Spokes.wiki Search About

Log — Business Messaging Wiki

Append-only history. Entries: ## [YYYY-MM-DD] <op> | <title> — <op> ∈ {split, ingest, query, lint}.

[2026-07-01] split | business-messaging-wiki created from _inbox (3 sources)

Spun out of the hub messaging-apis _inbox cluster once it reached 3 coherent, no-other-home sources, all WhatsApp-platform infrastructure:

Founding corpus is 100% WhatsApp; domain framed as business/conversational messaging platforms & their APIs with WhatsApp as the deepest corner, broadenable to other channels. Founding Thing pages: whatsapp-business-platform (the platform), whatsapp-business-account (WABA), message-template, business-solution-provider (BSP roles), meta (owner/publisher).

Adjacency, not re-filed: the routed whatsapp-chatgpt-bot (a GPT-4o WhatsApp bot — an agent product, stays in agentic-tooling) and wassenger (a WhatsApp BSP/API vendor) ride on this platform; cross-linked both ways. The two parked _inbox records were deleted.

[2026-07-01] ingest | Wapi.go (WhatsApp Business API Go SDK)

See wapi-go. Go SDK abstracting the WhatsApp Business (Cloud) API — inbound messages/status/webhooks, account/template/phone-number management, multi-number two-way messaging, no hand-built payloads. By WapiKit. T2 (first-party SDK docs; license/maintainer detail thin on the intro page). Touched whatsapp-business-platform (client-SDK-over-platform), meta (upstream API owner). Cross-linked the agentic-tooling bot as the consumer of such SDKs.

[2026-07-01] ingest | WhatsApp Business Platform / WABA docs (Meta)

See whatsapp-business-platform-docs. Meta's first-party Cloud API reference centered on the whatsapp-business-account: auth, phone numbers, message types, templates, webhooks, catalogs/commerce, calling, pricing, partner tiers. T1 vendor docs. Established the hub node whatsapp-business-platform and meta (owner).

[2026-07-01] ingest | WhatsApp Solution Partner / BSP get-started (Meta)

See whatsapp-solution-partner. Meta's onboarding for the Solution-Partner role: embedded signup, permissions/access tokens, credit line, Tech-Provider vs Solution-Partner split. T1 vendor docs. Fleshed out business-solution-provider; cross-linked wassenger as a real-world BSP-class vendor.

[2026-07-30] ingest | four channel/pricing docs, batched — WhatsApp pricing, RCS ×2, Apple Messages

Research pass at the human's request ("expand business messaging by means of research in net") — the first sources this spoke has taken since it spun out on 2026-07-01, and the first not supplied by the human. Targets were chosen from this wiki's own open questions: the unsourced pricing model, and whether non-WhatsApp channels exist to justify the broad domain.

Sources (all T1, all first-party operator documentation):

  • whatsapp-pricing-docs — Meta. Per-message billing replaced conversation-based pricing on 1 July 2025; category decides the rate; free inside the customer-service-window and in the 72-hour free entry point; volume tiers on utility/authentication.
  • rcs-business-messaging-docs — Google. Agents, APIs, Developer Console, and the carrier's own administration role.
  • rcs-brand-verification — Google. The authorization email to a named individual at the brand, Google- vs carrier-managed launch, the verified checkmark.
  • apple-messages-for-business-docs — Apple. Brand/MSP/Apple roles, registration, the one-to-two week experience review, entry points, Apple Pay and pickers.

New pages (7): 4 sources + rcs-business-messaging · apple-messages-for-business · per-message-pricing · customer-service-window · sender-verification. Entity: apple (Corporation); google linked cross-wiki to ../llm-providers-wiki, not duplicated (1.00 index match). Budget: 1 of 8.

Updated: whatsapp-business-platform (pricing model, window, the three-channel frame) · message-template (category is now the billing unit) · business-solution-provider (the intermediary role across all three channels — optional on WhatsApp, mandatory on Apple) · synthesis · index.

Two of the spoke's three open questions are now answered. Pricing is documented. And the breadth question resolves toward business messaging across channels: with three operators in hand, the gates turn out not to be WhatsApp-specific — each channel polices identity, content, and the intermediary, differing only in where it enforces (WhatsApp per message, Apple up front, RCS at launch with carriers holding a second veto). CLAUDE.md's domain line and the hub registry were updated to match. The agentic-tooling seam question is untouched — no borderline source arrived.

Recorded weaknesses. The comparison is thin: one to two sources per non-WhatsApp channel against four for WhatsApp, so the symmetry in synthesis rests on a small corpus. No pricing is sourced for RCS or Apple, which leaves the "billing is policy" claim resting on one channel. Capabilities commonly attributed to RCS (carousels, typing indicators, carrier reach) are not in Google's own docs and were deliberately left out; "server-to-server REST" for Apple is likewise a vendor gloss the primary page does not state.

avoid-ai-writing run over the new prose.

[2026-08-05] ingest | Chatery WhatsApp

Routed from the hub (runner-up: none — ../agentic-tooling-wiki was considered and does not apply, since the project contains no model). chatery-whatsapp — an Express.js 5 WhatsApp API backend over baileys 7.x: multi-session, Socket.IO events, group admin, media, queued bulk send, webhooks, OpenAPI docs, Docker. MIT, 214★, T3.

This is the spoke's first source from outside the sanctioned path, and it is the most consequential ingest since the multi-channel turn. No WABA, no template approval, no BSP, nothing metered — it QR-pairs an ordinary WhatsApp account as a linked device and speaks the WhatsApp Web protocol.

Created baileys as a Thing page: the library is the actual subject here, and it is the second door into WhatsApp that every page in this wiki had implicitly assumed away. Added the comparison table (account / auth / business-initiated message / cost / identity gate / sanctioned) against the Cloud API path.

synthesis gained a section on the unofficial path and its first genuine contradiction: the founding thesis calls WhatsApp "a permissioned, metered channel, not an open socket," and this source reaches the same users through something close to an open socket. Both recorded — they describe different doors. The thesis narrowed rather than broke: the gates sell permission and durability, not access. whatsapp-business-platform now carries the qualifier so the platform page cannot be read as the whole story.

Also resolved the open question about the agentic-tooling seam, though on an easy case — no model means transport by construction. Noted as such rather than as a real test.

Entities: created fajri-rinaldi-chan (Person; entity-index returned no match ≥ threshold). WhiskeySockets recorded as prose on the baileys page, not noded — a maintainer collective with no other evidence in the corpus. Budget: 1 of 8.

Caveat carried on both new pages: the README documents capabilities and is silent on the consequences — Meta's terms prohibit unofficial clients, and no ban-rate, deliverability or enforcement data exists anywhere in this corpus.

[2026-08-22] ingest | A2P 10DLC (via quality cycle)

a2p-10dlc and a thin twilio entity page, from Twilio's A2P compliance documentation and its published US SMS rate card. T3 — a vendor describing a registration regime it participates in and selling access to the channel it prices. The mechanics and the rate card are first-party fact of the kind the ladder grades T3; the benefit claim ("lower message filtering and higher messaging throughput") is the interested party and the only published source at once, and is flagged on the page rather than carried as a finding.

Closes coverage edge 4, which had said this wiki was comparing per-message-pricing and rcs-business-messaging to a baseline it did not hold. It closed cleanly on first contact, which is now the corpus's usual result for a coverage edge.

It also contradicted the spine. The thesis has said "billing is policy" — WhatsApp charges by template category, frees customer-initiated traffic inside the customer-service-window, and discounts the categories Meta considers legitimate. US carrier SMS does none of it: $0.0083 per outbound segment plus a $0.0035–$0.005 carrier surcharge, flat, whatever the message says and whoever started the conversation. Volume buys a ceiling (~3,000 / ~6,000 / unlimited segments a day by brand tier), not a discount. Both channels are gated; Meta puts the policy in the price and the US carriers put it in the registration. Open question 1 is partly answered as a result, and the general form of the claim is withdrawn on per-message-pricing itself.

What survives and strengthens: the gate has the same shape in both ecosystems. Brand registration is sender-verification; Campaign registration — declared purpose plus opt-in, opt-out and help, approved before traffic flows — is message-template approval one level up, approving the use case rather than the text. Two ecosystems sharing no technology arrived at the same structure.

Third finding, smaller: the destination network is a party to the price. Verizon charges more inbound than outbound and the three carriers differ on the same message, where WhatsApp has one counterparty and one price. That is the structure open question 1 was guessing at for RCS, observed next door.

New edge 4a records the residue — nobody has published a delivery or filtering rate for registered against unregistered traffic, so what registration buys is still only the reseller's word. Coverage edge 5 moves to one-third started. Dedup: no page in any spoke mentioned Twilio or 10DLC.

[2026-09-03] lint | Edge 6 struck, and telegram-bot-api given an inbound link

Quality cycle, no new source. Two bookkeeping repairs on yesterday's ingest.

telegram-bot-api was written 2026-09-02 and left orphaned — no other wiki/ page linked to it, so it was reachable only from index.md. Fixed by writing its finding into sender-verification, where it belongs: the three-channel identity-gate table now carries a fourth row's worth of contrast, a channel with no gate at all. That is the more useful place for it than a "see also" line, because it turns the spoke's universal-gates claim into a claim about operator policy.

Growth edge 6 still said "needs: platform documentation, one page each" a day after one of those pages was written. Marked a quarter closed with the residue named (WeChat, LINE, KakaoTalk). This is the third instance this week of a spoke edge outliving the source that closed it — recorded in the hub's ## Outstanding.

[2026-09-04] ingest | LINE Messaging API + LINE Official Account plans (research pass, growth edges 1 + 6)

Hunted for growth edge 6 — the markets where a different app is the default — which telegram-bot-api had opened on 2026-09-02 and left asking for WeChat, LINE and KakaoTalk. Two T1 first-party pages taken: LY Corporation's developer overview and its Japanese rate card, plus the verified-account page beside it.

The edge's own guess was wrong. It expected the East Asian super-apps to be the restrictive pole. LINE does not gate sending at all — open an account, push messages, no review, and no 24-hour window either. What it gates is being found: an account is issued unverified, and an unverified account does not appear in LINE's in-app search. The review takes 5–10 business days, demands the business's official name (unchangeable afterwards), and returns the green badge, searchability, friend-add advertising, coupon placement and invoice billing.

That is a third gate design, and sender-verification now carries it: block the sending (WhatsApp, RCS, Apple), gate nothing (Telegram), or sell discovery (LINE). The spine's symmetry paragraph was rewritten to say the gates are an operator's policy choice, which two channels have now demonstrated from opposite ends.

It also half-closed edge 1 (pricing beyond WhatsApp) without touching either channel that edge names. LINE charges a monthly subscription with a message quota — ¥0/200, ¥5,000/5,000, ¥15,000/30,000, overage only on the top plan at ¥3 down to ¥2.6 — and prices nothing by message category. Three channels, three billing philosophies: policy (WhatsApp), carriage (US A2P SMS), capacity (LINE). "Billing is policy" now has two counterexamples and is firmly a WhatsApp property.

Paged: line-messaging-api (WebAPI, T1), line-official-account-plans (TechArticle, T1), ly-corporation (Organization, same-as Wikipedia + Wikidata Q86734122 + lycorp.co.jp). Updated sender-verification, per-message-pricing, synthesis.md (spine + edges 1 and 6) and index.md.

Integrity fix found while here: telegram-bot-api was ingested on 2026-09-02 and never added to index.md. Added, with line-messaging-api beside it.

Caveat on the numbers: the rate card is the Japanese one; Taiwan and Thailand price separately and are not covered. Both pages are live vendor pages, so both are freshness: volatile.

Still open on edge 6: WeChat and KakaoTalk, the two the "restrictive" description actually points at.

[2026-09-22] quality | research pass — WeChat and KakaoTalk (growth edge 6 closed)

Via the hub quality cycle's research pass (Most-wanted #40; spoke never hunted before). Two T1 pages from first-party developer docs: wechat-official-account-messaging and kakao-bizmessage. Finding: both are restrictive, but on content and consent rather than identity — folded into the thesis, sender-verification (new section) and customer-service-window (WeChat's 48-hour, five-message window). Gaps left on the pages rather than filled: WeChat registration eligibility, both platforms' prices.

[2026-10-07] quality | staleness re-verification — baileys and apple-messages-for-business

Both were past the 60-day volatile window. baileys: 11.3k stars (+700), 3.4k forks (+126), and a breaking 7.0.0 release mid-migration to a new doc site (baileys.wiki) — a second, library-side source of the instability the spoke already named on the protocol side. apple-messages-for-business: docs now Version 3.1 (2026-07-21); unchanged gate and approval flow, one new use case ("Fan experiences & subscriptions," Apple Music links in-conversation) and an "Apple Account (formerly Apple ID)" terminology update.

Related pages

Links to