deduplication and more accurate token estimation for images

Author: orual

bsky_agent/agents/pattern_config.toml

@@ -115,7 +115,7 @@ chunk_size = 30
 summarization_model = "claude-3-5-haiku-20241022"
 
 [context]
-max_messages = 250
+max_messages = 200
 
 
 # Bluesky configuration for the constellation

crates/pattern_core/src/context/mod.rs

@@ -156,7 +156,7 @@ impl Default for ContextConfig {
             base_instructions: DEFAULT_BASE_INSTRUCTIONS.to_string(),
             memory_char_limit: DEFAULT_CORE_MEMORY_CHAR_LIMIT,
             max_context_messages: DEFAULT_MAX_CONTEXT_MESSAGES,
-            max_context_tokens: Some(128000),
+            max_context_tokens: Some(200_000),
             enable_thinking: true,
             tool_usage_rules: Vec::new(),
             tool_workflow_rules: Vec::new(),
@@ -187,7 +187,7 @@ impl Default for ModelAdjustments {
         Self {
             native_thinking: false,
             use_xml_tags: true,
-            max_context_tokens: Some(128_000),
+            max_context_tokens: Some(200_000),
             token_multiplier: 1.0, // Rough estimate: 1 token ≈ 0.75 words
         }
     }

crates/pattern_core/src/data_source/bluesky.rs

@@ -78,32 +78,32 @@ pub struct ThreadContext {
 impl ThreadContext {
     /// Collect all image URLs from the thread context
     pub fn collect_all_image_urls(&self, main_post: &BlueskyPost) -> Vec<String> {
-        let mut urls = Vec::new();
+        let mut unique_urls = std::collections::HashSet::new();
 
         // Collect from parent chain
         for (parent, siblings) in &self.parent_chain {
-            urls.extend(parent.collect_image_urls());
+            unique_urls.extend(parent.collect_image_urls());
             for sibling in siblings {
-                urls.extend(sibling.collect_image_urls());
+                unique_urls.extend(sibling.collect_image_urls());
             }
         }
 
         // Collect from root if different
         if let Some(root) = &self.root {
-            urls.extend(root.collect_image_urls());
+            unique_urls.extend(root.collect_image_urls());
         }
 
         // Collect from main post
-        urls.extend(main_post.collect_image_urls());
+        unique_urls.extend(main_post.collect_image_urls());
 
         // Collect from replies
         for replies in self.replies_map.values() {
             for reply in replies {
-                urls.extend(reply.collect_image_urls());
+                unique_urls.extend(reply.collect_image_urls());
             }
         }
 
-        urls
+        unique_urls.into_iter().collect()
     }
 
     /// Append full thread tree to buffer
@@ -1018,8 +1018,8 @@ impl BlueskyPost {
                                     if cid.starts_with("http") {
                                         cid
                                     } else {
-                                        // Use the parent post's DID (not the quoted post's DID)
-                                        format!("https://cdn.bsky.app/img/feed_fullsize/plain/{}/{}@jpeg", did, cid)
+                                        // Use the parent post's DID (not the quoted post's DID) - thumbnail for LLM
+                                        format!("https://cdn.bsky.app/img/feed_thumbnail/plain/{}/{}@jpeg", did, cid)
                                     }
                                 } else {
                                     // Use the parent post's DID for the fallback case too
@@ -4015,16 +4015,16 @@ fn convert_blob_to_url(blob_ref: &str, did: &str) -> String {
         if cid.starts_with("http") {
             cid
         } else {
-            // Convert CID to CDN URL (full size)
+            // Convert CID to CDN URL (thumbnail - better for LLM processing)
             format!(
-                "https://cdn.bsky.app/img/feed_fullsize/plain/{}/{}@jpeg",
+                "https://cdn.bsky.app/img/feed_thumbnail/plain/{}/{}@jpeg",
                 did, cid
             )
         }
     } else {
-        // Fallback: treat as plain CID
+        // Fallback: treat as plain CID (thumbnail)
         format!(
-            "https://cdn.bsky.app/img/feed_fullsize/plain/{}/{}@jpeg",
+            "https://cdn.bsky.app/img/feed_thumbnail/plain/{}/{}@jpeg",
             did, blob_ref
         )
     }

crates/pattern_core/src/message.rs

@@ -2223,8 +2223,20 @@ impl Message {
     /// Rough estimation of token count for this message
     ///
     /// Uses the approximation of ~4 characters per token
+    /// Images are estimated at 1600 tokens each
     pub fn estimate_tokens(&self) -> usize {
-        self.display_content().len() / 4
+        let text_tokens = self.display_content().len() / 4;
+
+        // Count images in the message
+        let image_count = match &self.content {
+            MessageContent::Parts(parts) => parts
+                .iter()
+                .filter(|part| matches!(part, ContentPart::Image { .. }))
+                .count(),
+            _ => 0,
+        };
+
+        text_tokens + (image_count * 1600)
     }
 }
 

crates/pattern_discord/src/bot.rs

@@ -1135,7 +1135,7 @@ impl DiscordBot {
 
             // Process attachments if any
             let mut attachment_content = String::new();
-            let mut image_markers = Vec::new();
+            let mut unique_image_urls = std::collections::HashSet::new();
             if !msg.attachments.is_empty() {
                 for attachment in &msg.attachments {
                     // Check if it's an image file
@@ -1150,8 +1150,8 @@ impl DiscordBot {
                             .map_or(false, |ct| ct.starts_with("image/"));
 
                     if is_image {
-                        // Add image marker for multimodal processing
-                        image_markers.push(format!("[IMAGE: {}]", attachment.url));
+                        // Add unique image URL for multimodal processing
+                        unique_image_urls.insert(attachment.url.clone());
                         attachment_content.push_str(&format!(
                             "\n\n[Image attachment: {} ({} bytes)]",
                             attachment.filename, attachment.size
@@ -1197,13 +1197,13 @@ impl DiscordBot {
                 }
             }
 
-            // Take only last 4 images to avoid token bloat
-            let selected_images: Vec<_> =
-                image_markers.iter().rev().take(4).rev().cloned().collect();
+            // Convert to vec and take only last 4 images to avoid token bloat
+            let all_images: Vec<String> = unique_image_urls.into_iter().collect();
+            let selected_images: Vec<_> = all_images.iter().rev().take(4).rev().cloned().collect();
 
             // Append image markers to attachment content
-            for image_marker in &selected_images {
-                attachment_content.push_str(&format!("\n{}", image_marker));
+            for image_url in &selected_images {
+                attachment_content.push_str(&format!("\n[IMAGE: {}]", image_url));
             }
 
             // Create framing prompt that makes responding optional