Summary: Obscura is worth cataloging not as just another copy-trading app or generic zk reputation pitch, but as a layered attempt to turn private trading history into portable, cryptographically verifiable reputation. The main mechanism is the separation between exchange-credential custody inside TEEs, trade-data ingestion, zero-knowledge performance proofs, follower execution plumbing, and a staged roadmap toward decentralized oracle and prover networks. That makes Obscura a useful comparison point for Nomis, Omnid, OpenRank, and other reputation systems on one side, and for social/copy-trading platforms on the other: it tries to make trading performance legible without fully exposing strategy or account data.
What it does:
Connects to trader accounts on centralized and decentralized exchanges and ingests trading activity for reputation reporting and copy-trading flows
Protects exchange credentials inside a dedicated confidential-compute layer (Citadel) rather than exposing raw keys to ordinary application services
Generates zero-knowledge proofs over aggregate trading performance through an Alchemist proving service so traders can publish proof-backed results without revealing individual trades
Stores or verifies proof-backed reports on Horizen L3, making reputation output more portable than platform-only dashboards
Exposes follower execution and white-label B2B copy-trading infrastructure, including pool creation, signal submission, subscriber risk modes, and multi-tenant API surfaces
Frames the current production stack as centralized bootstrap infrastructure while proposing later decentralization of trade ingestion, credential handling, and proof generation
Key claims:
The official GitHub profile and Medium announcement define the core pitch clearly: Obscura wants to decouple reputation from revelation by letting traders build a unified, verifiable track record across exchanges without exposing strategies, balances, or raw trade history.
The documentation README and privacy overview make the current stack decomposition explicit: Sentinel handles trade ingestion, Citadel handles credential custody, Alchemist handles ZK proving, and Horizen L3 anchors proof-backed reports on-chain. That separation is more analytically useful than filing Obscura as a generic trading app.
The current docs also surface an important implementation caveat: earlier public materials emphasize AWS Nitro Enclaves, while the later technical docs say Citadel currently runs on Nillion nilCC with AMD SEV-SNP and explicitly describe Nitro and Evervault as earlier migration stages. That difference matters because the present control surface is a hosted confidential-compute provider stack, not merely a timeless TEE abstraction.
The protocol-roadmap document is unusually explicit about current versus target architecture. Today, Sentinel is still a single FastAPI service and Alchemist is still a single prover with proofs covering batches of up to 10 trades. The future DON, MPC, and prover-market layers are roadmap claims rather than already-deployed decentralized infrastructure.
The ZK docs are concrete enough to clear the bar: they describe Noir circuits, BN254, Honk verification, per-symbol queue-based PnL accounting, report commitments, and specific Horizen testnet verifier / contract addresses. That gives the corpus a useful proof-of-performance comparison point beyond generic marketing copy.
The B2B integration guide shows Obscura is not only a reputation layer. It also exposes a white-label control plane for pools, follower subscriptions, signal queues, risk modes, and exchange execution. That means practical power may sit not only in proving and data ingestion, but also in pool administration, tenant isolation, queueing policy, and subscriber-risk defaults.
Obscura is therefore best read as a hybrid of reputation middleware and copy-trading execution infrastructure. The reusable insight is not simply private trading with ZK, but the full split between data source authority, credential-custody trust, proof generation, on-chain reputation anchoring, and downstream social-trading distribution.
Whitepaper: Obscura surfaces both a docs-hosted whitepaper summary and a linked full PDF whitepaper. The strongest reviewed primary materials for this pass are collected in ../whitepapers/obscura-primary-sources-2026-05-13.md.