The previous two attempts at allowlisting cdn.jsdelivr.net for
transformers.js's onnxruntime-web loader landed in shared-utils
security-headers.ts. The actual file change was correct (verified by
grep), the commits got pushed, the live security-headers.ts on disk
had the additions — but Vite's SSR module cache for cross-workspace-
package imports kept serving the OLD compiled shared-utils to
hooks.server.ts. Net effect: edits to hooks.server.ts hot-reloaded
fine (proven by the *.hf.co connect-src additions showing up
immediately) while edits to shared-utils/security-headers.ts did not.
A dev server restart should clear it but I'd rather not depend on
manual intervention every time we touch the shared CSP.
Move the jsdelivr allowlist out of the shared default and into
mana-web's hooks.server.ts via the existing scriptSrc + connectSrc
options. hooks.server.ts is in the SvelteKit app's own source tree so
it HMRs reliably, no SSR cache to fight. As a bonus this is also
architecturally cleaner: cdn.jsdelivr.net is only needed by mana-web
because mana-web is the only Mana app that bundles @mana/local-llm —
other apps get a slightly tighter CSP for free.
The pattern to remember: changes to packages/shared-utils that affect
SSR (response headers, server hooks) require either a dev server
restart OR a manual `rm -rf apps/.../node_modules/.vite` to take
effect. Client-side changes hot-reload fine.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two QuickActionsWidget files lived in parallel under different widget
systems — `dashboard/widgets/` (the user-customizable dashboard, i18n
keys, 3 actions: credits/feedback/profile) and `core/widgets/` (the
mana home screen, hardcoded German strings, 5 actions: todo/calendar/
contacts/context/times). The two rendered the same shape character-
for-character: optional emoji-prefixed title + a list of rounded-card
links each with icon + label + description. Only the data and a
slightly different padding/icon sizing differed.
Extract <QuickActionsList> in $lib/components that takes the actions
array directly (consumers resolve i18n before passing in). Both widget
files become thin wrappers — the dashboard one resolves $_(...) keys
and passes the result, the core one passes its hardcoded data with
`compact` set.
LOC: 110 → 102 across the 3 files (-8 net, plus the shared 70-LOC
molecule). Small numerically, but the bigger win is that future
changes to the link layout (hover state, padding, icon style) happen
once instead of twice — and the two widget files no longer accidentally
drift in sizing/spacing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Nine files (8 ListViews + todo's TaskList) reimplemented the same
context-menu state machinery character-for-character: a typed
$state object with visible/x/y/<itemKey>, a handleItemContextMenu
function that calls preventDefault and stuffs the click position
in, and a close handler that resets the entity field.
Extract `useItemContextMenu<T>()` in $lib/data/item-context-menu.svelte
that returns a reactive handle with `.state` (visible/x/y/target),
`.open(e, target)`, and `.close()`. Consumers derive their menu
items from `ctxMenu.state.target` and pass `ctxMenu.close` directly
to <ContextMenu onClose>.
Per file: ~10 LOC of state declaration + handler removed; consumer
items array switches from `ctxMenu.<entity>` to `ctxMenu.state.target`.
Across the 9 files this is ~−90 LOC of pure boilerplate; helper itself
is 50 LOC. Net small (~−40 LOC) but the boilerplate is gone and the
shape is one helper away from being adjustable globally.
Note: shared-ui already exports a `createContextMenuState` factory,
but it's a plain default-value object — not a Svelte 5 reactive
helper. This new wrapper composes with the existing `ContextMenuState<T>`
type from shared-ui rather than replacing it.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
After the WebLLM → transformers.js migration, the first attempt to load
Gemma 4 E2B in /llm-test was blocked by CSP at the *weight shard*
download step (tokenizer.json got through fine — it lives on
huggingface.co directly). HF has rolled out a new XET-backed CDN for
large model files at cas-bridge.xethub.hf.co, served from
*.xethub.hf.co (the parent zone is hf.co, NOT huggingface.co — so our
existing wildcard `*.huggingface.co` did not cover it).
Open the broader hf.co wildcard (`https://*.hf.co`) so future XET host
rotations don't bite us, plus the explicit cas-bridge.xethub.hf.co
entry for older CSP-strict browsers that want narrower matches first.
The legacy huggingface.co + cdn-lfs.huggingface.co entries stay in place
for repo metadata and any model still on the old LFS path.
Update the comment block above the CSP additions to reflect that the
package now uses transformers.js + ONNX shards rather than the old
WebLLM/MLC path, including a quick map of which HF domain serves what.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
End-to-end testing surfaced a 404 from the synth path. mana-llm
(services/mana-llm/src/main.py) mounts the OpenAI-compatible API at
/v1/* — there's no /api prefix.
The first quick-depth e2e run only worked because the planner is
skipped on quick (it just uses the question itself), so llmJson never
fired; only llmStream did, and the streaming path also used the wrong
prefix but the test happened to land before this was caught.
The other apps/api modules (chat, guides, context, traces) all use the
wrong /api/v1/ path too — that's a separate, pre-existing bug to be
addressed in their own commits.
Verified by re-running a standard-depth research run end-to-end against
mana-llm pointed at the GPU server's ollama with gemma3:4b/12b: plan +
retrieve + extract + synth all succeed.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Per-app web Dockerfiles do `FROM sveltekit-base:local` and do NOT re-COPY
packages/shared-* — those packages are baked into the base image. So a
change to packages/shared-utils, packages/shared-ui, etc. only reaches
the live web app if the base image is also rebuilt.
This bit us THREE times on 2026-04-08 alone:
1. CSP fix in shared-utils ('wasm-unsafe-eval') sat unused in production
for over an hour because every `build-app.sh mana-web` cheerfully
reused the cached base layer that still contained the old shared-utils.
2. Same problem with the BaseListView export in shared-ui after the
ListView consolidation refactor — mana-web's build failed because the
Rollup pass couldn't resolve the new symbol from the stale base.
3. Same shape, different package, repeatedly.
The pattern is identical every time and the manual workaround
(`build-app.sh --base` first) is something you only think to run if you
already know how the layering works. Make the script catch it.
New `is_base_image_stale` helper compares the base image's `Created`
timestamp against the latest git commit touching paths the base image
actually depends on:
- packages/ (all shared-* packages baked in)
- docker/Dockerfile.sveltekit-base
- pnpm-lock.yaml (transitive dep changes)
When building any *-web service, if the image is stale or missing, the
base is rebuilt automatically before the per-app build kicks off, with
the triggering commit's oneline printed for transparency.
Date parsing notes:
- macOS Docker emits the Created field with the local TZ offset
("...+02:00"), not Z. We strip the fractional + offset suffix and
parse the literal local clock time with BSD date (no -u), which is
what the original timestamp meant on this host. GNU date is the
fallback for Linux dev boxes and handles the full ISO directly.
- If parsing fails for any reason we conservatively force a rebuild
rather than risk shipping stale code.
Verified end-to-end against the live Mac Mini's current state earlier
today: image 55s newer than the last packages/ commit at the time →
"fresh, skip" (correct). When the next packages/ commit lands, the
script will see commit_epoch > image_epoch and trigger the base
rebuild automatically.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Four ListViews (calendar, contacts, places, todo) reimplemented the
same drag-drop tag append logic, and four matching DetailViews
reimplemented the same "remove tag with undo toast" logic. Extract
both into pure helpers in $lib/data/tag-mutations.ts that take a
store-agnostic update function — works for the standard tagIds
modules and for todo's metadata.labelIds via tasksStore.updateLabels.
Side win: places/views/DetailView's removeTag had no undo toast
(every other module did). Consolidating fixes the inconsistency.
zitare is the outlier — its drag target is a Quote, but the tags
live on a (possibly-not-yet-existing) Favorite record. Stays custom.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace the entire @mana/local-llm engine with a transformers.js-based
implementation backed by Google's Gemma 4 E2B (released 2026-04-02).
The external API of LocalLLMEngine — load(), generate(), prompt(),
extractJson(), classify(), onStatusChange(), isSupported() — is
preserved 1:1, so the /llm-test page, the playground module, and the
Svelte 5 reactive bindings in svelte.svelte.ts need no changes
beyond updating the default model key.
Why the engine swap: MLC has not (and as of today still hasn't)
published Gemma 4 builds for WebLLM. The webml-community team and
HuggingFace's onnx-community already have Gemma 4 E2B running in
the browser via transformers.js + WebGPU, with a documented
Gemma4ForConditionalGeneration class shipped in @huggingface/transformers
v4.0.0. Going through the ONNX route gets us the latest Google model
six days after release instead of waiting on MLC compilation.
Trade-offs accepted (discussed before this commit):
- transformers.js is a more generic ONNX runtime, so per-token
throughput will be ~20-40% lower than WebLLM would deliver for the
same model size. For a 2B model on a modern WebGPU device that's
still well above interactive latency.
- The JS bundle gains ~2-3 MB (the ONNX runtime). Negligible compared
to the 500 MB model download.
- transformers.js v4 is brand new (released alongside Gemma 4) so the
Gemma4ForConditionalGeneration code path has very little battle
testing yet. The risk is partially offset by webml-community's
reference implementation.
What changed file by file:
- packages/local-llm/package.json: drop @mlc-ai/web-llm, add
@huggingface/transformers ^4.0.0; bump version 0.1.0 → 0.2.0; rewrite
description.
- packages/local-llm/src/types.ts: add `dtype` field to ModelConfig
('fp32' | 'fp16' | 'q8' | 'q4' | 'q4f16') so each model can request
the quantization that matches its uploaded ONNX shards.
- packages/local-llm/src/models.ts: replace the old Qwen 2.5 + Gemma 2
registry with a single `gemma-4-e2b` entry pointing at
onnx-community/gemma-4-E2B-it-ONNX with q4f16 quantization. Future
models can be added by appending entries — the /llm-test picker
reads MODELS dynamically and picks them up automatically.
- packages/local-llm/src/cache.ts: replace the WebLLM-specific
hasModelInCache helper with a generic Cache API probe that looks for
`https://huggingface.co/{model_id}/resolve/main/tokenizer.json` in
any open cache. tokenizer.json is small, downloaded first, and
always present, so its presence is a reliable proxy for "model has
been loaded before".
- packages/local-llm/src/engine.ts: full rewrite. Internally we now
hold a transformers.js model + processor pair (created via
AutoProcessor.from_pretrained + Gemma4ForConditionalGeneration.from_pretrained
with `device: 'webgpu'`), and translate our LoadingStatus union from
the library's `progress_callback` shape. generate() applies Gemma's
chat template via the processor, runs model.generate() with optional
TextStreamer for streaming, then slices the prompt tokens off the
output tensor to compute per-call usage. The convenience methods
(prompt, extractJson, classify) are unchanged because they only call
generate() under the hood.
- packages/local-llm/src/generate.ts and status.svelte.ts: deleted.
These were orphaned from a much earlier engine API (referenced
`getEngine()` / `subscribe()` / `LlmState` symbols that haven't
existed for a while) and were never re-exported from index.ts —
they only showed up because `tsc --noEmit` was crawling the src
tree. Their functionality lives in engine.ts + svelte.svelte.ts now.
- apps/mana/apps/web/package.json: swap the direct dep from
@mlc-ai/web-llm to @huggingface/transformers. This is the same
trick we used for the previous adapter-node externals warning —
having it as a direct dep makes adapter-node's Rollup pass treat
it as external automatically.
- apps/mana/apps/web/vite.config.ts: swap ssr.external entry from
@mlc-ai/web-llm to @huggingface/transformers. Add a comment
explaining the why so the next person doesn't wonder.
- apps/mana/apps/web/src/routes/(app)/llm-test/+page.svelte: change
the default selectedModel from 'qwen-2.5-1.5b' to 'gemma-4-e2b'.
All other model display strings come from the MODELS registry, so
this is the single hard-coded reference that needed updating.
- pnpm-lock.yaml: regenerated. Confirmed @mlc-ai/web-llm is gone (0
references) and @huggingface/transformers is in (4 references).
CSP: no header changes needed. We already opened connect-src for
huggingface.co + cdn-lfs.huggingface.co + raw.githubusercontent.com
when fixing the WebLLM blockers earlier today, and 'wasm-unsafe-eval'
is already in script-src — both transformers.js (ONNX runtime) and
WebLLM (MLC runtime) need that. If transformers.js spawns its
inference into a Web Worker via a blob URL we may need to add
`worker-src 'self' blob:` once we hit the first runtime test, but
the existing CSP should be enough for the synchronous path.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Smoke-testing /api/v1/research/start with mana-search down surfaced a
crash: drizzle's .values([]) throws "values() must be called with at
least one value", which dropped the run into status='error' even though
the failure is a perfectly normal "no results" case.
Two changes:
- Guard the sources insert behind enriched.length > 0
- If retrieval returns nothing, short-circuit straight to status='done'
with an explicit German "keine Quellen gefunden" summary instead of
feeding an empty corpus to the synthesiser
The same path also triggers when every sub-query genuinely returns no
results (very specific question, niche domain) so this isn't just an
ops-failure case.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
End-to-end deep-research feature for the questions module: a fire-and-
forget orchestrator in apps/api that plans sub-queries with mana-llm,
retrieves sources via mana-search (with optional Readability extraction),
and streams a structured synthesis back to the web app over SSE.
Backend (apps/api/src/modules/research):
- schema.ts: pgSchema('research') with research_results + sources
- orchestrator.ts: three-phase pipeline (plan / retrieve / synthesise)
with depth-aware config (quick=1×, standard=3×, deep=6× sub-queries)
- pubsub.ts: in-process event bus, single-node, swappable for Redis
- routes.ts: POST /start (202, fire-and-forget), GET /:id/stream (SSE),
POST /start-sync (test only), GET /:id, GET /:id/sources
- Credit gating via @mana/shared-hono/credits — validate up-front,
consume best-effort on `done`. Failed runs cost nothing.
Helpers (apps/api/src/lib):
- llm.ts: llmJson() + llmStream() over mana-llm OpenAI-compat API
- search.ts: webSearch() + bulkExtract() over mana-search Go service
- responses.ts: shared errorResponse / listResponse / validationError
Schema deployment:
- drizzle.config.ts (research-scoped) + drizzle/research/0000_init.sql
hand-authored migration, deployable via psql -f or drizzle-kit push.
- drizzle-kit added as devDep with db:generate / db:push scripts.
Web client (apps/mana/apps/web/src/lib/api/research.ts):
- Typed start() / get() / listSources() / streamProgress(). The stream
uses fetch + ReadableStream (not EventSource) so we can attach the
JWT via Authorization header. Special-cases 402 for friendly toast.
- New PUBLIC_MANA_API_URL plumbing in hooks.server.ts + config.ts.
Module store (modules/questions/stores/answers.svelte.ts):
- New write-side store with createManual / startResearch / accept /
softDelete. startResearch creates an optimistic empty answer, opens
the SSE stream, debounces token deltas in 100ms batches into the
encrypted local row, and on `done` replaces the streamed text with
the parsed { summary, keyPoints, followUps } payload + citations
resolved against research.sources.id.
Citation rendering (modules/questions/components/AnswerCitations.svelte):
- Tokenises [n] markers in the answer body into clickable pills with
hover popovers showing title / host / snippet / external link.
- Lazy-loaded via a session-scoped source cache (stores/sources.svelte.ts)
that deduplicates concurrent fetches.
UI (routes/(app)/questions/[id]/+page.svelte):
- Recherche card with three-state button (start / cancel / re-run),
animated phase indicator, source counter.
- Confirmation dialog warning about web/LLM transmission since the
question itself is locally encrypted.
- Toasts for success / error / cancel via @mana/shared-ui/toast.
- Re-run flow soft-deletes prior research-driven answers but keeps
manual ones intact.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Every module's inline-editable DetailView reimplemented the same
plumbing: liveQuery → optional decryptRecord → reset on id change
→ focused/confirmDelete state → save-on-blur → deleteWithUndo via
toastStore. Plus ~150 LOC of duplicated scoped CSS for the
.detail-view / .title-input / .properties / .prop-row / .section /
.danger-zone style track.
Extract two pieces:
- useDetailEntity (svelte runes module, $lib/data/detail-entity.svelte.ts)
handles the JS plumbing: liveQuery + optional decrypt + reset
on id change + focused/confirmDelete state + deleteWithUndo.
Supports a custom `loader` for cross-table joins (events+timeBlocks,
timeEntries+timeBlocks, tasks+timeBlocks).
- DetailViewShell ($lib/components/DetailViewShell.svelte)
handles the visual scaffold: outer flex column with scroll,
loading/not-found state, body snippet, danger zone with confirm
flow. Exports the shared field/property/section/meta classes as
:global so consumer snippets can use them without redefining.
Migrated 16 of the 18 DetailViews. Skipped:
- zitare: no DB entity (quotes from bundled @zitare/content),
no edit/delete flow.
- events: different page shape (centered max-width, edit/view
modes, eventId via direct prop instead of params, nested guest
list / RSVP sections).
Side wins:
- 6 encrypted modules (storage, uload, music, questions, calendar,
todo) now route their decrypt logic through one path instead of
six separate `liveQuery + decryptRecord({ ...raw })` variations.
- times/views/DetailView had the same latent type bug as the
ListView (reading .date / .startTime / .endTime / .source off
LocalTimeEntry, which doesn't define them). Now uses toTimeEntry()
via the loader option for the joined TimeEntry shape.
Net impact: ~3640 LOC removed across the 16 files (~49% reduction),
~510 LOC added for shell + helper. Net ~3130 LOC saved.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Every workbench-style module ListView reimplemented the same
liveQuery + filter + scroll-area + empty-state shell. Extract a
shared <BaseListView> in @mana/shared-ui (with toolbar/header/
listHeader/item/empty snippets) and migrate the 17 modules whose
list templates fit the workbench tailwind track.
While here:
- migrate DeckCard onto the existing (previously unused) shared
Card atom from shared-ui/atoms.
- fix a latent type bug in times/ListView: it was reading .date /
.startTime / .isRunning off LocalTimeEntry, which doesn't define
them. Now uses the proper joined TimeEntry via toTimeEntry() like
the rest of the times module.
Modules with their own scoped-CSS layout track (calendar, finance,
contacts, notes, places, todo, photos, habits, automations, dreams,
cycles) and outliers (calc, events, playground, zitare) are left
alone — migrating them would be a visual rewrite, not a structural
shell swap.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two unrelated bugs in scripts/mac-mini/ensure-containers-running.sh,
both caught while debugging a mana-auth crash loop on 2026-04-08:
1. The recovery path passed --env-file "$PROJECT_ROOT/.env.macmini" to
docker compose, but that file has never existed on the server — only
.env does, and compose auto-loads it from the working directory. The
explicit --env-file silently caused recovered containers to start with
empty secrets (e.g. blank MANA_AUTH_KEK), which made mana-auth crash
the moment it came back up. The auto-recovery loop was therefore
self-defeating: it kept "fixing" auth into the same broken state
every 5 minutes for hours, with no notification because compose
exited 0. Drop --env-file entirely and cd into PROJECT_ROOT so
compose's standard .env discovery applies.
2. mana-infra-minio-init is a one-shot job container that legitimately
sits in "exited" state after running once. The script flagged it as
"stuck" every cycle, tried to "recover" it, and spammed the log with
ERROR lines. Add an explicit ONESHOT_INIT_CONTAINERS allowlist and
skip those names in both the initial scan and the post-recovery
verification.
Also tee compose output into the log so future failures actually leave
a breadcrumb instead of disappearing into the void.
Also: bump @mlc-ai/web-llm from a transitive dep (via @mana/local-llm)
to a direct dep of @mana/web. SvelteKit's adapter-node post-build
Rollup pass uses the web app's direct deps as its externals heuristic;
without this entry it warns "@mlc-ai/web-llm ... could not be resolved
- treating it as an external dependency" on every build. Functionally
harmless (the dynamic import in LocalLLMEngine only fires in the
browser), but the warning hid a real adapter-node misconfiguration
that would have bitten us if we'd ever tried to SSR /llm-test.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two more centred bottom-anchored toasts had the same problem the
EncryptionIntroBanner had in 2a437a586: their own position: fixed
with hardcoded bottom + transform centring put them in stacking
contexts that the QuickInputBar (z-index 90) either covered up
(SessionWarning, z-index 45 → hidden) or sat under (SuggestionToast,
z-index 9999 → covered the input bar instead).
Both moved into .bottom-stack as .bottom-stack-notification children
in (app)/+layout.svelte, with the parent handling positioning and
the components themselves stripped down to in-flow flex items.
- SessionWarning: was a free-floating element inside (app)/+layout
but outside the bottom-stack — moved into the stack, kept the
authStore.isAuthenticated gate so it only renders for logged-in
users
- SuggestionToast: was mounted in the ROOT layout, but its only
consumer (automationsStore) is an (app)-only module so the toast
never made sense on auth/landing pages. Moved into (app) bottom-
stack, removed from root layout
CSS cleanup in both: dropped position: fixed, bottom, left,
transform, max-width, z-index. Slide-up keyframes rewritten to use
translateY only (no more parent-transform-X to fight with).
Stack order in (app)/+layout.svelte from top to bottom now:
1. EncryptionIntroBanner (one-time)
2. NotificationBar (guest nudge, conditional)
3. SessionWarning (auth-only, conditional)
4. SuggestionToast (auto-dismissing, conditional)
5. QuickInputBar
6. TagStrip
7. PillNav
Corner-anchored toasts (PwaUpdatePrompt right-12px, SyncConflictToast
right-1rem) intentionally NOT moved — they live in different visual
real estate and don't compete with the centred stack column.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The one-time encryption intro banner used its own position: fixed at
bottom: 1.5rem with z-index: 60, mounted at the root layout level.
That put it in a stacking context the QuickInputBar covered up — the
search bar visually sat ON TOP of the banner instead of below it,
making the privacy claim half-readable and the dismiss X impossible
to click.
Same fix the guest nudge got in c8ed58b7d: move into the bottom-stack
flex container in (app)/+layout.svelte and let the parent handle
positioning. The banner is now the FIRST child of the stack so it
renders above the guest nudge / QuickInputBar / TagStrip / PillNav
and stays in flow as the stack reflows when nav collapses.
- Removed `<EncryptionIntroBanner />` from root +layout.svelte (it
doesn't belong above the (app) gate anyway since it self-checks
isVaultUnlocked() which is always false outside auth context)
- Mounted inside `.bottom-stack` as the first `.bottom-stack-notification`
child in (app)/+layout.svelte
- Stripped position: fixed / bottom / left / transform / max-width /
z-index from the banner CSS — now an in-flow flex item with
width: 100% (the wrapper centres + caps width via the existing
bottom-stack-notification rules)
- Slide-up animation rewritten to use translateY only since the
parent no longer transforms the banner
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Covers today's voice quick-add session: shared <VoiceCaptureBar>
across five modules, generic /api/v1/voice/{transcribe,parse-task,
parse-habit} endpoints, LLM-driven structure extraction with tag
matching, the mana-llm Ollama routing fix (Colima LAN-range RST
gotcha), gemma3:12b + few-shot prompt iteration, 49 unit tests,
the .env.secrets persistent dev-secret layer, and the two real
bugs found during prod deployment (SvelteKit prod-export
restriction + $env/dynamic/private PUBLIC_-prefix exclusion).
18 commits, ~+1.220 LOC net, voice quick-add now live on mana.how
for todo + habits with full structured extraction.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
WebLLM was blocked by connect-src — model config and weight shards live
on huggingface.co (+ cdn-lfs.* for LFS), and the WebGPU model_lib WASM
comes from raw.githubusercontent.com (binary-mlc-llm-libs). Also wires
Gemma 2 2B/9B into the model registry so /llm-test picks them up.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The playground route was previously a stub. This turns it into a proper
module:
- A streaming chat surface that talks to mana-llm's OpenAI-compatible
/v1/chat/completions and /v1/models. The SSE chunk parser is hand-rolled
in modules/playground/llm.ts (~30 lines) rather than pulling a dep —
the wire format is straight OpenAI and the playground is the only
consumer right now. If chat / todo enrichment / cycles insights end up
hitting the same surface, this lifts cleanly into $lib/data/llm-client.ts.
- A persisted **snippets** store: name + systemPrompt + (model, temperature)
defaults that the user can pin and reorder. Stateless chat history stays
out — that's what the chat module is for. Both `name` and `systemPrompt`
are encrypted (same pattern as notes/dreams), with a registry entry in
data/crypto/registry.ts and a Dexie schema in data/database.ts.
- Standard module wiring: collections.ts / queries.ts / types.ts /
stores/snippets.svelte.ts / module.config.ts, registered in
module-registry.ts alongside the other 30+ modules.
- ListView.svelte and the (app)/playground/+page.svelte route consume
the new store + the streaming client.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Phase 8 follow-up. Places carries GDPR-sensitive PII so it gets the same
treatment as the rest of Phase 7+8, with one deliberate carve-out:
- `places` encrypts the user-typed surface (name / description / address)
but leaves lat/lng plaintext so the proximity matcher in
tracking.svelte.ts can run during background geolocation logging without
a vault unlock. The trade-off is documented inline in registry.ts: a
handful of named POIs is much less sensitive than the full movement
trail.
- `locationLogs` IS the movement trail, so every coordinate field
(latitude, longitude, accuracy, altitude, speed, heading) is encrypted.
Indexed columns (timestamp, placeId, [placeId+timestamp]) stay plaintext
for the time-range scans in the log view.
- `placeTags` stays out of the registry — pure FK join table, no user
content, same pattern as manaLinks.
queries.useAllPlaces / useLocationLogs now decrypt before mapping to the
DTO. placesStore.create/update snapshot the plaintext DTO before
encryptRecord mutates the local in place — same pattern as
notes/dreams/contacts. trackingStore.logPosition decrypts the place set
before running the nearest-place match (the lat/lng carve-out means this
still works pre-unlock, but downstream consumers want the decrypted name).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The first prod deploy of voice quick-add (3b41b39a3) silently fell
back for every transcript: title=transcript verbatim, dueDate=null,
priority=null, labels=[]. The endpoint code was reaching the
fallback() path even though mana-llm was healthy and reachable from
inside the mana-web container.
Root cause: SvelteKit's $env/dynamic/private explicitly excludes any
env var that starts with the public prefix (default PUBLIC_). The
parse-task code read
env.MANA_LLM_URL || env.PUBLIC_MANA_LLM_URL || 'http://localhost:3025'
expecting to fall back to PUBLIC_MANA_LLM_URL when MANA_LLM_URL was
unset, but $env/dynamic/private treats PUBLIC_MANA_LLM_URL as if it
didn't exist on the server side. So it always fell through to
http://localhost:3025, which from inside mana-web is nothing,
fetch threw, and coerce returned the fallback shape.
Two fixes:
1. docker-compose.macmini.yml — set MANA_LLM_URL (no prefix) on
mana-web alongside PUBLIC_MANA_LLM_URL. The PUBLIC_ var is still
needed for the browser-side playground and status page; the
private one is what the parse endpoints actually read.
2. parse-task and parse-habit — drop the dead env.PUBLIC_MANA_LLM_URL
fallback so the next dev who reads the code doesn't think it'd
ever work. Add a comment explaining the SvelteKit gotcha so the
next person setting up a new env var doesn't repeat this mistake.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SvelteKit's production build forbids non-handler exports from a
+server.ts file — dev runs them fine but `pnpm build` errored with
"Invalid export 'coerce' in /api/v1/voice/parse-task" when trying to
deploy mana-web with the new unit tests.
Move ParseResult, fallback, DATE_TRIGGER_PATTERNS,
PRIORITY_TRIGGER_PATTERNS, transcriptMentions, coerce, and extractJson
into a sibling coerce.ts module. The +server.ts file imports from
there and only exports POST, which is the prod build's hard rule.
Tests now import from ./coerce instead of from the route handler,
which also drops the $env/dynamic/private resolution dance from the
test fast path — coerce.test.ts now runs in ~130ms instead of ~400ms
because it pulls in zero SvelteKit runtime.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Fills the gap between the last entry (2026-03-31) and today. The 2026-04-06
slot is intentionally skipped — no commits that day.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The deterministic guards in parse-task's coerce() are the load-bearing
defense against gemma3 hallucinating dueDate / priority on bare tasks.
The integration tests against the live LLM cover the happy path
end-to-end, but they go offline as soon as mana-llm is unreachable —
the unit tests cover the guard logic in isolation with synthetic LLM
responses, so a regression in the rules is caught even when the LLM
itself is dark.
22 cases:
- transcriptMentions: substring matching, case-insensitivity, empty
pattern list, the German + English date words from the few-shot
examples, and the negative cases ("Mülltonnen rausstellen",
"Buy milk") that must NOT trigger.
- coerce: fallback shape on garbage input, transcript-as-title when
the model omits one, time-component stripping ("2026-04-09T14:00:00"
→ "2026-04-09"), malformed dueDate rejection, the dueDate /
priority hallucination guards (drop when the transcript has no
trigger word), real-date / real-priority preservation, label
filtering (cap at 3, drop non-strings, empty array on non-array
input), invalid priority value rejection.
Helpers exported solely for the tests via a __test object — the
production endpoint goes through buildPrompt + coerce as before.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two related changes that fall out of real end-to-end testing against
the now-working local mana-llm.
1. Default model bumped from gemma3:4b to gemma3:12b for both
parse-task and parse-habit. The 4b model gets weekday math
off-by-one ("nächsten Montag" from a Wednesday → 2026-04-14
instead of 2026-04-13), aggressively shortens titles ("Anna
anrufen" → "Anrufen"), and frequently paraphrases habit names
instead of copying verbatim ("Joggen" instead of "Laufen") which
the verbatim-validation in coerce drops, costing an LLM round-trip
for nothing. The 12b variant is roughly 10% slower for these
tiny prompts (~1.1s vs ~1.0s on the GPU box) so the accuracy
win is essentially free.
2. parse-task prompt rewritten as few-shot. Pure rule descriptions
were *worse* than simple examples — the long "Rules — read
carefully" section in the previous prompt actually made the model
compute next Monday as 2026-04-14 even though a direct "what date
is next Monday?" prompt to the same model returned 2026-04-13.
The detailed rules were also priming the model to over-shorten
titles and over-eagerly tag filler words. Five worked examples
(including the previously-failing "Anna nächsten Montag anrufen"
case) plus one novel case ("Mama am Wochenende besuchen") all
come back correct now, including for the novel one.
The deterministic guards in coerce() are kept as a backstop for the
day the GPU box swaps in a weaker model — they're cheap and don't
hurt the happy path.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Real end-to-end testing against the now-working local mana-llm
surfaced two model behaviours the prompt couldn't talk down:
1. gemma3:4b stamps today's date on every task that doesn't have a
real time anchor. "Mülltonnen rausstellen" came back with
dueDate=2026-04-08 and priority=low even though the prompt
explicitly said "MUST be null when no date is mentioned". After
typing "Buy milk" the user would silently get a today-due task,
which is worse than no parsing at all.
2. The model occasionally returns dueDate as a full ISO timestamp
("2026-04-09T14:00:00") when the transcript mentions a time. The
coerce regex previously matched the prefix and let the timestamp
through unchanged, which then breaks the YYYY-MM-DD-shaped Dexie
field downstream.
Fix: deterministic post-processing in coerce. The prompt is also
tightened with explicit "ONLY when…" rules but the guards are the
load-bearing change since gemma3:4b ignores prompt restrictions.
- Strict YYYY-MM-DD extraction: a leading-anchor regex match keeps
only the date prefix even if the model adds a time component.
- DATE_TRIGGER_PATTERNS: substring scan over the original transcript
for German + English date words. If the LLM returned a dueDate but
the transcript has zero matches, drop the date — it was a
hallucination. False positives are preferable to false negatives:
letting through a fake date is more annoying than suppressing a
real one the user can re-type.
- PRIORITY_TRIGGER_PATTERNS: same idea for priority. The model thinks
taxes are inherently urgent; we don't want to inherit that opinion.
The labels field is left noisy on purpose — "müll", "unbedingt",
"erledigen" all come back from a single transcript and only the ones
that fuzzy-match an existing workspace tag end up on the task, so
filtering filler words at this layer would be wasted work.
Verified against five transcripts spanning bare/explicit/relative
date in DE + EN. Real LLM round-trip via http://localhost:5173 →
https://llm.mana.how → ollama gemma3:4b. Local mana-llm now reaches
its Ollama backend after the gpu-proxy routing fix in 7f382138a.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
A grep audit after the previous matrix removal commits found a handful
of stragglers in non-runtime files that the earlier sweeps missed:
- services/mana-llm/CLAUDE.md: removed matrix-ollama-bot from the
consumer-apps diagram and from the related-services table
- services/mana-video-gen/CLAUDE.md: removed "Matrix Bots" integration
bullet
- packages/notify-client/README.md: removed sendMatrix() doc entry
(the method itself was already gone in the prior cleanup)
- docker/grafana/dashboards/logs-explorer.json: dropped the "Matrix
Stack" log row that queried tier="matrix" (would show no data forever)
- docker/grafana/dashboards/master-overview.json: dropped the "Matrix
Bots" stat panel that counted up{job=~"matrix-.*-bot"}
- apps/mana/apps/landing/src/data/ecosystem-health.json: regenerated via
scripts/ecosystem-audit.mjs to drop matrix from the app list, icon
counts, file analytics, top offenders and authGuard missing list
- .gitignore: removed services/matrix-stt-bot/data/ pattern (the
service itself was deleted long ago)
Production-side stragglers also addressed (not in this commit):
- DROP USER synapse on prod Postgres (the parallel cleanup commit
2514831a3 dropped DATABASE matrix + DATABASE synapse but left the
role behind)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The /api/v1/voice/parse-task and /api/v1/voice/parse-habit endpoints
forwarded transcripts to mana-llm without an X-API-Key header. This
worked against the local mana-llm container (no auth) but silently
fell back to the no-LLM path when pointed at gpu-llm.mana.how, which
requires an API key — voice quick-add would look like it was running
in degraded mode forever with no signal that auth was the cause.
Now both endpoints read MANA_LLM_API_KEY from the server-side env and
attach it as X-API-Key when present, mirroring the pattern already
used by /api/v1/voice/transcribe for mana-stt. When the var is empty
the header is omitted, so local Docker setups without auth still work.
Plumbing: generate-env.mjs writes MANA_LLM_URL + MANA_LLM_API_KEY into
apps/mana/apps/web/.env, .env.development gets the new keys with empty
defaults, ENVIRONMENT_VARIABLES.md documents the gateway and where to
get a key.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two new test files lock in the matching boundary where free-text LLM
hints meet the user's actual workspace data — that's where bugs hide
silently. Both matchers are now pure-function-shaped (the production
wrappers just feed them Dexie data) so the tests run without
fake-indexeddb or any I/O.
todo: 16 cases for matchLabelsToTagsPure covering exact / case /
diacritic / substring / specificity rules + the "never invent tags"
guarantee.
habits: 11 cases for matchHabitToTranscript including the word-
boundary "Bier vs ausprobiert" false-positive, multi-word matching,
and a real bug the test surfaced on the first run:
Without specificity ranking, "Tee" would always beat "Grüner Tee"
because the first matching habit in input order won. The matcher
now collects all candidates and returns the one with the most
matched tokens, so multi-word habits beat single-word substrings
whenever both could fit the transcript.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit bundles two unrelated changes that were swept together by an
accidental `git add -A` in another working session. Documented here so the
history reflects what's actually inside.
═══════════════════════════════════════════════════════════════════════
1. fix(mana-auth): /api/v1/auth/login mints JWT via auth.handler instead
of api.signInEmail
═══════════════════════════════════════════════════════════════════════
Previous attempt (commit 55cc75e7d) tried to fix the broken JWT mint in
/api/v1/auth/login by switching the cookie name from `mana.session_token`
to `__Secure-mana.session_token` for production. That was necessary but
not sufficient: Better Auth's session cookie value isn't just the raw
session token, it's `<token>.<HMAC>` where the HMAC is derived from the
better-auth secret. Reconstructing the cookie from auth.api.signInEmail's
JSON response only gave us the raw token, so /api/auth/token's
get-session middleware still couldn't validate it and the JWT mint kept
silently failing.
Real fix: do the sign-in via auth.handler (the HTTP path) rather than
auth.api.signInEmail (the SDK path). The handler returns a real fetch
Response with a Set-Cookie header containing the fully signed cookie
envelope. We capture that header verbatim and forward it as the cookie
on the /api/auth/token request, which now passes validation and mints
the JWT correctly.
Verified end-to-end on auth.mana.how:
$ curl -X POST https://auth.mana.how/api/v1/auth/login \
-d '{"email":"...","password":"..."}'
{
"user": {...},
"token": "<session token>",
"accessToken": "eyJhbGciOiJFZERTQSI...", ← real JWT now
"refreshToken": "<session token>"
}
Side benefits:
- Email-not-verified path is now handled by checking
signInResponse.status === 403 directly, no more catching APIError
with the comment-noted async-stream footgun.
- X-Forwarded-For is forwarded explicitly so Better Auth's rate limiter
and our security log see the real client IP.
- The leftover catch block now only handles unexpected exceptions
(network errors etc); the FORBIDDEN-checking logic in it is dead but
harmless and left in for defense in depth.
═══════════════════════════════════════════════════════════════════════
2. chore: remove the entire self-hosted Matrix stack (Synapse, Element,
Manalink, mana-matrix-bot)
═══════════════════════════════════════════════════════════════════════
The Matrix subsystem ran parallel to the main Mana product without any
load-bearing integration: the unified web app never imported matrix-js-sdk,
the chat module uses mana-sync (local-first), and mana-matrix-bot's
plugins duplicated features the unified app already ships natively.
Keeping it alive cost a Synapse + Element + matrix-web + bot container
quartet, three Cloudflare routes, an OIDC provider plugin in mana-auth,
and a steady drip of devlog/dependency churn.
Removed:
- apps/matrix (Manalink web + mobile, ~150 files)
- services/mana-matrix-bot (Go bot with ~20 plugins)
- docker/matrix configs (Synapse + Element)
- synapse/element-web/matrix-web/mana-matrix-bot services in
docker-compose.macmini.yml
- matrix.mana.how/element.mana.how/link.mana.how Cloudflare tunnel routes
- OIDC provider plugin + matrix-synapse trustedClient + matrixUserLinks
table from mana-auth (oauth_* schema definitions also removed)
- MatrixService import path in mana-media (importFromMatrix endpoint)
- Matrix notification channel in mana-notify (worker, metrics, config,
channel_type enum, MatrixOptions handler)
- Matrix entries from shared-branding (mana-apps + app-icons),
notify-client, the i18n bundle, the observatory map, the credits
app-label list, the landing footer/apps page, the prometheus + alerts
+ promtail tier mappings, and the matrix-related deploy paths in
cd-macmini.yml + ci.yml
Devlog/manascore/blueprint entries that mention Matrix are left intact
as historical record. The oauth_* + matrix_user_links Postgres tables
stay on existing prod databases — code can no longer write to them, drop
them in a follow-up migration if you want them gone for real.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Users can now define multiple named layouts of the workbench homepage and
switch between them. Each scene holds its own openApps list with per-app
window state (minimized / maximized / size). Scene list syncs cross-device
via mana-sync; the active scene id is per-device (localStorage) so device
A doesn't pull device B into a different scene.
- new `workbenchScenes` Dexie table, registered in manaCoreConfig
- `workbenchScenesStore` (Dexie liveQuery) with scene CRUD + per-scene app
mutations; auto-seeds a default "Home" scene on first run
- SceneTabs pill bar above the carousel with dnd reorder + context menu
(rename / duplicate / delete); SceneRenameDialog and a reusable
ConfirmDialog for the destructive path
- workbench +page.svelte refactored to delegate all openApps mutations to
the store; the carousel itself is unchanged
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The parse-task endpoint already returns free-text label hints from the
LLM ("steuern", "haushalt", …). Now the todo store fuzzy-matches each
hint against the user's existing tags via tagCollection and assigns
the matched IDs to the task's metadata.labelIds.
Match policy is intentionally conservative:
- Normalize via NFD strip + lowercase + collapsed whitespace
- Exact normalized match wins
- Substring fallback only for ≥3 char strings (avoids "ab" hitting
every tag containing "ab")
- Never auto-creates a tag — even if the LLM is sure, an unknown topic
silently drops, because auto-creating would clutter the user's tag
list with one-off duplicates from voice transcripts
Both flows pick this up: voice always (transcripts almost always carry
topic hints) and typed only when there's structured payoff, same
asymmetry as before — typed quick-add now also enriches when the LLM
just finds a tag match without a date or priority.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Press Enter on "Steuererklärung morgen 14 Uhr hoch" and the task lands
instantly with your exact text as the title — then a background pass
through /api/v1/voice/parse-task swaps in dueDate + priority once
mana-llm answers. The title only gets rewritten when the LLM actually
finds structured info (dueDate or priority); for plain titles like
"Mülltonnen rausstellen" the typed text is left alone, since silently
"cleaning up" perfectly fine input is more annoying than helpful.
Pulled the parse + STT-then-parse plumbing apart so both flows share
parseTaskText() and only differ in policy: voice always applies the
LLM title (raw transcripts are noisy), typed only when there's
structured payoff.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The previous voice quick-add dumped the whole transcript into the task
title — fine for "Steuererklärung" but useless for "Steuererklärung
morgen 14 Uhr hoch", which should land as title="Steuererklärung",
dueDate=tomorrow, priority="high".
New endpoint /api/v1/voice/parse-task posts the transcript to mana-llm
(gemma3:4b, temperature 0) with a tight system prompt that asks for
strict JSON: { title, dueDate, priority, labels }. The endpoint coerces
the response back into the typed shape and falls through to
{ title: transcript, … } whenever anything goes wrong — mana-llm down,
JSON garbled, network timeout. Voice quick-add must never fail harder
than typed quick-add, so the fallback path is the rule, not the
exception.
Labels come back from the LLM as free-text topic hints and don't yet
map to the workspace's tag IDs — fuzzy matching against existing tags
is a follow-up.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The per-module /api/v1/memoro/transcribe and /api/v1/dreams/transcribe
endpoints were literal copies that proxied to mana-stt. Now that the
generic /api/v1/voice/transcribe endpoint exists (added with notes),
point both stores at it and delete the duplicates. -200 LOC, one place
to update STT auth or response shape from now on.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Speak a task and it lands in the list as a placeholder while mana-stt
transcribes it; the title swaps in once the transcript returns.
No date/priority/label parsing yet — that's a follow-up that needs an
LLM pass over the transcript. For now the whole transcript becomes the
task title and the user can edit inline.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Drop a mic into Notes — record, transcribe through the new generic
/api/v1/voice/transcribe proxy (mana-stt), then write the result back
into the placeholder note. The first transcript line becomes the title
when it fits in 80 chars, otherwise a generic 'Sprachnotiz' label.
The inline editor refreshes from the live note while the placeholder
'…' content is still on screen, so a transcript that arrives a moment
after the editor opens shows up automatically without overwriting
anything the user has typed.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Dreams and Memoro had two literal copies of the MediaRecorder boilerplate
plus parallel mic-button markup, error UI, and requireAuth gating. Lift
the recorder + bar into $lib/components/voice and add it to the memoro
workbench ListView (which had no mic at all). New voice-capture features
just drop in <VoiceCaptureBar> with idleLabel/feature/reason/onComplete.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The unified Mana app runs most modules in a "guest mode": you can
open a module, look around, type a quick note, etc. without an
account. But anything that touches an *encrypted* table (dreams
voice capture, memoro recordings, notes, todo, calendar events, …)
needs the user to be logged in — the encryption vault only unlocks
against a Mana Auth session, and writing to those tables without
it throws `VaultLockedError` at the very last step of the action.
Before this commit, every entry point into an encryption-required
action would silently let the guest go through the whole flow
(record audio, wait for transcription, open the dexie write) and
then explode with a stack-trace error. The user lost work and
didn't know why. The dreams voice capture flow surfaced this
during the 2026-04-08 STT debugging session.
The fix is a global imperative gate: `requireAuth({ feature, reason })`.
Call sites await it before the action; it returns immediately if the
user is already authenticated, otherwise pops a global modal that
asks the guest to log in or cancel. Promise-based, so callers
decide what to do with `false` (silent abort, restore state, own
toast).
$lib/auth/require-auth.svelte.ts new — store + helper
$lib/components/auth/AuthRequiredModal.svelte new — global modal
routes/+layout.svelte mount the modal once
packages/shared-utils/src/analytics.ts new ManaEvents.featureBlockedByAuth
event for conversion tracking
Wired into the two voice-capture entry points that actually exhibited
the bug:
modules/dreams/ListView.svelte → feature: 'dreams-voice-capture'
routes/(app)/memoro/+page.svelte → feature: 'memoro-voice-capture'
Both gate on `requireAuth()` BEFORE the mic permission request, so
guests see the friendly "Konto erforderlich" modal instead of
recording → transcribing → crashing.
Design choices documented in detail in the require-auth.svelte.ts
header comment:
- Imperative function (not a button wrapper component) so it
works in event handlers, store actions, keyboard shortcuts,
drag-drop handlers — anywhere async code runs.
- Single global modal mounted once in the root layout, no
portal/z-index gymnastics; two simultaneous prompts replace
each other (the most recent one wins).
- Checks `authStore.isAuthenticated`, not vault-unlocked state —
the user-facing concept is "I need an account", not "I need
a working encryption vault". Vault-unlock failures (network
error etc.) are a separate bug class with their own UX.
- The modal navigates to `/login?next=<current path>` so the
user lands back on the same page after logging in. The
Promise resolves `false` on navigation; the user re-clicks
the original button after coming back, and the second click
sees `isAuthenticated === true` and proceeds without a modal.
Re-triggering the original action across a navigation cycle
would require restoring half-recorded mic state — not worth
the complexity, and the second click is a clean UX.
How to wire a new entry point (4 lines):
import { requireAuth } from '$lib/auth/require-auth.svelte';
async function handleCreateThing() {
const ok = await requireAuth({
feature: 'create-thing',
reason: 'Things werden verschlüsselt gespeichert. Dafür brauchst du ein Mana-Konto.',
});
if (!ok) return;
// ...existing logic
}
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Three independent bugs that conspired to make the dreams + memoro mic
buttons completely unusable in production AND in dev. Each one alone
would have been the only blocker; they layered on top of each other so
fixing the top one just exposed the next.
1. Permissions-Policy header blocked the microphone API entirely.
`packages/shared-utils/src/security-headers.ts` set
`microphone=()` which means "no origin, including self, may use
the microphone". `getUserMedia()` throws a `Permissions policy
violation` and the browser never even shows the permission
dialog — no amount of OS / browser / site settings can override
it because the policy blocks the API at the document level.
Fix: change to `microphone=(self)` so mana.how itself can use
the API. Camera stays disallowed (no module needs it).
2. Notification permission was requested at layout mount time.
`(app)/+layout.svelte` called
`notificationService.requestPermission()` from `onMount()`. Modern
browsers require permission requests to come from a user gesture
— calling it without one queues the prompt until the next click.
That meant the user's FIRST click on any button (in this case the
dreams "Traum sprechen" mic button) showed the queued notifications
prompt instead of the action they actually clicked. Worse,
`getUserMedia()` was then silently dropped because Chrome only
shows one permission dialog at a time.
Fix: remove the mount-time call entirely. Notification permission
must be requested from a button the user explicitly clicks
("Benachrichtigungen aktivieren" toggle in Settings or first time
a reminder is created) — the reminder scheduler still runs without
permission, it just won't fire OS notifications until granted.
3. vite-plugin-pwa registered a service worker in dev that cached
the old layout chunks across reloads, so the fix for #2 was
invisible until the user manually unregistered the SW in DevTools.
`vite-plugin-pwa` defaults `devEnabled: true`, which is a
well-known footgun for fast iteration. Production still gets the
full SW (this only flips dev). The 2026-04-08 mic-button hunt
took an extra hour for exactly this reason.
Fix: pass `devEnabled: false` to createPWAConfig in vite.config.ts.
Verified: in a fresh incognito tab on `localhost:5173/`, opening the
Dreams app in the workbench and clicking the mic button now shows the
microphone permission dialog directly (no notifications hijack), and
recording → transcription works end-to-end against the production
mana-stt service on the GPU box.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
LoginPage cleanup:
- Drop dev pre-fill credentials and the secret logo-as-button trick
- Remove duplicate in-component theme toggle; accept isDark as a prop and let the (auth) layout's global theme toggle drive it
- Move passkey CTA below the password form so the primary flow stays primary
- Remove the dead "Angemeldet bleiben" checkbox (was bound but never forwarded to onSignIn)
- Fix the skip-to-form link to use sr-only/focus:not-sr-only so it only appears on keyboard focus
- Fix the "oder" divider to render its before/after hairlines by setting an explicit color on the parent
- Wire focus-visible outlines on all interactive controls
- Bump 0.6 → 0.75 opacity on subtitle text for AA contrast
- Drop opacity-60 from the headerControls wrapper
Robustness:
- Track all setTimeout IDs in a Set and clear them in an effect cleanup so navigation away doesn't fire stale callbacks (success redirects, error shake, focus restore)
- Replace (result as any) casts with the new typed AuthResult fields
- New resolveErrorCode() helper prefers result.errorCode and falls back to legacy string matching, so rate-limit / account-lock detection survives i18n
- WebAuthn Conditional UI: on mount, if PublicKeyCredential.isConditionalMediationAvailable(), call onSignInWithPasskey({ conditional: true }) so passkeys appear inline in the email autofill dropdown
- Extract the dismissible success-banner markup into a {#snippet successBanner} and reuse it for the verified / verification-sent / magic-link-sent cases (~50 lines of duplicate JSX out)
Page wrappers:
- login/+page.svelte passes isDark={theme.isDark} so the in-app theme store drives both layouts
- register/+page.svelte wraps trackGuestConversion() in queueMicrotask + try/catch so analytics can never block the success redirect
- Drop the dead baseSignupCredits={25} prop from register/+page.svelte (RegisterPage never accepted it)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two changes:
1. New BACKLOG_FILE_BYTES_ENCRYPTION.md captures everything I'd
want to know if I were picking up the file-bytes encryption
work cold in 6 months. ~370 lines, sits next to
DATA_LAYER_AUDIT.md for discoverability.
Sections:
- TL;DR + status (deferred, no production impact yet)
- Goal + non-goals
- Threat model delta table (mode-by-mode)
- Architecture: write path with ASCII flow diagram
- Architecture: read path with ASCII flow diagram
- The six hard parts:
1. Web Crypto AES-GCM doesn't stream → chunked-AEAD wrapper
2. Multipart uploads need coordinated chunking (S3 5 MB minimum
vs. our 1 MB AES-GCM chunks)
3. Resumable uploads + key persistence (new _pendingUploads
table for the in-flight content key)
4. No more server-side thumbnails (three options, recommended:
client-side resize before upload)
5. Sharing complicates the trust model (URL-fragment key
sharing, recommended; Mega.nz / Cryptpad pattern)
6. Migration of existing plaintext files (lazy on-read,
recommended)
- Schema delta (sql + Dexie additions)
- File map (~2200 LoC across 9 new files + 3 touched)
- Testing strategy (unit + integration + e2e per layer)
- Out-of-scope items explicitly listed
- Decision criteria for when to actually do this
- Five open questions for whoever picks it up
- Cross-references to related files
The doc is opinionated where I have a defensible recommendation
and explicit about uncertainty where I don't.
2. DATA_LAYER_AUDIT.md updates:
- Backlog "Offen" item #1 (File-Bytes-Encryption) now points
directly at the new plan doc with a one-line teaser.
- Backlog "Abgeschlossen" gains a row C for the Conflict
Visualization UI shipped in ed8ab4483 (was still listed as
open from the previous audit roll-up).
- List renumbered: Conflict-UI dropped from "Offen", remaining
items shifted up.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The "Gefällt es dir?" guest nudge was a free-floating fixed element at
bottom: 10rem, so it didn't follow the bottom-stack when the PillNav was
collapsed. Move it inside .bottom-stack as the first child so it shares
the stack's reflow.
NotificationBar now uses the elevation system (--color-surface-elevated,
--color-border-strong, --color-foreground) instead of hardcoded rgba so
it adapts to all themes. Bumped the CTA button (shadow + hover lift) and
container (stronger border, layered shadow) to be more visible.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Removed:
- apps/manacore/ — three Svelte files were byte-identical duplicates of
the apps/mana/ versions, leftover from the 2025 rename. Untracked .env
files in the same dir were also cleared.
- 21 empty apps/*/apps/web-archived/ directories — leftover from the
unification move, never tracked in git.
- services/it-landing/ — empty directory, picked up by the services/*
workspace glob for no reason.
- apps/news/apps/server-archived/ — empty.
Fixed:
- scripts/mac-mini/status.sh: COMPOSE_PROJECT_NAME fallback was still
manacore-monorepo from before the rename.
Documented:
- Root CLAUDE.md now describes apps/api/ (the @mana/api unified backend)
as a top-level peer to apps/mana/. It was completely missing from the
trimmed CLAUDE.md, which made the layout look frontend-only.
Closes backlog C from the Phase 9 audit. The data layer has had
real field-level LWW since Sprint 1, but when the server's value
beat a local edit, the user had no way to know. This commit adds
the missing UI piece: a toast that appears whenever applyServerChanges
overwrites a non-empty local field with a strictly newer server
value, with a one-click "restore my version" path.
sync.ts — detection
-------------------
Two new exports:
- SyncConflictPayload: per-field overwrite event shape
(tableName, recordId, field, wasLocal, nowServer, localTime,
serverTime).
- subscribeSyncConflicts(listener): in-module pub/sub. Returns
an unsubscribe function.
Both LWW branches in applyServerChanges (insert-as-update and the
canonical update-with-fields path) now call notifyConflict() when:
1. The server time is STRICTLY greater (not equal) than the local
field time → there's actually an edit window to lose
2. The local field value is non-null/undefined → user actually
typed something to overwrite
3. The values are not equal (cheap JSON-string compare for objects,
=== for primitives) → there's a real change, not an idempotent
server replay
Why a custom registry instead of CustomEvent + window.dispatchEvent?
The existing sync-telemetry + quota-detect helpers use
window.dispatchEvent which doesn't work in node-based vitest envs
(no DOM EventTarget). The conflict bus is small enough that a plain
Set<listener> is simpler than polyfilling EventTarget — and the
node test path matters because we need automated coverage of the
detection logic.
conflict-store.svelte.ts — UI state
-----------------------------------
Svelte 5 $state-backed store with three responsibilities:
1. Coalescing: a SyncConflict is keyed by `${tableName}|${recordId}`,
so a burst of N field-overwrites on the same record collapses
into ONE toast with all affected fields underneath. The original
wasLocal value is preserved across coalescing (we don't clobber
the user's first typed value if a later field event arrives).
2. Auto-dismiss: each conflict has a 30s TTL after which it
evicts itself. Manual dismiss trumps the timer.
3. Restore: writes wasLocal back to Dexie with a fresh updatedAt
that beats the server's serverTime, plus a __fieldTimestamps
patch so the field-LWW pass on the next sync round will let
our value win. Deferred via setTimeout(0) so it lands AFTER
applyServerChanges releases its per-table apply lock — running
before the lock release would silently drop the restore (the
hook suppression is per-table-set, not per-record).
FIFO eviction at MAX_VISIBLE=8 keeps a bursty server from growing
the visible array unbounded.
SyncConflictToast.svelte — the UI
---------------------------------
Mounts globally in +layout.svelte. Stacks bottom-right above the
OfflineIndicator. Each toast shows:
- Module label ("Aufgabe", "Notiz", "Termin", …) derived from a
table-name → German label map. Unknown tables fall through to
the bare table name.
- Field count summary ("Feld »title«" / "3 Felder") — we
deliberately do NOT render the actual values because some are
encrypted blobs and decrypting them in the toast would be
significant complexity for marginal UX gain. The user knows
what they were just editing.
- Two buttons: "Wiederherstellen" (calls conflictStore.restore)
and "Behalten" (calls dismiss).
Slide-in animation, dark-mode-aware styling, role="alertdialog"
for accessibility.
Wiring
------
data-layer-listeners.ts:
- Imports installConflictListener from conflict-store
- Calls it from installDataLayerListeners() right after the
quota + telemetry handlers
- Adds the disposeConflict() call to the cleanup return
+layout.svelte:
- Imports SyncConflictToast and mounts it next to SuggestionToast
so it inherits the same global-overlay positioning context
Tests
-----
Five new integration tests in sync.test.ts cover:
- Fires when server overwrites a non-empty local field with a
strictly newer value
- Does NOT fire when local field is null/undefined (no edit to lose)
- Does NOT fire when values are equal (idempotent replay)
- Fires once per overwritten field on a multi-field update
- Does NOT fire on a timestamp tie (LWW lets server win silently
when there's no real edit window)
All 25 sync tests + 138 total data-layer tests pass. The new
captureConflicts() helper subscribes via subscribeSyncConflicts()
which works in the node-vitest env without needing a DOM polyfill.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root CLAUDE.md: 1138 → 169 lines. Removed ghost apps-archived list,
Supabase env examples, duplicate mana-auth row, contradictory "Code
Quality TODO" block. Pushed search/storage/database/landing/manascore
howtos out to docs/ + .claude/guidelines/ pointers.
apps/mana/CLAUDE.md: 259 → 175 lines. Dropped non-existent workbench/
route from the routing diagram. Folded the auth section into a pointer
to root + the mana-specific current-user stamping pattern. Merged the
two module-system sections. Kept the data-flow ASCII diagram and the
encryption 3-step workflow (the part you actually need while writing
stores).
Five documentation surfaces gained encryption awareness in this
sweep. Before this commit, the only place anyone could learn about
the at-rest encryption layer or the zero-knowledge opt-in was the
internal DATA_LAYER_AUDIT.md. New contributors and self-hosters
would never discover one of the most important features of the
product just by reading the standard onboarding docs.
apps/docs/src/content/docs/architecture/security.mdx (NEW)
----------------------------------------------------------
First-class user-facing security page in the Starlight site,
slotted into the Architecture sidebar between Authentication and
Backend.
Sections:
- What's encrypted (overview table of 27 modules + the
intentional plaintext carve-outs)
- Standard mode flow with ASCII diagram
- "What Mana CAN see" trust statements per mode
- Zero-knowledge mode setup walkthrough (Steps component)
- Unlock flow on a new device
- Recovery code rotation
- Deployment requirements (the loud MANA_AUTH_KEK warning)
- Audit trail action vocabulary
- Threat model summary table
- Implementation file references with paths
services/mana-auth/CLAUDE.md
----------------------------
New "Encryption Vault" section under Key Endpoints, listing all 7
routes (status, init, key, rotate, recovery-wrap GET+DELETE,
zero-knowledge) with their HTTP method, path, error codes, and a
description. Mentions the three CHECK constraints + RLS + audit
table. Points readers at DATA_LAYER_AUDIT.md and the new
security.mdx for the deep dive.
Environment Variables block gains MANA_AUTH_KEK with a multi-line
comment explaining the openssl rand command + dev fallback warning.
apps/mana/CLAUDE.md
-------------------
Full rewrite. The existing file was from the Supabase era and
described things like @supabase/ssr, safeGetSession(), and a
five-table schema with users + organizations + teams that doesn't
exist any more. Replaced with the unified-app architecture:
- Module system layout (collections.ts / queries.ts / stores/)
- Mana Auth (Better Auth + EdDSA JWT) instead of Supabase
- Local-first data layer with the full pipeline diagram
- At-rest encryption section with the "when writing module code
that touches sensitive fields" 4-step guide
- Updated routing structure (no more separate /organizations,
/teams routes)
- Module store pattern code example
- Reference document table at the bottom pointing at the audit,
the new security.mdx, and the auth doc
Root CLAUDE.md
--------------
New "At-Rest Encryption (Phase 1–9)" subsection under the
Local-First Architecture section. Two-mode trust summary table,
production requirement for MANA_AUTH_KEK with the openssl command,
the "when writing module code" 4-step guide, and a reference
table. New contributors reading the root CLAUDE.md from top to
bottom now hit encryption naturally as part of the data layer
discussion.
.env.macmini.example
--------------------
MANA_AUTH_KEK was missing from the production env example
entirely — the macmini deployment would silently boot on the
32-zero-byte dev fallback if you copied this file. Added with a
multi-paragraph comment covering: how to generate, why it's
required, how to store securely (Docker secrets / KMS / Vault),
and the rotation caveat.
apps/docs/src/content/docs/deployment/self-hosting.mdx
------------------------------------------------------
Two changes:
1. Added MANA_AUTH_KEK to the mana-auth service block in the
Compose example with an inline comment pointing at the new
section below.
2. New "Encryption Vault Setup" H2 section with subsections:
- Generating a KEK (with a fake example value labelled DO NOT
USE — generate your own)
- Securing the KEK (Docker secrets, KMS, systemd
LoadCredential, anti-patterns)
- "What if I lose the KEK?" — explains the data is
unrecoverable by design and mitigation via zero-knowledge
mode opt-in
- KEK rotation — calls out the missing background re-wrap
job as a known limitation
apps/docs/astro.config.mjs
--------------------------
Added "Security & Encryption" entry to the Architecture sidebar
between Authentication and Backend so the new page is reachable
from the docs nav.
Astro check: 0 errors, 0 warnings, 0 hints across 4 .astro files.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Marks the four backlog items closed in this session — vault service
integration tests, recovery code rotation, pre-wired insert helpers
for future server-pushed records, and boards/boardItems encryption.
Updates the encrypted-tables list to 27 tables.
Updates
-------
1. Sprint table grows by 4 rows (BL1, BL2, BL3+4, BL5) with the
four backlog commits.
2. Test-Status line bumped:
21 web test files → 21 web + 2 mana-auth
78 vitest crypto tests + 39 bun mana-auth tests
"25+ tables" → "27 tables" (boards + boardItems added)
3. Section 5 encrypted-tables list grows by:
- boards (name, description)
- boardItems (textContent, only when itemType === 'text')
Both labelled "9 BL" in the Phase column to mark them as
backlog-sweep additions.
4. "Tabellen ohne Encryption (bewusst)" subsection: removed the
stale "boards/boardItems are a candidate for later" entry —
they're encrypted now. Added a redirect note pointing readers
at Section 6 where the actual decision is recorded.
5. Section 6 ("Backlog") completely restructured. The flat
"in priority order" list became two subsections:
"Abgeschlossen (Phase 9 Follow-Up Sweep)" — table with the four
commits + a one-line "what" notice each. Item 3+4 is explicitly
marked as a re-frame: the original "server pushes plaintext"
risk turned out to overstate the problem because the
generate/upload UIs are TODO stubs. The fix was pre-wired
insert() helpers, not a server-side rewrite.
"Offen" — five remaining items, reordered:
1. File-Bytes-Encryption (NEW: surfaced as "#4b" while
documenting that filesStore.insert() only protects metadata)
2. Image-Generation / File-Upload Wire-Up (NEW: ensures the
future UIs go through the helpers from #3+4)
3. Conflict Visualization UI (unchanged)
4. Composite Indexes für Multi-Account (unchanged)
5. V3 Migration Tests (unchanged)
6. Eckdaten line bumped from "25+ Tabellen aktiv" to "27 Tabellen
aktiv". Best Practices line for ZK gets the "+ rotate im
Active-State-Support" suffix.
7. Last-update header bumped to today.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Closes backlog #5 from the Phase 9 audit. Adds two new registry
entries (boards, boardItems) and wraps the boards store + queries
+ search provider so the moodboard names, descriptions and
text-item content are sealed at rest like every other user-typed
field.
Registry
--------
- boards: ['name', 'description']
- boardItems: ['textContent']
Inline comments explain that textContent is only set when
itemType === 'text' (image-type items have it null, encryptRecord
is a pass-through). Coordinates / dimensions / z-index / opacity
stay plaintext for the canvas renderer.
Boards store
------------
- createBoard: snapshots plaintext for the return value before
encryptRecord mutates the row in place
- updateBoard: encrypts the diff before update, then re-fetches +
decrypts for the return value (so the caller gets plaintext,
not the ciphertext we just wrote)
- duplicateBoard: NEW behaviour — explicitly decrypts the
original board first because the duplicate concatenates "(Kopie)"
onto the name string. Concatenating onto a "enc:1:..." prefix
would produce a malformed blob that fails to decrypt later.
The board items are spread directly because the duplicate
uses the SAME master key, so the existing ciphertext stays
valid; encryptRecord is idempotent on already-encrypted strings
so it's a no-op safety check.
Reads
-----
- useAllBoards: decrypts the visible board set before mapping. The
item count map only reads structural fields (deletedAt + boardId)
so it doesn't need a decrypt pass for boardItems.
- allBoards$ raw observable: same pattern
- search/providers/picture: decrypts before substring scoring
against the user query
The unified mana app currently has no UI that renders boardItems
.textContent (the seed data in collections.ts is exported as
PICTURE_GUEST_SEED but never imported anywhere — dead code), so
no item-side reader needs touching for this commit. When a future
canvas editor lands it'll go through the existing decryptRecord
helpers naturally.
78/78 crypto tests still pass (registry shape unchanged at the API
level).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Closes backlog #3+4 from the Phase 9 audit. The original framing —
"server-pushed records bypass client-side encryption" — turned out
to overstate the problem after a code audit:
- apps/mana/apps/web/src/routes/(app)/picture/generate/+page.svelte
is currently a TODO stub. The handleGenerate() function returns
"requires connection to Picture-Server (port 3006)" without
inserting anything.
- There is no fileTable.add() call site anywhere in the unified
mana app. File uploads still happen via the standalone storage
server in apps/storage and arrive via legacy mana-sync push.
So the production code path that would write plaintext images or
files to the user's IndexedDB doesn't yet exist. The risk only
materialises when someone wires up the in-app generate / upload
UI in the unified app.
The right action is to leave behind a clearly-labelled, encryption-
aware insert() helper on each store so the future implementation
has an obvious "do the right thing" path to call. This commit does
exactly that.
picture/stores/images.svelte.ts
-------------------------------
New imagesStore.insert(image: LocalImage) method:
- Calls encryptRecord('images', image) to seal `prompt` +
`negativePrompt` (the two registered encrypted fields)
- Calls imageTable().add(image)
- Fires the PictureEvents.imageCreated analytic (replaces the
old plain-table-add path)
A long doc comment on the method explains the architectural
reasoning: the server cannot encrypt under the user's master key
(the key only lives in the browser), so the generation flow MUST
round-trip through the client store even if the AI call itself
happens server-side. The pattern is documented as:
1. Client posts { prompt, negativePrompt, ... } to image-gen API
2. Server returns { storagePath, generationId, dimensions, ... }
3. Client calls imagesStore.insert(...) with both halves
4. encryptRecord seals the prompt fields before the IndexedDB write
The mixed-state guarantee from picture/queries.ts already covers
the migration window where some images came in via legacy
server-side push and others through this path — decryptRecord
passes plaintext through and unwraps ciphertext blobs.
storage/stores/files.svelte.ts
------------------------------
New filesStore.insert(file: LocalFile) method:
- Calls encryptRecord('files', file) to seal `name` +
`originalName`
- Calls fileTable.add(file)
Same architectural reasoning applies. The doc comment also flags a
SEPARATE concern that this commit does NOT address: encrypting the
actual file *bytes* on S3 (so the storage provider can't read the
content) needs streaming AES-GCM and is a much bigger lift. Tracked
as "backlog #4b" in the comment for whoever picks it up next.
(No analytic call yet on the storage side because StorageEvents
doesn't have a fileUploaded() event — the upload UI is unbuilt, so
adding the analytic event is up to whoever lands the UI.)
Pre-existing TS error on line 46 of images.svelte.ts (the
`toggleField(imageTable(), ...)` Drizzle/Dexie type variance bug)
is unchanged — it predates Phase 9 and is not introduced by this
commit.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Closes backlog #2 from the Phase 9 audit. Lets a user replace their
recovery code without going through the disable→generate→re-enable
dance. Works in BOTH standard and zero-knowledge modes.
vault-client
------------
New rotateRecoveryCode() method on the VaultClient interface.
Returns RecoveryCodeSetupResult, identical shape to setupRecoveryCode.
Branches on the current vault state via getStatus():
Standard mode:
Re-fetches the plaintext MK from the server (same path as the
initial setupRecoveryCode), generates a fresh 32-byte recovery
secret, derives the new wrap key via HKDF, seals the MK, posts
the wrap to /recovery-wrap (idempotent server-side, replaces
the existing row in place).
Zero-knowledge mode:
Server can't hand out the plaintext MK any more, so we use the
cachedUnwrappedMkBytes that unlockWithRecoveryCode stashed when
the user typed in their old recovery code earlier this session.
Throws with a clear message if the cache is empty (e.g. user
landed on the page via init rather than recovery-unlock):
"sign out and back in with your current recovery code first"
so the cache gets repopulated.
Both branches:
- Wipe the raw MK reference after sealing
- Wipe the recovery secret after format
- Return the formatted code for the UI to display
The OLD recovery code is now permanently invalid. Using it on a
future unlock attempt will fail with the standard generic
"wrong recovery code" error.
Settings UI
-----------
New rotateStep state machine ('idle' / 'rotated') runs alongside
the existing zkSetupStep so the user can rotate without leaving the
active-state UI.
In the active-mode card (zkSetupStep === 'enabled'):
- Two side-by-side buttons:
"🔁 Recovery-Code rotieren" + "Zero-Knowledge-Modus wieder deaktivieren …"
- When the user clicks rotate, handleRotateRecoveryCode() runs the
flow and renders an inline "Neuer Recovery-Code" subsection
(same .recovery-code monospace block + Copy button as the
initial setup) with explicit warning that the old code is now
invalid.
- "Ich habe den neuen Code gesichert" button wipes the displayed
code and drops back to idle.
- The disable flow stays available (the rotate UI hides itself
when the user has clicked into the disable confirmation path).
The 28 vault integration tests still pass (39 total in
encryption-vault/, including the existing 11 KEK tests). The new
rotateRecoveryCode method reuses the already-tested
setRecoveryWrap server endpoint, so no new server-side tests are
needed for this commit.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
