Event: Anthropic acquired Stainless 2026-05-18 per https://www.anthropic.com/news. Confirmed via Rich-observation 2026-05-22T~11:55 BST that https://app.stainless.com/signup displays “stainless is no longer accepting new sign-ups as it is joining anthropic”.
Why: Stainless’s product surface (SDKs in 9 stable languages + MCP server generation + Apache 2.0 published SDKs + contract testing) is exactly the technology category Anthropic needs to deepen its agent-facing platform play. SDK gen + MCP server is the bridge between Anthropic’s models and developer ecosystems; owning that bridge in-house lets Anthropic guide the developer experience end-to-end.
How to apply (load-bearing for future TT strategic-decision sessions):
1. Stainless paths are closed for IK.
- Cannot create new Stainless accounts (app.stainless.com/signup blocked)
opensource@stainless.comFOSS path effectively dead during acquisition-transition- Gmail draft
r-6430428868353364735to opensource@stainless.com: DELETE-MANUALLY-BY-RICH - Spike-Y (Stainless empirical validation) PERMANENTLY OUT-OF-SCOPE per A-275
2. Commercial market for Arazzo-aware SDK gen + MCP has contracted to ONE vendor (Speakeasy).
- Fern eliminated (Docs $150/mo vs SDKs Free-50-endpoint-cap-or-Enterprise-custom; no Arazzo support in SDK product)
- APIMatic eliminated ($300/mo per language; no Arazzo)
- Stainless absorbed into Anthropic
- Speakeasy ~$720/mo per language is now the de facto sole commercial choice
- Vendor-concentration risk is now a NEW factor in Q15 lock analysis (alongside cost)
3. Free-stack-first rationale strengthens on TWO axes.
- Cost-saving (£0 vs ~£11.4-13.7K/yr Phase-1 + ~£28.5-34.1K/yr Phase-1.5+) — already established
- Vendor-concentration-risk-mitigation (open-source-stack independence vs sole-remaining-commercial-vendor dependence) — NEW
- Spike-X-prime is now the SOLE alternative-evaluation track (per A-275)
4. Strategic-positioning two-sided implication for build-and-sell-to-Anthropic.
Anthropic just acquired SDK-gen + MCP-server-gen technology. This is direct context for project-strategic-direction-locked-build-and-sell-2026-05-20 which was locked 2 days AFTER the Stainless acquisition. The memory’s body may or may not have implicitly factored this in; surface it explicitly as load-bearing context regardless.
Side A — Complementary (positions IK as acquisition-candidate-with-unique-value):
- IK’s inheritance-specific domain expertise + canonical-primitives + Catala formal-verification + faith-tradition rule-bodies + 23 SKOS classifier schemes + 22 A-21 CI gates are NOT in Stainless’s scope
- Stainless gave Anthropic the SDK-gen + MCP generation tech; IK could give Anthropic the vertical-domain content (inheritance/probate/trusts/succession) that runs on top of that infrastructure
- “Anthropic now has the SDK plumbing; IK provides the inheritance-domain canonical schema and Catala-verified rule bodies that consume it” is a coherent pitch
Side B — Competitive/redundant (risks “we built this internally already” rejection):
- IK’s SDK + MCP + Arazzo layer partially duplicates what Anthropic-via-Stainless now has internally
- If Anthropic’s view is “we have SDK gen sorted in-house with Stainless; we need vertical-domain partners not technology partners”, IK’s SDK/MCP/Arazzo work is wrong-shaped for the pitch
- Defensible pivot: emphasise IK’s schema + rule-bodies + formal-verification + faith-tradition + 23 SKOS classifier work; de-emphasise the SDK/MCP infrastructure layer (or pitch IK’s vertical-domain SDK as a Stainless-internal customer-shaped use case)
Rich-decision required at next strategic checkpoint (NOT now): which side of the two-sided implication is dominant? May affect:
- BUILD-PLAN emphasis (SDK polish vs schema-and-rule-body breadth)
- Pitch deck framing for any Anthropic conversation
- Q15 lock long-term (free-stack independence reinforces the schema-and-rule-body-as-the-product positioning; Speakeasy commercial-dependence weakens it)
5. Timing observation worth noting.
Acquisition was announced 2026-05-18 (Monday). Rich’s strategic-direction lock was 2026-05-20 (Wednesday). The build-and-sell-to-Anthropic memory was authored 2 days AFTER the Stainless acquisition. Whether Rich was tracking the news + factored it in, or made the strategic decision independently and the timing is coincidental, is unclear from the memory body alone. Future-Claude reading this memory together with the build-and-sell memory has the information to make the connection.
Provenance:
- Rich-report 2026-05-22T~11:55 BST: app.stainless.com/signup observation
- WebFetch https://www.anthropic.com/news: 2026-05-18 “Anthropic acquires Stainless” news item (canonical source)
- A-275 substrate-correction amendment in inherit-v2-architecture-state.md v4.67 §15 (companion arch-state lock)
- IK arch-state v1.10 §SDK Languages footnote update (Stainless CLOSED-NEVER-FIRED)
- Ledger task-164 (v2.09)
Cross-link memories:
- project-strategic-direction-locked-build-and-sell-2026-05-20 — parent strategic-direction memory; SDK-gen-as-strategic-for-Anthropic signal load-bearing for this analysis
- feedback_research_artefact_forward_traceability — discipline reason this memory exists at all (durable per-topic project memory routing)
- feedback_check_t_files_first_for_any_inherit_v2_work — T9-OPENAPI-ARAZZO-notes.md substrate the Stainless ecosystem entry came from
Speculation: future Anthropic-API availability of Stainless tech
Added 2026-05-22T12:15 BST per Rich-prediction “i imagine their functionality will become available within anthropic api”.
Three plausible shapes Stainless tech could land in at Anthropic, in order of how fast they typically arrive:
| Shape | Likelihood | Timing | IK impact |
|---|---|---|---|
| Internal tooling — Anthropic uses Stainless tech to regenerate Anthropic’s own SDKs (Python / TS / Go / Java / Ruby) at higher quality and cadence | High — basically guaranteed | Already underway; visible 3-6 months as Anthropic SDK quality lifts | Indirect — Anthropic SDKs get nicer; IK neither gains nor loses |
| Console feature — “Generate SDK for your API” in console.anthropic.com (upload OpenAPI → download idiomatic SDKs + possibly MCP server) | Medium-high | 6-12 months typical for acquired tech → console feature; could be faster if it’s a re-skin of Stainless’s existing app | Direct Q15 alternative. Cheaper Speakeasy-equivalent emerges inside Anthropic’s pricing model; likely bundled with Claude API spend, possibly free at low tiers |
| Agent-platform integration — Arazzo workflow → MCP tool conversion + contract testing baked into Claude Agent SDK + Claude Code skill ecosystem | Speculative | 12-24 months | Major Q15 reframe. If Anthropic makes “Arazzo-gated MCP” a first-class platform primitive, that’s exactly Q15’s specification — but free/bundled |
Watch-triggers for future-Claude re-evaluating this:
- Anthropic blog / news for “SDK generation” / “Stainless integration” announcements (https://www.anthropic.com/news)
- console.anthropic.com adds new sections or workbench features beyond eval/agent
- Claude API release notes mention SDK-gen primitives or Arazzo support
- New Anthropic Skills mention Stainless tooling (e.g.,
anthropic-sdk-gen-skill) - Spike-X-prime closure outcome (2 sprints from 2026-05-22) — natural re-evaluation point regardless
When to re-act:
- If Spike-X-prime closes VALIDATED → Stainless-via-Anthropic becomes a “nice to have, evaluate later” option, not a blocker
- If Spike-X-prime closes with ≥3 FAIL → Speakeasy is default fallback, BUT this watch-item becomes load-bearing — re-evaluate Q15 once Anthropic ships their version
- If Anthropic ships SDK-gen in Console + Arazzo support is included → file substrate-correction amendment (likely A-280-ish range) to re-evaluate Q15 against the new option
Practical baseline: assume no public Anthropic SDK-gen feature is available before BUILD-PLAN M1 Sprint 1 (M7-M8 calendar). Spike-X-prime outcome drives Q15 disposition; Anthropic-Stainless integration is a future-watch item, not an immediate decision factor.