feat: consolidate SiteContext into SITE.md with auto-refresh

data-machine.php

@@ -500,6 +500,10 @@ function datamachine_activate_for_site() {
 	// Migrate USER.md to network-scoped paths and create NETWORK.md on multisite (idempotent).
 	datamachine_migrate_user_md_to_network_scope();
 
+	// Regenerate SITE.md with enriched content and clean up legacy SiteContext transient.
+	datamachine_regenerate_site_md();
+	delete_transient( 'datamachine_site_context_data' );
+
 	// Clean up legacy per-agent-type log level options (idempotent).
 	foreach ( array( 'pipeline', 'chat', 'system' ) as $legacy_agent_type ) {
 		delete_option( "datamachine_log_level_{$legacy_agent_type}" );

inc/Core/WordPress/SiteContext.php

@@ -1,6 +1,14 @@
 <?php
 /**
- * Cached WordPress site metadata for AI context injection.
+ * Cached WordPress site metadata — DEPRECATED.
+ *
+ * Site context is now provided by SITE.md (auto-regenerated on disk via
+ * datamachine_regenerate_site_md()). This class remains for backward
+ * compatibility with code that references the class name or its static
+ * methods. It will be removed in a future major version.
+ *
+ * @package DataMachine\Core\WordPress
+ * @deprecated 0.48.0 Site context is now provided by SITE.md auto-regeneration.
  */
 
 namespace DataMachine\Core\WordPress;
@@ -12,137 +20,19 @@ class SiteContext {
 	const CACHE_KEY = 'datamachine_site_context_data';
 
 	/**
-	 * Get site context data with automatic caching.
+	 * Get site context data.
 	 *
-	 * Plugins can extend the context data via the 'datamachine_site_context' filter.
-	 * Note: Filtering bypasses cache to ensure dynamic data is always fresh.
-	 *
-	 * @return array Site metadata, post types, and taxonomies
+	 * @deprecated 0.48.0 Site context is now in SITE.md. This method remains for backward compat.
+	 * @return array Site metadata (empty — context is now file-based).
 	 */
 	public static function get_context(): array {
-		$cached = get_transient( self::CACHE_KEY );
-
-		// Clear stale cache if date has changed (ensures current_date is always accurate).
-		if ( false !== $cached && isset( $cached['site']['current_date'] ) ) {
-			if ( wp_date( 'Y-m-d' ) !== $cached['site']['current_date'] ) {
-				delete_transient( self::CACHE_KEY );
-				$cached = false;
-			}
-		}
-
-		if ( false !== $cached ) {
-			return $cached;
-		}
-
-		$context = array(
-			'site'       => self::get_site_metadata(),
-			'post_types' => self::get_post_types_data(),
-			'taxonomies' => self::get_taxonomies_data(),
-		);
-
-		/**
-		 * Filter site context data before caching.
-		 *
-		 * Plugins can use this hook to inject custom context data (e.g., events,
-		 * analytics, custom post type summaries). Note: When this filter is used,
-		 * caching is bypassed to ensure dynamic data remains fresh.
-		 *
-		 * @param array $context Site context data with 'site', 'post_types', 'taxonomies' keys
-		 * @return array Modified context data
-		 */
-		$context = apply_filters( 'datamachine_site_context', $context );
-
-		set_transient( self::CACHE_KEY, $context, 0 ); // 0 = permanent until invalidated
-
-		return $context;
-	}
-
-	/**
-	 * Get site metadata.
-	 *
-	 * @return array Site name, URL, language, timezone, current_date
-	 */
-	private static function get_site_metadata(): array {
-		return array(
-			'name'         => get_bloginfo( 'name' ),
-			'tagline'      => get_bloginfo( 'description' ),
-			'url'          => home_url(),
-			'admin_url'    => admin_url(),
-			'language'     => get_locale(),
-			'timezone'     => wp_timezone_string(),
-			'current_date' => wp_date( 'Y-m-d' ),
-		);
-	}
-
-	/**
-	 * Get public post types with published counts.
-	 *
-	 * @return array Post type labels, counts, and hierarchy status
-	 */
-	private static function get_post_types_data(): array {
-		$post_types_data = array();
-		$post_types      = get_post_types( array( 'public' => true ), 'objects' );
-
-		foreach ( $post_types as $post_type ) {
-			$count           = wp_count_posts( $post_type->name );
-			$published_count = $count->publish ?? 0;
-
-			$post_types_data[ $post_type->name ] = array(
-				'label'          => $post_type->label,
-				'singular_label' => ( is_object( $post_type->labels ) && isset( $post_type->labels->singular_name ) )
-					? $post_type->labels->singular_name
-					: $post_type->label,
-				'count'          => (int) $published_count,
-				'hierarchical'   => $post_type->hierarchical,
-			);
-		}
-
-		return $post_types_data;
-	}
-
-	/**
-	 * Get public taxonomies with metadata and term counts.
-	 *
-	 * Returns taxonomy structure without individual term listings to keep
-	 * context payload small. Use search_taxonomy_terms tool for term discovery.
-	 *
-	 * @return array Taxonomy labels, term counts, hierarchy, post type associations
-	 */
-	private static function get_taxonomies_data(): array {
-		$taxonomies_data = array();
-		$taxonomies      = get_taxonomies( array( 'public' => true ), 'objects' );
-
-		foreach ( $taxonomies as $taxonomy ) {
-			if ( \DataMachine\Core\WordPress\TaxonomyHandler::shouldSkipTaxonomy( $taxonomy->name ) ) {
-				continue;
-			}
-
-			$term_count = wp_count_terms(
-				array(
-					'taxonomy'   => $taxonomy->name,
-					'hide_empty' => false,
-				)
-			);
-			if ( is_wp_error( $term_count ) ) {
-				$term_count = 0;
-			}
-
-			$taxonomies_data[ $taxonomy->name ] = array(
-				'label'          => $taxonomy->label,
-				'singular_label' => ( is_object( $taxonomy->labels ) && isset( $taxonomy->labels->singular_name ) )
-					? $taxonomy->labels->singular_name
-					: $taxonomy->label,
-				'term_count'     => (int) $term_count,
-				'hierarchical'   => $taxonomy->hierarchical,
-				'post_types'     => $taxonomy->object_type ?? array(),
-			);
-		}
-
-		return $taxonomies_data;
+		return array();
 	}
 
 	/**
 	 * Clear site context cache.
+	 *
+	 * @deprecated 0.48.0 SITE.md regeneration handles freshness. Cleans up legacy transient.
 	 */
 	public static function clear_cache(): void {
 		delete_transient( self::CACHE_KEY );
@@ -151,29 +41,10 @@ public static function clear_cache(): void {
 	/**
 	 * Register automatic cache invalidation hooks.
 	 *
-	 * Clears cache when posts, terms, or site options change.
-	 * Comprehensive invalidation hooks eliminate need for time-based expiration.
+	 * @deprecated 0.48.0 Use datamachine_register_site_md_invalidation() instead.
 	 */
 	public static function register_cache_invalidation(): void {
-		add_action( 'save_post', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'delete_post', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'wp_trash_post', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'untrash_post', array( __CLASS__, 'clear_cache' ) );
-
-		add_action( 'create_term', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'edit_term', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'delete_term', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'set_object_terms', array( __CLASS__, 'clear_cache' ) );
-
-		add_action( 'user_register', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'delete_user', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'set_user_role', array( __CLASS__, 'clear_cache' ) );
-
-		add_action( 'switch_theme', array( __CLASS__, 'clear_cache' ) );
-
-		add_action( 'update_option_blogname', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'update_option_blogdescription', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'update_option_home', array( __CLASS__, 'clear_cache' ) );
-		add_action( 'update_option_siteurl', array( __CLASS__, 'clear_cache' ) );
+		// No-op. SITE.md invalidation hooks are registered in bootstrap.php.
+		// This method remains to avoid fatal errors from code calling it directly.
 	}
 }

inc/Engine/AI/Directives/SiteContextDirective.php

@@ -1,81 +1,52 @@
 <?php
 /**
- * Site Context Directive - Priority 80 (Lowest Priority)
+ * Site Context Directive — DEPRECATED.
  *
- * Injects WordPress site context information as the final directive in the
- * AI directive system. Provides comprehensive site metadata including
- * posts, taxonomies, users, and configuration. Toggleable via settings.
+ * This directive previously injected a JSON blob of WordPress site metadata
+ * at priority 80. Site context is now provided entirely by SITE.md via the
+ * CoreMemoryFilesDirective at priority 20.
  *
- * Priority Order in Directive System:
- * 1. Priority 10 - Plugin Core Directive
- * 2. Priority 20 - Core Memory Files (SITE.md, RULES.md, SOUL.md, MEMORY.md, USER.md)
- * 3. Priority 40 - Pipeline Memory Files (per-pipeline selectable)
- * 4. Priority 50 - Pipeline System Prompt
- * 5. Priority 60 - Pipeline Context Files
- * 6. Priority 70 - Tool Definitions and Workflow Context
- * 7. Priority 80 - WordPress Site Context (THIS CLASS)
+ * This class remains as a no-op for backward compatibility with code that
+ * references the class name (e.g., datamachine_site_context_directive filter
+ * consumers). It will be removed in a future major version.
+ *
+ * @package DataMachine\Engine\AI\Directives
+ * @since   0.30.0
+ * @deprecated 0.48.0 Site context is now provided by SITE.md. This directive is a no-op.
  */
 
 namespace DataMachine\Engine\AI\Directives;
 
-use DataMachine\Core\PluginSettings;
-use DataMachine\Core\WordPress\SiteContext;
-use DataMachine\Engine\AI\Directives\DirectiveInterface;
-
 defined( 'ABSPATH' ) || exit;
 
 class SiteContextDirective implements DirectiveInterface {
 
+	/**
+	 * Returns empty outputs — site context is now in SITE.md.
+	 *
+	 * @deprecated 0.48.0
+	 *
+	 * @param string      $provider_name AI provider name.
+	 * @param array       $tools         Available tools.
+	 * @param string|null $step_id       Pipeline step ID.
+	 * @param array       $payload       Additional payload data.
+	 * @return array Always returns empty array.
+	 */
 	public static function get_outputs( string $provider_name, array $tools, ?string $step_id = null, array $payload = array() ): array {
-		if ( ! self::is_site_context_enabled() ) {
-			return array();
-		}
-
-		$context_data = SiteContext::get_context();
-		if ( empty( $context_data ) || ! is_array( $context_data ) ) {
-			do_action( 'datamachine_log', 'warning', 'Site Context Directive: Empty context generated' );
-			return array();
-		}
-
-		return array(
-			array(
-				'type'  => 'system_json',
-				'label' => 'WORDPRESS SITE CONTEXT',
-				'data'  => $context_data,
-			),
-		);
+		return array();
 	}
 
 	/**
-	 * Check if site context injection is enabled in plugin settings.
+	 * Check if site context injection is enabled.
 	 *
-	 * @return bool True if enabled, false otherwise
+	 * @deprecated 0.48.0 Use the site_context_enabled setting directly. Controls SITE.md auto-refresh.
+	 * @return bool
 	 */
 	public static function is_site_context_enabled(): bool {
-		return PluginSettings::get( 'site_context_enabled', true );
+		return \DataMachine\Core\PluginSettings::get( 'site_context_enabled', true );
 	}
 }
 
-/**
- * Allow plugins to override the site context directive class.
- * datamachine-multisite uses this to replace single-site context with multisite context.
- *
- * @param string $directive_class The directive class to use for site context
- * @return string The filtered directive class
- */
-$datamachine_site_context_directive = apply_filters( 'datamachine_site_context_directive', SiteContextDirective::class );
-
-// Register the filtered directive for global context (applies to all AI agents - allows replacement by multisite plugin)
-if ( $datamachine_site_context_directive ) {
-	add_filter(
-		'datamachine_directives',
-		function ( $directives ) use ( $datamachine_site_context_directive ) {
-			$directives[] = array(
-				'class'    => $datamachine_site_context_directive,
-				'priority' => 80,
-				'contexts' => array( 'all' ),
-			);
-			return $directives;
-		}
-	);
-}
+// The directive is no longer registered in the directive system.
+// The class exists purely for backward compatibility with code that
+// references it by name (e.g., datamachine_site_context_directive filter).

inc/bootstrap.php

@@ -104,9 +104,10 @@
 	'label'       => 'Network Context',
 	'description' => 'WordPress multisite network topology and shared resources.',
 ) );
-// SiteContext is autoloaded (Core\WordPress\SiteContext) — register its cache invalidation hook here.
-add_action( 'init', array( \DataMachine\Core\WordPress\SiteContext::class, 'register_cache_invalidation' ) );
-require_once __DIR__ . '/Engine/AI/Directives/SiteContextDirective.php';
+// SITE.md auto-regeneration — replaces the former SiteContext + SiteContextDirective system.
+// SITE.md is now the single source of truth for site context, auto-refreshing on structural changes.
+add_action( 'init', 'datamachine_register_site_md_invalidation' );
+
 require_once __DIR__ . '/Engine/AI/Directives/DailyMemorySelectorDirective.php';
 require_once __DIR__ . '/Api/Chat/ChatContextDirective.php';
 require_once __DIR__ . '/Api/System/SystemContextDirective.php';