Compare commits

...

12 Commits

Author SHA1 Message Date
39e448e807 Merge pull request 'feat/r42-accuracy-and-clear-history' (#34) from feat/r42-accuracy-and-clear-history into main
All checks were successful
On Push to Main / test (push) Successful in 34s
On Push to Main / publish (push) Successful in 1m10s
On Push to Main / deploy-dev (push) Successful in 3m10s
Reviewed-on: #34
2026-07-13 12:42:12 +00:00
RaymondVerhoef
17cb30003d feat: add guide for calculating and processing pension premiums in proforma payslips
All checks were successful
On Pull Request to Main / test (pull_request) Successful in 44s
On Pull Request to Main / publish (pull_request) Successful in 1m14s
On Pull Request to Main / deploy-dev (pull_request) Successful in 3m15s
2026-07-13 14:40:43 +02:00
RaymondVerhoef
85452f66a7 feat(r42): improve KB grounding accuracy and add clear-history
R42 was missing knowledge-graph information (e.g. pension questions)
because retrieval and context-building dropped relevant facts:

- retrieval: exact-token TF-IDF could not match Dutch compound words,
  so a "pensioen" query scored 0 against "pensioenregeling" /
  "partnerpensioen" and never retrieved them. Add a compound-word
  fallback (shared >=6-char stem or containment, 0.4x weight) alongside
  exact matching.
- rag: deep article content was only injected for verbatim-mentioned
  topics; retrieved topics contributed just a 200-char description.
  Inject ~1000 chars of content for up to 5 topics (mentions first,
  then top-ranked retrieved) and widen the description snippet to 320.
- prompts: add a NAUWKEURIGHEID block (use all relevant facts, call
  lookup_topic before giving up) and relax the 4-sentence cap for
  detail/list answers so complete facts aren't summarised away.

Also add a clear-history control: a trash button in the chat header
(confirm dialog) wipes chat🧵{userId} and reseeds the greeting
via clearThread() in useChat.

Tests: compound-word matching + rag deep-content injection. Spec updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-13 14:25:08 +02:00
e73700763d Merge pull request 'Fix #32: gestringifieerde arrays in onboarding-overview output normaliseren' (#33) from fix/issue-32-onboarding-schema into main
All checks were successful
On Push to Main / test (push) Successful in 35s
On Push to Main / publish (push) Successful in 1m12s
On Push to Main / deploy-dev (push) Successful in 3m15s
Reviewed-on: #33
2026-07-13 07:46:53 +00:00
RaymondVerhoef
182fb2357f fix: coerce stringified arrays in onboarding overview output (#32)
All checks were successful
On Pull Request to Main / test (pull_request) Successful in 41s
On Pull Request to Main / publish (pull_request) Successful in 1m15s
On Pull Request to Main / deploy-dev (pull_request) Successful in 3m17s
Opening an onboarding theme could fail with "LLM output failed schema
validation": the fast-tier model occasionally emits the key_points and
topics_covered tool fields as JSON-encoded strings instead of real
arrays. Same failure class the quiz schema already guards against with
z.preprocess.

- onboardingOverviewSchema: parse "[...]" strings back to arrays before
  validation; split a bullet/newline string of key points as a fallback;
  keep the first 5 key points on overage instead of failing the user.
  Genuinely bad output (too few points, non-JSON strings, wrong types)
  still fails validation.
- emit_onboarding_overview tool: descriptions now state the fields must
  be JSON arrays, never strings (prompt-side nudge).
- 5 regression tests reproducing the exact #32 payload shapes
  (npm test 139/139).

Closes #32

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-13 09:38:46 +02:00
b1d3686d35 Merge pull request 'Onboarding-track: 5-daagse thema-introductie vanuit de "New here?"-kaart (#30)' (#31) from feat/onboarding-track into main
All checks were successful
On Push to Main / test (push) Successful in 34s
On Push to Main / publish (push) Successful in 1m11s
On Push to Main / deploy-dev (push) Successful in 3m19s
Reviewed-on: #31
2026-07-13 07:20:00 +00:00
RaymondVerhoef
b1330ebe3e ci: validate the production Caddyfile inside the built image (#26)
All checks were successful
On Pull Request to Main / test (pull_request) Successful in 41s
On Pull Request to Main / publish (pull_request) Successful in 1m10s
On Pull Request to Main / deploy-dev (pull_request) Successful in 3m15s
The caddy-validate step added for #26 (and merged to main via #29) used a
`-v "$PWD":/workspace` bind mount, which does not propagate to the sibling
Docker daemon in the Gitea act runner — the container saw an empty dir and
every PR failed with "open Caddyfile: no such file or directory".

Validate instead inside the freshly built image, where the production
Caddyfile lives at /etc/caddy/Caddyfile: no bind mount, and it checks the
exact file that ships. Caddyfile.test keeps being covered by the homepage
smoke test (its container must serve 200).

Closes #26

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-13 09:13:12 +02:00
RaymondVerhoef
5214c9db3b feat: 5-day theme-level onboarding track from the "New here?" card (#30)
Some checks failed
On Pull Request to Main / test (pull_request) Failing after 24s
On Pull Request to Main / publish (pull_request) Has been skipped
On Pull Request to Main / deploy-dev (pull_request) Has been skipped
A self-paced onboarding track that introduces a new employee to every KB
theme in breadth (not depth), so they grasp how Respellion works day to
day and week to week. Offered as a CTA inside the Dashboard "New here?"
explainer card; always available regardless of enrollment.

Design:
- Theme is the trackable unit; the 5 "days" are a read-time presentation
  grouping, so re-chunking never loses progress. Completion is stored per
  theme in onboarding_completions.
- Per-theme overview generated lazily on first open (fast-tier
  emit_onboarding_overview tool), cached in onboarding_overviews keyed by
  theme + a topics_fingerprint that triggers regeneration when the theme's
  topic set changes.
- Reachable via /onboarding-track using the existing skipEnrollmentGate
  prop, decoupled from the 26-week curriculum (distinct from /onboarding,
  the enrollment page).

Backend:
- pb_migrations/1781200000_created_onboarding.js: two collections with
  authenticated-only rules and unique indexes; TEXT team_member_id (no
  relation) per the post-#18/#27 convention. Mirrored in
  scripts/setup-pb-collections.mjs.
- src/lib/onboardingService.js: pure helpers (orderThemes,
  distributeThemesIntoDays, computeTopicsFingerprint,
  computeOnboardingProgress, buildOnboardingPlan) + generation + I/O.
- db.js onboarding helpers use pb.filter() bindings (theme is free text).
- LLM tool + Zod schema + registry + simulation stub.

Frontend:
- src/pages/OnboardingTrack.jsx (day list, per-theme overview, completion
  banner, progress ring/day bar).
- Dashboard "New here?" card CTA + X/5-days progress chip (hidden when the
  KB has no themes).

Docs: data-model, generation-spec (§D), frontend-spec updated.

Verified: 22 new unit tests (npm test 134/134), eslint clean on changed
files, npm run build OK, PocketBase v0.30.4 boot applies the migration
(collections + unique indexes + authed rules confirmed), and a backend
contract check (upsert idempotency, unique-index guard, special-char
theme filtering).

Closes #30

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-13 09:08:38 +02:00
48715df147 Merge pull request 'Fix #27: TeamManager-rosterbeheer, role-escalatie gedicht, escalate-only resync' (#29) from fix/issue-27-teammanager-admin-rules into main
Some checks failed
On Push to Main / test (push) Failing after 25s
On Push to Main / publish (push) Has been skipped
On Push to Main / deploy-dev (push) Has been skipped
Reviewed-on: #29
2026-07-12 19:04:43 +00:00
85204938df Merge pull request 'CI: caddy validate van beide Caddyfiles in de test-job' (#28) from fix/issue-26-ci-caddy-validate into main
Some checks failed
On Push to Main / publish (push) Has been cancelled
On Push to Main / deploy-dev (push) Has been cancelled
On Push to Main / test (push) Has been cancelled
Reviewed-on: #28
2026-07-12 19:04:29 +00:00
RaymondVerhoef
cbce4555ff fix: TeamManager admin rules, close role self-escalation, escalate-only resync (#27)
Some checks failed
On Pull Request to Main / test (pull_request) Failing after 24s
On Pull Request to Main / publish (pull_request) Has been skipped
On Pull Request to Main / deploy-dev (pull_request) Has been skipped
TeamManager could not manage the roster: updateRule allowed self-update
only and deleteRule was superuser-only. The same self-update rule also
left a privilege escalation open — it did not restrict fields, so any
authenticated user could PATCH their own role to "admin".

- team_members rules (migration 1781000003 + base migration + setup
  script mirror):
    update: (@request.auth.id = id && @request.body.role:isset = false)
            || @request.auth.role = "admin"
    delete: @request.auth.role = "admin"
  Self-service onboarding keeps working (it never touches role); the
  role field is admin-only; deletion is admin-only.
- pb_hooks/team_members.pb.js: the ENTRA_ADMIN_EMAILS resync is now
  escalate-only — it guarantees admin for allow-list members on every
  login but no longer demotes, otherwise every TeamManager promotion
  would revert on the member's next sign-in (ADR-004 amended).
- TeamManager.jsx: info text updated to the new behaviour.

Verified with a two-identity mock-OIDC matrix (11/11): allow-list vs
regular login, self-escalation blocked while onboarding self-update
still works, cross-user update/delete blocked, admin promote/demote/
delete work, promotion survives the member's next login, allow-list
admin stays admin. Regression: #22 matrix 9/9, #24 matrix 8/8, #18
harness 15/15, npm test 112/112, eslint clean.

Closes #27

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-12 21:01:28 +02:00
8decdd454f Merge pull request 'Fix #22: OAuth2 sign-up toegestaan via @request.context-rule (eerste Microsoft-login werkt weer)' (#23) from fix/issue-22-oauth2-signup-rule into main
All checks were successful
On Push to Main / test (push) Successful in 33s
On Push to Main / publish (push) Successful in 1m4s
On Push to Main / deploy-dev (push) Successful in 3m3s
Reviewed-on: #23
2026-07-12 15:11:24 +00:00
31 changed files with 1679 additions and 61 deletions

View File

@@ -11,21 +11,25 @@ jobs:
- name: Checkout
uses: actions/checkout@v4
# The test image below swaps in Caddyfile.test, so the production
# Caddyfile is otherwise never exercised by CI — an invalid config then
# only surfaces when the deploy health-gate fails, after the broken
# container already replaced the healthy one (issue #20/#26). Validate
# both files up front so a parse error fails the PR check in seconds.
- name: Validate Caddyfiles
run: |
docker run --rm -v "$PWD":/workspace -w /workspace caddy:2-alpine \
caddy validate --config Caddyfile --adapter caddyfile
docker run --rm -v "$PWD":/workspace -w /workspace caddy:2-alpine \
caddy validate --config Caddyfile.test --adapter caddyfile
- name: Build Docker image
run: docker build -t learning-platform .
# The production Caddyfile is otherwise never exercised by CI — the test
# image below swaps in Caddyfile.test — so a parse error only surfaced at
# deploy time, after the broken container had already replaced the healthy
# one (issue #20/#26). Validate it here, inside the freshly built image
# where it lives at /etc/caddy/Caddyfile. This runs caddy against the
# exact file that ships and needs no bind mount — a `-v "$PWD":...` mount
# does not propagate to the sibling Docker daemon in the Gitea runner,
# which is why the earlier mount-based step failed with
# "open Caddyfile: no such file or directory" (#26).
# Caddyfile.test needs no separate check: the "Test homepage" step below
# runs the test image and fails if that config can't serve.
- name: Validate production Caddyfile
run: |
docker run --rm --entrypoint caddy learning-platform \
validate --adapter caddyfile --config /etc/caddy/Caddyfile
- name: Build test image
run: |
cat > Dockerfile.test <<'EOF'

View File

@@ -39,13 +39,14 @@ Browser (SPA) PocketBase (auth) Entra ID
| 001 | OIDC-provider tegen Entra v2-endpoints i.p.v. eigen MSAL-flow of header-trust | Sluit aan op "PocketBase = enige backend"; secret server-side; auto-provisioning gratis; werkt los van de perimeter-implementatie |
| 002 | `team_members` hergebruikt als auth-collection (zelfde naam) | Alle `user_id`-referenties blijven werken; minimale blast-radius |
| 003 | Provisioning + admin-rol via `pb_hooks/team_members.pb.js` | Rol-logica server-side, niet in de client |
| 004 | Admin-rol via e-mail-allowlist (`ENTRA_ADMIN_EMAILS`), re-synced elke login | Geen handmatig beheer; allowlist-wijziging werkt bij volgende login |
| 004 | Admin-rol via e-mail-allowlist (`ENTRA_ADMIN_EMAILS`), **escalate-only** bij elke login (aangepast in #27) | Allow-list garandeert admin (promotie werkt bij volgende login) maar demoteert niet meer — anders zou elke TeamManager-promotie bij de volgende login worden teruggedraaid. Demotie via TeamManager; allow-list-leden demoteer je door ze van de lijst te halen |
| 005 | API-rules → `@request.auth.id != ""` op alle app-collecties | Geen anonieme toegang meer; admins/users zijn beide geauthenticeerd |
| 006 | Baseline-ledger-sync migratie voor out-of-band geprovisionede DB's | Labs/prod draaide historisch zonder `--migrationsDir`; replay van de historie crashte PB (issue #18). De vroegst-sorterende migratie markeert de historie als applied wanneer schema bestaat maar de ledger leeg is |
| 007 | Migratie = structuur, hook = configuratie | De OIDC-provider wordt bij elke start (en per cron-minuut) gereconcilieerd vanuit `ENTRA_*` env door `pb_hooks/entra_oidc.pb.js`; een one-shot migratie die van deploy-time env afhangt kan OAuth2 anders permanent uitschakelen |
| 008 | Hook-helpers via `require(`${__hooks}/utils.js`)` | De JSVM draait elke hook-callback als geïsoleerd programma; top-level functies zijn níet in scope op call-time |
| 009 | `createRule: '@request.context = "oauth2"'` op `team_members` | PocketBase past de createRule toe op de OAuth2-sign-up (eerste login maakt het record aan); `null` blokkeerde élke eerste login met 403 (issue #22). De context-rule staat alléén de OAuth2-flow toe — anonieme REST-creates blijven geweigerd |
| 010 | Claims uit het Entra **id_token** (`userInfoURL: ""`) + UPN-fallback voor e-mail | Graph `/oidc/userinfo` geeft géén `email`-claim voor accounts zonder mail-attribuut → 400 bij eerste login (issue #24). Het id_token bevat altijd `preferred_username` (UPN = primair e-mailadres bij Respellion); `entra_oidc.pb.js` gebruikt die als fallback (regex-guard tegen guest-UPN's) en logt gefaalde OAuth2-pogingen naar de containerlog |
| 011 | `updateRule: '(@request.auth.id = id && @request.body.role:isset = false) \|\| @request.auth.role = "admin"'`, `deleteRule: '@request.auth.role = "admin"'` | TeamManager-rosterbeheer werkt nu voor app-admins (issue #27). De `role:isset`-guard sluit tegelijk de privilege-escalatie waarbij een gebruiker zijn eigen `role` naar admin kon patchen; onboarding raakt `role` niet en blijft self-service |
## Bestanden
@@ -55,6 +56,7 @@ Browser (SPA) PocketBase (auth) Entra ID
| `pb_migrations/1781000000_team_members_to_auth.js` | Converteert relation-velden naar text (data blijft behouden), dropt de oude PIN-collection en maakt `team_members` als auth-collection. Idempotent; provider wordt alleen als fast-path meegenomen als de env al aanwezig is. |
| `pb_migrations/1781000001_tighten_api_rules.js` | Zet alle niet-systeem-collecties op `@request.auth.id != ""`. |
| `pb_migrations/1781000002_allow_oauth2_signup.js` | Zet `createRule = '@request.context = "oauth2"'` op reeds-gemigreerde omgevingen (issue #22). |
| `pb_migrations/1781000003_team_members_admin_rules.js` | Admin-rosterbeheer + role-escalatie-guard op reeds-gemigreerde omgevingen (issue #27). |
| `pb_hooks/entra_oidc.pb.js` | Reconcilieert de OIDC-provider vanuit `ENTRA_*` env bij elke start + cron-tick (compare-before-save). |
| `pb_hooks/utils.js` | Gedeelde helpers (`resolveRole`, `reconcileEntraOidc`) — per callback binnenhalen via `require()`. |
| `pb_hooks/team_members.pb.js` | Auto-provisioning (`role`, `enrollment_status`, naam-fallback) + admin-rol re-sync. |

View File

@@ -212,6 +212,34 @@ Best-effort telemetry for every Anthropic call (written by `callLLM`).
---
### `onboarding_overviews`
Cached, breadth-first per-theme overview for the onboarding track (issue #30).
One row per theme, shared across all users. Generated lazily on first open.
| Field | Type | Notes |
|---|---|---|
| theme | text | canonical theme name (from `topics.theme`); **unique** |
| content | json | validated `onboardingOverviewSchema` payload (title, what_it_is, why_it_matters, key_points, topics_covered) |
| topics_fingerprint | text | stable hash of the theme's sorted topic ids; a mismatch triggers regeneration |
Unique index on `(theme)`. Rules: authenticated-only (`@request.auth.id != ""`).
---
### `onboarding_completions`
Per-user, per-theme completion marker for the onboarding track (issue #30).
| Field | Type | Notes |
|---|---|---|
| team_member_id | text | the employee — **plain text**, not a relation (admins can delete members; orphan rows are harmless) |
| theme | text | the theme marked complete |
Unique index on `(team_member_id, theme)` → idempotent. Rules: authenticated-only.
Completion is tracked per **theme**, not per day; the 5 "days" are a read-time
presentation grouping, so re-chunking never loses progress.
---
## Dropped / legacy collections
These existed in earlier iterations and have been removed. Their `db.js` helpers

View File

@@ -10,8 +10,9 @@ A React 19 SPA built with Vite, routed by React Router 7. Entry: `src/main.jsx`
| Route | Screen | Access |
|---|---|---|
| `/login` | `Login.jsx` | public |
| `/onboarding` | `Onboarding.jsx` | logged-in, not yet enrolled |
| `/onboarding` | `Onboarding.jsx` | logged-in, not yet enrolled (enrollment page) |
| `/` | `Dashboard.jsx` | enrolled user |
| `/onboarding-track` | `OnboardingTrack.jsx` | any logged-in user, **any enrollment** (`skipEnrollmentGate`) |
| `/learn` | `Leren.jsx` | enrolled user |
| `/test` | `Testen.jsx` | enrolled user |
| `/leaderboard` | `Leaderboard.jsx` | enrolled user |
@@ -20,9 +21,14 @@ A React 19 SPA built with Vite, routed by React Router 7. Entry: `src/main.jsx`
`ProtectedRoute`:
- redirects to `/login` if not authenticated;
- redirects to `/onboarding` if `enrollment_status !== 'active'`**except** admins
heading to the admin panel, who are exempt;
heading to the admin panel, and routes passing `skipEnrollmentGate` (the onboarding
track), which are exempt;
- enforces `requireAdmin` for `/admin`.
> `/onboarding` (enrollment) and `/onboarding-track` (the 5-day theme intro) are
> distinct. The track is decoupled from enrollment on purpose so it also works as a
> refresher for existing staff.
Navigation chrome (top bar + mobile bottom nav) is rendered by `ProtectedRoute`.
`ChatLauncher` (R42) is mounted globally.
@@ -51,7 +57,17 @@ enrolled are redirected to `/`. See `docs/curriculum-spec.md`.
## Employee screens
- **Dashboard** — current cycle/week, assigned topic, cycle progress ring, quick
links to Learn and Test, mini leaderboard, recent activity.
links to Learn and Test, mini leaderboard, recent activity. The dismissible
"New here?" explainer card also carries the **onboarding-track CTA** (a progress
chip `X/5 days` + a Start/Continue/Review button linking to `/onboarding-track`),
shown only when the KB has themes to introduce.
- **Onboarding Track (`/onboarding-track`)** — a self-paced, breadth-first tour of
every theme, grouped into ≤5 day cards (`OnboardingTrack.jsx`). Opening a theme
lazily generates a short overview (what it is / why it matters for daily & weekly
work / key points / topics) and offers "Mark as done"; a progress ring + day bar
track completion, and a completion banner shows when every theme is done. Available
to any logged-in user regardless of enrollment; completion is stored per theme in
`onboarding_completions`. See `docs/generation-spec.md §D`.
- **Learning Station (`/learn`)** — the week's required topic + the rest of the
knowledge library; opening a topic shows the micro-learning selector
(`src/components/micro_learning/`). Completing ≥1 micro-learning marks the week done.

View File

@@ -74,6 +74,29 @@ explanation, difficulty }`.
---
## D. Onboarding overviews — `src/lib/onboardingService.js`
Powers the 5-day onboarding track (issue #30): a short, breadth-first overview of
**one theme** for a brand-new employee — deliberately light, not a deep lesson.
- **Tool:** `emit_onboarding_overview` · **tier:** `fast` (Haiku) · `maxTokens: 1500`,
60s timeout · schema `onboardingOverviewSchema`.
- **Shape:** `{ title, what_it_is, why_it_matters, key_points[35], topics_covered[{topic_id,label}] }`.
`why_it_matters` is framed around day-to-day / week-to-week work at Respellion.
- **Cache:** `onboarding_overviews`, one row per theme, keyed by theme name plus a
`topics_fingerprint` (stable hash of the theme's sorted topic ids).
`getOrGenerateOnboardingOverview(theme, topics, {force})` returns the cached row when
the fingerprint matches; a mismatch (topic added/removed) or `force` regenerates.
- **Simulation:** `emit_onboarding_overview` has a stub in `SIMULATION_TOOL_STUBS`.
Theme ordering + day grouping are pure helpers in the same module
(`orderThemes`, `distributeThemesIntoDays`, `computeTopicsFingerprint`,
`computeOnboardingProgress`, `buildOnboardingPlan`), unit-tested in
`src/lib/__tests__/onboardingService.test.js`. Completion is recorded per **theme**
in `onboarding_completions` (`{ team_member_id, theme }`), not per day.
---
## Shared infrastructure (`src/lib/llm.js`)
- **Tiers:** `fast` (Haiku 4.5), `standard` (Sonnet 4.6), `reasoning` (Opus 4.7);

View File

@@ -18,6 +18,8 @@ client-side and is grounded by local TF-IDF retrieval — **no vector database**
context is truncated with a notice.
- A greeting message seeds an empty thread.
- Each turn calls `callLLM` (fast/standard Claude tier — low latency matters for chat).
- The chat header has a **clear** button (trash icon). It confirms, then wipes
`chat:thread:{userId}` and reseeds the greeting via `clearThread` in `useChat.js`.
---
@@ -25,13 +27,21 @@ client-side and is grounded by local TF-IDF retrieval — **no vector database**
`buildKbContext` in `rag.js`:
1. Build / reuse the TF-IDF index over `topics` (`src/lib/retrieval.js`).
2. Retrieve the top **10** topics for the user's message.
2. Retrieve the top **10** topics for the user's message. Scoring is exact-token
TF-IDF **plus a compound-word fallback**: an unmatched query token (≥6 chars)
also matches a document term when they share a ≥6-char stem or one contains
the other, at a reduced weight. This recovers Dutch compounds — e.g. a
`pensioen` query matches `pensioenregeling` and `partnerpensioen`.
3. Always include topics whose `id` or `label` appears verbatim in the message.
4. Include relations only when **both** endpoints are in the retrieved set.
5. For explicitly mentioned topics, inject up to ~1200 chars of their generated
content.
5. Inject up to ~1000 chars of generated content for up to **5** topics —
verbatim-mentioned first, then the highest-ranked retrieved ones — so a query
that never names a topic exactly still gets rich content for what it matched.
6. Append a short KB hash so the cached context busts when topics change.
If the summarised context is still too thin, R42 can call the `lookup_topic`
tool to pull a topic's full description and learning content on demand.
The system prompt (`prompts.js`) is assembled as cacheable blocks: a stable
preamble (role, tasks, style, "answer only from the KB"), the KB context block, and
a per-turn tail with the user's name and admin/non-admin flag.

View File

@@ -0,0 +1,118 @@
# pensioenpremie berekenen en verwerken in een proforma-loonstrook
## doel
Handleiding voor het handmatig berekenen en verwerken van de pensioenpremie bij het
opstellen van een proforma-salarisspecificatie in NMBRS, wanneer de pensioengrondslag
niet automatisch op de proforma-berekening verschijnt.
## wanneer van toepassing
Bij het opstellen van een proforma-loonstrook (bijvoorbeeld ter voorbereiding van een
aanbod aan een kandidaat) toont NMBRS geen automatische pensioenpremie-regel. Deze moet
dan handmatig worden toegevoegd volgens onderstaande methode.
## benodigde gegevens vooraf
Voordat je begint moet je de volgende vier parameters hebben. Geen van deze staat op de
proforma zelf.
| parameter | waar te vinden |
| --- | --- |
| pensioenfranchise (Respellion / a.s.r. Doen Pensioen) | `pension-scheme-and-benefits.md` of het a.s.r.-polisblad; kan afwijken van het wettelijk minimum |
| verdeelsleutel werkgever/werknemer | `pension-scheme-and-benefits.md` of het a.s.r.-contract |
| definitie pensioengevend salaris | a.s.r.-polisvoorwaarden — bevestig welke looncomponenten meetellen (onkostenvergoedingen in elk geval niet) |
| fiscaal maximum pensioengevend salaris (aftoppingsgrens) | jaarlijkse publicatie belastingdienst/SZW, zie tabel hieronder |
Deze vier punten waren bij het opstellen van dit document nog niet allemaal bevestigd
voor Respellion specifiek — zie "openstaande vragen" onderaan.
## rekenstappen
1. **pensioengevend salaris (jaarbasis) bepalen**
Standaard: 12 × maandsalaris + vakantietoeslag (8%).
`pensioengevend salaris = 12 × bruto maandsalaris × 1,08`
Onkostenvergoedingen (reiskosten, telefoonvergoeding) tellen niet mee.
2. **aftoppen op het fiscaal maximum**
`gemaximeerd pensioengevend salaris = min(pensioengevend salaris, aftoppingsgrens)`
3. **franchise aftrekken**
`pensioengrondslag = gemaximeerd pensioengevend salaris franchise`
4. **parttimefactor toepassen**
`pensioengrondslag (parttime) = pensioengrondslag × parttimefactor`
Parttimefactor = contracturen / 40 (of de fulltime-norm die Respellion hanteert).
5. **premie berekenen**
`jaarpremie = pensioengrondslag (parttime) × 5%`
`maandpremie = jaarpremie / 12`
6. **verdelen werkgever/werknemer**
`werknemersdeel = maandpremie × werknemerspercentage`
`werkgeversdeel = maandpremie werknemersdeel`
## fiscale kernbedragen 2026
| bedrag | waarde 2026 | bron |
| --- | --- | --- |
| aftoppingsgrens pensioengevend salaris | € 137.800 | belastingdienst/SZW, voorlopige bedragen 2026 |
| minimale AOW-franchise (Wtp-regelingen / beschikbare premieregelingen) | € 19.172 | belastingdienst/SZW, voorlopige bedragen 2026 |
Deze twee bedragen worden jaarlijks (eind december) opnieuw gepubliceerd. Controleer bij
elke proforma-berekening of de bedragen nog actueel zijn.
## verwerking in NMBRS — fiscale nuance
Het werknemersdeel van de pensioenpremie is een pre-tax inhouding:
- het verlaagt de grondslag voor de loonheffing (de kolom "Tabel" op de proforma-uitdraai)
- het verlaagt **niet** het SV-loon (de kolom "SVW" blijft ongewijzigd)
Trek het werknemersdeel dus niet rechtstreeks af van het bestaande netto-bedrag. Voer de
premie in met dezelfde paycode die in de reguliere (niet-proforma) loonrun voor
pensioenpremie werknemer/werkgever is ingericht — controleer de codeconfiguratie in
NMBRS (niet de bedragen) op de vlag "aftrekbaar voor loonheffing, niet voor SVW". Laat
NMBRS de loonheffing zelf herberekenen over de verlaagde grondslag; reken dit niet
handmatig na.
Het werkgeversdeel is een werkgeverslast (vergelijkbaar met de bestaande WGA/Aof-regels
op de proforma) en raakt het netto van de werknemer niet, wel de totale loonkosten.
## rekenvoorbeeld
Uitgangspunt: proforma fulltime (40 uur/week), bruto maandsalaris € 6.700, 100%
dienstverband.
1. Pensioengevend salaris (jaar): 12 × 6.700 × 1,08 = € 86.832
2. Aftopping: 86.832 < 137.800 → geen aftopping
3. Franchise (voorbeeld met het wettelijk minimum € 19.172 — niet bevestigd als
Respellion-waarde): 86.832 19.172 = € 67.660
4. Parttimefactor: 100% → geen aanpassing
5. Premie: 67.660 × 5% = € 3.383 per jaar = € 281,92 per maand
6. Verdeling (voorbeeld 50/50 — verdeelsleutel niet bevestigd): € 140,96 werknemer,
€ 140,96 werkgever
Dit rekenvoorbeeld gebruikt aannames voor franchise en verdeelsleutel die nog niet
bevestigd zijn voor Respellion — zie "openstaande vragen".
## openstaande vragen
- Exacte pensioenfranchise van de Respellion a.s.r. Doen Pensioen-regeling (mogelijk
hoger dan het wettelijk minimum van € 19.172).
- Verdeelsleutel werkgever/werknemer voor de 5% premie.
- Of vakantietoeslag standaard meetelt in het pensioengevend salaris volgens de
a.s.r.-polis, of dat er een afwijkende definitie geldt.
- Exacte NMBRS-paycode voor pensioenpremie werknemer/werkgever (te vinden in een
bestaande productierun of de paycode-lijst).
Zodra deze vier punten bevestigd zijn, kan dit document worden aangevuld met de
definitieve waarden in plaats van de wettelijke minimum-aannames.
## bronnen
- Proforma-salarisspecificatie Respellion B.V. (voorbeeldcasus, dit document)
- Belastingdienst/SZW: voorlopige bedragen AOW-franchise en maximum pensioengevend
salaris 2026
- `pension-scheme-and-benefits.md` (Respellion-handboek) — te raadplegen voor de
resterende openstaande vragen

View File

@@ -41,14 +41,17 @@ onRecordCreate(function (e) {
e.next();
}, "team_members");
// On every successful auth (incl. OIDC): keep the admin role in sync with the
// allow-list, so promoting/demoting an admin only requires an env change.
// On every successful auth (incl. OIDC): guarantee the admin role for
// allow-list members. ESCALATE-ONLY (#27): the allow-list grants admin, it no
// longer demotes — otherwise every role promotion made through TeamManager
// would be reverted on the member's next login. Demotion runs through
// TeamManager; allow-list members cannot be demoted (the list wins on their
// next login).
onRecordAuthRequest(function (e) {
const { resolveRole } = require(`${__hooks}/utils.js`);
const email = e.record.get("email") || "";
const want = resolveRole(email);
if (e.record.get("role") !== want) {
e.record.set("role", want);
if (resolveRole(email) === "admin" && e.record.get("role") !== "admin") {
e.record.set("role", "admin");
e.app.save(e.record);
}
e.next();

View File

@@ -151,8 +151,10 @@ migrate((app) => {
listRule: '@request.auth.id != ""',
viewRule: '@request.auth.id != ""',
createRule: '@request.context = "oauth2"',
updateRule: '@request.auth.id = id',
deleteRule: null,
// Self-update may not touch `role` (closes self-service privilege
// escalation); admins manage the roster incl. roles and deletion (#27).
updateRule: '(@request.auth.id = id && @request.body.role:isset = false) || @request.auth.role = "admin"',
deleteRule: '@request.auth.role = "admin"',
authRule: "",
manageRule: null,

View File

@@ -0,0 +1,52 @@
/// <reference path="../pb_data/types.d.ts" />
//
// Issue #27 — TeamManager roster management + close role self-escalation.
//
// The collection shipped with `updateRule: '@request.auth.id = id'` and
// `deleteRule: null`. That broke the TeamManager admin panel (admins could
// not change roles or remove members) AND left a privilege escalation open:
// the self-update rule did not restrict fields, so any authenticated user
// could PATCH their own `role` to "admin".
//
// New rules:
// update — a user may update their own record but NOT the `role` field
// (onboarding only touches enrollment fields); admins may update
// anyone, including `role`.
// delete — admins only.
//
// Environments that have not applied 1781000000 yet get these rules directly
// from that (updated) migration; this follow-up exists for databases where it
// already ran (Labs). Applying both is harmless (same values).
const UPDATE_RULE = '(@request.auth.id = id && @request.body.role:isset = false) || @request.auth.role = "admin"';
const DELETE_RULE = '@request.auth.role = "admin"';
migrate((app) => {
let col;
try {
col = app.findCollectionByNameOrId("team_members");
} catch (_) {
console.log("team_members_admin_rules: team_members does not exist — skipping.");
return;
}
if (col.type !== "auth") {
console.log("team_members_admin_rules: team_members is not an auth collection — skipping.");
return;
}
col.updateRule = UPDATE_RULE;
col.deleteRule = DELETE_RULE;
app.save(col);
}, (app) => {
// Down: restore the previous (self-update only, superuser delete) rules.
let col;
try {
col = app.findCollectionByNameOrId("team_members");
} catch (_) {
return;
}
if (col.type !== "auth") {
return;
}
col.updateRule = '@request.auth.id = id';
col.deleteRule = null;
app.save(col);
});

View File

@@ -0,0 +1,189 @@
/// <reference path="../pb_data/types.d.ts" />
//
// Issue #30 — Onboarding track (5-day, theme-level introduction).
//
// Two collections:
// onboarding_overviews — cached per-theme overview content (one row per theme),
// keyed by the theme name, with a topics_fingerprint for
// staleness detection / regeneration.
// onboarding_completions — per-user, per-theme completion marker.
//
// Design notes (consistent with issues #18/#27):
// * team_member_id is a plain TEXT field, NOT a relation — admins can delete
// team_members, and a relation with cascadeDelete would otherwise block that
// delete / drag the collection into the migration graph. Completions are keyed
// by (team_member_id, theme); orphan rows after a member delete are harmless.
// * API rules are set EXPLICITLY to authenticated-only. The tighten-rules
// migration (1781000001) already ran against the collections that existed then,
// so new collections must lock themselves down or they default to public.
const AUTHED = '@request.auth.id != ""';
migrate((app) => {
const overviews = new Collection({
"id": "pbc_onboarding_overviews0",
"name": "onboarding_overviews",
"type": "base",
"system": false,
"fields": [
{
"autogeneratePattern": "[a-z0-9]{15}",
"hidden": false,
"id": "text_id_oov",
"max": 15,
"min": 15,
"name": "id",
"pattern": "^[a-z0-9]+$",
"presentable": false,
"primaryKey": true,
"required": true,
"system": true,
"type": "text"
},
{
"system": false,
"id": "text_theme_oov",
"name": "theme",
"type": "text",
"required": true,
"presentable": false,
"max": 0,
"min": 0,
"pattern": ""
},
{
"system": false,
"id": "json_content_oov",
"name": "content",
"type": "json",
"required": true,
"presentable": false
},
{
"system": false,
"id": "text_fingerprint_oov",
"name": "topics_fingerprint",
"type": "text",
"required": true,
"presentable": false,
"max": 0,
"min": 0,
"pattern": ""
},
{
"hidden": false,
"id": "autodate_created_oov",
"name": "created",
"onCreate": true,
"onUpdate": false,
"presentable": false,
"system": true,
"type": "autodate"
},
{
"hidden": false,
"id": "autodate_updated_oov",
"name": "updated",
"onCreate": true,
"onUpdate": true,
"presentable": false,
"system": true,
"type": "autodate"
}
],
"indexes": [
"CREATE UNIQUE INDEX `idx_onboarding_overviews_theme` ON `onboarding_overviews` (`theme`)"
],
"listRule": AUTHED,
"viewRule": AUTHED,
"createRule": AUTHED,
"updateRule": AUTHED,
"deleteRule": AUTHED
});
app.save(overviews);
const completions = new Collection({
"id": "pbc_onboarding_completions0",
"name": "onboarding_completions",
"type": "base",
"system": false,
"fields": [
{
"autogeneratePattern": "[a-z0-9]{15}",
"hidden": false,
"id": "text_id_ocp",
"max": 15,
"min": 15,
"name": "id",
"pattern": "^[a-z0-9]+$",
"presentable": false,
"primaryKey": true,
"required": true,
"system": true,
"type": "text"
},
{
"system": false,
"id": "text_ocp_team_member",
"name": "team_member_id",
"type": "text",
"required": true,
"presentable": false,
"max": 0,
"min": 0,
"pattern": ""
},
{
"system": false,
"id": "text_ocp_theme",
"name": "theme",
"type": "text",
"required": true,
"presentable": false,
"max": 0,
"min": 0,
"pattern": ""
},
{
"hidden": false,
"id": "autodate_created_ocp",
"name": "created",
"onCreate": true,
"onUpdate": false,
"presentable": false,
"system": true,
"type": "autodate"
},
{
"hidden": false,
"id": "autodate_updated_ocp",
"name": "updated",
"onCreate": true,
"onUpdate": true,
"presentable": false,
"system": true,
"type": "autodate"
}
],
"indexes": [
"CREATE UNIQUE INDEX `idx_onboarding_completions_user_theme` ON `onboarding_completions` (`team_member_id`, `theme`)"
],
"listRule": AUTHED,
"viewRule": AUTHED,
"createRule": AUTHED,
"updateRule": AUTHED,
"deleteRule": AUTHED
});
app.save(completions);
}, (app) => {
const completions = app.findCollectionByNameOrId("onboarding_completions");
if (completions) {
app.delete(completions);
}
const overviews = app.findCollectionByNameOrId("onboarding_overviews");
if (overviews) {
app.delete(overviews);
}
})

View File

@@ -14,6 +14,8 @@ if (!ADMIN_EMAIL || !ADMIN_PASSWORD) {
}
const OPEN_RULES = { listRule: '', viewRule: '', createRule: '', updateRule: '', deleteRule: '' };
const AUTHED = '@request.auth.id != ""';
const AUTHED_RULES = { listRule: AUTHED, viewRule: AUTHED, createRule: AUTHED, updateRule: AUTHED, deleteRule: AUTHED };
// PocketBase v0.23: created/updated must be explicitly defined as autodate fields
const AUTODATE_FIELDS = [
@@ -114,6 +116,10 @@ const COLLECTIONS = [
// Mirror of pb_migrations/1781000002: creation only via the OAuth2
// sign-up flow (issue #22).
createRule: '@request.context = "oauth2"',
// Mirror of pb_migrations/1781000003: self-update without role changes;
// roster management (incl. role/delete) is admin-only (issue #27).
updateRule: '(@request.auth.id = id && @request.body.role:isset = false) || @request.auth.role = "admin"',
deleteRule: '@request.auth.role = "admin"',
passwordAuth: { enabled: false, identityFields: ['email'] },
fields: [
{ name: 'name', type: 'text', required: false },
@@ -200,6 +206,32 @@ const COLLECTIONS = [
...AUTODATE_FIELDS,
],
},
// Onboarding track (issue #30). Owned by pb_migrations/1781200000; these are a
// fallback for envs where migrations have not run. Authenticated-only rules
// (mirrors the migration) — do NOT use OPEN_RULES here.
{
name: 'onboarding_overviews',
type: 'base',
...AUTHED_RULES,
fields: [
{ name: 'theme', type: 'text', required: true },
{ name: 'content', type: 'json', required: true },
{ name: 'topics_fingerprint', type: 'text', required: true },
...AUTODATE_FIELDS,
],
indexes: ['CREATE UNIQUE INDEX `idx_onboarding_overviews_theme` ON `onboarding_overviews` (`theme`)'],
},
{
name: 'onboarding_completions',
type: 'base',
...AUTHED_RULES,
fields: [
{ name: 'team_member_id', type: 'text', required: true },
{ name: 'theme', type: 'text', required: true },
...AUTODATE_FIELDS,
],
indexes: ['CREATE UNIQUE INDEX `idx_onboarding_completions_user_theme` ON `onboarding_completions` (`team_member_id`, `theme`)'],
},
];
async function post(path, body, token) {

View File

@@ -7,6 +7,7 @@ import ChatLauncher from './components/chat/ChatLauncher'
import Login from './pages/Login'
import Onboarding from './pages/Onboarding'
import OnboardingTrack from './pages/OnboardingTrack'
import Dashboard from './pages/Dashboard'
import Admin from './pages/Admin'
@@ -106,6 +107,10 @@ function App() {
<Route path="/login" element={<Login />} />
<Route path="/onboarding" element={<Onboarding />} />
<Route path="/" element={<ProtectedRoute><Dashboard /></ProtectedRoute>} />
{/* Onboarding track is available to any logged-in user, regardless of
enrollment — hence skipEnrollmentGate (distinct from /onboarding, the
enrollment page). */}
<Route path="/onboarding-track" element={<ProtectedRoute skipEnrollmentGate><OnboardingTrack /></ProtectedRoute>} />
<Route path="/learn" element={<ProtectedRoute><Leren /></ProtectedRoute>} />
<Route path="/test" element={<ProtectedRoute><Testen /></ProtectedRoute>} />
<Route path="/topic-test" element={<ProtectedRoute><TopicTest /></ProtectedRoute>} />

View File

@@ -8,10 +8,11 @@ import { useApp } from '../../store/AppContext';
// Users are provisioned automatically on first Azure (Entra ID) login — admins
// no longer create them by hand. This panel lets an admin review the roster,
// switch a member between user/admin, and remove members. The admin role is
// also re-synced from the ENTRA_ADMIN_EMAILS allow-list on every login (see
// pb_hooks/team_members.pb.js), so a manual change here can be overridden on
// the member's next sign-in if their e-mail is on/off that list.
// switch a member between user/admin, and remove members. The
// ENTRA_ADMIN_EMAILS allow-list is escalate-only (see
// pb_hooks/team_members.pb.js): it guarantees admin for its members on every
// login, so promotions made here stick, but demoting an allow-list member is
// undone on their next sign-in — remove them from the list instead.
const TeamManager = () => {
const { state } = useApp();
const [users, setUsers] = useState([]);
@@ -67,8 +68,9 @@ const TeamManager = () => {
<p className="text-sm text-fg-muted flex items-start gap-2">
<Info size={16} className="mt-0.5 shrink-0 text-teal" />
Team members are created automatically when they first sign in with their
Microsoft account. Admins are granted via the <code>ENTRA_ADMIN_EMAILS</code>{' '}
allow-list and re-synced on each login.
Microsoft account. Members of the <code>ENTRA_ADMIN_EMAILS</code> allow-list
are always admin; promotions made here stick, but demoting an allow-list
member is undone on their next sign-in.
</p>
</Card>

View File

@@ -1,4 +1,5 @@
import { useCallback, useEffect, useRef, useState } from 'react';
import { Trash2 } from 'lucide-react';
import Mark from '../ui/Mark';
import ChatMessage from './ChatMessage';
import { useChat } from './useChat';
@@ -6,7 +7,7 @@ import { kbStore } from '../../lib/kbStore';
import { BOT_NAME, STRINGS } from './prompts';
export default function ChatWindow({ user, isAdmin, onClose }) {
const { messages, isThinking, send } = useChat({ user, isAdmin });
const { messages, isThinking, send, clearThread } = useChat({ user, isAdmin });
const [draft, setDraft] = useState('');
const bodyRef = useRef(null);
const inputRef = useRef(null);
@@ -60,6 +61,14 @@ export default function ChatWindow({ user, isAdmin, onClose }) {
setDecided(prev => ({ ...prev, [msgId]: 'rejected' }));
}, []);
const handleClear = useCallback(() => {
if (isThinking) return;
if (!window.confirm(STRINGS.clearConfirm)) return;
setDecided({});
clearThread();
inputRef.current?.focus();
}, [isThinking, clearThread]);
const renderedMessages = messages.map(m => {
if (!m.suggestion) return m;
const status = decided[m.id] || m.suggestion.status || 'pending';
@@ -81,14 +90,26 @@ export default function ChatWindow({ user, isAdmin, onClose }) {
<div className="r42-window-hd-name">{BOT_NAME}</div>
<div className="r42-window-hd-status"><i /> {STRINGS.status}</div>
</div>
<button
type="button"
className="r42-window-hd-x"
onClick={onClose}
aria-label={STRINGS.closeAria}
>
×
</button>
<div className="r42-window-hd-actions">
<button
type="button"
className="r42-window-hd-clear"
onClick={handleClear}
disabled={isThinking}
aria-label={STRINGS.clearAria}
title={STRINGS.clearAria}
>
<Trash2 size={16} />
</button>
<button
type="button"
className="r42-window-hd-x"
onClick={onClose}
aria-label={STRINGS.closeAria}
>
×
</button>
</div>
</header>
<div className="r42-window-body" ref={bodyRef}>

View File

@@ -0,0 +1,63 @@
import { beforeEach, describe, expect, it, vi } from 'vitest';
// In-memory KB the mocked db serves from.
const store = {
topics: [],
relations: [],
content: new Map(),
};
vi.mock('../../../lib/db', () => ({
getTopics: vi.fn(async () => store.topics),
getRelations: vi.fn(async () => store.relations),
getContent: vi.fn(async (id) => store.content.get(id) ?? null),
}));
import { buildKbContext } from '../rag';
beforeEach(() => {
store.topics = [];
store.relations = [];
store.content = new Map();
});
describe('buildKbContext', () => {
it('reports an empty graph', async () => {
const { context, allTopics } = await buildKbContext('pensioen');
expect(context).toMatch(/leeg/);
expect(allTopics).toEqual([]);
});
it('injects deep content for a retrieved topic that is not named verbatim', async () => {
store.topics = [
{ id: 'pensioenregeling', label: 'Pensioenregeling', type: 'concept', description: 'De beschikbare premieregeling.' },
{ id: 'onboarding-buddy', label: 'Onboarding Buddy', type: 'role', description: 'Begeleidt nieuwe medewerkers.' },
];
store.content.set('pensioenregeling', {
article: 'De premie is 10% van de pensioengrondslag; werkgever en werknemer betalen elk 50%.',
});
// "pensioen" is never a verbatim topic id/label, but the compound-word
// matching should retrieve pensioenregeling and pull its article body in.
const { context, retrievedTopics } = await buildKbContext('wat dekt mijn pensioen?');
expect(retrievedTopics.map(t => t.id)).toContain('pensioenregeling');
expect(context).toMatch(/DIEPERE INHOUD/);
expect(context).toMatch(/10% van de pensioengrondslag/);
});
it('only includes relations whose endpoints are both in the selection', async () => {
store.topics = [
{ id: 'pensioenregeling', label: 'Pensioenregeling', type: 'concept', description: 'De beschikbare premieregeling.' },
{ id: 'partnerpensioen', label: 'Partnerpensioen', type: 'concept', description: 'Uitkering aan de partner.' },
];
store.relations = [
{ source: 'partnerpensioen', target: 'pensioenregeling', type: 'part_of' },
{ source: 'pensioenregeling', target: 'iets-anders', type: 'related_to' },
];
const { context } = await buildKbContext('pensioen');
expect(context).toMatch(/partnerpensioen --part_of--> pensioenregeling/);
expect(context).not.toMatch(/iets-anders/);
});
});

View File

@@ -87,8 +87,13 @@
border-radius: 999px;
background: var(--sage);
}
.r42-window-hd-x {
.r42-window-hd-actions {
margin-left: auto;
display: flex;
align-items: center;
gap: 2px;
}
.r42-window-hd-x {
color: rgba(236, 233, 233, 0.7);
background: transparent;
border: none;
@@ -99,6 +104,18 @@
border-radius: var(--r-sm);
}
.r42-window-hd-x:hover { background: rgba(236, 233, 233, 0.1); }
.r42-window-hd-clear {
color: rgba(236, 233, 233, 0.7);
background: transparent;
border: none;
cursor: pointer;
padding: 6px;
border-radius: var(--r-sm);
display: grid;
place-items: center;
}
.r42-window-hd-clear:hover { background: rgba(236, 233, 233, 0.1); }
.r42-window-hd-clear:disabled { opacity: 0.4; cursor: default; }
.r42-window-body {
flex: 1;

View File

@@ -21,6 +21,8 @@ export const STRINGS = {
suggestionDismissed: 'Oké, niets gedaan.',
closeAria: 'Sluit chatvenster',
openAria: 'Open R42 chatbot',
clearAria: 'Wis gesprek',
clearConfirm: 'Dit gesprek wissen? Dit kan niet ongedaan worden gemaakt.',
};
const STABLE_PREAMBLE = [
@@ -35,11 +37,16 @@ const STABLE_PREAMBLE = [
`JE KENNIS:`,
`Je kennis is beperkt tot de Respellion-kennisgraaf die hieronder volgt. Als een vraag duidelijk buiten dit bereik valt, zeg dat dan eerlijk en stel voor dat de gebruiker de bron toevoegt via Admin → Sources.`,
``,
`NAUWKEURIGHEID (belangrijk):`,
`- Baseer je antwoord uitsluitend op de KENNISGRAAF en DIEPERE INHOUD hieronder; verzin niets.`,
`- Gebruik ALLE relevante feiten die daar staan. Bij een vraag om details, bedragen, percentages, voorwaarden of een opsomming: noem elk relevant feit — vat niet samen ten koste van volledigheid.`,
`- Als de samenvattende KENNISGRAAF te dun is om de vraag volledig te beantwoorden, roep dan éérst de tool "lookup_topic" aan (met het exacte topic-id) voordat je concludeert dat je het niet weet.`,
``,
`KENNISGRAAF VERFIJNEN:`,
`Wanneer de gebruiker iets noemt dat duidelijk een nieuw topic, nieuwe relatie, proces of rol is — en dat nog niet in de kennisgraaf staat — gebruik dan de tool "propose_graph_delta" om een voorstel te maken. Verzin niets: stel alleen iets voor als de gebruiker het concreet noemt. Stel maximaal 3 topics en 5 relaties per beurt voor.`,
``,
`STIJL:`,
`- Houd antwoorden onder de 4 zinnen tenzij de gebruiker om uitleg vraagt.`,
`- Zo kort als kan, zo volledig als nodig: houd eenvoudige antwoorden onder de 4 zinnen, maar som bij details- of opsommingsvragen álle relevante feiten op (desnoods als korte lijst met streepjes).`,
`- Geen markdown-headers; gewone Nederlandse tekst.`,
`- Bij onzekerheid: "Ik weet het niet zeker — controleer dit in het handboek."`,
].join('\n');

View File

@@ -2,6 +2,13 @@ import * as db from '../../lib/db';
import { buildIndex, retrieveTopK } from '../../lib/retrieval';
const TOP_K = 10;
// How many topics get their full article body injected (not just the short
// description). Verbatim-mentioned topics come first, then the highest-ranked
// retrieved ones, so a query that never names a topic exactly still gets rich
// content for what it matched.
const DEEP_CONTENT_LIMIT = 5;
const DEEP_SNIPPET_CHARS = 1000;
const DESC_SNIPPET_CHARS = 320;
async function sha256Hex(input) {
const enc = new TextEncoder().encode(input);
@@ -71,7 +78,7 @@ export async function buildKbContext(userMessage = '') {
const included = [...includedById.values()];
const topicLines = included.map(t => {
const desc = (t.description || '').replace(/\s+/g, ' ').trim().slice(0, 200);
const desc = (t.description || '').replace(/\s+/g, ' ').trim().slice(0, DESC_SNIPPET_CHARS);
return `- ${t.id} (${t.type || 'concept'}) "${t.label}": ${desc}`;
});
@@ -85,19 +92,30 @@ export async function buildKbContext(userMessage = '') {
}
}
const mentionedDeepContent = [];
for (const id of mentionedIds) {
const t = includedById.get(id);
if (!t) continue;
const content = await db.getContent(t.id).catch(() => null);
if (!content) continue;
let raw;
if (typeof content === 'string') raw = content;
else if (content.article) raw = typeof content.article === 'string' ? content.article : JSON.stringify(content.article);
else raw = JSON.stringify(content);
const snippet = raw.replace(/\s+/g, ' ').trim().slice(0, 1200);
mentionedDeepContent.push(`### ${t.label}\n${snippet}`);
// Pick which topics get their full article body: verbatim mentions first,
// then the highest-ranked retrieved topics, capped at DEEP_CONTENT_LIMIT.
const deepIds = [];
for (const id of mentionedIds) deepIds.push(id);
for (const t of retrieved) {
if (deepIds.length >= DEEP_CONTENT_LIMIT) break;
if (!mentionedIds.has(t.id)) deepIds.push(t.id);
}
const deepBlocks = await Promise.all(
deepIds.slice(0, DEEP_CONTENT_LIMIT).map(async (id) => {
const t = includedById.get(id);
if (!t) return null;
const content = await db.getContent(id).catch(() => null);
if (!content) return null;
let raw;
if (typeof content === 'string') raw = content;
else if (content.article) raw = typeof content.article === 'string' ? content.article : JSON.stringify(content.article);
else raw = JSON.stringify(content);
const snippet = raw.replace(/\s+/g, ' ').trim().slice(0, DEEP_SNIPPET_CHARS);
if (!snippet) return null;
return `### ${t.label}\n${snippet}`;
}),
);
const mentionedDeepContent = deepBlocks.filter(Boolean);
const context = [
`KENNISGRAAF — RELEVANTE TOPICS (top ${included.length} van ${allTopics.length}):`,
@@ -106,7 +124,7 @@ export async function buildKbContext(userMessage = '') {
`KENNISGRAAF — RELATIES (binnen deze selectie):`,
relLines.length ? relLines.join('\n') : '(geen relaties binnen deze selectie)',
mentionedDeepContent.length
? `\n\nDIEPERE INHOUD (voor genoemde topics):\n${mentionedDeepContent.join('\n\n')}`
? `\n\nDIEPERE INHOUD (volledige leerinhoud van de meest relevante topics — gebruik álle feiten hieruit die de vraag beantwoorden):\n${mentionedDeepContent.join('\n\n')}`
: '',
``,
`Als de relevante context hierboven te beperkt is, gebruik dan de tool "lookup_topic" om de volledige beschrijving en eventuele leerinhoud van een specifiek topic op te halen.`,

View File

@@ -92,6 +92,19 @@ export function useChat({ user, isAdmin }) {
setMessages(prev => prev.map(m => (m.id === id ? { ...m, ...patch } : m)));
}, []);
/** Wipe the persisted thread and reset to a fresh greeting. */
const clearThread = useCallback(() => {
seenDeltaKeys.current = new Set();
const greeting = {
id: `m_${Date.now()}`,
role: 'assistant',
content: STRINGS.greeting(user?.name || 'daar'),
ts: Date.now(),
};
setMessages([greeting]);
if (threadKey) storage.set(threadKey, [greeting]);
}, [user, threadKey]);
const send = useCallback(async (text) => {
const trimmed = (text || '').trim();
if (!trimmed || !user) return;
@@ -225,5 +238,6 @@ export function useChat({ user, isAdmin }) {
errored,
send,
updateMessage,
clearThread,
};
}

View File

@@ -9,6 +9,7 @@ import {
customTopicSchema,
graphActionsSchema,
proposeGraphDeltaSchema,
onboardingOverviewSchema,
} from '../llmSchemas';
const sampleTopic = {
@@ -192,3 +193,56 @@ describe('proposeGraphDeltaSchema', () => {
).toThrow();
});
});
describe('onboardingOverviewSchema (issue #32 — fast-tier stringified arrays)', () => {
const base = {
title: 'Privacy',
what_it_is: 'How we handle data.',
why_it_matters: 'You touch personal data weekly.',
};
const points = ['Point one', 'Point two', 'Point three'];
const topics = [{ topic_id: 'avg', label: 'AVG' }];
it('accepts well-formed output', () => {
const r = onboardingOverviewSchema.parse({ ...base, key_points: points, topics_covered: topics });
expect(r.key_points).toEqual(points);
expect(r.topics_covered).toEqual(topics);
});
it('coerces JSON-stringified arrays back to arrays (the exact #32 failure)', () => {
const r = onboardingOverviewSchema.parse({
...base,
key_points: JSON.stringify(points),
topics_covered: JSON.stringify(topics),
});
expect(r.key_points).toEqual(points);
expect(r.topics_covered).toEqual(topics);
});
it('splits a bullet/newline string of key points as a fallback', () => {
const r = onboardingOverviewSchema.parse({
...base,
key_points: '- Point one\n• Point two\n3. Point three',
topics_covered: topics,
});
expect(r.key_points).toEqual(points);
});
it('keeps the first 5 key points instead of failing on overage', () => {
const seven = Array.from({ length: 7 }, (_, i) => `P${i + 1}`);
const r = onboardingOverviewSchema.parse({ ...base, key_points: seven, topics_covered: topics });
expect(r.key_points).toEqual(['P1', 'P2', 'P3', 'P4', 'P5']);
});
it('still rejects genuinely bad output', () => {
expect(() =>
onboardingOverviewSchema.parse({ ...base, key_points: ['only', 'two'], topics_covered: topics }),
).toThrow();
expect(() =>
onboardingOverviewSchema.parse({ ...base, key_points: points, topics_covered: 'not json at all' }),
).toThrow();
expect(() =>
onboardingOverviewSchema.parse({ ...base, key_points: 42, topics_covered: topics }),
).toThrow();
});
});

View File

@@ -0,0 +1,152 @@
import { describe, expect, it } from 'vitest';
import {
ONBOARDING_DAY_COUNT,
orderThemes,
distributeThemesIntoDays,
computeTopicsFingerprint,
computeOnboardingProgress,
buildOnboardingPlan,
} from '../onboardingService';
const sizes = (days) => days.map((d) => d.themes.length);
describe('onboardingService pure helpers', () => {
it('exposes a 5-day guideline', () => {
expect(ONBOARDING_DAY_COUNT).toBe(5);
});
describe('orderThemes', () => {
const kb = ['Governance', 'Culture', 'Privacy', 'Process'];
it('orders by first appearance in the schedule, then appends the rest alphabetically', () => {
const schedule = [{ theme: 'Privacy' }, { theme: 'Governance' }];
expect(orderThemes(kb, schedule)).toEqual(['Privacy', 'Governance', 'Culture', 'Process']);
});
it('de-duplicates repeated schedule labels', () => {
const schedule = [{ theme: 'Privacy' }, { theme: 'Privacy' }, { theme: 'Culture' }];
expect(orderThemes(kb, schedule)).toEqual(['Privacy', 'Culture', 'Governance', 'Process']);
});
it('ignores schedule labels that are not real KB themes (e.g. merged-week labels)', () => {
const schedule = [{ theme: 'Privacy & Governance (merged)' }, { theme: 'Culture' }];
expect(orderThemes(kb, schedule)).toEqual(['Culture', 'Governance', 'Privacy', 'Process']);
});
it('falls back to pure alphabetical when the schedule is null/empty', () => {
expect(orderThemes(kb, null)).toEqual(['Culture', 'Governance', 'Privacy', 'Process']);
expect(orderThemes(kb, [])).toEqual(['Culture', 'Governance', 'Privacy', 'Process']);
});
});
describe('distributeThemesIntoDays', () => {
const themes = (n) => Array.from({ length: n }, (_, i) => `T${i + 1}`);
it('splits 10 themes evenly across 5 days', () => {
expect(sizes(distributeThemesIntoDays(themes(10)))).toEqual([2, 2, 2, 2, 2]);
});
it('front-loads the remainder (12 -> 3,3,2,2,2)', () => {
expect(sizes(distributeThemesIntoDays(themes(12)))).toEqual([3, 3, 2, 2, 2]);
});
it('handles 7 -> 2,2,1,1,1', () => {
expect(sizes(distributeThemesIntoDays(themes(7)))).toEqual([2, 2, 1, 1, 1]);
});
it('uses fewer days than the cap when there are fewer themes', () => {
const days = distributeThemesIntoDays(themes(3));
expect(sizes(days)).toEqual([1, 1, 1]);
expect(days.map((d) => d.day)).toEqual([1, 2, 3]);
});
it('returns [] for no themes', () => {
expect(distributeThemesIntoDays([])).toEqual([]);
});
it('preserves order and loses nothing', () => {
const input = themes(12);
const flat = distributeThemesIntoDays(input).flatMap((d) => d.themes);
expect(flat).toEqual(input);
});
});
describe('computeTopicsFingerprint', () => {
it('is order-independent', () => {
const a = computeTopicsFingerprint([{ id: 'x' }, { id: 'y' }, { id: 'z' }]);
const b = computeTopicsFingerprint([{ id: 'z' }, { id: 'x' }, { id: 'y' }]);
expect(a).toBe(b);
});
it('changes when a topic is added or removed', () => {
const base = computeTopicsFingerprint([{ id: 'x' }, { id: 'y' }]);
expect(computeTopicsFingerprint([{ id: 'x' }])).not.toBe(base);
expect(computeTopicsFingerprint([{ id: 'x' }, { id: 'y' }, { id: 'z' }])).not.toBe(base);
});
it('handles empty input', () => {
expect(typeof computeTopicsFingerprint([])).toBe('string');
});
});
describe('computeOnboardingProgress', () => {
const days = [
{ day: 1, themes: ['A', 'B'] },
{ day: 2, themes: ['C'] },
];
it('does not count a partially-done day as complete', () => {
const p = computeOnboardingProgress(days, new Set(['A']));
expect(p).toMatchObject({ themesTotal: 3, themesDone: 1, dayCount: 2, daysCompleted: 0, allDone: false });
expect(p.percentage).toBe(33);
});
it('counts a fully-done day', () => {
const p = computeOnboardingProgress(days, new Set(['A', 'B']));
expect(p).toMatchObject({ themesDone: 2, daysCompleted: 1, allDone: false });
});
it('reports allDone only when every theme is complete', () => {
const p = computeOnboardingProgress(days, new Set(['A', 'B', 'C']));
expect(p).toMatchObject({ themesDone: 3, daysCompleted: 2, percentage: 100, allDone: true });
});
it('accepts an array as well as a Set', () => {
expect(computeOnboardingProgress(days, ['A', 'B', 'C']).allDone).toBe(true);
});
it('returns zeros for an empty plan', () => {
expect(computeOnboardingProgress([], new Set())).toEqual({
themesTotal: 0, themesDone: 0, dayCount: 0, daysCompleted: 0, percentage: 0, allDone: false,
});
});
});
describe('buildOnboardingPlan', () => {
const topics = [
{ id: 't1', label: 'T1', theme: 'Privacy', complexity_weight: 2 },
{ id: 't2', label: 'T2', theme: 'Culture', complexity_weight: 1 },
{ id: 'f1', label: 'F1', theme: 'Privacy', type: 'fact' }, // excluded by buildThemeTopicMap
];
it('builds themes (schedule-ordered), days, and a theme→topics map', () => {
const activeVersion = { schedule: [{ theme: 'Privacy' }, { theme: 'Culture' }] };
const plan = buildOnboardingPlan(topics, activeVersion);
expect(plan.themes).toEqual(['Privacy', 'Culture']);
expect(sizes(plan.days)).toEqual([1, 1]);
expect(plan.themeTopicMap.get('Privacy').map((t) => t.id)).toEqual(['t1']); // fact excluded
});
it('tolerates a stringified schedule and a missing version', () => {
const stringified = { schedule: JSON.stringify([{ theme: 'Culture' }]) };
expect(buildOnboardingPlan(topics, stringified).themes).toEqual(['Culture', 'Privacy']);
expect(buildOnboardingPlan(topics, null).themes).toEqual(['Culture', 'Privacy']);
});
it('returns an empty plan for an empty KB', () => {
const plan = buildOnboardingPlan([], null);
expect(plan.themes).toEqual([]);
expect(plan.days).toEqual([]);
});
});
});

View File

@@ -50,6 +50,31 @@ describe('buildIndex / retrieveTopK', () => {
expect(retrieveTopK(idx, 'kwantumfysica raketten')).toEqual([]);
});
it('matches Dutch compound words on a shared stem', () => {
const pensionTopics = [
{ id: 'pensioenregeling', label: 'Pensioenregeling', description: 'De beschikbare premieregeling bij a.s.r. Doen Pensioen.' },
{ id: 'partnerpensioen', label: 'Partnerpensioen', description: 'Uitkering aan de partner bij overlijden.' },
{ id: 'reiskostenvergoeding', label: 'Reiskostenvergoeding', description: 'EUR 0,23 per kilometer voor woon-werkverkeer.' },
];
const idx = buildIndex(pensionTopics);
// "pensioen" never appears as a standalone token in a label, yet the stem is
// a prefix of "pensioenregeling" and an infix of "partnerpensioen".
const hits = retrieveTopK(idx, 'wat dekt mijn pensioen?', 3).map(h => h.id);
expect(hits).toContain('pensioenregeling');
expect(hits).toContain('partnerpensioen');
expect(hits).not.toContain('reiskostenvergoeding');
});
it('does not partial-match on short shared prefixes', () => {
const topics = [
{ id: 'onderhoud', label: 'Onderhoud', description: 'Technisch beheer van systemen.' },
];
const idx = buildIndex(topics);
// "onderneming" shares only "onder" (5) with "onderhoud" — below the overlap
// needed for a query token this size to count.
expect(retrieveTopK(idx, 'onderneming')).toEqual([]);
});
it('caches the index per topics array reference', () => {
const idx1 = buildIndex(sampleTopics);
const idx2 = buildIndex(sampleTopics);

View File

@@ -406,6 +406,49 @@ export async function setThemeSessionCompletion(userId, themeSessionId, sessionW
);
}
// ── Onboarding track ─────────────────────────────────────────────────────────
/**
* Cached onboarding overview for a theme, or null. `theme` is free text →
* use pb.filter() bindings so quotes/specials can't break the filter.
*/
export async function getOnboardingOverview(theme) {
try {
return await pb.collection('onboarding_overviews').getFirstListItem(
pb.filter('theme = {:theme}', { theme }),
);
} catch {
return null;
}
}
/** Upsert the cached overview for a theme (keyed by theme). */
export async function setOnboardingOverview(theme, content, fingerprint) {
return pbUpsert(
'onboarding_overviews',
pb.filter('theme = {:theme}', { theme }),
{ content, topics_fingerprint: fingerprint },
{ theme, content, topics_fingerprint: fingerprint },
);
}
/** All onboarding-completion rows for a user (each has a `theme`). */
export async function getCompletedOnboardingThemes(userId) {
return pb.collection('onboarding_completions').getFullList({
filter: pb.filter('team_member_id = {:userId}', { userId }),
});
}
/** Mark a theme complete for the user. Idempotent (unique on user+theme). */
export async function setOnboardingCompletion(userId, theme) {
return pbUpsert(
'onboarding_completions',
pb.filter('team_member_id = {:userId} && theme = {:theme}', { userId, theme }),
{ team_member_id: userId, theme },
{ team_member_id: userId, theme },
);
}
// ── Leaderboard ───────────────────────────────────────────────────────────────
export async function getLeaderboard() {

View File

@@ -213,6 +213,13 @@ const SIMULATION_TOOL_STUBS = {
},
emit_graph_actions: { merges: [], deletions: [], newRelations: [], relevanceUpdates: [] },
set_intro: { intro: 'Bijgewerkte intro (simulatie).' },
emit_onboarding_overview: {
title: 'Voorbeeldthema',
what_it_is: 'Een korte simulatie-omschrijving van dit thema.',
why_it_matters: 'In simulatiemodus laat dit zien hoe de onboarding-overview eruitziet zonder de API te raken.',
key_points: ['Kernpunt één', 'Kernpunt twee', 'Kernpunt drie'],
topics_covered: [{ topic_id: 'sim-topic', label: 'Simulatie onderwerp' }],
},
};
function stubResponse({ stopReason = 'end_turn', text = '', toolUses = [] }) {

View File

@@ -229,6 +229,45 @@ export const themeSessionSchema = z.object({
keyTakeaways: z.array(z.string().min(1)).min(3),
});
// Fast-tier models occasionally emit array tool fields as a JSON-encoded
// string ("[\"a\",\"b\"]") instead of a real array (issue #32). Parse those
// back to arrays before validating; leave anything else untouched so real
// type errors still fail validation.
function coerceStringifiedArray(v) {
if (typeof v === 'string') {
const s = v.trim();
if (s.startsWith('[')) {
try { return JSON.parse(s); } catch { /* keep original, let Zod report */ }
}
}
return v;
}
const onboardingKeyPoints = z.preprocess((v) => {
let out = coerceStringifiedArray(v);
// Fallback: a bullet/newline list as one string → split into points.
if (typeof out === 'string' && out.includes('\n')) {
out = out
.split('\n')
.map((line) => line.replace(/^\s*[-•*\d.]+\s*/, '').trim())
.filter(Boolean);
}
// Overage is trivial — keep the first 5 rather than failing the user.
if (Array.isArray(out) && out.length > 5) out = out.slice(0, 5);
return out;
}, z.array(z.string().min(1)).min(3).max(5));
export const onboardingOverviewSchema = z.object({
title: z.string().min(1),
what_it_is: z.string().min(1),
why_it_matters: z.string().min(1),
key_points: onboardingKeyPoints,
topics_covered: z.preprocess(
coerceStringifiedArray,
z.array(z.object({ topic_id: z.string().min(1), label: z.string().min(1) })).min(1),
),
});
/**
* Registry mapping known tool names to their input schemas. `callLLM`
* consults this when the caller does not pass an explicit `toolSchemas`
@@ -252,4 +291,5 @@ export const toolSchemaRegistry = {
remove_section: removeSectionPatchSchema,
replace_takeaways: replaceTakeawaysPatchSchema,
emit_theme_session: themeSessionSchema,
emit_onboarding_overview: onboardingOverviewSchema,
};

View File

@@ -454,3 +454,39 @@ export const EMIT_THEME_SESSION_TOOL = {
},
};
// ── Onboarding overview (breadth-first, one theme) ───────────────────────────
export const EMIT_ONBOARDING_OVERVIEW_TOOL = {
name: 'emit_onboarding_overview',
description: 'Return a SHORT, breadth-first onboarding overview of ONE theme for a brand-new Respellion employee. This is a light introduction, NOT a deep lesson: what the theme is in plain language, why it matters for day-to-day and week-to-week work at Respellion, a few key points, and which topics belong to it. Keep it skimmable.',
input_schema: {
type: 'object',
properties: {
title: { type: 'string', description: 'Learner-facing theme title, 26 words.' },
what_it_is: { type: 'string', description: '12 plain-language sentences defining what this theme is about.' },
why_it_matters: { type: 'string', description: '13 sentences on why this theme matters for a new employee\'s daily and weekly work at Respellion.' },
key_points: {
type: 'array',
items: { type: 'string' },
minItems: 3,
maxItems: 5,
description: '35 short, concrete takeaways a newcomer should remember about this theme. Must be a JSON array of strings — never a single string.',
},
topics_covered: {
type: 'array',
description: 'The topics that make up this theme. Reuse the exact topic_id you were given so the UI can link back. Must be a JSON array of objects — never a string.',
items: {
type: 'object',
properties: {
topic_id: { type: 'string', description: 'The id of the topic (must match one of the ids provided).' },
label: { type: 'string', description: 'The topic label, as provided.' },
},
required: ['topic_id', 'label'],
},
minItems: 1,
},
},
required: ['title', 'what_it_is', 'why_it_matters', 'key_points', 'topics_covered'],
},
};

View File

@@ -0,0 +1,233 @@
import * as db from './db';
import { callLLM, cachedSystem } from './llm';
import { buildThemeTopicMap } from './curriculumService';
import { EMIT_ONBOARDING_OVERVIEW_TOOL } from './llmTools';
// The onboarding track introduces a new employee to every KB theme in breadth,
// spread over a guideline of 5 "days". A theme is the unit we track completion
// on; "days" are only a presentation grouping computed at read time, so
// re-chunking (e.g. when the theme set changes) never loses a user's progress.
export const ONBOARDING_DAY_COUNT = 5;
// ── Pure helpers (unit-tested) ───────────────────────────────────────────────
/**
* Order theme names by their first appearance in the active curriculum schedule
* (a pedagogical order), then append any KB themes the schedule never names,
* alphabetically. Schedule labels that aren't real KB themes (e.g. merged-week
* labels) are ignored. De-duplicated.
*
* @param {string[]} themeNames — canonical theme names present in the KB
* @param {Array<{theme?: string}>|null} schedule — active curriculum schedule
* @returns {string[]}
*/
export function orderThemes(themeNames, schedule) {
const known = new Set(themeNames);
const ordered = [];
const seen = new Set();
if (Array.isArray(schedule)) {
for (const week of schedule) {
const theme = week && week.theme;
if (theme && known.has(theme) && !seen.has(theme)) {
seen.add(theme);
ordered.push(theme);
}
}
}
const remaining = themeNames
.filter((t) => !seen.has(t))
.sort((a, b) => a.localeCompare(b));
return [...ordered, ...remaining];
}
/**
* Split an ordered theme list into balanced, order-preserving day buckets.
* Uses at most `dayCount` days; with fewer themes than days it uses `N` days.
* The first `remainder` days get one extra theme.
*
* @param {string[]} orderedThemes
* @param {number} [dayCount=5]
* @returns {Array<{ day: number, themes: string[] }>} non-empty days only
*/
export function distributeThemesIntoDays(orderedThemes, dayCount = ONBOARDING_DAY_COUNT) {
const themes = Array.isArray(orderedThemes) ? orderedThemes : [];
const n = themes.length;
if (n === 0) return [];
const effectiveDays = Math.min(dayCount, n);
const base = Math.floor(n / effectiveDays);
const remainder = n % effectiveDays;
const days = [];
let idx = 0;
for (let d = 0; d < effectiveDays; d++) {
const size = base + (d < remainder ? 1 : 0);
days.push({ day: d + 1, themes: themes.slice(idx, idx + size) });
idx += size;
}
return days;
}
/**
* Stable, order-independent fingerprint over a theme's topic ids. Changes when
* a topic is added to or removed from the theme, so cached overviews can be
* detected as stale and regenerated.
*
* @param {Array<{id?: string}>} topics
* @returns {string}
*/
export function computeTopicsFingerprint(topics) {
const ids = (Array.isArray(topics) ? topics : [])
.map((t) => t && t.id)
.filter(Boolean)
.sort();
const joined = ids.join(',');
let h = 5381;
for (let i = 0; i < joined.length; i++) {
h = ((h << 5) + h + joined.charCodeAt(i)) >>> 0; // djb2, unsigned
}
return `${ids.length}:${h.toString(16)}`;
}
/**
* Derive progress from the plan's days and the set of completed themes.
*
* @param {Array<{ day: number, themes: string[] }>} days
* @param {Set<string>|string[]} completedThemeSet
* @returns {{ themesTotal:number, themesDone:number, dayCount:number, daysCompleted:number, percentage:number, allDone:boolean }}
*/
export function computeOnboardingProgress(days, completedThemeSet) {
const set = completedThemeSet instanceof Set
? completedThemeSet
: new Set(completedThemeSet || []);
const allThemes = (days || []).flatMap((d) => d.themes);
const themesTotal = allThemes.length;
const themesDone = allThemes.filter((t) => set.has(t)).length;
const dayCount = (days || []).length;
const daysCompleted = (days || []).filter(
(d) => d.themes.length > 0 && d.themes.every((t) => set.has(t)),
).length;
const percentage = themesTotal === 0 ? 0 : Math.round((themesDone / themesTotal) * 100);
const allDone = themesTotal > 0 && themesDone === themesTotal;
return { themesTotal, themesDone, dayCount, daysCompleted, percentage, allDone };
}
/**
* Build the onboarding plan from already-loaded topics + active curriculum version.
*
* @param {Array} topics — all topics (db.getTopics())
* @param {object|null} activeVersion — active curriculum_versions row (for schedule order)
* @returns {{ themes: string[], days: Array<{day:number,themes:string[]}>, themeTopicMap: Map<string, Array> }}
*/
export function buildOnboardingPlan(topics, activeVersion) {
const themeTopicMap = buildThemeTopicMap(topics || []);
let schedule = activeVersion && activeVersion.schedule ? activeVersion.schedule : null;
if (typeof schedule === 'string') {
try { schedule = JSON.parse(schedule); } catch { schedule = null; }
}
const themes = orderThemes([...themeTopicMap.keys()], schedule);
const days = distributeThemesIntoDays(themes);
return { themes, days, themeTopicMap };
}
// ── I/O + generation ─────────────────────────────────────────────────────────
const ONBOARDING_OVERVIEW_SYSTEM = `You are an onboarding guide for Respellion, an internal IT company.
You introduce a brand-new employee to ONE theme from the company knowledge base.
This is a breadth-first introduction, not a deep lesson: keep it short, plain and skimmable, and always connect the theme to how work actually happens day-to-day and week-to-week at Respellion.
Write in clear, professional English. Emit content only through the emit_onboarding_overview tool.`;
function buildOverviewPrompt(theme, topics) {
const topicLines = topics
.map((t, idx) => `${idx + 1}. id=${t.id} | label="${t.label}" | type=${t.type || '—'} | description=${t.description || '—'}`)
.join('\n');
return `Theme: ${theme}
Topics that make up this theme (reuse the exact topic_id for each entry in topics_covered):
${topicLines || '(no topics listed)'}
Write a short, breadth-first onboarding overview of this theme for a new employee's first week: what it is, why it matters for their daily and weekly work at Respellion, 35 key points, and the list of topics it covers.`;
}
/**
* Load everything needed to render the track.
* @returns {Promise<{themes:string[], days:Array, themeTopicMap:Map}>}
*/
export async function getOnboardingPlan() {
const [topics, activeVersion] = await Promise.all([
db.getTopics(),
db.getActiveCurriculumVersion(),
]);
return buildOnboardingPlan(topics, activeVersion);
}
/**
* Get the cached onboarding overview for a theme, or generate + cache it.
* Regenerates when the theme's topic set has changed (fingerprint mismatch).
*
* @param {string} theme
* @param {Array} topicsForTheme — topics belonging to the theme
* @param {object} [opts]
* @param {boolean} [opts.force=false]
*/
export async function getOrGenerateOnboardingOverview(theme, topicsForTheme, { force = false } = {}) {
if (!theme) throw new Error('Onboarding overview requires a theme.');
const topics = Array.isArray(topicsForTheme) ? topicsForTheme : [];
const fingerprint = computeTopicsFingerprint(topics);
if (!force) {
const cached = await db.getOnboardingOverview(theme);
if (cached && cached.topics_fingerprint === fingerprint) return cached;
}
const result = await callLLM({
task: 'onboarding.overview',
tier: 'fast',
system: cachedSystem(ONBOARDING_OVERVIEW_SYSTEM),
user: buildOverviewPrompt(theme, topics),
tools: [EMIT_ONBOARDING_OVERVIEW_TOOL],
toolChoice: { type: 'tool', name: EMIT_ONBOARDING_OVERVIEW_TOOL.name },
maxTokens: 1500,
timeoutMs: 60_000,
});
const emitted = result.toolUses[0]?.input;
if (!emitted) throw new Error('AI did not return an onboarding overview. Please try again.');
return db.setOnboardingOverview(theme, emitted, fingerprint);
}
/**
* Set of theme names the user has marked complete.
* @param {string} userId
* @returns {Promise<Set<string>>}
*/
export async function getCompletedThemes(userId) {
const rows = await db.getCompletedOnboardingThemes(userId);
return new Set(rows.map((r) => r.theme));
}
/**
* Mark a theme complete for the user. Idempotent.
*/
export async function markThemeCompleted(userId, theme) {
return db.setOnboardingCompletion(userId, theme);
}
/**
* Progress summary for the Dashboard chip.
* @param {string} userId
*/
export async function getOnboardingSummary(userId) {
const [plan, completed] = await Promise.all([
getOnboardingPlan(),
getCompletedThemes(userId),
]);
return computeOnboardingProgress(plan.days, completed);
}

View File

@@ -63,6 +63,29 @@ export function buildIndex(topics) {
return index;
}
// Compound-word matching. Dutch is heavily compounding, so a user's word
// (`pensioenafspraken`) is a *different* token than the graph's labels
// (`pensioenregeling`, `partnerpensioen`), even though they share the stem
// `pensioen`. Exact TF-IDF scores those pairs at 0, so the relevant topics are
// never retrieved. These heuristics recover that recall at a reduced weight,
// so exact matches still dominate the ranking.
const PARTIAL_MIN_QUERY_LEN = 6; // only expand meaty query tokens
const PARTIAL_MIN_OVERLAP = 6; // shared stem / substring must be this long
const PARTIAL_WEIGHT = 0.4; // discount vs. an exact term hit
/** True when two distinct tokens share a long stem or one contains the other. */
function partialMatch(q, d) {
if (q === d) return false;
const shorter = q.length <= d.length ? q : d;
const longer = q.length <= d.length ? d : q;
if (shorter.length < PARTIAL_MIN_OVERLAP) return false;
if (longer.includes(shorter)) return true;
let n = 0;
const m = shorter.length;
while (n < m && q[n] === d[n]) n++;
return n >= PARTIAL_MIN_OVERLAP;
}
export function retrieveTopK(index, query, k = 10) {
if (!index || !index.N || !query) return [];
const qTokens = tokenize(query);
@@ -80,8 +103,19 @@ export function retrieveTopK(index, query, k = 10) {
let s = 0;
for (const t of qTokens) {
const f = tf.get(t);
if (!f) continue;
s += (1 + Math.log(f)) * idf(t);
if (f) {
s += (1 + Math.log(f)) * idf(t);
continue;
}
// No exact hit — try a compound-word match against this doc's terms.
if (t.length < PARTIAL_MIN_QUERY_LEN) continue;
let best = 0;
for (const [term, tf2] of tf) {
if (!partialMatch(t, term)) continue;
const w = PARTIAL_WEIGHT * (1 + Math.log(tf2)) * idf(term);
if (w > best) best = w;
}
s += best;
}
scores[i] = s;
}

View File

@@ -1,6 +1,6 @@
import { useState, useEffect } from 'react';
import { Link } from 'react-router-dom';
import { BookOpen, CheckSquare, Trophy, Sparkles, X, HelpCircle } from 'lucide-react';
import { BookOpen, CheckSquare, Trophy, Sparkles, X, HelpCircle, Rocket, ArrowRight } from 'lucide-react';
import { useApp } from '../store/AppContext';
import Card from '../components/ui/Card';
import Button from '../components/ui/Button';
@@ -9,6 +9,7 @@ import * as db from '../lib/db';
import { storage } from '../lib/storage';
import { getAssignedTopic } from '../lib/learningService';
import { getYearProgress, getCurriculumCycle, getCurriculumWeek, getActiveVersion } from '../lib/curriculumService';
import { getOnboardingSummary } from '../lib/onboardingService';
const Dashboard = () => {
const { state } = useApp();
@@ -25,6 +26,7 @@ const Dashboard = () => {
yearProgress: null,
hasCurriculum: false,
theme: '',
onboarding: null,
});
useEffect(() => {
@@ -69,6 +71,15 @@ const Dashboard = () => {
console.warn('[Dashboard] Could not load curriculum data:', e.message);
}
// Onboarding-track progress (independent of enrollment/curriculum). Guarded
// so a missing collection or empty KB never breaks the dashboard.
let onboarding = null;
try {
onboarding = await getOnboardingSummary(currentUser.id);
} catch (e) {
console.warn('[Dashboard] Could not load onboarding summary:', e.message);
}
setDashData({
topic,
learnDone,
@@ -80,13 +91,19 @@ const Dashboard = () => {
yearProgress,
hasCurriculum: curriculumExists,
theme: topic?.theme || '',
onboarding,
});
};
load();
}, [currentUser, weekNumber]);
const { topic, learnDone, testResult, top3, myRank, myPoints, activity, yearProgress, hasCurriculum: curriculumActive, theme } = dashData;
const { topic, learnDone, testResult, top3, myRank, myPoints, activity, yearProgress, hasCurriculum: curriculumActive, onboarding } = dashData;
const onboardingLabel = onboarding
? (onboarding.allDone
? 'Review onboarding'
: onboarding.daysCompleted > 0 ? 'Continue onboarding' : 'Start onboarding')
: 'Start onboarding';
const currentCycle = getCurriculumCycle(weekNumber);
const currWeek = getCurriculumWeek(weekNumber);
@@ -174,6 +191,31 @@ const Dashboard = () => {
</div>
))}
</div>
{/* Onboarding track CTA — only when there are themes to introduce. */}
{onboarding && onboarding.themesTotal > 0 && (
<div className="mt-5 pt-5 border-t border-bg-warm flex flex-col sm:flex-row sm:items-center justify-between gap-3">
<div className="flex items-center gap-3">
<div className="w-9 h-9 rounded-[var(--r-org)] bg-teal/10 text-teal flex items-center justify-center shrink-0">
<Rocket size={18} />
</div>
<div>
<p className="font-bold flex items-center gap-2">
New here? Take the 5-day onboarding
<Tag variant={onboarding.allDone ? 'success' : 'accent'} className="text-[10px]">
{onboarding.allDone ? 'Onboarding complete' : `${onboarding.daysCompleted}/${onboarding.dayCount} days`}
</Tag>
</p>
<p className="text-sm text-fg-muted">A light tour of every theme how Respellion works day to day and week to week.</p>
</div>
</div>
<Link to="/onboarding-track" className="shrink-0">
<Button variant="outline" className="whitespace-nowrap">
{onboardingLabel} <ArrowRight size={16} className="ml-1" />
</Button>
</Link>
</div>
)}
</Card>
)}

View File

@@ -0,0 +1,326 @@
import { useEffect, useMemo, useState } from 'react';
import { Link } from 'react-router-dom';
import {
Loader, Rocket, ArrowLeft, ChevronRight, CheckCircle2, Circle,
Sparkles, ListChecks, PartyPopper,
} from 'lucide-react';
import Card from '../components/ui/Card';
import Button from '../components/ui/Button';
import Tag from '../components/ui/Tag';
import { useApp } from '../store/AppContext';
import {
getOnboardingPlan,
getCompletedThemes,
getOrGenerateOnboardingOverview,
markThemeCompleted,
computeOnboardingProgress,
} from '../lib/onboardingService';
function normalizeContent(raw) {
if (!raw) return null;
if (typeof raw === 'string') {
try { return JSON.parse(raw); } catch { return null; }
}
return raw;
}
// ── Per-theme overview view ──────────────────────────────────────────────────
function OnboardingThemeView({ theme, topics, done, onBack, onDone }) {
const [record, setRecord] = useState(null);
const [loading, setLoading] = useState(true);
const [error, setError] = useState(null);
const [marked, setMarked] = useState(done);
useEffect(() => {
let cancelled = false;
setLoading(true);
setError(null);
getOrGenerateOnboardingOverview(theme, topics)
.then((rec) => { if (!cancelled) setRecord(rec); })
.catch((err) => { if (!cancelled) setError(err.message || 'Failed to load the overview.'); })
.finally(() => { if (!cancelled) setLoading(false); });
return () => { cancelled = true; };
}, [theme]); // eslint-disable-line react-hooks/exhaustive-deps
const content = normalizeContent(record?.content);
const handleDone = async () => {
if (marked) return;
setMarked(true);
await onDone(theme);
};
return (
<div className="space-y-6">
<button
type="button"
onClick={onBack}
className="inline-flex items-center text-teal font-medium text-sm hover:underline"
>
<ArrowLeft size={16} className="mr-1" /> Back to overview
</button>
{loading && (
<Card className="w-full text-center py-16">
<Loader size={48} className="mx-auto text-teal animate-spin mb-4" />
<p className="font-medium text-lg">Preparing this theme</p>
<p className="text-fg-muted text-sm mt-2">This may take 1030 seconds the first time the result is cached.</p>
</Card>
)}
{!loading && error && (
<Card className="w-full border border-red-200 bg-red-50 text-red-900 p-6">
<p className="font-bold mb-1">Could not load this theme</p>
<p className="text-sm">{error}</p>
</Card>
)}
{!loading && !error && content && (
<>
<Card className="w-full p-6">
<div className="flex items-center gap-2 mb-3">
<Sparkles size={18} className="text-teal" />
<Tag variant="dark" className="text-[10px] uppercase tracking-wide">Theme</Tag>
{marked && <Tag variant="success" className="text-[10px]">Done</Tag>}
</div>
<h2 className="text-2xl md:text-3xl font-bold text-teal mb-2">{content.title}</h2>
<p className="text-fg-muted">{content.what_it_is}</p>
</Card>
<Card className="w-full p-6">
<h3 className="text-lg font-bold mb-2">Why it matters here</h3>
<p className="text-fg-muted">{content.why_it_matters}</p>
</Card>
<Card className="w-full p-6">
<h3 className="text-lg font-bold mb-3">Key points</h3>
<ul className="list-disc pl-5 space-y-1 text-sm">
{(content.key_points || []).map((p, i) => <li key={i}>{p}</li>)}
</ul>
</Card>
{Array.isArray(content.topics_covered) && content.topics_covered.length > 0 && (
<Card className="w-full p-6">
<h3 className="text-lg font-bold mb-3">Topics in this theme</h3>
<div className="flex flex-wrap gap-2">
{content.topics_covered.map((t) => (
<Tag key={t.topic_id} variant="default" className="text-xs">{t.label}</Tag>
))}
</div>
</Card>
)}
<div className="flex justify-end">
<Button onClick={handleDone} disabled={marked}>
{marked
? <span className="flex items-center"><CheckCircle2 size={16} className="mr-2" /> Marked as done</span>
: 'Mark as done'}
</Button>
</div>
</>
)}
</div>
);
}
// ── Progress ring ─────────────────────────────────────────────────────────────
function ProgressRing({ percentage }) {
return (
<div className="relative w-20 h-20 shrink-0">
<svg viewBox="0 0 36 36" className="w-20 h-20 -rotate-90">
<circle cx="18" cy="18" r="15.9155" fill="none" stroke="var(--bg-warm)" strokeWidth="3" />
<circle
cx="18" cy="18" r="15.9155" fill="none" stroke="var(--teal)" strokeWidth="3"
strokeDasharray={`${percentage} 100`} strokeLinecap="round"
/>
</svg>
<div className="absolute inset-0 flex items-center justify-center text-sm font-bold">{percentage}%</div>
</div>
);
}
// ── Page ──────────────────────────────────────────────────────────────────────
export default function OnboardingTrack() {
const { state } = useApp();
const { currentUser } = state;
const [plan, setPlan] = useState(null);
const [completed, setCompleted] = useState(new Set());
const [loading, setLoading] = useState(true);
const [error, setError] = useState(null);
const [selectedTheme, setSelectedTheme] = useState(null);
useEffect(() => {
if (!currentUser) return;
let cancelled = false;
setLoading(true);
Promise.all([getOnboardingPlan(), getCompletedThemes(currentUser.id)])
.then(([p, done]) => { if (!cancelled) { setPlan(p); setCompleted(done); } })
.catch((err) => { if (!cancelled) setError(err.message || 'Failed to load the onboarding track.'); })
.finally(() => { if (!cancelled) setLoading(false); });
return () => { cancelled = true; };
}, [currentUser]);
const progress = useMemo(
() => computeOnboardingProgress(plan?.days || [], completed),
[plan, completed],
);
const handleThemeDone = async (theme) => {
await markThemeCompleted(currentUser.id, theme);
setCompleted((prev) => new Set(prev).add(theme));
setSelectedTheme(null);
};
if (loading) {
return (
<div className="p-6 md:p-10">
<Card className="w-full text-center py-16">
<Loader size={48} className="mx-auto text-teal animate-spin mb-4" />
<p className="font-medium text-lg">Loading your onboarding track</p>
</Card>
</div>
);
}
if (error) {
return (
<div className="p-6 md:p-10">
<Card className="w-full border border-red-200 bg-red-50 text-red-900 p-6">
<p className="font-bold mb-1">Could not load the onboarding track</p>
<p className="text-sm">{error}</p>
</Card>
</div>
);
}
// Empty KB → friendly empty state.
if (!plan || plan.themes.length === 0) {
return (
<div className="p-6 md:p-10 space-y-6">
<header>
<h1 className="text-3xl md:text-4xl font-bold mb-2 flex items-center gap-3"><Rocket size={28} className="text-teal" /> Onboarding</h1>
</header>
<Card className="w-full p-8 text-center">
<p className="text-fg-muted">There are no themes to introduce yet. Once the knowledge base has content, your 5-day onboarding track will appear here.</p>
<div className="mt-4"><Link to="/"><Button variant="outline">Back to dashboard</Button></Link></div>
</Card>
</div>
);
}
const themeTopics = (theme) => plan.themeTopicMap.get(theme) || [];
if (selectedTheme) {
return (
<div className="p-6 md:p-10 animate-in fade-in slide-in-from-bottom-4 duration-500">
<OnboardingThemeView
theme={selectedTheme}
topics={themeTopics(selectedTheme)}
done={completed.has(selectedTheme)}
onBack={() => setSelectedTheme(null)}
onDone={handleThemeDone}
/>
</div>
);
}
return (
<div className="p-6 md:p-10 space-y-8 animate-in fade-in slide-in-from-bottom-4 duration-500">
<header className="flex items-start justify-between gap-4">
<div>
<h1 className="text-3xl md:text-4xl font-bold mb-2 flex items-center gap-3"><Rocket size={28} className="text-teal" /> Onboarding</h1>
<p className="text-fg-muted text-lg max-w-2xl">
A light, self-paced tour of every theme at Respellion, spread over about five days.
It gives you the breadth you need to understand how we work day to day and week to week
you can go faster if you like.
</p>
</div>
</header>
<Card className="border border-bg-warm">
<div className="flex items-center gap-5">
<ProgressRing percentage={progress.percentage} />
<div className="flex-1">
<p className="font-bold text-lg">
{progress.allDone
? 'Onboarding complete'
: `${progress.daysCompleted}/${progress.dayCount} days complete`}
</p>
<p className="text-fg-muted text-sm">{progress.themesDone} of {progress.themesTotal} themes done</p>
<div className="flex gap-1.5 mt-3">
{plan.days.map((d) => {
const dayDone = d.themes.every((t) => completed.has(t));
return (
<div
key={d.day}
title={`Day ${d.day}`}
className={`h-2 flex-1 rounded-full ${dayDone ? 'bg-teal' : 'bg-bg-warm'}`}
/>
);
})}
</div>
</div>
</div>
</Card>
{progress.allDone && (
<Card className="w-full p-6 border border-teal/30 bg-sage/40">
<div className="flex items-start gap-3">
<PartyPopper size={24} className="text-teal shrink-0 mt-0.5" />
<div>
<h3 className="text-lg font-bold">You've completed the onboarding track 🎉</h3>
<p className="text-fg-muted text-sm mt-1">
You've been introduced to every theme. You're ready to pick up a first assignment —
and you can revisit any theme below whenever you want a refresher.
</p>
<div className="mt-3"><Link to="/"><Button>Back to dashboard</Button></Link></div>
</div>
</div>
</Card>
)}
<div className="space-y-6">
{plan.days.map((d) => {
const doneCount = d.themes.filter((t) => completed.has(t)).length;
return (
<Card key={d.day} className="p-0 border border-bg-warm overflow-hidden">
<div className="flex items-center justify-between px-5 py-3 bg-bg-warm/40 border-b border-bg-warm">
<div className="flex items-center gap-2">
<Tag variant="dark" className="text-[10px] uppercase tracking-wide">Day {d.day}</Tag>
<ListChecks size={16} className="text-fg-muted" />
</div>
<span className="text-sm text-fg-muted">{doneCount}/{d.themes.length} done</span>
</div>
<div className="divide-y divide-bg-warm">
{d.themes.map((theme) => {
const isDone = completed.has(theme);
const count = themeTopics(theme).length;
return (
<button
key={theme}
type="button"
onClick={() => setSelectedTheme(theme)}
className="w-full flex items-center justify-between p-4 text-left hover:bg-bg-warm/30 transition-colors"
>
<div className="flex items-center gap-3">
{isDone
? <CheckCircle2 size={20} className="text-teal shrink-0" />
: <Circle size={20} className="text-fg-muted/40 shrink-0" />}
<div>
<p className="font-medium">{theme}</p>
<p className="text-sm text-fg-muted">{count} {count === 1 ? 'topic' : 'topics'}</p>
</div>
</div>
<ChevronRight size={18} className="text-fg-muted" />
</button>
);
})}
</div>
</Card>
);
})}
</div>
</div>
);
}