Secure capability acquisition

Skill alidevai/chiron_agent/.claude/skills/secure-capability-acquisition

Eksik bir skill/MCP/tool'u guvenli sekilde edinme prosedurunun tam akisi. "Hemen bulup kur" yerine "hemen arastir, hemen sandbox'a al, hemen test et, risk uygunsa ekle; degilse onay iste veya reddet" ilkesini uygular. Yeni bir yetenek kurulmasi gerektiginde kullanilir.From its SKILL.md

Install

npx -y skills add alidevai/chiron_agent --skill secure-capability-acquisition

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

3 things to look at

24 days oldThe repository was created 24 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.6 KB, ~1.1k tokens by cl100k_base, as published. Nobody here has run it

Secure Capability Acquisition Pipeline

Platformun en kritik guvenlik akisidir. Internette bulunan HICBIR skill veya MCP dogrudan kurulmaz. Agent arastirmada ve sandbox testinde otonomdur; kalici kurulumda politika ve insan onayina tabidir.

Zorunlu akis (asla atlanmaz)

1. Ihtiyac dogrulama    -> gercekten yeni yetenek mi gerekli?
2. Kesif + puanlama     -> capability-manager subagent (kaynak oncelikli arama)
3. Staging + statik tarama -> python -m core stage ...   (kritik bulgu = otomatik red)
4. Bagimsiz guvenlik    -> security-gatekeeper subagent  (kor, dusmanca inceleme)
5. Sandbox eval         -> python -m core eval <id>       (olculebilir testler)
6. Politika karari      -> python -m core promote <id>
                            - allow            -> otomatik kurulur (yalnizca dusuk risk)
                            - require_approval -> onay paketi olusur -> INSAN
                            - deny             -> reddedilir
7. Kurulum sonrasi      -> registry'de izlenir, periyodik yeniden dogrulanir

Adim adim

1. Ihtiyac dogrulama

Once sor: Mevcut bir skill parametre/ornek eklenerek yeterli olur mu? Ihtiyac tek seferlik mi (o zaman kalici kurulum yerine tek seferlik sandbox kullanimi)? Yeni bagimlilik teknik borc yaratir mi? Cevap "yeni yetenek gerekli" ise devam et.

2. Kesif ve puanlama

capability-manager subagent'ini cagir. O, kaynak oncelik sirasiyla (resmi repo

dogrulanmis koleksiyon > buyuk kurum > akademik > topluluk listesi > genel arama kendin uret) aday bulur, 100 uzerinden puanlar ve en iyisini staging'e alir.

3. Staging + statik tarama

python -m core stage <indirilen-dizin> --id <ad> --risk <seviye> \
    --domains <alan1,alan2> --source <kaynak-url>

Tarama otomatik calisir. Kritik bulgu (prompt injection, exfiltration, pipe-to-shell, audit/policy kurcalama...) varsa aday ANINDA reddedilir ve registry'ye revoked yazilir.

4. Bagimsiz guvenlik incelemesi

security-gatekeeper subagent'ini cagir. Makine taramasi tek basina yeterli degildir; gatekeeper dosyalari insan gibi okur, izin manifesti cikarir, en kotu senaryoyu tanimlar. REJECT verirse dur.

5. Sandbox eval

python -m core eval <id>

Eval skoru >= 0.90 ve kritik hata = 0 olmali. Gecerse durum otomatik sandbox-validated olur.

6. Politika karari

python -m core promote <id>

allow: yalnizca dusuk riskli, ag erisimsiz, salt-okunur yetenekler. Otomatik .claude/skills/ altina kurulur.
require_approval: orta/yuksek risk. approvals/pending/<id>.md onay paketi olusur. Kullaniciya bu paketi ozetle ve python -m core approve <id> --by "AD" komutunu INSANIN interaktif terminalde calistirmasi gerektigini soyle. Sen (agent) bu komutu calistiramazsin; guard hook engeller.
deny: kritik risk. Reddedilir.

Kesin kurallar

Staging'deki hicbir sey aktif degildir; yalnizca .claude/skills/ altindakiler yuklenır.
Indirilen kod yalnizca sandbox-run / eval uzerinden calistirilir, dogrudan degil.
Yuksek riskli yetenegi ureten/bulan agent, kendi guvenlik onayini VEREMEZ.
Emin degilsen reddet (fail-closed).

Kaynaklar

Pipeline: core/lifecycle.py, core/scanner.py
Ilgili: [[capability-gap-analysis]], [[mcp-security-review]], [[skill-creator-safe]]

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most security skills give in ~1.1k tokens

Counted across 648 of the 828 authors here whose files we hold, read 2026-08-07

Parameterize all database queriesin 68 of 648, across 51 files
Hash passwords using bcrypt, scrypt, or argon2in 49 of 648, across 36 files
Apply rate limiting to authentication endpointsin 48 of 648, across 24 files
Configure security headersin 35 of 648, across 19 files
Validate all inputsin 32 of 648, across 24 files
Validate all external input at the system boundaryin 29 of 648, across 19 files
Run containers as a non-root userin 28 of 648, across 15 files
Use httponly secure samesite cookies for sessionsin 26 of 648, across 15 files
Run dependency audits before every releasein 21 of 648, across 10 files
Encode output to prevent cross-site scriptingin 21 of 648, across 11 files
Copy dependencies before source codein 20 of 648, across 9 files
Store secrets in environment variablesin 20 of 648, across 18 files

Said here and by no other author read

Verify if a new capability is truly required
Search for candidates via the capability-manager
Stage and statically scan the candidate
Run independent security review via security-gatekeeper
Run sandbox evaluation via core eval
Run promotion policy via core promote

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.