates an existing rule; tighten the existing rule instead.
- Hooks / permissions: read `~/.claude/settings.json`. Settings changes go through the `update-config` skill if available.

Drop suggestions that turn out to be already done, and say why.

## 4. Let the user choose

Ask once, with a multi-select listing the surviving suggestions, each labelled new vs update and the target path. Do not apply anything they didn't pick.

## 5. Apply

For each picked item:

- **New skill**: `<skills-dir>/<name>/SKILL.md` with `name` and a `description` that says when to trigger it. Write the steps from what the user actually typed (quote the repeated prompt's intent), not generic advice. Keep it short.
- **Skill update**: minimal edit to the existing SKILL.md; keep its structure and voice.
- **CLAUDE.md**: one or two lines in the matching section.
- **Hook / permission**: via `update-config`.
- If the config lives in a git repo, show `git diff` for the touched files. Do not commit or push unless the user asks; follow that repo's commit conventions if they do.
- If the skills dir is populated by an install script (symlinks into a repo), tell the user to re-run it for new skills to appear.

## 6. Report

List what changed (paths), what was skipped and why, and the HTML report path. Suggest re-running monthly; `/schedule` can automate the report part, but applying changes always waits for the user.

Wersje

W bazie nie ma innych skilli o nazwie ccretro.

Migawka bazy z 13.11.2026 · wersja paczki b1002954c72e (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

  • 100retro
    Robi retrospektywę sesji Claude Code: znajduje błędy i przeoczenia, a potem proponuje zmiany w pamięci, CLAUDE.md, skillach lub hookach.
  • 90improve
    Poprawia konfigurację Claude Code na podstawie powtarzających się błędów i proponuje zmiany w CLAUDE.md, skillach i hookach.
  • 100install-skill-tracker
    Instaluje system śledzenia użycia skilli w Claude Code oparty na hookach: loguje użycie, czas wykonania, tokeny i prompty do późniejszej analizy wzorców i kosztów.
  • 95claude-code-skills
    Wyjaśnia instalację, użycie i tworzenie skilli Claude Code, w tym ścieżki i konwencje specyficzne dla Claude Code.
  • 98coograph-retro
    Zamienia sygnały z poprzednich sesji w OpenSpec z dowodowo uzasadnionymi zmianami w instrukcjach, hookach i skillach.
  • 100audit
    Ocenia gotowość repozytorium i ustawień dewelopera do pracy z agentami AI i rekomenduje zmiany w CLAUDE.md, testach, skillach i hookach.

Uwagi i zgłoszenia

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