The operating system behind the evaluation.
CheckDragonfly is an account-intelligence service designed to organize observable public X account evidence into a traceable deterministic evaluation, then provide a supplemental AI interpretation of that stored evidence.
1. Purpose
CheckDragonfly separates evidence from interpretation. The service retrieves public account and activity data available through the X API, applies the frozen AAE-1.0 deterministic methodology, records the evaluation, and presents the result through an instrument-style interface.
The service is not identity verification, a background check, KYC, or a guarantee that an account is genuine. Its output is an automated assessment of observable signals at a point in time.
2. How to use CheckDragonfly
Authentication establishes the requesting user session and entitlement.
Ordinary users evaluate their authenticated account. Authorized administrators may enter a different target account.
The current Privacy Policy, Terms of Service, and Disclaimer must be affirmatively approved before evaluation.
The interface presents the deterministic result, evidence dimensions, recent activity, the traceable record, and—where entitled—AI Analysis.
3. System architecture
Authentication, consent, server-side authorization, rate limiting, evaluation orchestration, credential handling, AI orchestration, and API responses.
Persistent evaluation records, evaluation-run audit data, public activity evidence, consent records, credentials, and AI-analysis records.
Source of the public account and activity evidence used by the evaluator.
Supplemental interpretation layer using stored evaluation evidence. It does not replace the deterministic evaluator.
4. AAE-1.0
AAE-1.0 is the authoritative deterministic methodology presented to users. It produces the production score, decision, confidence, evidence record, and credential eligibility from observable account signals.
The presentation layer may map the AAE-1.0 0–100 result onto a 300–850 display scale. That presentation mapping does not modify the underlying AAE-1.0 result.
5. AAE-1.1 shadow layer
AAE-1.1 is a separate internal shadow evaluation path. It is non-authoritative, does not replace AAE-1.0, is not the public user-facing score, and is not used to alter existing evaluation decisions or credential issuance.
The shadow layer exists for controlled observation and future methodology development without changing the frozen production authority of AAE-1.0.
6. Evidence model
An evaluation may retain the returned X account record, the most recent public post, seven-day activity aggregates, and the posts returned for the evaluation’s 30-day public activity corpus.
The interface separates account evidence, recent activity, deterministic signals, contextual evidence, and the traceable evaluation record rather than collapsing them into a single opaque display.
7. AI Analysis
Eligible users may receive an automated AI interpretation after evaluation. The AI layer reads the stored evaluation outcome, deterministic signal evidence, account facts, seven-day activity aggregates, and the stored 30-day public post corpus.
AI Analysis is supplemental. It does not recalculate, override, or replace the deterministic AAE-1.0 score, decision, confidence, or methodology. Generated analysis may be cached for authorized retrieval.
8. Credentials
When an evaluation satisfies the applicable approval threshold, CheckDragonfly may issue a one-time verification credential. Credentials are time-limited and single-use.
A credential demonstrates that CheckDragonfly performed and recorded an evaluation at a particular time. It is not a confirmation of the account holder’s real-world identity and does not guarantee the underlying evaluation is correct.
9. Administrator mode
Authorized administrators are identified server-side from the authenticated user record. Administrators may evaluate a target account other than the account used to sign in. Ordinary users remain restricted to their authenticated account.
The browser’s administrator state is only a user-interface reflection. Authorization is enforced on the server.
10. Design
CheckDragonfly is designed as a digital instrument rather than a conventional dashboard. The interface uses restrained typography, sparse neutral surfaces, analytical blue accents, explicit provenance cues, evidence-oriented sections, and motion that communicates state changes.
Evaluation progress, AI processing, and incoming result states are intentionally distinguishable from static information.
11. Mission
The mission of CheckDragonfly is to make account-intelligence workflows more inspectable: observable evidence should remain distinguishable from deterministic scoring, supplemental interpretation, and human judgment.
The product is intentionally bounded. It does not present automated output as certainty or claim knowledge beyond the available evidence.
12. Current disclosures
The current disclosure version is 2026-09-26-v1. See the Privacy Policy, Terms of Service, and Disclaimer for the operative disclosures governing use of the service.