From 19786166c114baad62e2e1bf604954c9096f9a24 Mon Sep 17 00:00:00 2001
From: Goon <duy@wearetopgroup.com>
Date: Fri, 13 Mar 2026 23:01:59 +0700
Subject: [PATCH] fix(agent): enrich <media:image> tags with persisted media
 IDs for Discord image attachments (#179)

* fix(agent): enrich <media:image> tags with persisted media IDs for Discord image attachments

Discord image attachments were downloaded and persisted correctly, but
<media:image> tags in the message content remained bare (no ID attribute),
unlike <media:audio> and <media:video> tags which get enriched with media
IDs. This made it harder for the LLM to confirm image receipt and
reference specific images.

Add enrichImageIDs() that embeds persisted media IDs into <media:image>
tags, matching the existing enrichAudioIDs()/enrichVideoIDs() pattern.
Iterates refs in reverse order to correctly map multiple image refs to
their positional tags when users attach several images at once.

Closes #178

https://claude.ai/code/session_01KkE9UxcNB8eXpqRiJHRqeB

* style(agent): add missing continue in enrichImageIDs for consistency with enrichVideoIDs

https://claude.ai/code/session_01KkE9UxcNB8eXpqRiJHRqeB

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Viet Tran <viettranx@gmail.com>
---
 internal/agent/loop.go  |  8 ++++++--
 internal/agent/media.go | 39 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 2 deletions(-)

diff --git a/internal/agent/loop.go b/internal/agent/loop.go
index 319e0077..75f645e5 100644
--- a/internal/agent/loop.go
+++ b/internal/agent/loop.go
@@ -297,8 +297,12 @@ func (l *Loop) runLoop(ctx context.Context, req RunRequest) (*RunResult, error)
 		l.enrichVideoIDs(messages, mediaRefs)
 	}
 
-	// 2e. Cross-session recovery: recover stale locked tasks and notify team leads
-	// about orphaned pending tasks and in-progress tasks being handled by delegates.
+	// 2e. Enrich <media:image> tags with persisted media IDs so the LLM
+	// knows images were received and stored (consistent with audio/video enrichment).
+	l.enrichImageIDs(messages, mediaRefs)
+
+	// 2f. Cross-session recovery: notify team leads about orphaned pending tasks
+	// and in-progress tasks being handled by delegates.
 	// Safe because Bước 1 (early ClaimTask) ensures running tasks are in_progress,
 	// so only truly un-spawned tasks remain pending.
 	if l.teamStore != nil && l.agentUUID != uuid.Nil {
diff --git a/internal/agent/media.go b/internal/agent/media.go
index e2c280c3..a687c68b 100644
--- a/internal/agent/media.go
+++ b/internal/agent/media.go
@@ -244,6 +244,45 @@ func (l *Loop) enrichVideoIDs(messages []providers.Message, refs []providers.Med
 	messages[lastIdx].Content = content
 }
 
+// enrichImageIDs updates the last user message to embed persisted media IDs
+// in <media:image> tags so the LLM knows images were received and stored.
+// Without this, the LLM sees plain <media:image> and cannot confirm image availability.
+// Iterates refs in reverse order so that when multiple images are present,
+// each ref maps to the correct positional tag (last ref → last tag, etc.).
+func (l *Loop) enrichImageIDs(messages []providers.Message, refs []providers.MediaRef) {
+	if len(messages) == 0 {
+		return
+	}
+	lastIdx := -1
+	for i := len(messages) - 1; i >= 0; i-- {
+		if messages[i].Role == "user" {
+			lastIdx = i
+			break
+		}
+	}
+	if lastIdx < 0 {
+		return
+	}
+
+	content := messages[lastIdx].Content
+	// Iterate in reverse so first ref matches first tag when using LastIndex replacements.
+	for i := len(refs) - 1; i >= 0; i-- {
+		ref := refs[i]
+		if ref.Kind != "image" {
+			continue
+		}
+		idAttr := fmt.Sprintf(" id=%q", ref.ID)
+
+		// Replace the LAST bare <media:image> with <media:image id="uuid">
+		bare := "<media:image>"
+		if idx := strings.LastIndex(content, bare); idx >= 0 {
+			content = content[:idx] + "<media:image" + idAttr + ">" + content[idx+len(bare):]
+			continue
+		}
+	}
+	messages[lastIdx].Content = content
+}
+
 // mediaKindFromMime returns the media kind ("image", "video", "audio", "document")
 // based on MIME type prefix.
 func mediaKindFromMime(mime string) string {