From 2173da727cabdfcef7e97405ccdd384161f810e9 Mon Sep 17 00:00:00 2001
From: FilipBl4gojevic <84703395+FilipBl4gojevic@users.noreply.github.com>
Date: Thu, 30 Jul 2026 18:55:38 +0200
Subject: [PATCH] docs: correct the audit-#10 rollout claim, testnet-first
 connect example, close a stray vocabulary residue

---
 README.md                                                    | 5 +++--
 docs/architecture/non-custodial-split.md                     | 2 +-
 .../test/integration/self-improvement-keyless.test.ts        | 2 +-
 3 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index 8b35f1c..4678eaf 100644
--- a/README.md
+++ b/README.md
@@ -23,10 +23,11 @@ Hosted instances run on both networks, exposing all 24 tools:
 > custodial: assume they may still accept a mnemonic as a tool-call parameter, and prefer
 > self-hosting this release or the testnet instance instead.
 
-Connect from Claude Code in one command:
+Connect from Claude Code in one command (testnet shown - swap in the mainnet URL once the
+cutover deploy described above ships):
 
 ```bash
-claude mcp add --transport sse vector-mcp https://mcp.vector.mainnet.apexfusion.org/sse
+claude mcp add --transport sse vector-mcp https://mcp.vector.testnet.apexfusion.org/sse
 ```
 
 > **Security notice: as of this release, the non-custodial migration is complete in this
diff --git a/docs/architecture/non-custodial-split.md b/docs/architecture/non-custodial-split.md
index 3fbd4f7..1c5dced 100644
--- a/docs/architecture/non-custodial-split.md
+++ b/docs/architecture/non-custodial-split.md
@@ -190,7 +190,7 @@ TDD throughout: failing test before implementation, every PR.
 | **5** | Builder hardening: auth (enforced when configured), per-identity rate limits, `submit_transaction`, `await_transaction` | no |
 | **6** | **Family 1 — wallet/tx keyless.** Mnemonics removed; `build_*` tools land ✅ **(landed — this PR)** | **yes** |
 | **7** | **Family 2 — agent registry keyless.** Drops `@ts-nocheck` from `agent-network.ts` ✅ **(landed — this PR)** | yes |
-| **8** | **Family 3 — self-improvement keyless.** Plus env-driven ref-UTxO config and a CBOR encoding fix ✅ **(landed — this PR)** | yes |
+| **8** | **Family 3 — self-improvement keyless.** Ref-UTxO env config carried over from the custodial code; the suspected CBOR encoding issue closed as no-bug (audit #10, golden-pinned against a live on-chain token) ✅ **(landed — this PR)** | yes |
 | **9** | Mainnet cutover: README, security-page correction, deliberate `workflow_dispatch` | — |
 
 **PR 4 is the review that matters.** It is additive, self-contained, and *is* the security boundary. It breaks nothing while it is reviewed.
diff --git a/packages/builder/test/integration/self-improvement-keyless.test.ts b/packages/builder/test/integration/self-improvement-keyless.test.ts
index e024653..751d3da 100644
--- a/packages/builder/test/integration/self-improvement-keyless.test.ts
+++ b/packages/builder/test/integration/self-improvement-keyless.test.ts
@@ -9,7 +9,7 @@ import {
 } from '../../src/vector/gov-build.ts';
 
 // Tier-1 self-improvement integration: LIVE testnet, ZERO secrets. Build-only -
-// nothing is signed, nothing is submitted, no funds move. critique, endorse,
+// nothing is signed, nothing is submitted, no value moves. critique, endorse,
 // and proposal_lock are all plain locks (a single unsigned payment to a
 // script address, no script spend), so this tier can fully exercise their
 // datum-construction path against real chain data (a live-discovered agent