> ## 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. # Dune vault readiness scope # Dune Scope & Cost Audit — Stablecoin Vault Readiness **Date:** 2026-06-23 **Branch:** `audit/dune-vault-readiness-scope` **Method:** static analysis only — derived from `api/*.js` route handlers, `src/lib/**`, `dune/**`, `.env.example`, the existing Dune audits, and the guardrail code/tests. **No Dune queries were executed, refreshed, scheduled, exported, written, or test-run to produce this audit. No runtime Dune behavior was changed.** This audit re-anchors every Dune query reference against the **stablecoin vault readiness platform** direction established by the alignment series (PRs #1233–#1244). The existing [dune-query-inventory.md](./dune-query-inventory.md) (#1128) and [dune-product-dependency-map.md](./dune-product-dependency-map.md) (#1130) were last audited **2026-06-20** — *before* that alignment work. This document does **not** duplicate them; it adds a third lens on top of them: > For each Dune query, **how relevant is it to stablecoin vault readiness**, and what is the > recommended **keep / reduce / review / deprecate** action given that direction and the > RED credit-budget posture? See also: * [dune-query-inventory.md](./dune-query-inventory.md) — authoritative per-env-var inventory (status, query IDs, files) * [dune-product-dependency-map.md](./dune-product-dependency-map.md) — per-surface runtime dependency + degradation * [../operations/dune-query-guardrails.md](../operations/dune-query-guardrails.md) — read-only/cache-only operating contract * [../operations/dune-agent-safety.md](../operations/dune-agent-safety.md) — agent/CLI execution lockdown * [../operations/dune-budget-baseline.md](../operations/dune-budget-baseline.md) — 🔴 RED credit-budget baseline (#1131) * [../operations/dune-schedule-triage.md](../operations/dune-schedule-triage.md) — owner-side Dune-UI cadence triage * [api-cost-control-map.md](./api-cost-control-map.md) — per-provider credit-risk map *** ## 1. Executive summary * **The vault-readiness alignment added ZERO new Dune spend paths.** The new shared model, adapter, packet builder, Migration Readiness wiring, and state refinement (`src/lib/stablecoin-vault-readiness-model.js`, `buildStablecoinVaultReadiness.js`, `buildVaultReadinessPacket.js`, `migration-readiness.js`) are **pure functions** — each documents "no Dune query execution, no I/O." They consume Dune-derived facts that already flow through the existing cached read paths (`/api/quantum-readiness`, `/api/quantum-exposure` → `api/_quantum-facts.js`). No new query IDs, no new routes, no execution. * **Guardrails hold and are unchanged.** Dune is read-only by default; `executeAndPoll` is fail-closed behind a four-factor gate; reads and writes use separate keys; `dune:verify` * four guardrail test files + the CLI pre-flight guard enforce the contract; CI and tests hard-block execution. Verified by reading the code and running the static + mocked checks (0 credits) — see §7. * **The direct vault-readiness signal set is mostly cheap.** The queries that *directly* feed vault/wallet readiness (migration readiness, dormant quantum exposure, counterparty context, adversarial heuristics, value-at-risk) are weekly or scheduled snapshots. The expensive surfaces are the **discovery/analytics** queries (48h active wallets, whale trades, 12-week activity, and the generic Coinstellation/Insights trio), several of which are only *indirectly* aligned or **not aligned at all** with the vault direction. * **Highest-leverage cost action is unchanged from #1131 triage:** pause stale Dune-UI schedules and slow the 6h/12h cadences. This audit adds product justification: the generic analytics queries are now **deprecate candidates** under the vault-readiness direction, and the stale/mis-named slots should be resolved. Classification totals (by logical query; see §4): | Classification | Count | | ------------------------------------------------------------------- | -----------------------------: | | **Keep** — critical for stablecoin vault readiness | 9 | | **Keep but cache/reduce** — useful but cost-sensitive | 7 | | **Review** — unclear or partially aligned | 6 | | **Deprecate candidate** — stale or generic-analytics direction | 4 | | **Guardrail-only / reference** — must not execute / already removed | 4 removed + execution-gate set | *** ## 2. Current Dune posture | Dimension | State | | ------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Execution mode (app)** | Read-only `GET /query/:id/results` (0 credits). No `/execute` on any request path. | | **Execution gate** | `executeAndPoll` fail-closed behind `assertDuneExecutionAllowed` (4 factors: `ALLOW_DUNE_EXECUTION` + `DUNE_EXECUTION_ACK` + `DUNE_WRITE_API_KEY` + not CI/test). | | **Key separation** | Reads: `DUNE_READONLY_API_KEY` → `DUNE_API_KEY` (fallback). Execution: `DUNE_WRITE_API_KEY` only. A read key can never authorize a write. | | **Caching** | Redis (`dune:v1::default`) + in-process fallback; in-flight read coalescing; TTLs 6h/12h/24h as named constants asserted in tests. | | **Budget** | 🔴 **RED** per [dune-budget-baseline.md](../operations/dune-budget-baseline.md): \~11,064 / 4,000 credits at mid-cycle (≈2.77×), projecting ≈5.2× month-end. Spend originates from **Dune-UI scheduled executions**, not the app. | | **Vault path** | `/vault` (`StablecoinVaultPage.jsx`) is mock/rehearsal-only — **no Dune reads on the vault route itself**. Vault *readiness* is computed from quantum/migration facts read elsewhere. | *** ## 3. Query inventory (vault-readiness lens) One row per **logical query** (by Dune query ID). Env-var-level detail (aliases, exact files, `.env` vs `.env.example`) lives in [dune-query-inventory.md](./dune-query-inventory.md); this table adds the **product surface (new architecture)**, **vault-readiness relevance**, **cost/scan risk**, and **classification**. Relevance legend: **Direct** = feeds the wallet/vault readiness, migration, or evidence computation. **Indirect** = supports a discovery/context surface that funnels into readiness or frames the stablecoin product. **None** = generic analytics with no vault-readiness role. | Query ID | Primary env var | Product surface (new) | Vault relevance | Cost / scan risk | Cadence | Classification | | ------------------------------------- | ------------------------------------------------------------------------------------ | ------------------------------------------------------- | -------------------------------------------- | --------------------------------------------------------- | ----------------- | ------------------------------------------ | | 7605290 | `DUNE_QUERY_WALLET_SIGNATURE_EXPOSURE` | Wallet Readiness (PQ signature) | **Direct** | Med (per-wallet snapshot, 6h TTL — aggressive) | scheduled / 6h | Keep but cache/reduce | | 7605304 | `DUNE_QUERY_QUANTUM_VALUE_AT_RISK` (alias `…_VALUE_AT_RISK`) | Wallet Readiness (value-at-risk) | **Direct** | Low–Med | scheduled | Keep | | 7605315 | `DUNE_QUERY_WALLET_MIGRATION_READINESS` | Migration Readiness | **Direct** | Low (weekly) | weekly | Keep | | 7605194 | `DUNE_QUERY_DORMANT_QUANTUM_EXPOSURE` | Wallet Readiness + Holder Wall (shared) | **Direct** | Med (global snapshot) | weekly | Keep | | 7634380 | `DUNE_QUERY_QUANTUM_COUNTERPARTY_CONTEXT` | Readiness Packet / evidence | **Direct** | Low–Med | scheduled | Keep | | 7634433 | `DUNE_QUERY_QUANTUM_ADVERSARIAL_HEURISTICS` | Readiness Packet / evidence | **Direct** | Low–Med | scheduled | Keep | | 7724327 | `DUNE_QUERY_DORMANT_QUANTUM_CANDIDATES` | Holder Wall → vault-candidate discovery | **Direct** (heuristic proxy) | Med (broad receiver scan) | scheduled | Keep | | 7641256 | `DUNE_QUERY_STABLE_PEG_HISTORY` | Stablecoin Intelligence | **Indirect** (vault asset context) | Low (daily) | daily | Keep | | 7641531 | `DUNE_QUERY_STABLE_VOLUME_TREND` | Stablecoin Intelligence | **Indirect** | Low–Med | scheduled | Keep | | 7641537 | `DUNE_QUERY_STABLE_WHALE_FLOWS` | Stablecoin Intelligence / treasury | **Indirect** | Med (network flow scan) | scheduled | Keep | | 7521767 | `DUNE_QUERY_48H_ACTIVE_WALLETS` (alias `DUNE_QUERY_ID`) | Holder Wall / Wallet Activity | **Indirect** | **High** (broad scan, 6h, multi-surface, fallback-backed) | scheduled / 6h | Keep but cache/reduce | | 7560801 | `DUNE_QUERY_WHALE_TRADES` | Holder Wall + Coinstellation (shared) | **Indirect** (Holder Wall) / None (Insights) | **High** (network-wide ≥\$100k scan, multi-surface) | scheduled | Keep but cache/reduce | | 7583106 | `DUNE_QUERY_12WK_ACTIVE_WALLETS` | Whale Watcher / treasury monitoring | **Indirect** | **High** (12-week window scan) | 24h | Keep but cache/reduce | | 7641114 | `DUNE_QUERY_WALLET_PORTFOLIO` | Wallet Readiness (holdings → VaR context) | **Indirect** | Med (per-wallet) | scheduled | Keep but cache/reduce | | 7641495 | `DUNE_QUERY_WALLET_ACTIVITY_BREAKDOWN` | Wallet Profile | **Indirect** | Med (per-wallet, 12h) | 12h | Keep but cache/reduce | | 7641515 | `DUNE_QUERY_WALLET_PROTOCOL_AFFINITY` | Wallet Profile | **Indirect** | Med (per-wallet, 12h) | 12h | Keep but cache/reduce | | 7746246 | `DUNE_QUERY_WALLET_CHAIN_AFFINITY` (mis-named `DUNE_QUERY_CHAIN_AFFINITY` in `.env`) | Wallet Readiness (Quantum Exposure Clock bridge driver) | **Direct** (intended) | Low–Med (weekly, no route reads it yet) | weekly (intended) | Review | | 7673583 | `DUNE_QUERY_STABLECOIN_MARKET_DAILY` | Stablecoin Intelligence (reserved) | **Indirect** (potential) | Unknown — **no route reads it** | n/a | Review | | 7641521 | `DUNE_QUERY_RECENT_NOTABLE_WALLETS` | Leaderboards | **Indirect** (generic discovery) | Low–Med | scheduled | Review | | (empty) | `DUNE_QUERY_DEX_ID` | Wallet Profile (DEX enrichment, optional) | None | Low (optional, unset) | scheduled | Review | | 7560798 | `DUNE_QUERY_PROTOCOL_VOLUME` | Coinstellation / Insights | **None** (generic DeFi analytics) | Med | scheduled | Deprecate candidate | | 7560799 | `DUNE_QUERY_TOKEN_FLOWS` | Coinstellation / Insights | **None** | Med | scheduled | Deprecate candidate | | 7560800 | `DUNE_QUERY_CHAIN_ACTIVITY` | Coinstellation / Insights | **None** | Med | scheduled | Deprecate candidate | | (empty) | `DUNE_QUERY_MULTICHAIN_QUANTUM_EXPOSURE` | (stale placeholder, no route) | None | n/a (empty) | n/a | Deprecate candidate | | 7658207 / 7658215 / 7658220 / 7658222 | `DUNE_QUERY_SOLANA_STABLECOIN_*` (4) | Stable Seer (Solana — retired) | None | n/a (removed) | n/a | Guardrail-only / reference (removed #1130) | | — | `DUNE_WRITE_API_KEY` + `ALLOW_DUNE_EXECUTION` + `DUNE_EXECUTION_ACK` | Execution gate (admin one-off only) | n/a | **Credit-spending if opened** | n/a | Guardrail-only / reference | > **Aliases / config drift** (carried from the inventory; counted under their canonical query): > `DUNE_QUERY_ID` → 7521767, `DUNE_QUERY_VALUE_AT_RISK` → 7605304, > `DUNE_QUERY_CHAIN_AFFINITY` (mis-named) → 7746246. These are **Review** items (retire / > rename), not independent queries. *** ## 4. Classification summary | Classification | Queries | | ------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Keep** (9) | Signature exposure¹, value-at-risk, migration readiness, dormant quantum exposure, counterparty context, adversarial heuristics, dormant quantum candidates, stable peg history, stable volume trend, stable whale flows — *(10 listed; signature exposure is grouped under Keep-but-reduce, see ¹)* | | **Keep but cache/reduce** (7) | `WALLET_SIGNATURE_EXPOSURE` (6h→12–24h), `48H_ACTIVE_WALLETS` (6h→reduce/pause; BigQuery primary), `WHALE_TRADES` (shared, ensure single cache), `12WK_ACTIVE_WALLETS` (large window), `WALLET_PORTFOLIO`, `WALLET_ACTIVITY_BREAKDOWN` (12h→24h), `WALLET_PROTOCOL_AFFINITY` (12h→24h) | | **Review** (6) | `WALLET_CHAIN_AFFINITY` (name mismatch + no route), `STABLECOIN_MARKET_DAILY` (reserved, no route), `RECENT_NOTABLE_WALLETS` (generic), `DEX_ID` (optional/empty), `DUNE_QUERY_ID` (alias), `DUNE_QUERY_VALUE_AT_RISK` (alias) | | **Deprecate candidate** (4) | `PROTOCOL_VOLUME`, `TOKEN_FLOWS`, `CHAIN_ACTIVITY` (generic Coinstellation/Insights analytics, not vault-aligned), `MULTICHAIN_QUANTUM_EXPOSURE` (empty stale placeholder) | | **Guardrail-only / reference** | 4× retired Solana stablecoin vars (removed #1130); the execution-gate env set (`DUNE_WRITE_API_KEY`, `ALLOW_DUNE_EXECUTION`, `DUNE_EXECUTION_ACK`) which must never be set in app/dev/CI | > ¹ **Stable totals:** the **Keep** bucket is 9 (the three Stable Seer queries + value-at-risk, > migration readiness, dormant exposure, counterparty context, adversarial heuristics, dormant > candidates). `WALLET_SIGNATURE_EXPOSURE` is **Direct/critical** but its *recommended action* > is cadence reduction, so it is counted under **Keep but cache/reduce** to avoid double-count. > This reconciles with the inventory's 18 **Active** env vars + 2 aliases + 1 candidate-removal > > * 3 stale + the reserved market-daily slot. *** ## 5. Mapping to the new product architecture | Product pillar | Direct Dune queries | Indirect Dune queries | Notes | | ----------------------------------------- | ---------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------ | | **Stablecoin Intelligence** (Stable Seer) | — | `STABLE_PEG_HISTORY`, `STABLE_VOLUME_TREND`, `STABLE_WHALE_FLOWS`, *(reserved `STABLECOIN_MARKET_DAILY`)* | Frames the vault's stablecoin asset context; keep, cheap-ish. | | **Wallet Readiness** | `WALLET_SIGNATURE_EXPOSURE`, `QUANTUM_VALUE_AT_RISK`, `DORMANT_QUANTUM_EXPOSURE`, *(intended `WALLET_CHAIN_AFFINITY`)* | `WALLET_PORTFOLIO` | Core PQ/exposure signals feeding `buildStablecoinVaultReadiness`. | | **Migration Readiness** | `WALLET_MIGRATION_READINESS` | — | Weekly; powers `/api/quantum-readiness` + MigrationPathPanel adapter (#1242). | | **Readiness Packet / evidence** | `QUANTUM_COUNTERPARTY_CONTEXT`, `QUANTUM_ADVERSARIAL_HEURISTICS` | — | Source-backed evidence facts; consumed by `buildVaultReadinessPacket` (#1237). | | **Holder Wall / concentration discovery** | `DORMANT_QUANTUM_CANDIDATES` (vault-candidate proxy) | `48H_ACTIVE_WALLETS`, `WHALE_TRADES` | Discovery funnel into readiness; highest-cost surface. | | **Whale Watcher / treasury monitoring** | — | `12WK_ACTIVE_WALLETS` | Large-window scan; monitor cadence. | | **Legacy / generic analytics** | — | — (None-relevance) | `PROTOCOL_VOLUME`, `TOKEN_FLOWS`, `CHAIN_ACTIVITY`, `RECENT_NOTABLE_WALLETS`, `DEX_ID` — not part of the vault-readiness path. | **How readiness consumes Dune (no execution):** `/api/quantum-readiness` + `/api/quantum-exposure` → `api/_quantum-facts.js` (`getOrCache` / `readOrCache`, params ignored for execution, rows filtered by wallet) → `src/lib/quantum-vault-readiness.js` → **pure** `buildStablecoinVaultReadiness` / `buildVaultReadinessPacket` / `migration-readiness`. The pure layer never touches Dune; it only shapes already-read snapshot facts. *** ## 6. Cost / risk notes **Highest-risk queries (cost / scan breadth / blast radius):** 1. **`DUNE_QUERY_48H_ACTIVE_WALLETS` (7521767)** — 6h cadence, broad network activity scan, **multi-surface** (Holder Wall + Wallet Activity), and **fallback-backed** (BigQuery is the primary source). Reducing or pausing its Dune schedule is the single highest-leverage, lowest-risk spend cut. 2. **`DUNE_QUERY_WHALE_TRADES` (7560801)** — network-wide ≥\$100k transfer scan, **shared** across Holder Wall (aligned) and Coinstellation/Insights (not aligned). One cached snapshot serves both; if Insights is deprecated, the Holder Wall consumer alone justifies it. 3. **`DUNE_QUERY_12WK_ACTIVE_WALLETS` (7583106)** — 12-week window is an inherently large scan; already 24h cadence, but confirm it is not a top consumer. 4. **`DUNE_QUERY_WALLET_SIGNATURE_EXPOSURE` (7605290)** — core *direct* readiness signal but on an aggressive **6h** cadence; signature exposure changes slowly, so 12–24h is safe. 5. **Generic Insights trio (`PROTOCOL_VOLUME` / `TOKEN_FLOWS` / `CHAIN_ACTIVITY`)** — scheduled credit spend for a surface with **no vault-readiness role**. Prime deprecate/pause candidates. **Multi-surface / shared-snapshot queries** (one cached read serves N surfaces — good; but a schedule change affects all consumers): `WHALE_TRADES` (Holder Wall + Insights), `DORMANT_QUANTUM_EXPOSURE` (Wallet Readiness + Holder Wall), `48H_ACTIVE_WALLETS` (Holder Wall + Wallet Activity). **Queries that should require cached snapshots (and already do):** every active query. All reads go through `readOrCache` / `getOrCache` / `readLatestResults`; per-wallet routes share one snapshot per query ID and filter rows locally. No per-render reads; no per-wallet cache fan-out. **Accidental-execution exposure (agents/tests/CI):** **None reachable.** `executeAndPoll` is the only credit-spending path and is fail-closed behind the four-factor gate; `NODE_ENV=test` and CI env detection hard-block it regardless of other vars; `dune:verify` blocks `/execute` and mutating CLI verbs outside `api/_dune.js`; tests are fully mocked. The historical spike came from **human/agent Dune-UI or CLI executions**, now contained — see [dune-agent-safety.md](../operations/dune-agent-safety.md). **Queries that must never execute during CI:** all of them — CI never needs execution. Enforced by `isAutomatedEnv()` in the guard + the `dune:verify` CI-workflow scan. *** ## 7. Guardrails verified Verified by reading the code and running the **read-only / mocked** checks (0 credits — no query execution). Results in §"Tests run" of the PR. | # | Guardrail | Evidence | Status | | -- | -------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------ | | 1 | Public routes are read-only (`GET /results`, 0 credits) | `api/_dune.js` `readLatestResults`/`readOrCache`/`getOrCache`; `_dune-feature-route.js`, `_quantum-facts.js` use only read funnels | ✅ | | 2 | `executeAndPoll` fail-closed behind 4-factor gate | `api/_dune.js:169` calls `assertDuneExecutionAllowed`; `api/_dune-execution-guard.js` requires `ALLOW_DUNE_EXECUTION` + `DUNE_EXECUTION_ACK` + `DUNE_WRITE_API_KEY` + not CI/test | ✅ | | 3 | Read/write key separation | reads use `DUNE_READONLY_API_KEY`/`DUNE_API_KEY`; execution uses `DUNE_WRITE_API_KEY` only; read key can never authorize a write | ✅ | | 4 | No `/execute` or write path reachable from public routes | `scripts/verify-dune-usage.mjs` checks 1–2, 5b(A–D); only `api/_dune.js` may reference `executeAndPoll`/`/execute` | ✅ | | 5 | CLI mutating verbs denylisted | `api/_dune-execution-guard.js` `DENYLISTED_DUNE_COMMANDS`; `scripts/dune-guard.mjs` (`npm run dune:guard`) | ✅ | | 6 | CI/tests cannot execute | `isAutomatedEnv()` blocks `CI`/`GITHUB_ACTIONS`/`VERCEL`/`NODE_ENV=test`/`VITEST`; `dune:verify` fails CI workflows that set gate vars | ✅ | | 7 | No React/client-side Dune access | `verify-dune-usage.mjs` §3 + `dune-usage-guardrails.test.mjs` §6 block `_dune.js` imports and `api.dune.com` in `src/**` | ✅ | | 8 | New alignment libs add no execution path | `buildStablecoinVaultReadiness.js`, `buildVaultReadinessPacket.js`, `stablecoin-vault-readiness-model.js`, `migration-readiness.js` are PURE (documented + no `_dune` import); `quantum-vault-readiness-display.js` *redacts* query IDs from display | ✅ | | 9 | API key is read-only / documented as read-only | `.env.example` comments: "Dune READ key … read-only/metadata calls (0 credits) … NEVER put a write/execution-capable key in either of these"; write key documented as human-only, never in Vercel/CI | ✅ | | 10 | Agents warned not to write/execute | `docs/operations/dune-agent-safety.md` "hard rule"; guardrails doc anti-patterns | ✅ | **No write-capable Dune path exists in the app.** The only write/execute path (`executeAndPoll`) is gated, key-separated, and unreachable from routes, tests, CI, and agents. *** ## 8. Recommended actions **Repo-side (safe, no Dune execution; future PRs):** 1. **Retire confirmed aliases** once primary names are set in all deployed envs: `DUNE_QUERY_ID` → `DUNE_QUERY_48H_ACTIVE_WALLETS`; `DUNE_QUERY_VALUE_AT_RISK` → `DUNE_QUERY_QUANTUM_VALUE_AT_RISK`. 2. **Resolve chain-affinity config drift** — local `.env` uses the mis-named `DUNE_QUERY_CHAIN_AFFINITY`; code/`.env.example`/guardrails use `DUNE_QUERY_WALLET_CHAIN_AFFINITY`. Rename in `.env` (no code change) **or** decide whether to wire the lib into a route or drop it. 3. **Decide the generic Insights trio** (`PROTOCOL_VOLUME` / `TOKEN_FLOWS` / `CHAIN_ACTIVITY`): if Coinstellation/Insights is not part of the vault-readiness product, pause their Dune-UI schedules and mark the env vars deprecated. **No blind deletion** — confirm Coinstellation's product status first. 4. **Decide `STABLECOIN_MARKET_DAILY` and `MULTICHAIN_QUANTUM_EXPOSURE`:** wire into a surface or deprecate; both currently have no route. 5. **Declare `DUNE_QUERY_DEX_ID` intentionally optional** in docs so operators don't treat the empty value as missing config. **Owner-side (Dune UI / billing — outside this repo, per #1131):** 6. Execute the [dune-schedule-triage.md](../operations/dune-schedule-triage.md) checklist: pause §4a stale schedules, drop the two 6h cadences (`48H_ACTIVE_WALLETS`, `WALLET_SIGNATURE_EXPOSURE`) to 12–24h, move 12h wallet-profile enrichments to 24h, then re-baseline via the 0-credit Usage API. 7. Get **per-query credit attribution** (Dune billing/usage export) to confirm the top consumers match the High-risk list in §6. *** ## 9. Follow-up issues / PRs * **Dune query reduction PR** (repo): retire aliases (#1), resolve chain-affinity drift (#2), mark the generic Insights trio + stale placeholders deprecated (#3–#4) — docs/env/test only, no runtime behavior change. * **Owner Dune-UI cadence triage** (no repo change): execute schedule-triage §4–§7; flip the launch-checklist "Dune usage stays within budget plan" P1 once projection ≤ budget. * **Holder Wall tile signal polish** (product): now that `DORMANT_QUANTUM_CANDIDATES` is the vault-candidate discovery signal, tighten the tile signal mapping to the readiness vocabulary — separate, product-side follow-up. *** ## 10. Confirmation * **No runtime Dune behavior was changed by this audit** — docs + a structural guardrail test only. * **No Dune queries were executed, refreshed, scheduled, exported, written, or test-run.** * Every **Direct/Indirect** entry is backed by a confirmed runtime/cache reference in an `api/*.js` route or `src/lib/*` module; nothing was marked aligned by inference. * **Review / Deprecate candidate** entries are flagged for an owner decision, **not** auto-deleted. * Cross-checked against `npm run dune:verify`, the four Dune guardrail test files, and the existing inventory + dependency map (whose per-env-var status counts remain authoritative).