Skip to content
Back to skills

Knowledge Auto Update

FSecurity

Agent-local knowledge dosyalarini gorev sonunda ogrenilenlerle guncelle. Triggers: knowledge update, agent knowledge refresh, ogrendigini kaydet.

  • 3 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added May 27, 2026
data-aigobashapi

Works with

  • api

Security analysis

F10/100
  • criticalPipes output to a shell interpreter
  • highPerforms destructive filesystem operations
  • criticalDownloads and executes remote scripts — classic supply chain attack
  • criticalDownloads and executes remote scripts — classic supply chain attack

Pro shows the line behind each finding and how to fix it

Scanned May 27, 2026

npx -y skills add SkyWalker2506/claude-config --skill knowledge-auto-update --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Knowledge Auto Update?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Knowledge Auto Update
[![Security: F — Skills Directory](https://www.skillsdirectory.com/api/skills/skywalker2506-knowledge-auto-update/badge)](https://www.skillsdirectory.com/skills/skywalker2506-knowledge-auto-update)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: knowledge-auto-update
description: "Agent-local knowledge dosyalarini gorev sonunda ogrenilenlerle guncelle. Triggers: knowledge update, agent knowledge refresh, ogrendigini kaydet."
argument-hint: "[agent-id] [--auto]"
---

# /knowledge-auto-update — Per-Agent Knowledge-First Auto-Update

Bir agent non-trivial bir gorevi bitirdikten sonra kendi `knowledge/` dizinini (sadece kendisininkini) ogrendikleriyle gunceller. Patterns, trial-and-error dersleri, domain bulgulari — hepsi ekleme (additive patch) seklinde.

## Kapsam (zorunlu)

**Agent-local only.** Bu skill SADECE `~/Projects/claude-config/agents/<agent>/knowledge/*.md` altinda yazar.

- Global skill olusturma **YASAK** (SessionStart sisirir — Video 1 / Video 4 uyarisi)
- Baska agent'in knowledge'ini degistirme
- `~/.claude/` altina dogrudan yazma — install.sh ile mirror edilecek
- CLAUDE.md / charter / harness — dokunma

Manuel muadiller: `agent-sharpen` (web'den arastir), `agent-refine-knowledge` (Opus ile kalite review). Bu skill otomatik "gorevden ogrenilen" yakalayici; onlarin yerini tutmaz.

## Tetikleme modlari

| Mod | Nasil |
|-----|-------|
| Manuel | `/knowledge-auto-update <agent-id>` — kullanici veya Jarvis cagirir |
| Auto | Agent kendini `>10 tool call` sonrasi self-trigger — opsiyonel hook wiring (su an devre disi) |

Auto hook wiring bu skill kapsamina **dahil degil** — ayri karar. Su an sadece manuel + explicit trigger.

## Akis

### 1. Agent tespit

- Argument verildiyse `<agent-id>` onu kullan
- Yoksa son dispatch edilen agent'i `/tmp/watchdog/current_dispatch.json`'dan oku
- Hic yoksa → kullaniciya sor, dur

### 2. Gorev ozeti cikar

Son turn'deki (veya son dispatch'teki) tool call'lari ve sonuclari tara:
- Hangi dosyalara dokunuldu?
- Hangi komutlar basarisiz oldu, neden?
- Hangi non-obvious pattern kullanildi?
- Hangi domain gercegi kesfedildi? (API quirk, lib davranisi, repo yapisi)

Max 8 tool call bu adimda.

### 3. Kategorize et

Bulgulari 3 bucket'a ayir:

| Bucket | Ornek |
|--------|-------|
| **Patterns** | "Bu tur task'ta once X, sonra Y yap" |
| **Trial-and-error** | "Z denedim, W yuzunden patladi, V ile calisti" |
| **Domain facts** | "Library A'nin method B'si default'u C degil D" |

Trivial olani at (1 satirlik obvious seyleri kaydetme).

### 4. Safety guard

Her bulguyu yazmadan once kontrol et — asagidakilerden biri varsa **reddet**:

- `rm -rf`, `sudo rm`, `dd if=` gibi destructive komut ornekleri
- `curl ... | bash`, `wget ... | sh` gibi remote-exec patterns
- API key, token, password, secret deger (env var adi OK, deger DEGIL)
- Musteri/kisisel veri (email, telefon, kimlik)
- Baska kullanici/agent isimleri + kisisel yorumlar

Reddet = kaydetme + kullaniciya tek satir bildir: `[GUARD] rejected N entries (reason: secret|destructive|pii).`

### 5. Additive patch

`~/Projects/claude-config/agents/<agent>/knowledge/` altinda:

- Uygun dosya varsa (ornek `patterns.md`, `lessons.md`, `domain.md`) → dosyanin sonuna **ekle**, ustune yazma
- Yoksa yeni dosya olustur: `knowledge/<kategori>.md`
- Her yeni entry'e tarih + gorev kisa ozeti prefix'i: `<!-- 2026-04-23 — task: {ozet} -->`

### 6. Index guncelle

`knowledge/_index.md` dosyasini ac:
- Yeni dosya eklendiyse → index'e 1 satir ekle
- Mevcut dosyaya entry eklendiyse → (opsiyonel) entry count guncelle
- 200 satir limitini koru — asarsa kullaniciya `agent-refine-knowledge` oner

### 7. Rapor

```
## Knowledge Auto-Update — <agent-id>

- Patterns: +N entries → patterns.md
- Lessons: +M entries → lessons.md
- Domain: +K entries → domain.md
- Rejected by guard: R (reasons: ...)
- Index: updated

Next: install.sh gerekirse calistir (runtime mirror icin).
```

## Kurallar

- Max 15 tool call toplam
- ASLA global skill/agent/plugin olusturma
- ASLA baska agent'in klasorune yazma
- Safety guard **opt-out edilemez**
- Dil: entries Turkce+Ingilizce karisik OK (mevcut knowledge stiline uy)
- Duplicate kontrolu yap — ayni pattern zaten varsa ekleme, skip

## When NOT to Use
- Trivial gorev (<10 tool call, cok net yol) — ogrenilecek yeni birsey yok
- Sadece okuma yapan gorev (investigation) — patterns disinda kayit sey yok
- Hatali/yarida kalan gorev — yanlis dersler yazmaktansa atla

## Red Flags
- Ayni entry iki iterasyonda tekrar ekleniyor → duplicate check calismiyor, dur
- Guard > actual saves oranı yuksek → prompt secret/dangerous icerikle kirli
- Knowledge dosyasi 500 satiri astı → `agent-refine-knowledge` oner, ekleme yapma

## Error Handling
- Agent klasoru yok → dur, `agent-setup` oner
- Write izni yok → dur, path'i kullaniciya goster
- Guard tetiklendi → entry'yi sessizce at, final raporda say

## Verification
- [ ] Sadece hedef agent'in knowledge/ altina yazildi
- [ ] Safety guard calisti, reject sayisi raporlandi
- [ ] _index.md guncel
- [ ] Global scope'a dokunulmadi (skill/plugin/agent yaratilmadi)
- [ ] Duplicate kontrol yapildi

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.