Skip to content

Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.

docs(ops): record that no credential is readable from the service config any more

nirholasce976b07ee3c
diff --git a/docs/ops/wallet-key-migration.md b/docs/ops/wallet-key-migration.md
index 10e29b53c..22d370560 100644
--- a/docs/ops/wallet-key-migration.md
+++ b/docs/ops/wallet-key-migration.md
@@ -170,6 +170,43 @@ Manager. It is now a `secretKeyRef`.
Read the current state at any time with `node scripts/migrate-plaintext-secrets.mjs --verify`.
+### The rest of the service went with it
+
+The master key was not the only credential sitting in the config. The same sweep moved
+**57 more**, in two updates (`three-ws-api-00406-nlg`, then `00407-m7c` for one that a
+killed earlier run had left as a version-less secret shell). The service now carries **81
+Secret Manager references and zero plaintext credentials**; `node
+scripts/migrate-plaintext-secrets.mjs --verify` reports `Verify: clean` and exits 0.
+
+Four vars were pointed at a secret that already existed rather than a fresh copy, because
+duplicating a live credential means a later rotation can update one copy and silently leave
+the other serving: `ECONOMY_MASTER_SECRET_BASE58` and `LAUNCHER_MASTER_SECRET_KEY_B64` and
+`X402_TREASURY_SECRET_BASE58` to their `wallet-*` secrets, and `DATABASE_URL` to the
+existing `DATABASE_URL` secret. The tool reports every other value collision it finds but
+never adopts one on its own: several of the matches are secrets belonging to OTHER services
+in this project (`agent-orders-*`, `sniper-*`), and pointing this service at one of those
+would couple two rotations that have no business being coupled.
+
+126 vars stayed as plaintext config, correctly: public identifiers (`ECONOMY_MASTER_ADDRESS`,
+`X402_FEE_PAYER_SOLANA`, every `SIGNER_*` threshold and `X402_*` cap), handles rather than
+values (every `*_KEY_ID`, which names a credential without being one), and 19 provider URLs. The URLs were the one
+judgment call worth making by measurement rather than by name: each value was checked for an
+embedded credential (a `?api-key=`-style parameter, or userinfo before the host) and only
+`DATABASE_URL` carried one, so only it moved. That check is what the tool applies now, so a
+future URL that does embed a key is classified as the credential it is.
+
+Two things the sweep surfaced that are NOT fixed here, because both need an owner decision:
+
+- `WALLET_ENCRYPTION_KEY` and `JWT_SECRET` hold the **same value** on production. The
+ custodial-wallet guard in `secret-box.js` requires a dedicated key and refuses the
+ `JWT_SECRET` fallback, and a var set to the same string passes that guard while defeating
+ its point. Changing it is a re-key, not a config edit: see the top of this file for what
+ happens when a wallet-encryption key changes without carrying the old one forward.
+- `X402_SEED_SOLANA_SECRET_BASE58` holds the same value as `LAUNCHER_MASTER_SECRET_KEY_B64`,
+ though `scripts/wire-master-wallet.mjs` assigns those two slots to different wallets. They
+ were given separate secrets rather than one shared secret so the drift can be corrected
+ later without a rotation touching both.
+
### The tool, and the Cloud Run gotcha it encodes
`scripts/migrate-plaintext-secrets.mjs` does the whole move: it classifies every env var on
diff --git a/prompts/finish/production-100-00-INDEX.md b/prompts/finish/production-100-00-INDEX.md
index 070f82465..6c72e0931 100644
--- a/prompts/finish/production-100-00-INDEX.md
+++ b/prompts/finish/production-100-00-INDEX.md
@@ -68,7 +68,6 @@ run in parallel, H waits on owner gates, I is anytime, J is the parallel swarm.
|---|---|
| [../backlog/01-x402-settle-runway.md](backlog-01-x402-settle-runway.md) | Code complete; outcome false: settle 5.9%, `cause: sponsor_floor`. The dry-run reclaim line it was waiting on is fixed (`afd349790`, pinned by tests). Capital is OWNER-ACTIONS row 2. |
| [../backlog/11-agent-index-lag.md](backlog-11-agent-index-lag.md) | New 2026-09-01: healthz reports `agent_index` down (1,092 of 1,602 Solana agents erroring, EVM cursors 3,039 hours stale) and nothing owned it. |
-| [03-master-key-hygiene.md](production-100-03-master-key-hygiene.md) | Open; the runbook section was never written and the cloud half needs a live `gcloud`. |
### C. Repo and product defects
diff --git a/prompts/finish/production-100-03-master-key-hygiene.md b/prompts/finish/production-100-03-master-key-hygiene.md
deleted file mode 100644
index 807bfb039..000000000
--- a/prompts/finish/production-100-03-master-key-hygiene.md
+++ /dev/null
@@ -1,90 +0,0 @@
-# P100-03: Get the master wallet secret out of plaintext env
-
-**How to run this:** paste this whole file into a fresh Claude Code chat opened in
-`/workspaces/three.ws`, or say "execute `prompts/finish/production-100-03-master-key-hygiene.md`".
-Read [00-INDEX.md](production-100-00-INDEX.md) and `CLAUDE.md` first.
-
-## Binding operating clause
-
-1. Finish 100%. Never end with a question or an unexecuted plan. Delete this file when the
- definition of done is verified; write follow-ups for any remainder first.
-2. Storage migration of an existing secret is config work and pre-approved (config-only
- service updates, per CLAUDE.md). **Rotation is not:** generating a new master wallet and
- moving its funds is stop-and-ask gate 1. This order migrates storage and prepares
- rotation; it never rotates.
-3. Env-var traps, both of which have burned real sessions: `--set-env-vars` REPLACES the
- whole env set (never use it); `--update-env-vars` and `--update-secrets` merge. And
- `vercel env pull` returns empty secrets; never trust it.
-
-## The problem, measured 2026-08-02 (re-measure in step 0)
-
-`ECONOMY_MASTER_SECRET_BASE58`, the base58 secret key of the economy master wallet, sits as
-a PLAINTEXT env var on the `three-ws-api` Cloud Run service, while its sibling
-`X402_FEE_PAYER_SECRET_BASE58` is a Secret Manager reference. Any principal with
-`run.services.get` on the project can read the master wallet's key from the service config.
-
-## Step 0: re-derive the state
-
-```bash
-export PATH="$HOME/google-cloud-sdk/bin:$PATH"
-gcloud run services describe three-ws-api --region us-central1 \
- --project aerial-vehicle-466722-p5 --format=yaml | grep -n -A2 "SECRET\|_KEY" | head -60
-gcloud secrets list --project aerial-vehicle-466722-p5
-curl -s https://three.ws/api/healthz | head -c 1500 # baseline before touching env
-```
-
-Confirm which vars are literal `value:` entries versus `valueFrom.secretKeyRef`. If the
-master secret is already a secretRef, skip to task 3 (the sweep), then task 4.
-
-## Tasks
-
-1. **Migrate the master secret.** Create a Secret Manager secret (name it in the style of
- the existing ones), add the current value as version 1, grant the runtime service account
- (`three-ws@...`) `roles/secretmanager.secretAccessor` on that one secret (per-secret
- grant, matching how the existing secrets are wired, never project-wide), then flip the
- service: `gcloud run services update three-ws-api --region us-central1 --project
- aerial-vehicle-466722-p5 --update-secrets ECONOMY_MASTER_SECRET_BASE58=<secret>:latest`.
- That single update both attaches the secret and removes the plaintext literal; verify in
- the post-update YAML that no literal value remains anywhere (including old revisions'
- traffic: confirm 100% of traffic is on the new revision).
-2. **Prove nothing broke.** The new revision must serve: re-read healthz and compare to the
- step 0 baseline, and confirm a signing path that uses the master key still works (the
- treasury tick or a settle, observed via `/api/healthz` counters or
- `/api/ops/payment-outcomes`, not assumed). If the service degrades, roll back traffic to
- the prior revision first, then diagnose.
-3. **Sweep for siblings.** From the step 0 YAML, list every env var whose name looks
- secret-bearing (`_SECRET`, `_KEY`, `_TOKEN`, `_PRIVATE`) and is a plaintext literal.
- Migrate each the same way in one batch update. Judgment call per CLAUDE.md: non-secret
- config that merely matches the pattern (public keys, ids) stays put; say so in the report.
-4. **Document and hand the rotation to the owner.** Extend `docs/ops/wallet-key-migration.md`
- (it already holds the encryption-key rotation runbook) with a short section: where the
- master secret now lives, how to add a new version, and the owner-gated rotation plan
- (generate new wallet, move funds, retire old version) with exact commands. Add one row to
- [OWNER-ACTIONS.md](production-100-OWNER-ACTIONS.md): rotate-or-accept decision, since the old value was
- readable in plaintext for some window.
-5. No changelog entry: security-internal, nothing user-visible. Log in
- [PROGRESS.md](production-100-PROGRESS.md).
-
-## Definition of done
-
-- [ ] Post-update `gcloud run services describe` YAML shows the master secret (and every
- migrated sibling) as a secretRef; no plaintext secret literals remain on the serving
- revision, and 100% of traffic is on it.
-- [ ] Healthz and a real master-key signing path verified no worse than baseline.
-- [ ] Runbook section written; `npm run audit:docs` clean.
-- [ ] Rotation row in [OWNER-ACTIONS.md](production-100-OWNER-ACTIONS.md); outcome in
- [PROGRESS.md](production-100-PROGRESS.md); this file deleted.
-
-## Never blocked
-
-| Blocker | Resolution |
-|---|---|
-| `gcloud` missing / auth dead | PATH export above; revive with `gcloud auth login --no-launch-browser` through a fifo; verify with a real read. |
-| Fear of breaking settles mid-flight | The flip creates a new revision; the old one keeps serving until ready. Baseline first, compare after, roll back traffic on regression. |
-| A var is ambiguous (secret or config?) | Trace where the code reads it (`grep -rn "<NAME>" api/ server/`). Migrate real credentials; leave identifiers; record each call in the report. |
-| Secret Manager quota or IAM denial | The build/runtime SAs and per-secret grant pattern already exist on this project; copy an existing secret's binding. If genuinely denied, file the fix, migrate what you can, and follow-up-file the rest. |
-
-## Report format
-
-Table of migrated vars (name, secret name, revision), the verification reads (healthz delta,
-signing proof), siblings deliberately left with reasons, and the one-line owner ask.
diff --git a/prompts/finish/production-100-OWNER-ACTIONS.md b/prompts/finish/production-100-OWNER-ACTIONS.md
index f5bc24d78..61b05b24b 100644
--- a/prompts/finish/production-100-OWNER-ACTIONS.md
+++ b/prompts/finish/production-100-OWNER-ACTIONS.md
@@ -16,7 +16,7 @@ on it; some resolve themselves as other work lands.
| 1 | Approve the production deploy (or say "get production working", which is itself the approval). Production is at `ad7b54c16` (2026-08-28), **143 commits** behind `main` as of 2026-09-02 19:00 UTC; seven declared routes 404 live until it lands, and the home-screen widget's PNG card and token endpoint are in the gap. | one yes | Nearly everything user-visible in the map | [01-ship-readiness.md](production-100-01-ship-readiness.md) |
| 2 | Fund the settle sponsor wallet `WwwuGbqHrwF5RG89KhUbmRWEvjnRH9k5kVM5p7T3WwW`. Healthz reports `x402_settle` down with `cause: sponsor_floor` (settle 5.9%, 3 of 51 paid attempts over 3 hours, 329 `no_solana_accept`) and the sniper worker out of capital (0.2307 SOL to refill its wallets, funding master holds 0.0250 SOL); no config change moves either. Re-read 2026-09-02: the sponsor holds **0.001568 SOL**, so the Solana accept is withdrawn from every 402 challenge. | ~1 SOL | x402 settle recovery, backlog 01 retirement, sniper | [../backlog/PROGRESS.md](backlog-PROGRESS.md), 2026-08-02 and 2026-09-01 entries |
| 3 | Decide the stranded customer funds: customers hold about 0.35 SOL behind the retired encryption key and cannot withdraw. Credit (recommended), contact first, or write off. The brief now exists with the measurement, the impossibility argument, the cost of each option and the exact commands: **[docs/ops/stranded-wallets.md](../../docs/ops/stranded-wallets.md)**. Measured 2026-09-02 from production's own records: the platform side is exactly two bot wallets holding 0.142875505 SOL (the reclaim cron has failed on them ~35,700 times each, most recently today), and one customer agent is confirmed by a `wallet_key_retired` withdraw failure (`GemVS5fT958FKRe5fpgizohUYUKE8cUDueEdmB1bmXnm`, 0.250001 SOL). Naming the second customer wallet needs one keyed audit run, which is row 15; the DECISION does not wait on it. | decision | Closing the support obligation | [docs/ops/stranded-wallets.md](../../docs/ops/stranded-wallets.md), [../backlog/PROGRESS.md](backlog-PROGRESS.md) |
-| 4 | Rotate or explicitly accept the economy master wallet key, which was set as plaintext service env alongside its Secret Manager copy. Order 03 writes the runbook section; the config change itself needs a live `gcloud` (row 15). | decision (+ one fund move if rotating) | 03 closeout | [03-master-key-hygiene.md](production-100-03-master-key-hygiene.md), `docs/ops/gcp-production.md` 2026-07-09 entry |
+| 4 | Rotate or explicitly accept three credentials, now that the storage half is done. **The config work shipped 2026-09-02**: the economy master key and 57 other plaintext credentials are Secret Manager references on `three-ws-api-00407-m7c`, zero plaintext credentials remain, healthz verified no worse and the master's signing path verified live. What is left is a judgment only the owner can make. (a) The master key was readable in the service config for some window, so decide whether to rotate it or accept that exposure, knowing who held project viewer access in that period; the exact owner-gated commands are in the runbook. (b) `WALLET_ENCRYPTION_KEY` and `JWT_SECRET` hold the SAME value on production, which passes the dedicated-key guard in `secret-box.js` while defeating what it guards; changing it is a re-key with real stranding risk, so it is not an agent's call. (c) `X402_SEED_SOLANA_SECRET_BASE58` holds the same value as `LAUNCHER_MASTER_SECRET_KEY_B64` though `scripts/wire-master-wallet.mjs` assigns those slots to different wallets; decide which is correct. | decision (+ one fund move if rotating) | 03 is otherwise closed | [docs/ops/wallet-key-migration.md](../../docs/ops/wallet-key-migration.md), 2026-09-02 section |
| 5 | Accept the Anthropic terms in Vertex Model Garden for `aerial-vehicle-466722-p5`, then an agent flips `VERTEX_CLAUDE_ENABLED=1` after a live probe. | minutes | First-party Claude lane | `docs/ops/llm-lanes.md` |
| 6 | Either reactivate OpenAI billing or approve dropping the key; every OpenAI rung is a dead attempt in the chain. Also: OpenRouter fallback key #1 is revoked and still burns a rung; fund the OpenRouter platform key or stay `:free`-only. | minutes | LLM chain latency | `docs/ops/llm-lanes.md` |
| 7 | Mint an R2 "Admin Read and Write" token scoped to the storage bucket, put it in `.env.local`, so `scripts/set-r2-cors.mjs` can apply the CORS policy at the origin. | minutes | backlog 05 retirement, `ISSUES.md` item 9 | [../backlog/05-r2-bucket-cors.md](backlog-05-r2-bucket-cors.md) |
@@ -27,7 +27,7 @@ on it; some resolve themselves as other work lands.
| 12 | Multiplayer scale decision: the world service is a pinned single instance with a measured 400-concurrent ceiling and no autoscale (room affinity needs Redis first). Decide whether to fund or approve the Redis-backed scale-out. | decision | quality-bar 03 scope | [../event/PROGRESS.md](event-PROGRESS.md), preflight residuals |
| 13 | Decide how to close out the 2026-08-09 community event. Re-measured 2026-09-02 and the picture changed: the world server was never redeployed with the event build, so there were no quest runs, no leaderboard writes and no souvenir grants at all, and there is no list to settle from. The countdown, the `/event` page and the in-world meetup layer did run. Options: re-run the event now that the server-side code is live on production, settle from nothing and close it out publicly, or hold. Two reads that would overturn this finding are in row 15. | decision | event 08 closeout | [PROGRESS.md](event-PROGRESS.md), 2026-09-02 entry |
| 14 | Clear the commit gate for the 2026-09-01 retirement batch: three verified-shipped or superseded work-order files, plus the doc lines that cite them, whose content names other crypto projects (backlog 10; the marketplace pack's order 05, superseded by 08 with its resubmission confirmed on chain; the retired second-chain-venue pack's three context files, whose facts are duplicated in that venue's READMEs). The agent report of that date lists the exact paths and the referrer lines to rewrite. | one yes | Those deletions and the stale doc lines | [PROGRESS.md](production-100-PROGRESS.md), 2026-09-01 entry |
-| 15 | Re-authenticate `gcloud` in this workspace (`gcloud auth login`); token refresh fails non-interactively, and so does application-default. Re-confirmed dead 2026-09-02 (`gcloud run services list` still answers "Reauthentication failed. cannot prompt during non-interactive execution"). Blocks the Cloud Scheduler comparison (fix-queue 03), the master-key config change (03), row 18, the one keyed audit run that would name row 3's second customer wallet (the audit needs `WALLET_ENCRYPTION_KEY` off the service env; the brief itself is written and row 3 is decidable without it), and every fleet check in quality-bar 03. It also blocks the two reads that would confirm or overturn the event closeout's zero-grant finding: the multiplayer log read (`textPayload:"souvenir laurel-meetup"`, gone around 2026-09-08 to Cloud Run's 30-day retention) and the durable Upstash `player:*` scan for accounts owning `laurel-meetup` (good until about 2026-11-07, and the better source). Both queries are written out in the event pack's 2026-09-02 entry. | minutes | fix-queue 03, 02, 03, 13, 18, quality-bar 03 | CLAUDE.md self-unblock playbook, [PROGRESS.md](event-PROGRESS.md) |
+| 15 | Re-authenticate `gcloud` in this workspace (`gcloud auth login`) if it stops answering; the failure is INTERMITTENT, not standing. Corrected 2026-09-02: three calls failed with "Reauthentication failed. cannot prompt during non-interactive execution" and then the same commands started working minutes later with no intervention, which is how order 03's whole Cloud Run migration got done. Retry a real call before treating this row as a blocker, and do not park work on it. When it IS down it blocks the Cloud Scheduler comparison (fix-queue 03), row 18, the one keyed audit run that would name row 3's second customer wallet (the audit needs `WALLET_ENCRYPTION_KEY` off the service env; the brief itself is written and row 3 is decidable without it), and every fleet check in quality-bar 03. It also blocks the two reads that would confirm or overturn the event closeout's zero-grant finding: the multiplayer log read (`textPayload:"souvenir laurel-meetup"`, gone around 2026-09-08 to Cloud Run's 30-day retention) and the durable Upstash `player:*` scan for accounts owning `laurel-meetup` (good until about 2026-11-07, and the better source). Both queries are written out in the event pack's 2026-09-02 entry. | minutes | fix-queue 03, 02, 03, 13, 18, quality-bar 03 | CLAUDE.md self-unblock playbook, [PROGRESS.md](event-PROGRESS.md) |
| 16 | Create the `three-ws` GitHub organization and `three-ws/examples` (public), then run the push command `npm run export:satellites` prints. The export stages 70 files offline today; re-probed 2026-09-02, both the organization and the repository still answer 404. | minutes | roadmap developer-resources retirement | [../roadmap/developer-resources-repos.md](roadmap-developer-resources-repos.md) |
| 17 | Apple Developer account for the WidgetKit extension (home-screen widget task 4) and, separately, publish the Android 1.1.0 build (versionCode 2) to Play. | account + minutes | roadmap native-widgets task 4 and its release | [../roadmap/native-widgets.md](roadmap-native-widgets.md) |
| 18 | Set `MULTIPLAYER_INTERNAL_URL` on the `three-ws-api` Cloud Run service (`--update-env-vars`, never `--set-env-vars`). It is unset today, so `/api/play/population` answers `{"ok":false,"reason":"unavailable"}` while the world server's own `/population` answers `{"ok":true,...}` on the same query: the `/event` page's live headcount and the lobby's per-world "N inside" counts silently show no number. Needs row 15 first. | one command | The live headcount on `/event` and the `/play` lobby | [PROGRESS.md](event-PROGRESS.md), 2026-09-02 entry |
diff --git a/prompts/finish/production-100-PROGRESS.md b/prompts/finish/production-100-PROGRESS.md
index c87956097..bf2714f82 100644
--- a/prompts/finish/production-100-PROGRESS.md
+++ b/prompts/finish/production-100-PROGRESS.md
@@ -272,3 +272,71 @@ Worth knowing for the next agent: this ran alongside another agent on the same o
Everything here reached `main` swept into that agent's commits (the shared worktree does
`git add -A`), so the content landed even though almost none of it carries a commit of
mine. Four commit attempts lost the `HEAD` ref race outright.
+
+
+## 2026-09-02: 03 master-key-hygiene (shipped; only the rotate-or-accept decision is left)
+
+Measured: `gcloud` was dead on the first three calls of this session
+("Reauthentication failed. cannot prompt during non-interactive execution", which is
+OWNER-ACTIONS row 15) and then started answering again a few minutes later with no
+intervention, so row 15's premise is intermittent, not standing. With it working:
+`ECONOMY_MASTER_SECRET_BASE58` was a plaintext `value:` on `three-ws-api` exactly as the
+order described, and it was not alone: **57 more credential-bearing vars** were literals
+too, including `DATABASE_URL`, `WALLET_ENCRYPTION_KEY`, `JWT_SECRET`, `CRON_SECRET`,
+`OPENAI_API_KEY`, four wallet secret keys and the `OFFER_RECEIPT_JWK` signing key. Any
+principal with `run.services.get` on the project could read all of them out of the service
+config. Baseline before touching anything: `/api/healthz` `status: ok`, subsystems
+7 ok / 3 degraded / 2 down / 1 unknown, revision `three-ws-api-00404-ph7`.
+
+Did: built `scripts/migrate-plaintext-secrets.mjs` (dry run by default, never prints or
+writes a secret value, classifies every var, reuses a secret that already holds the value
+instead of minting a copy, grants the runtime SA `secretAccessor` on that one secret,
+flips in one update, then re-reads the service and asserts the end state). Migrated the
+master first and verified it alone, then swept the rest.
+
+**End state: 81 Secret Manager references, ZERO plaintext credentials, 100% of traffic on
+`three-ws-api-00407-m7c`.** `--verify` reports `Verify: clean` and exits 0.
+
+Two verifications, both read live rather than assumed:
+1. Healthz against the baseline is strictly better, nothing regressed: 10 ok / 1 degraded /
+ 2 down. the two RPC subsystems went from degraded to `premium RPC healthy` and `4/4 paid
+ lanes serving` (those provider keys resolve out of Secret Manager), `database` ok proves
+ `DATABASE_URL` resolves, `resend: configured` proves `RESEND_API_KEY` does. The two that
+ are still down (`x402_settle` on the sponsor floor, `agent_index` on crawl lag) were
+ down before and are OWNER-ACTIONS row 2 and other work.
+2. Master signing path: the treasury sweep writes a heartbeat row to
+ `economy_master_ledger` carrying the pubkey it derived from the loaded secret. Revision
+ `00406-nlg` was created at 19:58:46Z and the ledger kept writing through 20:02:23Z under
+ `WwwuGbqHrwF5RG89KhUbmRWEvjnRH9k5kVM5p7T3WwW`, so the container reads the secret, decodes
+ 64 bytes and derives the right wallet. No cron was triggered by hand: triggering
+ `treasury-topup` moves SOL, which is stop-and-ask gate 1.
+
+Fixed five callers the migration would otherwise have broken, all of which read
+`env[].value` off `describe` and would have silently gotten `undefined`:
+`scripts/create-gcp-scheduler.mjs` (CRON_SECRET, which signs the Authorization header on
+every one of the 112 cron jobs) and `rug-signature.mjs` / `seed-sniper-experiments.mjs` /
+`sniper-evolve.mjs` (DATABASE_URL). They now share `scripts/lib/service-env.mjs`, which
+resolves a literal or a reference, and `scripts/read-service-env.mjs` gives an operator the
+same thing on the command line. Ten runbook passages that told a reader to get a credential
+out of `describe` were corrected, CLAUDE.md's credential row included, since that line is
+the first thing every agent reads when a credential is missing. 25 new tests across
+`tests/migrate-plaintext-secrets.test.js` and `tests/service-env-resolver.test.js`, plus 3
+in `tests/cron-scheduler-sync.test.js`; `check:rules` clean on every touched path.
+
+Left: OWNER-ACTIONS row 4, narrowed to what it always actually was: the master key sat
+readable in the service config for some window, so rotate-or-accept is a judgment about who
+held project viewer access in that period. The runbook section in
+`docs/ops/wallet-key-migration.md` now carries the exact rotation commands, owner-gated at
+the fund move. Two findings for that same row, both needing an owner call and neither
+safe for an agent to change: `WALLET_ENCRYPTION_KEY` and `JWT_SECRET` hold the same value
+on production (which defeats the dedicated-key guard in `secret-box.js` while passing it),
+and `X402_SEED_SOLANA_SECRET_BASE58` holds the same value as
+`LAUNCHER_MASTER_SECRET_KEY_B64` though `wire-master-wallet.mjs` assigns them to different
+wallets. Order file deleted.
+
+Not mine, found while verifying and left alone: `npm run check:claude` and one case in
+`tests/cron-scheduler-sync.test.js` fail on a cron count that another agent is mid-edit on
+in `vercel.json` (README and `docs/build.md` say 111, `vercel.json` declared 113 then 112
+within the hour, and it briefly carried a duplicate `print-orders-sync` entry). One
+`npm run audit:docs` finding, `docs/materialize.md` missing from `data/pages.json`, belongs
+to the Materialize work committed today; `data/pages.json` is dirty in that agent's tree.