v0.6.2: deterministic peptide auto-linker

New Peptide_Linker class scans generated HTML and wraps every mention
of a known peptide in a hyperlink to its /peptides/ database page.
Runs as post-processing in Publisher — no LLM involvement, 100% reliable.
Handles case-insensitive matching, hyphen/space variants, skips existing
<a> tags. Gracefully no-ops if PR Core is not active.

Removed prompt-based peptide linking from Content_Prompts (unreliable).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

ARCHITECTURE.md

@@ -134,6 +134,7 @@ prautoblogger/
 │   │   ├── class-semantic-dedup.php   # Embedding-based cosine similarity dedup with keyword fallback
 │   │   ├── class-content-generator.php# Writer agent — manages the generation pipeline
 │   │   ├── class-content-prompts.php  # Builds all LLM prompts for content generation (system, stage, linking rules)
+│   │   ├── class-peptide-linker.php   # Deterministic post-processor: hyperlinks peptide mentions to /peptides/ pages
 │   │   ├── class-chief-editor.php     # Editor agent — LLM-powered editorial review
 │   │   ├── class-publisher.php        # Creates WordPress posts from approved content
 │   │   ├── class-post-assembler.php   # Post-creation helpers: taxonomy, log linking, images, sanitization

CHANGELOG.md

@@ -5,6 +5,23 @@ All notable changes to PRAutoBlogger will be documented in this file.
 Format follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project uses [Semantic Versioning](https://semver.org/).
 
+## [0.6.2] — 2026-04-20
+
+### Added
+- **Deterministic peptide auto-linker.** New `Peptide_Linker` class scans
+  generated HTML and wraps every mention of a known peptide in a hyperlink
+  to its `/peptides/{slug}/` database page. Runs as a post-processing step
+  in Publisher before `wp_insert_post` — no LLM involvement, 100% reliable.
+  Handles case-insensitive matching, hyphen/space variants (e.g., "BPC-157"
+  and "BPC 157"), and skips text already inside `<a>` tags. Gracefully
+  no-ops if PR Core is not active.
+
+### Changed
+- **Removed prompt-based peptide linking.** The system prompt no longer asks
+  the LLM to link peptides (unreliable). Peptide links are now injected
+  deterministically by `Peptide_Linker` after content generation. The prompt
+  still provides article links and the "never fabricate URLs" rule.
+
 ## [0.6.1] — 2026-04-20
 
 ### Added

includes/core/class-content-prompts.php

@@ -167,13 +167,10 @@ class PRAutoBlogger_Content_Prompts {
 		$rules  = "--- LINKING RULES ---\n";
 		$rules .= "NEVER fabricate or invent URLs. You do NOT have access to external sources.\n";
 		$rules .= "Only use the internal links listed below when linking within the article.\n";
-		$rules .= "If no listed page is relevant to a section, do not insert a link.\n";
-		$rules .= "IMPORTANT: Whenever ANY peptide name from the peptide database list below\n";
-		$rules .= "is mentioned in the article, ALWAYS link the FIRST mention to its database page.\n\n";
+		$rules .= "If no listed article is relevant to a section, do not insert a link.\n";
+		$rules .= "Do NOT add links for peptide names — those are injected automatically after generation.\n\n";
 
 		$rules .= self::build_article_links();
-		$rules .= "\n";
-		$rules .= self::build_peptide_links();
 
 		$rules .= "--- END LINKING RULES ---";
 
@@ -211,42 +208,4 @@ class PRAutoBlogger_Content_Prompts {
 		return implode( "\n", $lines ) . "\n";
 	}
 
-	/**
-	 * Build a reference list of all published peptide database pages.
-	 *
-	 * Every peptide mentioned in an article must link to its database page
-	 * on first mention. This gives the model the real URLs to use.
-	 *
-	 * @return string Formatted peptide link list, or empty if CPT unavailable.
-	 */
-	private static function build_peptide_links(): string {
-		// PR Core may not be active — guard against missing post type.
-		if ( ! post_type_exists( 'pr_peptide' ) ) {
-			return '';
-		}
-
-		$peptides = get_posts( [
-			'post_type'      => 'pr_peptide',
-			'post_status'    => 'publish',
-			'posts_per_page' => 200,
-			'orderby'        => 'title',
-			'order'          => 'ASC',
-			'fields'         => 'ids',
-		] );
-
-		if ( empty( $peptides ) ) {
-			return '';
-		}
-
-		$lines = [ "Peptide database pages — ALWAYS link the first mention of each peptide:\n" ];
-		foreach ( $peptides as $peptide_id ) {
-			$title = get_the_title( $peptide_id );
-			$url   = get_permalink( $peptide_id );
-			if ( $title && $url ) {
-				$lines[] = "- {$title}: {$url}";
-			}
-		}
-
-		return implode( "\n", $lines ) . "\n";
-	}
 }

includes/core/class-peptide-linker.php

@@ -0,0 +1,149 @@
+<?php
+declare(strict_types=1);
+
+/**
+ * Deterministic post-processor that hyperlinks peptide mentions in HTML content.
+ *
+ * Scans generated article HTML for any peptide name that exists in the PR Core
+ * database and wraps every mention in an <a> tag pointing to its /peptides/ page.
+ * Skips text that is already inside an existing <a> tag to avoid nested links.
+ *
+ * Triggered by: PRAutoBlogger_Publisher::create_post() (before wp_insert_post).
+ * Dependencies: PR Core plugin (pr_peptide CPT). Gracefully no-ops if PR Core is inactive.
+ *
+ * @see core/class-publisher.php — Calls inject_links() on post_content.
+ * @see peptide-repo-core/includes/cpt/class-pr-core-peptide-cpt.php — Peptide CPT.
+ */
+class PRAutoBlogger_Peptide_Linker {
+
+	/**
+	 * Inject peptide hyperlinks into HTML content.
+	 *
+	 * Every occurrence of a known peptide name is wrapped in a link to its
+	 * database page, except when the text is already inside an <a> tag.
+	 *
+	 * @param string $html Article HTML content.
+	 * @return string HTML with peptide names hyperlinked.
+	 */
+	public static function inject_links( string $html ): string {
+		if ( ! post_type_exists( 'pr_peptide' ) ) {
+			return $html;
+		}
+
+		$link_map = self::build_link_map();
+		if ( empty( $link_map ) ) {
+			return $html;
+		}
+
+		// Build a single regex alternation of all peptide names, longest first
+		// so "BPC-157" matches before a hypothetical "BPC" would.
+		$names = array_keys( $link_map );
+		usort( $names, static function ( string $a, string $b ): int {
+			return strlen( $b ) - strlen( $a );
+		} );
+
+		$escaped = array_map( static function ( string $name ): string {
+			return preg_quote( $name, '/' );
+		}, $names );
+
+		// Word-boundary match, case-insensitive.
+		$pattern = '/\b(' . implode( '|', $escaped ) . ')\b/iu';
+
+		// Split HTML into segments: inside <a>…</a> tags vs. outside.
+		// Only replace in segments that are NOT inside an anchor tag.
+		return self::replace_outside_links( $html, $pattern, $link_map );
+	}
+
+	/**
+	 * Build a map of peptide display names → permalink URLs.
+	 *
+	 * Includes the canonical title and common aliases (e.g., "BPC-157" and
+	 * "BPC 157") so both hyphenated and space-separated forms get linked.
+	 *
+	 * @return array<string, string> Name → URL pairs, case-preserved keys.
+	 */
+	private static function build_link_map(): array {
+		$peptides = get_posts( [
+			'post_type'      => 'pr_peptide',
+			'post_status'    => 'publish',
+			'posts_per_page' => 200,
+			'orderby'        => 'title',
+			'order'          => 'ASC',
+			'fields'         => 'ids',
+		] );
+
+		$map = [];
+		foreach ( $peptides as $peptide_id ) {
+			$title = get_the_title( $peptide_id );
+			$url   = get_permalink( $peptide_id );
+			if ( ! $title || ! $url ) {
+				continue;
+			}
+
+			// Canonical name (e.g., "BPC-157").
+			$map[ $title ] = $url;
+
+			// If the name contains hyphens, also match the space-separated
+			// variant (e.g., "BPC 157" → same URL as "BPC-157").
+			if ( strpos( $title, '-' ) !== false ) {
+				$space_variant = str_replace( '-', ' ', $title );
+				$map[ $space_variant ] = $url;
+			}
+		}
+
+		return $map;
+	}
+
+	/**
+	 * Replace peptide names with links, but only outside existing <a> tags.
+	 *
+	 * Splits HTML into chunks by <a …>…</a> boundaries. Replacement only
+	 * runs on chunks that are outside anchors, preventing nested <a> tags.
+	 *
+	 * @param string              $html     Full HTML content.
+	 * @param string              $pattern  Regex pattern matching peptide names.
+	 * @param array<string,string> $link_map Name → URL map (case-preserved keys).
+	 * @return string HTML with replacements applied.
+	 */
+	private static function replace_outside_links( string $html, string $pattern, array $link_map ): string {
+		// Split on anchor tags, keeping the delimiters.
+		// This regex captures full <a …>…</a> blocks as array elements.
+		$parts = preg_split( '/(<a\s[^>]*>.*?<\/a>)/is', $html, -1, PREG_SPLIT_DELIM_CAPTURE );
+
+		if ( false === $parts ) {
+			return $html;
+		}
+
+		// Build a lowercase → URL lookup for case-insensitive matching.
+		$lower_map = [];
+		foreach ( $link_map as $name => $url ) {
+			$lower_map[ mb_strtolower( $name ) ] = $url;
+		}
+
+		$result = '';
+		foreach ( $parts as $part ) {
+			// If this chunk is an anchor tag, pass it through untouched.
+			if ( preg_match( '/^<a\s/i', $part ) ) {
+				$result .= $part;
+				continue;
+			}
+
+			// Replace peptide names with hyperlinks.
+			$result .= preg_replace_callback(
+				$pattern,
+				static function ( array $matches ) use ( $lower_map ): string {
+					$matched_text = $matches[0];
+					$key          = mb_strtolower( $matched_text );
+					$url          = $lower_map[ $key ] ?? '';
+					if ( '' === $url ) {
+						return $matched_text;
+					}
+					return '<a href="' . esc_url( $url ) . '">' . esc_html( $matched_text ) . '</a>';
+				},
+				$part
+			);
+		}
+
+		return $result;
+	}
+}

includes/core/class-publisher.php

@@ -12,6 +12,7 @@ declare(strict_types=1);
  *               WordPress wp_insert_post(), post meta API.
  *
  * @see core/class-post-assembler.php   — Post-creation helpers (taxonomy, images, logs).
+ * @see core/class-peptide-linker.php   — Injects peptide database hyperlinks into content.
  * @see core/class-chief-editor.php     — Produces the editorial review we consume.
  * @see core/class-content-generator.php — Produces the content we publish.
  * @see ARCHITECTURE.md                  — Data flow step 6.
@@ -74,9 +75,15 @@ class PRAutoBlogger_Publisher {
 		string $post_status,
 		?string $run_id = null
 	): int {
+		// Clean LLM artifacts, then inject peptide hyperlinks deterministically
+		// before the content enters WordPress. Peptide linker no-ops gracefully
+		// if PR Core is not active.
+		$clean_content  = PRAutoBlogger_Post_Assembler::sanitize_llm_content( $content );
+		$linked_content = PRAutoBlogger_Peptide_Linker::inject_links( $clean_content );
+
 		$post_data = [
 			'post_title'   => sanitize_text_field( $idea->get_suggested_title() ),
-			'post_content' => wp_kses_post( PRAutoBlogger_Post_Assembler::sanitize_llm_content( $content ) ),
+			'post_content' => wp_kses_post( $linked_content ),
 			'post_status'  => $post_status,
 			'post_type'    => 'post',
 			'post_author'  => PRAutoBlogger_Post_Assembler::get_default_author_id(),

prautoblogger.php

@@ -8,7 +8,7 @@ declare(strict_types=1);
  * Plugin Name:       PRAutoBlogger
  * Plugin URI:        https://peptiderepo.com/prautoblogger
  * Description:       Monitors social media for trending topics, generates SEO-friendly blog posts using AI, and publishes them on a daily schedule with full cost tracking and self-improvement metrics.
- * Version:           0.6.1
+ * Version:           0.6.2
  * Requires at least: 6.0
  * Requires PHP:      7.4
  * Author:            PeptideRepo
@@ -35,7 +35,7 @@ if ( ! defined( 'ABSPATH' ) ) {
 | and limits without magic strings.
 */
 
-define( 'PRAUTOBLOGGER_VERSION', '0.6.1' );
+define( 'PRAUTOBLOGGER_VERSION', '0.6.2' );
 define( 'PRAUTOBLOGGER_DB_VERSION', '1.1.0' );
 define( 'PRAUTOBLOGGER_PLUGIN_FILE', __FILE__ );
 define( 'PRAUTOBLOGGER_PLUGIN_DIR', plugin_dir_path( __FILE__ ) );