v1 · design partnersoc · console
managed agent + pledge
§ faq

The questions sales gets repeatedly.

Most of these come up in the first 30 minutes of an enterprise-eval call. If yours isn't here, it should be — drop a note via /contact and we'll add it.

§ protocol
What's the relationship between OrangeCheck Console and the open protocols?

The protocols (oc-agent, oc-pledge, etc.) are public goods. They are MIT-licensed, fully specified, with reference implementations on npm. Console is the company that operates a managed deployment of those protocols and sells the operational tier (signing service, OC Stamp anchor pipeline, Nostr relay redundancy, audit-bundle assembly, operator dashboard) to enterprises.

If we disappear tomorrow, every receipt we ever issued continues to verify offline against Bitcoin headers using @orangecheck/agent-core. We sell operations, not lock-in.

Is this a fork of OAuth?

No. OAuth solves a different problem (federated user-to-app authorization with an opaque scope string). OC Agent is designed for autonomous principals: BIP-322-bound delegation, structured scope grammar enforced pre-action, content-addressed action receipts, deterministic time-ordering of scope changes against revocations via Bitcoin block height. The two can coexist — see /migrate for an incremental adoption path.

Why Bitcoin? Why not Ethereum / Solana / a private blockchain?

Authority and reputation must survive the issuer disappearing. Bitcoin is the strongest "the issuer cannot disappear" guarantee humanity has built. It is the public clock all the protocols anchor to (via OpenTimestamps, surfaced through OC Stamp). No on-chain transaction is required to use OC Agent — the protocols are off-chain envelopes anchored to Bitcoin block headers, so cost and throughput are not constraints.

What is "agent-action envelope (kind 30084)" actually?

A canonical-JSON object: BIP-322 signature by the agent's Bitcoin address over a domain-separated message, plus a content hash of the tool call, plus a reference to the active delegation, plus an optional OpenTimestamps proof. Published on Nostr kind 30084 with d-tag oc-agent-act:<id>. The shape is fully specified in oc-agent-protocol; you can recompute the envelope id from the canonical bytes on any laptop.

§ product
What's the difference between Console and just running the protocol myself?

Self-host gives you the cryptography. Console gives you the operations: the OC Stamp anchor pipeline (3 OpenTimestamps calendars, redundancy on calendar failure, anchor SLA), 4 Nostr relays for publication redundancy, audit-bundle assembly + signed export, operator dashboard, integration adapters, support, SOC 2 attestation (in progress). You can self-host any time — Enterprise tier customers can run the managed components on their own infrastructure with our help.

Is the dashboard mock data right now?

v1.1 of Console (the current public release) renders the dashboard surface against in-memory mock data. Real backend persistence lands with the v1.2 design-partner cohort. Every shape the dashboard renders matches the eventual API; the wallet-signing flows for delegations + scope edits + revocations also wire up in v1.2. See /roadmap for the in-flight items.

Can I see a live envelope without signing up?

Yes — /demo generates a real OC Agent delegation envelope in your browser using @orangecheck/agent-core (the same package an MCP server uses to verify). Edit any field; the canonical message changes; the SHA-256 envelope id changes. No fake hashes; no server-side magic.

§ security
Do you hold customer keys?

No. Wallet signing happens entirely in the user's wallet (UniSat, Xverse, Leather, OKX, Phantom, Alby, or paste-manual). Console never sees private keys, ever. See /security for the full data-flow diagram and what we structurally cannot have.

Do you custody funds?

No, ever. Bonded reputation under OC Pledge is attestation-of-unspent — sats stay in the customer's wallet; enforcement of broken pledges is by public exposure, never by slashing. We do not operate a custodial wallet of any kind. This is committed publicly on /charter.

What's your SOC 2 status?

SOC 2 Type 1 in progress, targeted before general-availability launch. Type 2 follows Type 1 by approximately 12 months of observation. Enterprise tier customers receive the report and continuous-compliance attestation. See /trust.

How do I report a vulnerability?

Email security@ochk.io directly. Don't open public issues for exploitable bugs. Acknowledged within 48 hours; verifier-impacting issues get priority over everything else. Full disclosure policy on /security.

§ pricing
Why both sats and dollars on the pricing page?

Lightning is a first-class rail (BTCPay Server, no custodian — instant settlement). Stripe is the dollar fallback for teams whose finance ops aren't ready for Bitcoin payments yet. Both prices are shown on the same row at parity because the strategic signal matters: this is infrastructure that closes its economic loop on Bitcoin, not a SaaS that bolts on crypto for marketing. We do not surcharge either path.

Do you have a free tier?

Yes — Community is free forever for solo developers and OSS agent projects evaluating the protocol. 3 agents, 5,000 actions / month, OC Stamp anchored audit (with watermark on exports). No credit card. See /pricing.

What if I exceed my action quota?

Pro overage is billed at the same per-action rate as the included tier — no exponential punishment for usage spikes. We surface live usage in /billing with sat and USD totals side-by-side. Enterprise tier is unlimited.

§ integration
Which AI agent stacks does Console support today?

MCP shipped (the @orangecheck/agent-mcp adapter is on npm; runnable examples in oc-agent-examples). Anthropic Tool Use, OpenAI Responses + Chat Completions, Vercel AI SDK, and LangGraph adapters are published at v0.0.1 (in-design — locked canonicalization, production wiring lands with the design-partner cohort). See /integrations.

My team is on a stack you don't list. What now?

Adapter prioritization is design-partner-driven. If a named partner asks for a stack, it moves from "planned" to "shipping" inside a quarter — that's how Vercel AI SDK and the others got onto the roadmap. Tell us your stack.

How long does a typical integration take?

For the MCP adapter: ~30 minutes for the basic install + first signed envelope, plus a half-day pairing session with an OrangeCheck engineer to wire it into your existing MCP server's auth path (design-partner cohort gets this for free). For other adapters: similar shape, longer pairing if the adapter is still pre-production.

Yours not here?

Send it to contact@ochk.io or use the form. Patterns we see twice land on this page within a week.