AI (Phase 7)

CreatorOS uses Anthropic Claude for text and a placeholder renderer for graphics. Everything the AI creates is a draft: inactive products, status = draft posts, disabled automations, media assets flagged ai_generated. Publishing, activating, pricing and deleting always require a person in the dashboard.

Configuration

Variable	Effect
`ANTHROPIC_API_KEY`	Blank → `MockAIProvider` (deterministic, every output labelled "(mock)"), dashboard shows an amber banner.
`ANTHROPIC_MODEL`	Model id sent to the Messages API (default `claude-sonnet-5`; `claude-opus-5` recommended for copilot quality).
`IMAGE_PROVIDER`	Only `mock` exists today (GD gradient + headline). Real provider is an open decision (CLAUDE.md §15).

Plan limits: tenants.limits.ai_monthly (free 20 / creator 500 / network 5000). Each generation inserts one ai_usage row (tokens + estimated cost in micros); exceeding the limit returns HTTP 429 "AI limit reached". GET /dashboard/ai/usage shows the log.

Code map

app/Providers/AI/AIProvider.php — interface. AnthropicProvider (raw Messages API via App\Core\Http: system, adaptive thinking, tools + runTools() tool-result loop, output_config.format JSON schema with prompt fallback, refusal handling). MockAIProvider for keyless dev/test.
app/Providers/AI/prompts/*.vN.md — versioned templates loaded by PromptLoader with {{var}} substitution. Bump the version suffix instead of editing a template in place.
app/Providers/Image/ — ImageProvider, MockImageProvider, Presets (ig_post, ig_story, facebook, tiktok, yt_thumb, discord_banner, twitch_panel, kick, x_header).
app/Modules/AI/Services/AIService.php — bound as $app->make('ai'). Methods: complete, structured, generateContent, adaptPost(tenantId, body, providers), contentPlan, summarizeCreator, serverSetup, graphicsBrief, renderGraphic, copilot.
app/Modules/AI/Services/CopilotTools.php — tool definitions and executors (all reads via TenantScope; writes are drafts and audited).

Routes (auth + tenant + role:staff; POSTs throttled `throttle:ai`)

Route	Purpose
`GET/POST /dashboard/copilot`, `POST /dashboard/copilot/reset`	Chat copilot; conversation in session per tenant (last 20 turns). JSON in/out.
`GET/POST /dashboard/ai/content`, `POST /dashboard/ai/content/send`	Content studio (captions, hashtags, titles, descriptions, CTAs, thumbnail text, short-form). `?plan=1` renders a 7-day plan. "→ Publisher" creates a draft post.
`GET/POST /dashboard/ai/graphics`, `GET /dashboard/ai/gallery`	Preset-size graphics saved to storage + `media_assets`.
`GET /dashboard/onboarding-wizard`, `POST /dashboard/onboarding-wizard/{1-5}`	Creator type → accounts → AI summary (Apply buttons) → community/server → "Launch your community" campaign drafts.
`GET/POST /dashboard/ai/server-setup`, `POST /dashboard/ai/server-setup/apply`	Plain-language description → launch kit (VIP product inactive, posts draft, automation disabled, graphic, FAQ/rules proposals).
`GET /dashboard/ai/usage`	Metering.

Copilot tools

get_overview, list_products, analyze_sales (read-only) · create_draft_post, create_draft_product, generate_graphic, create_automation_draft (draft-only, staff role required, audited as ai.*). The system prompt (copilot.system.v1.md) instructs the model to refuse publish/price/refund requests and point at the dashboard.

Adding a real image provider