Vai al contenuto
TakoAPI
Torna alle skill

didit-face-search

di rosasalberto

Search & Research
175Download riportati da clawskills.sh
0Visualizzazioni
0 likeAccedi per mettere like a questa skill
—Non ancora valutato
La tua valutazione

Sintesi

Integrate Didit Face Search standalone API to perform 1:N facial search against all previously verified sessions.

Descrizione

Cosa fa questa skill

Performs 1:N facial search by comparing a submitted photo against all previously approved verification sessions in a Didit application. Returns ranked matches with similarity scores from 0-100. Blocklist matches trigger automatic decline.

Unlike 1:1 face comparison, it searches the entire approved session database in one call, so no prior session ID is needed to detect duplicates or blocklist hits.

Quando usarla

  • Blocking duplicate account registrations during onboarding
  • Checking a new user's face against a fraud blocklist
  • Flagging users who attempt to re-register under a different identity
  • Reviewing borderline similarity matches for manual investigation
  • Running facial deduplication as part of a multi-step verification pipeline

Esempio di workflow

INPUT

User asks: Check if this selfie already exists in our verified user database

AGENT
  1. 1

    Validates the image meets format and size requirements (JPEG/PNG/WebP/TIFF, max 5MB)

  2. 2

    POSTs the image to https://verification.didit.me/v3/face-search/ with the DIDIT_API_KEY header

  3. 3

    Parses the response status: Approved (no matches), In Review (possible duplicates), or Declined (blocklist hit)

  4. 4

    Extracts similarity_percentage and vendor_data from each match to identify overlapping accounts

  5. 5

    Flags matches above 90% as likely duplicates and those in the 70-89% range for manual review

OUTPUT

List of matching sessions with similarity scores, partial user details, and a recommendation on whether to allow or block the registration

Requisiti

  • DIDIT_API_KEY environment variable
  • Didit account with API credits (business.didit.me)

Installazione

clawhub install rosasalberto/didit-face-search

Esegui il comando CLI oppure incolla il prompt in Claude Code, Cursor, Codex o nell'agente di coding che usi.

Link