> ## Documentation Index
> Fetch the complete documentation index at: https://docs.walletwall.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog

> Documentation and security-hardening updates to WalletWall's public docs, in date order.

<Note>
  This changelog covers changes to WalletWall's **public documentation site** and the
  static guards that protect it — not application releases, not scoring or provider
  changes, and not production milestones. Each entry states its category explicitly.
</Note>

<Update label="July 25, 2026" description="PR #1847 · Documentation">
  **Navigation icons completed; emoji removed from published pages.**

  30 of the 36 pages in the sidebar had no icon, and 7 had no frontmatter at all —
  rendering a path-derived label instead of their real title. Every navigation page now
  declares a verified Font Awesome icon name. Along the way, a pre-existing broken icon
  (`git-fork`, a Lucide name absent from this deployment's Font Awesome set) was fixed to
  `code-fork`.

  The one emoji on a published page was replaced with plain text. Internal,
  unpublished audit records that use check-mark and cross-mark emoji as
  load-bearing table values were left untouched — that is data, not decoration.

  Added a static guard (`test/docs-navigation-icons-and-emoji.test.mjs`) so a future
  page can never ship without a verified icon, and no published page can reintroduce
  emoji.
</Update>

<Update label="July 25, 2026" description="PR #1846 · Security hardening">
  **Published `.mdx` pages brought into the static security scan.**

  The repository's static-analysis lint walked every `.md` file under `docs/` for
  injection sinks (`dangerouslySetInnerHTML`, `eval`, raw HTML injection) and committed
  secrets — but silently skipped every `.mdx` file, even though Mintlify publishes
  `.mdx` pages as live pages that can contain executable JSX.

  The scan now covers `.mdx` too, with a rule split that keeps it accurate: sink rules
  scan only the *executable* body (fenced code samples and inline code spans are
  display-only and never run), while credential rules always scan the full raw text —
  a committed secret is a leak whether or not it happens to sit inside a code fence.

  This is a documentation-tooling change. No application code, provider, or scoring
  behavior changed.
</Update>

<Update label="July 25, 2026" description="PR #1841 · Documentation">
  **Three interactive explainer components added to the docs.**

  The Wallet Evidence Model, Quantum Intelligence, and PQ Evidence & Rehearsal pages
  each gained a small, deterministic, client-only React component readers can use to
  explore the underlying contract directly:

  * **Evidence State Explorer** — shows why an unavailable reading is never rendered as
    a confident `$0`.
  * **Quantum Exposure Calculator** — mirrors the production `SCORE_WEIGHTS` scoring
    curve, pinned by test against the canonical scoring modules so it cannot silently
    drift from what the application actually computes.
  * **Proof Gate Matrix** — demonstrates that PQ evidence, proof-artifact status,
    testnet rehearsal, and production readiness are independent gates, and that no
    combination of states can produce a "quantum-secure" claim or imply the platform
    is ready for production.

  Every component is read-only: no network call, no wallet address, no wallet
  connection, and no signing. The existing static prose, tables, and diagrams these
  components illustrate were kept in place — the components are supplemental, not a
  replacement.
</Update>