Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add khalilbenaz/claude-skills-collection --skill face-matching-kycgit clone --depth 1 https://github.com/khalilbenaz/claude-skills-collectionWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/khalilbenaz/claude-skills-collection/face-matching-kyc)<a href="https://agentmods.dev/skills/khalilbenaz/claude-skills-collection/face-matching-kyc"><img src="https://agentmods.dev/badge/skills/khalilbenaz/claude-skills-collection/face-matching-kyc/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/khalilbenaz/claude-skills-collection/face-matching-kyc"><img src="https://agentmods.dev/badge/skills/khalilbenaz/claude-skills-collection/face-matching-kyc.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00113 | $0.02801 |
| Opus 5 | $0.00056 | $0.01401 |
| Sonnet 5 | $0.00023 | $0.00560 |
| Haiku 4.5 | $0.00011 | $0.00280 |
Grade A, and why
face-matching-kyc scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 258 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Face Matching KYC
Workflow KYC — étapes numérotées
1. Détection et alignement de visage
Librairies recommandées (par priorité) :
| Contexte | Librairie | Points repères |
|---|---|---|
| Mobile Android/iOS (offline) | MediaPipe FaceMesh | 468 points |
| Qualité maximale (backend) | RetinaFace | 5 points |
| Équilibre vitesse/qualité | MTCNN | 5 points |
Transform affine sur 5 points (yeux G/D, nez, coins bouche) → normalisation 112×112 px.
# Python – alignement avec insightface
from insightface.app import FaceAnalysis
app = FaceAnalysis(allowed_modules=["detection", "recognition"])
app.prepare(ctx_id=0, det_size=(640, 640))
def align_and_embed(img_bgr):
faces = app.get(img_bgr)
if not faces:
raise ValueError("Aucun visage détecté")
return faces[0].normed_embedding # float32 array 512-D, L2-normalisé
Critères de rejet image avant traitement :
- Résolution < 224×224 px →
IMAGE_QUALITY_LOW - Confiance détection < 0.90 →
NO_FACE_DETECTED - Plus d'un visage détecté →
MULTIPLE_FACES - Blur score (Laplacien) < 100 →
BLURRY_IMAGE
2. Extraction d'embedding facial
Modèle par contexte :
| Modèle | Embedding | FAR @ threshold 0.75 | Cas d'usage |
|---|---|---|---|
| ArcFace R100 | 512-D | ~0.05% | KYC bancaire, onboarding critique |
| MobileFaceNet | 128-D | ~0.3% | Mobile, contrainte latence |
| FaceNet InceptionResNetV1 | 128-D | ~0.1% | Backend général |
Règle : toujours L2-normaliser l'embedding avant le calcul de similarité.
import numpy as np
def cosine_similarity(emb1: np.ndarray, emb2: np.ndarray) -> float:
# Les embeddings insightface/ArcFace sont déjà L2-normalisés
return float(np.dot(emb1, emb2)) # dot product = cosine similarity si normalisés
3. Seuil de décision — similarity score
Score cosinus Décision Cas d'usage
───────────── ────────────────── ─────────────────────────────────
< 0.65 REJECT Blocage systématique
0.65 – 0.75 REVIEW File de revue manuelle obligatoire
0.75 – 0.85 PASS (medium) Login biométrique, flux secondaires
≥ 0.85 PASS (high) Ouverture compte, transaction critique
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 258 lines · 0 tokens per session scan A a2fd44f89d9b
face-matching-kyc is a skill published in the GitHub repository khalilbenaz/claude-skills-collection (22 stars, last pushed 18d ago), licensed MIT. It adds 113 tokens to every session and 2,801 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
hipaa-validate
HIPAA validator: PHI exposure, audit logging, encryption, access control, BAA refs. Triggers: HIPAA, PHI, healthcare compliance, audit log, BAA.
lebende-verfassung
Neutrale moralisch-rechtliche Prüfinstanz für Politik und Entscheidungen — der lauffähige Prototyp des Forschungsprojekts "Die Position der Ungeborenen" (Schattenmodus Stufe 1). Nutze diesen Skill, wann immer eine politische Entscheidung, ein Gesetz(entwurf), eine Reform, ein Haushaltsbeschluss oder eine…
full-after-care
Tiefe Pflegerunde für ein veröffentlichtes GitHub-Repository (Stufe 2): enthält den vollständigen surface-after-care-Durchlauf und ergänzt ihn um drei teure Schritte — rechtliche Ersteinschätzung über die Law-Checker mit Wiedervorlage nach einem Jahr (Gutachten bleibt gitignored im Repo), Querverweise zu verwandten…
repo-publish-check
Nutzerneutrale Prüfung von Repositories vor einer Veröffentlichung oder bei einer erneuten öffentlichen Prüfung. Kontrolliert Privacy, Geheimnisse, Lizenzen, Drittinhalte, Dokumentation und Freigabestatus, ohne die Veröffentlichung selbst vorzunehmen.
law-checker
Points to the standalone module law-checker ("Legal Department"): source-grounded AI first-look legal assessments for German law with a statute registry and a statute-embodiment agent. Use this skill when a situation, contract, official notice, or legal question under German law should be checked with exact citations…
cold-email-outreach
Cold email and outreach sequence design with personalization, deliverability optimization, follow-up cadence, and compliance (CAN-SPAM/GDPR). Use when the user asks about cold email, outreach sequences, email prospecting, cold outreach, or email deliverability for outbound.