Add the trust and safety subsystem and the App Store compliance work

Implements the moderation and consent obligations a social platform carries,
so the web version and any client that speaks to it enforce the same rules.

Moderation core (services/moderation/, database/moderation.py): a reportable
target registry, the content filter and its choke points, the report queue with
atomic resolution, enforcement actions, consent tracking, maturity gating, and
account deletion with a grace window.

Surfaces: POST /reports plus the member report list, /admin/moderation and the
per-report admin view, /workspaces, terms acceptance at /auth/terms, consent and
account deletion under /profile, the report button and dialog partials, the
maturity gate, and the moderation stylesheet and ReportDialog client.

Every user-generated surface stays reportable by construction: new content tables
are registered in REPORTABLE_TARGETS or listed in UNREPORTABLE_TABLES with a
reason, and the registry test fails the suite on anything left unclassified.

Docs: community guidelines, content moderation, intellectual property, privacy,
terms, contact, and the admin-only moderation operations page, plus the
moderation API group and the Devii moderation actions.

Compliance record: applecomp.md is the requirement register, applechanges.md the
gap analysis against this codebase, and appleimpl.md the implementation design
they resolve to.

Tests cover the report flow, admin moderation, consent, account deletion, terms
acceptance, workspaces, and the registry invariant across the unit, api, and e2e
tiers.

README.md

@@ -87,7 +87,10 @@ devplacepy/
 | `/bookmarks` | Save/unsave content; `/bookmarks/saved` personal list |
 | `/polls` | Vote on post-attached polls |
 | `/follow` | Follow/unfollow users |
-| `/block` | Block/unblock a user: hides all of their posts, comments and messages from you everywhere except their own profile, and stops them notifying you |
+| `/reports` | Content reporting: `POST /reports/{target_type}/{target_uid}` files a report against any user-generated surface, `GET /reports/mine` lists the reports you filed and their outcome, `GET /reports/reasons` serves the reason registry so every client renders the same dialog |
+| `/admin/moderation` | Admin **Moderation** queue: reported content oldest-open-first with the response-window badge, one report per detail page with the offender's history, triage (`/status`) and decisions (`/decide`) |
+| `/workspaces/index` | Public index of every workspace published to the ingress proxy, with owner, project, maturity label and direct link |
+| `/block` | Block/unblock a user: hides all of their posts, comments and messages from you everywhere except their own profile, and stops them notifying you. Also reachable directly from every content action bar |
 | `/mute` | Mute/unmute a user: stops them creating notifications for you while their content stays visible |
 | `/leaderboard` | Contributor ranking by total stars earned |
 | `/game` | **Code Farm** cooperative idle game (member-only): plant projects that build over real time, harvest coins and XP, upgrade CI, buy plots, and water friends' builds at `/game/farm/{username}`. Live over pub/sub; every endpoint negotiates JSON |
@@ -203,6 +206,20 @@ The farm refreshes live over the pub/sub bus (a watered build appears on the own
 
 XP awards are wired at the existing content-creation, vote, and follow hook points in the routers and centralized in `award_xp()` / `check_milestone_badges()` (`devplacepy/utils/`). Existing accounts have their XP and levels backfilled once from prior activity at startup (`init_db()`).
 
+## Trust and safety
+
+DevPlace is open in the sense that it does not editorialise technical opinion. It enforces a short, fixed list of prohibited categories, published as the [Community Guidelines](/docs/community-guidelines.html), and it has the machinery to make that enforcement real.
+
+- **Automated filtering at post time.** Every user-authored text passes a classifier at the five places content is created or changed (content creation, content editing, comments, direct messages, profile and signup fields), covering every surface with no per-route code. The default mode is `review`, not `block`: a match is published **and** raises a report for a human, because a developer platform discusses exploits, malware analysis and violent subject matter as its work and a machine that suppressed that would destroy the product. Only sexual and exploitative content is refused outright. Classification failure falls back to review, never to publication. Tunable live at `/admin/settings` (`moderation_filter_mode`: `off`, `label`, `review`, `block`).
+- **A report control on every surface.** Posts, comments, gists, projects, project files, news, uploads, direct messages, quizzes, polls, awards, profiles, issues, workspaces and assistant output are all reportable through one polymorphic endpoint keyed on `(target_type, target_uid)`, from one dialog, with one reason list. Reports are private to the reporter; the reported person is never told who filed.
+- **One queue with a published response window.** `/admin/moderation` is worked oldest-open-first and carries a badge showing the age of the oldest unresolved report against `moderation_sla_hours` (default 24), so a breach is visible rather than assumed. Resolution is a single atomic conditional update: two administrators deciding at once produce exactly one decision, the second gets a 409.
+- **Real enforcement, with a statement of reasons.** Remove or restore content, warn, suspend for a stated period, ban, lift, dismiss, or escalate. Every decision writes a permanent `moderation_actions` row plus an audit event, and tells the affected user what was decided and why. A suspended account can still read, still see why, still report, and still delete itself; it cannot create. A junior administrator can never action a senior one.
+- **Blocking**, unchanged and independent of all of the above, is now reachable from the content itself as well as from a profile.
+- **Age and maturity.** Signup collects a date of birth, derives an age band, and **discards the date**; accounts below `moderation_minimum_age` (default 16) are refused. Content labelled mature is hidden behind an interstitial until the viewer explicitly opts in, and the reveal is never offered to a minor age band for restricted content.
+- **Consent.** Five versioned, independently withdrawable consents (`terms`, `privacy`, `ai_third_party`, `activity_recording`, `container_credentials`) with full history, managed at `/profile/{username}?tab=privacy` and changeable **only by the account holder** - an administrator reads the record but never grants or withdraws on someone else's behalf. **No content is sent to a third-party AI provider without `ai_third_party` consent**, enforced once at the gateway; the per-feature AI toggles remain as preferences subordinate to it. Withdrawing `activity_recording` stops presence recording. While recording is on, an indicator says so on every page. `container_credentials` is what lets software another member runs in a container receive your API key; running your own container never asks.
+- **Account deletion.** Self-service at `/profile/{username}/delete`, reauthenticated with the account password. Sessions and tokens are revoked, the profile is anonymised immediately, and all content is removed under one deletion event that stays restorable for `account_deletion_grace_hours` (default 24) before `devplace accounts prune` (and the Moderation housekeeping service) purges it permanently. The devRant `DELETE /api/users/me` routes into the same cascade.
+- **Legal pages**: [Terms of Service](/docs/terms.html), [Community Guidelines](/docs/community-guidelines.html), [Privacy Policy](/docs/privacy.html), [How moderation works](/docs/content-moderation.html), [Notice and takedown](/docs/intellectual-property.html) and [Contact](/docs/contact.html). All six are indexed in the sitemap and reachable from the docs index; the footer of every page links Terms, Privacy, Community Guidelines and Contact. Contact details come from the `contact_email`, `contact_phone` and `contact_address` settings, so the in-product page and any app-store trader declaration have one source of truth.
+
 ## Vibe coding (Alpha, admin only)
 
 Build software by talking to an AI agent instead of typing every line. Create a project for storage, attach a container to it (the shared `ppy` image, your files mounted at `/app`), start it, and open a terminal. The whole flow is drivable conversationally through Devii. Inside every container three agents ship preinstalled and run on **your own API key**, so all AI usage is metered to your account: **DevPlace Code (`dpc`)**, a coding agent in the same class as Claude Code; **`botje.py`**, a plug-and-play DevPlace bot you can copy and customise; and **`pagent`**, a minimal zero-dependency agent. Each container is launched with `DEVPLACE_BASE_URL`, `DEVPLACE_OPENAI_URL`, `DEVPLACE_API_KEY`, `DEVPLACE_USER_UID`, `DEVPLACE_CONTAINER_NAME`, `DEVPLACE_CONTAINER_UID`, and `DEVPLACE_INGRESS_URL` already set. Publish a container port to a public URL at `/p/<slug>` by setting an `ingress_slug` and `ingress_port` (ask Devii to do it at create time). The feature is in **Alpha** and currently limited to administrators; the full walkthrough, including a tutorial that vibes a web app and puts it online, is at `/docs/getting-started-vibing.html`.