DOCS/PROJECT
REFERENCE9 MIN READUPDATED 03 SEP 2026

PROJECT / 04

Direct answers before you connect anything.

Privacy, hosting, AI models, agents, wallets, payments, tokens, receipts, and production-readiness answers.
Start here when a claim or boundary is unclear. If an answer depends on deployment-specific evidence, the documentation says so rather than treating a plan as a live capability.
PRIVACY

What does SHREDDAR protect?

01

DOES THE PUBLIC PLAYGROUND UPLOAD MY FILE?

The current proof reads supported files in browser memory and performs deterministic transformations in the tab. Use synthetic or authorized data and inspect the outbound panel.

02

IS PSEUDONYMIZED DATA ANONYMOUS?

No. Protected context can remain linkable or personal data when combined with outside knowledge. SHREDDAR reduces exposure; it does not guarantee anonymity.

03

CAN SHREDDAR MISS SENSITIVE DATA?

Yes. The alpha detector can produce false negatives and false positives. Evaluate it against an authorized corpus for the actual workflow before production use.

04

DO RECEIPTS CONTAIN THE PROMPT?

No by design. They contain policy, counts, route, timing, reduction, and integrity fields—not source content or Vault mappings.

DEPLOYMENT + MODELS

Where does processing happen?

01

IS THERE A HOSTED PRIVATE-DATA API?

Not for anonymous use. Operate the gateway inside the environment that owns the source. The public deployment is an authenticated operator proof.

02

DOES SHREDDAR CALL AN AI MODEL?

The browser proof and default demo mode do not. A self-hosted operator can configure a non-streaming OpenAI-compatible upstream, which receives protected messages only.

03

ARE PRESIDIO, SPACY, OR ONNX LIVE?

No. They are researched contextual-extension paths. No production weights have been selected, benchmarked, or shipped.

04

IS IT PRODUCTION-READY?

No universal production claim is made. Each deployment needs quality, security, legal, data-governance, backup, recovery, and incident-response review.

AGENTS + INTEGRATIONS

How do agents use it safely?

01

WHAT DOES MCP EXPOSE?

Exactly analyze, protect, shard, get_receipt, and get_usage. Reconstruction and Vault credentials are intentionally absent from model-visible tools.

02

WHAT DOES THE AGENT SKILL ADD?

It makes authorization, purpose binding, fail-closed routing, output inspection, and truthful receipt reporting part of the agent's operating procedure.

03

WHICH HOSTS HAVE GUIDES?

OpenClaw, Hermes Agent, OpenAI Codex, Claude Code, generic MCP hosts, and OpenAI-compatible clients.

04

IS VIRTUALS ACP LIVE?

No. The offering blueprint exists, but registration, identity, signer, pricing, escrow, sandbox lifecycle, and recovery are held.

WALLETS + PAYMENT

Does SHREDDAR require a wallet or $SHRED?

01

IS A WALLET REQUIRED?

No. The website is walletless and core product paths do not require a wallet. App connection is optional account context and is not payment authorization.

02

HOW ARE MULTIPLE WALLETS HANDLED?

When the SHREDDAR Reown project is configured, selecting Connect wallet opens AppKit's EVM Connect view, where the user chooses an installed or WalletConnect-compatible wallet. The adjacent person icon opens only the separate SHREDDAR usage profile.

03

CAN I BUY OR HOLD $SHRED?

No verified contract exists. Any address presented before the official launch record should be treated as unrelated or fraudulent.

04

HOW IS USE PRICED?

The hypothesis is strict PAYG measured by input-context tokens, so larger workloads cost more at a posted rate. No rate is published and no cost is charged today.

05

IS X402 ACTIVE?

No. The status endpoint returns enabled false until rate, asset, chain, facilitator, recipient, recovery, legal, and receipt evidence exists.

SHREDDAR DOCSYOUR DOC. YOUR DATA. YOUR CONTROL.

Public alpha documentation. Capability labels describe implementation state, not certification or guaranteed suitability.