Skip to content

CHANGELOG.md

Latest commit

 

History

History
226 lines (176 loc) · 11.8 KB

CHANGELOG.md

File metadata and controls

226 lines (176 loc) · 11.8 KB

Changelog

All notable changes to Numen are documented here.

[Unreleased]

Added

AI Pipeline Templates & Preset Library (Issue #36)

Reusable AI pipeline templates for accelerated content creation workflows, featuring 8 built-in templates, community library, space-scoped templates, one-click install wizard, template versioning, and plugin registration hooks.

Features:

  • 8 built-in templates: Blog Post Pipeline, Social Media Campaign, Product Description, Email Newsletter, Press Release, Landing Page, Technical Documentation, Video Script
  • Template library API: Discover, rate, and install templates with metadata
  • Space-scoped templates: Custom templates per content space with full RBAC support
  • Install wizard: Auto-configures personas, stages, and input variables from template schema
  • Template versioning: Track changes, publish/unpublish versions, rollback support
  • Template packs: Plugin system for registering template collections
  • Community ratings: Rate and provide feedback on templates
  • Metadata support: Categories, icons, author info, and schema versioning
  • Security: Space-scoped installs, RBAC permission gates, audit logging

Endpoints:

  • GET /api/v1/spaces/{space}/pipeline-templates — List templates (paginated)
  • POST /api/v1/spaces/{space}/pipeline-templates — Create custom template
  • GET /api/v1/spaces/{space}/pipeline-templates/{template} — Get template details
  • PATCH /api/v1/spaces/{space}/pipeline-templates/{template} — Update template
  • DELETE /api/v1/spaces/{space}/pipeline-templates/{template} — Delete template
  • POST /api/v1/spaces/{space}/pipeline-templates/{template}/publish — Publish template
  • POST /api/v1/spaces/{space}/pipeline-templates/{template}/unpublish — Unpublish template
  • GET /api/v1/spaces/{space}/pipeline-templates/{template}/versions — List versions
  • POST /api/v1/spaces/{space}/pipeline-templates/{template}/versions — Create version
  • GET /api/v1/spaces/{space}/pipeline-templates/{template}/versions/{version} — Get version
  • POST /api/v1/spaces/{space}/pipeline-templates/installs/{version} — Install template (rate-limited 5/min)
  • PATCH /api/v1/spaces/{space}/pipeline-templates/installs/{install} — Update install
  • DELETE /api/v1/spaces/{space}/pipeline-templates/installs/{install} — Remove install
  • GET /api/v1/spaces/{space}/pipeline-templates/{template}/ratings — List ratings
  • POST /api/v1/spaces/{space}/pipeline-templates/{template}/ratings — Rate template

Plugin Hooks:

  • registerTemplateCategory(array $category) — Register custom template categories
  • registerTemplatePack(array $pack) — Register template collections from plugins

Models:

  • PipelineTemplate — Template metadata (name, slug, category, icon, author)
  • PipelineTemplateVersion — Versioned template definitions with JSON schema
  • PipelineTemplateInstall — Track template usage per space
  • PipelineTemplateRating — Community feedback (1-5 stars)

New environment variables:

  • TEMPLATE_LIBRARY_ENABLED=true
  • TEMPLATE_INSTALL_RATE_LIMIT=5 (per minute)

See docs/pipeline-templates.md for complete documentation.

Added

  • Webhooks admin UI — manage webhook endpoints, event subscriptions, delivery logs, and secret rotation directly from the admin panel (Settings → Webhooks)

Added — #37 Competitor-Aware Content Differentiation

Infrastructure

  • 6 new database tables with ULID PKs and idempotent migrations:
    • competitor_sources — crawlable competitor feeds (RSS, sitemap, scrape, API)
    • competitor_content_items — crawled articles with dedup by content hash
    • content_fingerprints — morphic TF-IDF/keyword fingerprints for similarity
    • differentiation_analyses — LLM-assisted differentiation scoring
    • competitor_alerts — configurable alert rules
    • competitor_alert_events — fired alert history with notification log

Crawler Infrastructure (Chunk 1)

  • CrawlerService — orchestrates RSS, sitemap, scrape, and API crawlers
  • RssCrawler, SitemapCrawler, ScrapeCrawler, ApiCrawler — pluggable crawlers
  • CrawlCompetitorSourceJob — queued job with retries + stale-check

Fingerprinting & Similarity (Chunk 2-3)

  • ContentFingerprintService — TF-IDF vectorization over content body
  • SimilarityCalculator — cosine similarity between fingerprint vectors
  • SimilarContentFinder — finds the top-N most similar competitor items

Differentiation Analysis Engine (Chunk 4)

  • DifferentiationAnalysisService — LLM-powered angle/gap/recommendation extraction
  • DifferentiationResult — typed value object for analysis output
  • Pipeline stage CompetitorAnalysisStage — integrates into the content pipeline

Pipeline Integration (Chunk 5)

  • CompetitorAnalysisStage wired into StageRegistry
  • Automatic enrichment of ContentBrief with competitor insights on pipeline run

Alert System (Chunk 6)

  • CompetitorAlertService — evaluates active alerts against new competitor content
  • CheckCompetitorAlertsJob — queued job dispatched post-crawl
  • CompetitorAlertNotification — Laravel notification (email channel)
  • SlackChannel — Block Kit Slack webhook notifications
  • WebhookChannel — generic HTTP webhook with structured JSON payload
  • Alert types: new_content, keyword, high_similarity

Knowledge Graph Integration (Chunk 7)

  • CompetitorGraphIndexer — creates virtual nodes + competitor_similarity edges
  • Reuses existing content_graph_nodes / content_graph_edges tables from #14
  • Competitor items indexed with deterministic node IDs (SHA-1 prefix)

REST API (Chunk 8)

  • CompetitorSourceController — CRUD for competitor sources
  • CompetitorController — content listing, crawl trigger, alert CRUD
  • DifferentiationController — analysis listing + summary endpoint
  • Form requests with full validation
  • JSON:API-style resources

Security Hardening (Chunk 9)

  • Input validation on all competitor source URLs (must match protocol/domain whitelist)
  • Rate limiting on crawlers (500 req/day per source)
  • Auth: All endpoints require manage-competitors permission
  • CORS disabled for competitor data (internal only)
  • All components use Composition API + TypeScript

Monitoring & Retention (Chunk 10)

  • CrawlerHealthMonitor — detects stale/high-error sources, logs warnings
  • RetentionPolicyService — prunes old content/analyses/events on configurable schedule
  • Scheduler entries: health check (hourly), retention prune (weekly Sun 02:00)
  • OpenAPI 3.1 spec: docs/competitor-differentiation-api.yaml
  • Blog post: docs/blog-competitor-differentiation.md

Configuration

COMPETITOR_ANALYSIS_ENABLED=true
COMPETITOR_SIMILARITY_THRESHOLD=0.25
COMPETITOR_MAX_ANALYZE=5
COMPETITOR_AUTO_ENRICH_BRIEFS=true
COMPETITOR_CONTENT_RETENTION_DAYS=90
COMPETITOR_ANALYSIS_RETENTION_DAYS=180
COMPETITOR_ALERT_EVENT_RETENTION_DAYS=30

[0.8.0] — 2026-03-15

Added

Media Library & Digital Asset Management (Discussion #4)

A complete digital asset management (DAM) system for organizing, tagging, editing, and serving media assets. Built for multi-format content delivery and CDN integration.

Features:

  • Folders & Collections — Organize assets hierarchically using adjacency-list folders. Create smart collections with powerful filtering and bulk operations.
  • Drag-and-drop Upload — Metadata extraction (MIME type, dimensions, file size, duration) on ingest. Progress tracking and batch upload support.
  • AI Auto-tagging (opt-in) — Enable MEDIA_AI_TAGGING environment variable to automatically tag images using Claude vision. Powered by Anthropic API; all costs logged to AIGenerationLog.
  • Image Editing — Crop, rotate, and resize images via MediaEditController. Changes create new variants; originals are preserved.
  • Automatic Variant Generation — On upload, generate thumb (150×150), medium (600×600), and large (1600×1600) variants. WebP format with configurable quality. Stored locally or on S3 (via FILESYSTEM_DISK).
  • Usage Tracking — Query which content items reference a specific asset. Prevents accidental deletion of in-use media.
  • Public Headless API — /v1/public/media endpoints (no auth required) with throttle protection (120 req/min). Perfect for headless frontends and CDN edge caching.
  • Full REST API — Complete CRUD operations on assets, folders, and collections. Bearer token auth via Sanctum.
  • MediaPicker Vue Component — Integrates with content editor for seamless asset selection during content creation.

Environment Variables (new):

  • MEDIA_AI_TAGGING — Enable automatic AI-based image tagging (default: false)
  • CDN_ENABLED — Enable public CDN delivery endpoints (default: true)

API Endpoints:

Authenticated (requires Bearer token):

  • GET /v1/media — List all assets
  • POST /v1/media — Upload asset (20 req/min throttle)
  • GET /v1/media/{asset} — Fetch asset details
  • PATCH /v1/media/{asset} — Update asset metadata
  • DELETE /v1/media/{asset} — Delete asset
  • PATCH /v1/media/{asset}/move — Move to folder
  • GET /v1/media/{asset}/usage — Show usage in content
  • POST /v1/media/{asset}/edit — Edit (crop/rotate/resize)
  • GET /v1/media/{asset}/variants — List generated variants
  • GET|POST /v1/media/folders — CRUD folders
  • PATCH /v1/media/folders/{folder}/move — Move folder
  • GET|POST|PATCH|DELETE /v1/media/collections — CRUD collections
  • POST|DELETE /v1/media/collections/{collection}/items — Manage collection items

Public (no auth):

  • GET /v1/public/media — List public assets (120 req/min throttle)
  • GET /v1/public/media/{asset} — Fetch public asset
  • GET /v1/public/media/collections/{collection} — Fetch collection

Planned

  • Remove legacy numen.anthropic config block (duplicates numen.providers.anthropic)
  • AgentContract interface extracted from Agent abstract class

[0.9.0] — 2026-03-15

Added

Multi-Language & i18n Support (Discussion #7)

Full content localization with AI-powered translation, space-level locale management, and intelligent fallback chains.

Locale Management:

  • Space-level locale configuration: add/remove/reorder locales, set default locale
  • Intelligent 5-step fallback chain (exact match → language prefix → fallback config → space default → "en")
  • Prevents invalid locale codes with Locale validation class (BCP 47 compliant)

AI-Powered Translation:

  • Tone-aware translation using existing Persona system — respects content creator's voice
  • Async job queue (TranslateContentJob on ai-pipeline queue) for background processing
  • Translation status tracking: pending → completed/failed with error logging
  • Job retry support with configurable backoff

Translation Workflow:

  • Translation matrix view showing per-content, per-locale coverage and status
  • Side-by-side translation editor for reviewing AI-generated translations
  • Manual translation support via API
  • Batch translation operations with progress reporting

REST API Endpoints:

  • Locale Management: GET/POST/PATCH/DELETE /api/v1/locales (space-scoped)
  • Translation Workflow: POST /api/v1/content/{content}/translate, GET /api/v1/content/{content}/translations
  • Translation Matrix: GET /api/v1/translations/matrix with pagination and status filters
  • Supported Locales: GET /api/v1/locales/supported for available BCP 47 codes

Locale Awareness:

  • Middleware: SetLocaleFromRequest respects Accept-Language header, ?locale= query param, and X-Locale header
  • API responses include current locale context; content delivery selects best-match locale automatically