diff --git a/pkg/appview/readme/fetcher.go b/pkg/appview/readme/fetcher.go
index 47d81f0..fa621d3 100644
--- a/pkg/appview/readme/fetcher.go
+++ b/pkg/appview/readme/fetcher.go
@@ -70,22 +70,7 @@ func NewFetcher() *Fetcher {
 // FetchAndRender fetches a README from a URL and renders it as HTML
 // Returns the rendered HTML and any error
 func (f *Fetcher) FetchAndRender(ctx context.Context, readmeURL string) (string, error) {
-	// Validate URL
-	if readmeURL == "" {
-		return "", fmt.Errorf("empty README URL")
-	}
-
-	parsedURL, err := url.Parse(readmeURL)
-	if err != nil {
-		return "", fmt.Errorf("invalid README URL: %w", err)
-	}
-
-	// Only allow HTTP/HTTPS
-	if parsedURL.Scheme != "http" && parsedURL.Scheme != "https" {
-		return "", fmt.Errorf("invalid URL scheme: %s", parsedURL.Scheme)
-	}
-
-	// Fetch content
+	// Fetch content (includes URL validation, Content-Type check, and HTML detection)
 	content, baseURL, err := f.fetchContent(ctx, readmeURL)
 	if err != nil {
 		return "", err
@@ -100,8 +85,36 @@ func (f *Fetcher) FetchAndRender(ctx context.Context, readmeURL string) (string,
 	return html, nil
 }
 
+// FetchRaw fetches raw README content from a URL without rendering
+// Returns raw bytes with Content-Type and HTML validation
+// Use this when you need to store the raw markdown (e.g., in PDS records)
+func (f *Fetcher) FetchRaw(ctx context.Context, readmeURL string) ([]byte, error) {
+	// Fetch content (includes URL validation, Content-Type check, and HTML detection)
+	content, _, err := f.fetchContent(ctx, readmeURL)
+	if err != nil {
+		return nil, err
+	}
+
+	return content, nil
+}
+
 // fetchContent fetches the raw content from a URL
 func (f *Fetcher) fetchContent(ctx context.Context, urlStr string) ([]byte, string, error) {
+	// Validate URL
+	if urlStr == "" {
+		return nil, "", fmt.Errorf("empty README URL")
+	}
+
+	parsedURL, err := url.Parse(urlStr)
+	if err != nil {
+		return nil, "", fmt.Errorf("invalid README URL: %w", err)
+	}
+
+	// Only allow HTTP/HTTPS
+	if parsedURL.Scheme != "http" && parsedURL.Scheme != "https" {
+		return nil, "", fmt.Errorf("invalid URL scheme: %s", parsedURL.Scheme)
+	}
+
 	req, err := http.NewRequestWithContext(ctx, "GET", urlStr, nil)
 	if err != nil {
 		return nil, "", fmt.Errorf("failed to create request: %w", err)
@@ -120,6 +133,15 @@ func (f *Fetcher) fetchContent(ctx context.Context, urlStr string) ([]byte, stri
 		return nil, "", fmt.Errorf("unexpected status code: %d", resp.StatusCode)
 	}
 
+	// Reject HTML content types (catches proper error pages)
+	contentType := resp.Header.Get("Content-Type")
+	if contentType != "" {
+		ct := strings.ToLower(contentType)
+		if strings.Contains(ct, "text/html") || strings.Contains(ct, "application/xhtml") {
+			return nil, "", fmt.Errorf("unsupported content type: %s (expected markdown or plain text)", contentType)
+		}
+	}
+
 	// Limit content size to 1MB
 	limitedReader := io.LimitReader(resp.Body, 1*1024*1024)
 	content, err := io.ReadAll(limitedReader)
@@ -127,12 +149,35 @@ func (f *Fetcher) fetchContent(ctx context.Context, urlStr string) ([]byte, stri
 		return nil, "", fmt.Errorf("failed to read response body: %w", err)
 	}
 
+	// Detect HTML content by checking for common markers (catches soft 404s)
+	if LooksLikeHTML(content) {
+		return nil, "", fmt.Errorf("detected HTML content instead of markdown")
+	}
+
 	// Get base URL for relative link resolution
 	baseURL := getBaseURL(resp.Request.URL)
 
 	return content, baseURL, nil
 }
 
+// LooksLikeHTML checks if content appears to be HTML rather than markdown
+// Exported for use by other packages that fetch README content
+func LooksLikeHTML(content []byte) bool {
+	if len(content) == 0 {
+		return false
+	}
+
+	// Check first 512 bytes for HTML markers
+	checkLen := min(len(content), 512)
+
+	trimmed := strings.TrimSpace(string(content[:checkLen]))
+	lower := strings.ToLower(trimmed)
+
+	return strings.HasPrefix(lower, "<!doctype") ||
+		strings.HasPrefix(lower, "<html") ||
+		strings.HasPrefix(lower, "<?xml")
+}
+
 // renderMarkdown renders markdown content to sanitized HTML
 func (f *Fetcher) renderMarkdown(content []byte, baseURL string) (string, error) {
 	var buf bytes.Buffer
diff --git a/pkg/appview/readme/fetcher_test.go b/pkg/appview/readme/fetcher_test.go
index d929c72..c968252 100644
--- a/pkg/appview/readme/fetcher_test.go
+++ b/pkg/appview/readme/fetcher_test.go
@@ -1,7 +1,11 @@
 package readme
 
 import (
+	"context"
+	"net/http"
+	"net/http/httptest"
 	"net/url"
+	"strings"
 	"testing"
 )
 
@@ -305,4 +309,270 @@ func containsSubstringHelper(s, substr string) bool {
 	return false
 }
 
-// TODO: Add README fetching and caching tests
+func TestLooksLikeHTML(t *testing.T) {
+	tests := []struct {
+		name     string
+		content  string
+		expected bool
+	}{
+		{
+			name:     "empty content",
+			content:  "",
+			expected: false,
+		},
+		{
+			name:     "markdown content",
+			content:  "# Hello World\n\nThis is a README.",
+			expected: false,
+		},
+		{
+			name:     "plain text",
+			content:  "Just some plain text without any HTML.",
+			expected: false,
+		},
+		{
+			name:     "doctype html",
+			content:  "<!DOCTYPE html>\n<html><body>Page</body></html>",
+			expected: true,
+		},
+		{
+			name:     "doctype html lowercase",
+			content:  "<!doctype html>\n<html><body>Page</body></html>",
+			expected: true,
+		},
+		{
+			name:     "html tag only",
+			content:  "<html><head></head><body>Page</body></html>",
+			expected: true,
+		},
+		{
+			name:     "html tag with whitespace",
+			content:  "  \n  <html>\n<body>Page</body></html>",
+			expected: true,
+		},
+		{
+			name:     "xml declaration",
+			content:  "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<html>...</html>",
+			expected: true,
+		},
+		{
+			name:     "soft 404 page",
+			content:  "<!DOCTYPE html><html><head><title>Page Not Found</title></head><body><h1>404</h1></body></html>",
+			expected: true,
+		},
+		{
+			name:     "markdown with inline html",
+			content:  "# Title\n\nSome text with <strong>bold</strong> inline.",
+			expected: false,
+		},
+		{
+			name:     "markdown starting with hash",
+			content:  "## Section\n\nContent here.",
+			expected: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := LooksLikeHTML([]byte(tt.content))
+			if result != tt.expected {
+				t.Errorf("looksLikeHTML(%q) = %v, want %v", tt.content, result, tt.expected)
+			}
+		})
+	}
+}
+
+func TestFetcher_FetchRaw(t *testing.T) {
+	fetcher := NewFetcher()
+
+	tests := []struct {
+		name        string
+		handler     http.HandlerFunc
+		wantErr     bool
+		errContains string
+		wantContent string
+	}{
+		{
+			name: "successful markdown fetch",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.Header().Set("Content-Type", "text/plain")
+				w.Write([]byte("# Hello World\n\nThis is markdown."))
+			},
+			wantErr:     false,
+			wantContent: "# Hello World",
+		},
+		{
+			name: "rejects HTML content type",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.Header().Set("Content-Type", "text/html; charset=utf-8")
+				w.Write([]byte("<html><body>Error</body></html>"))
+			},
+			wantErr:     true,
+			errContains: "unsupported content type",
+		},
+		{
+			name: "rejects soft 404 HTML content",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.Header().Set("Content-Type", "text/plain")
+				w.Write([]byte("<!DOCTYPE html><html><body>404 Not Found</body></html>"))
+			},
+			wantErr:     true,
+			errContains: "detected HTML content",
+		},
+		{
+			name: "rejects 404 status",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.WriteHeader(http.StatusNotFound)
+				w.Write([]byte("Not Found"))
+			},
+			wantErr:     true,
+			errContains: "unexpected status code: 404",
+		},
+		{
+			name: "rejects 500 status",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.WriteHeader(http.StatusInternalServerError)
+				w.Write([]byte("Internal Server Error"))
+			},
+			wantErr:     true,
+			errContains: "unexpected status code: 500",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			server := httptest.NewServer(tt.handler)
+			defer server.Close()
+
+			content, err := fetcher.FetchRaw(context.Background(), server.URL)
+
+			if tt.wantErr {
+				if err == nil {
+					t.Errorf("FetchRaw() expected error containing %q, got nil", tt.errContains)
+					return
+				}
+				if !strings.Contains(err.Error(), tt.errContains) {
+					t.Errorf("FetchRaw() error = %q, want error containing %q", err.Error(), tt.errContains)
+				}
+				return
+			}
+
+			if err != nil {
+				t.Errorf("FetchRaw() unexpected error: %v", err)
+				return
+			}
+
+			if !strings.Contains(string(content), tt.wantContent) {
+				t.Errorf("FetchRaw() content = %q, want content containing %q", string(content), tt.wantContent)
+			}
+		})
+	}
+}
+
+func TestFetcher_FetchRaw_URLValidation(t *testing.T) {
+	fetcher := NewFetcher()
+
+	tests := []struct {
+		name        string
+		url         string
+		errContains string
+	}{
+		{
+			name:        "empty URL",
+			url:         "",
+			errContains: "empty README URL",
+		},
+		{
+			name:        "invalid URL scheme",
+			url:         "ftp://example.com/README.md",
+			errContains: "invalid URL scheme",
+		},
+		{
+			name:        "file URL scheme",
+			url:         "file:///etc/passwd",
+			errContains: "invalid URL scheme",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			_, err := fetcher.FetchRaw(context.Background(), tt.url)
+			if err == nil {
+				t.Errorf("FetchRaw(%q) expected error, got nil", tt.url)
+				return
+			}
+			if !strings.Contains(err.Error(), tt.errContains) {
+				t.Errorf("FetchRaw(%q) error = %q, want error containing %q", tt.url, err.Error(), tt.errContains)
+			}
+		})
+	}
+}
+
+func TestFetcher_FetchAndRender(t *testing.T) {
+	fetcher := NewFetcher()
+
+	tests := []struct {
+		name        string
+		handler     http.HandlerFunc
+		wantErr     bool
+		errContains string
+		wantContain string
+	}{
+		{
+			name: "renders markdown to HTML",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.Header().Set("Content-Type", "text/plain")
+				w.Write([]byte("# Hello World\n\nThis is **bold** text."))
+			},
+			wantErr:     false,
+			wantContain: "<strong>bold</strong>",
+		},
+		{
+			name: "rejects HTML content type",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.Header().Set("Content-Type", "text/html")
+				w.Write([]byte("<html><body>Error</body></html>"))
+			},
+			wantErr:     true,
+			errContains: "unsupported content type",
+		},
+		{
+			name: "rejects soft 404",
+			handler: func(w http.ResponseWriter, r *http.Request) {
+				w.Header().Set("Content-Type", "text/plain")
+				w.Write([]byte("<!doctype html><html><body>Not Found</body></html>"))
+			},
+			wantErr:     true,
+			errContains: "detected HTML content",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			server := httptest.NewServer(tt.handler)
+			defer server.Close()
+
+			html, err := fetcher.FetchAndRender(context.Background(), server.URL)
+
+			if tt.wantErr {
+				if err == nil {
+					t.Errorf("FetchAndRender() expected error containing %q, got nil", tt.errContains)
+					return
+				}
+				if !strings.Contains(err.Error(), tt.errContains) {
+					t.Errorf("FetchAndRender() error = %q, want error containing %q", err.Error(), tt.errContains)
+				}
+				return
+			}
+
+			if err != nil {
+				t.Errorf("FetchAndRender() unexpected error: %v", err)
+				return
+			}
+
+			if !strings.Contains(html, tt.wantContain) {
+				t.Errorf("FetchAndRender() = %q, want HTML containing %q", html, tt.wantContain)
+			}
+		})
+	}
+}
diff --git a/pkg/appview/storage/manifest_store.go b/pkg/appview/storage/manifest_store.go
index b6abf4c..7b4bc33 100644
--- a/pkg/appview/storage/manifest_store.go
+++ b/pkg/appview/storage/manifest_store.go
@@ -424,23 +424,30 @@ func (s *ManifestStore) notifyHoldAboutManifest(ctx context.Context, manifestRec
 	return nil
 }
 
-// ensureRepoPage creates or updates a repo page record in the user's PDS if needed
+// ensureRepoPage creates or updates a repo page record in the user's PDS
 // This syncs repository metadata from manifest annotations to the io.atcr.repo.page collection
-// Only creates a new record if one doesn't exist (doesn't overwrite user's custom content)
+// Always updates the description on push (since users can't edit it via appview yet)
+// Preserves user's avatar if they've set one via the appview
 func (s *ManifestStore) ensureRepoPage(ctx context.Context, manifestRecord *atproto.ManifestRecord) {
-	// Check if repo page already exists (don't overwrite user's custom content)
 	rkey := s.ctx.Repository
-	_, err := s.ctx.ATProtoClient.GetRecord(ctx, atproto.RepoPageCollection, rkey)
-	if err == nil {
-		// Record already exists - don't overwrite
-		slog.Debug("Repo page already exists, skipping creation", "did", s.ctx.DID, "repository", s.ctx.Repository)
-		return
-	}
 
-	// Only continue if it's a "not found" error - other errors mean we should skip
-	if !errors.Is(err, atproto.ErrRecordNotFound) {
+	// Check for existing record to preserve user's avatar
+	var existingAvatarRef *atproto.ATProtoBlobRef
+	var existingRecord *atproto.RepoPageRecord
+	record, err := s.ctx.ATProtoClient.GetRecord(ctx, atproto.RepoPageCollection, rkey)
+	if err == nil && record != nil {
+		// Unmarshal the Value to get the RepoPageRecord
+		var repoPage atproto.RepoPageRecord
+		if unmarshalErr := json.Unmarshal(record.Value, &repoPage); unmarshalErr == nil {
+			existingRecord = &repoPage
+			existingAvatarRef = repoPage.Avatar
+			slog.Debug("Found existing repo page, will update", "did", s.ctx.DID, "repository", s.ctx.Repository, "hasExistingAvatar", existingAvatarRef != nil)
+		} else {
+			slog.Warn("Failed to unmarshal existing repo page", "did", s.ctx.DID, "repository", s.ctx.Repository, "error", unmarshalErr)
+		}
+	} else if err != nil && !errors.Is(err, atproto.ErrRecordNotFound) {
+		// Unexpected error - log and continue (will create new record)
 		slog.Warn("Failed to check for existing repo page", "did", s.ctx.DID, "repository", s.ctx.Repository, "error", err)
-		return
 	}
 
 	// Get annotations (may be nil if image has no OCI labels)
@@ -458,29 +465,37 @@ func (s *ManifestStore) ensureRepoPage(ctx context.Context, manifestRecord *atpr
 		description = annotations["org.opencontainers.image.description"]
 	}
 
-	// Try to fetch and upload icon from io.atcr.icon annotation
-	var avatarRef *atproto.ATProtoBlobRef
+	// Determine avatar: prefer new icon from annotations, otherwise keep existing
+	avatarRef := existingAvatarRef
 	if iconURL := annotations["io.atcr.icon"]; iconURL != "" {
-		avatarRef = s.fetchAndUploadIcon(ctx, iconURL)
+		if newAvatar := s.fetchAndUploadIcon(ctx, iconURL); newAvatar != nil {
+			avatarRef = newAvatar
+		}
 	}
 
-	// Create new repo page record with description and optional avatar
+	// Create/update repo page record with description and avatar
 	repoPage := atproto.NewRepoPageRecord(s.ctx.Repository, description, avatarRef)
 
-	slog.Info("Creating repo page from manifest annotations", "did", s.ctx.DID, "repository", s.ctx.Repository, "descriptionLength", len(description), "hasAvatar", avatarRef != nil)
+	isUpdate := existingRecord != nil
+	action := "Creating"
+	if isUpdate {
+		action = "Updating"
+	}
+	slog.Info(action+" repo page from manifest annotations", "did", s.ctx.DID, "repository", s.ctx.Repository, "descriptionLength", len(description), "hasAvatar", avatarRef != nil)
 
 	_, err = s.ctx.ATProtoClient.PutRecord(ctx, atproto.RepoPageCollection, rkey, repoPage)
 	if err != nil {
-		slog.Warn("Failed to create repo page", "did", s.ctx.DID, "repository", s.ctx.Repository, "error", err)
+		slog.Warn("Failed to "+strings.ToLower(action)+" repo page", "did", s.ctx.DID, "repository", s.ctx.Repository, "error", err)
 		return
 	}
 
-	slog.Info("Repo page created successfully", "did", s.ctx.DID, "repository", s.ctx.Repository)
+	slog.Info("Repo page "+strings.ToLower(action)+"d successfully", "did", s.ctx.DID, "repository", s.ctx.Repository)
 }
 
 // fetchReadmeContent attempts to fetch README content from external sources
 // Priority: io.atcr.readme annotation > derived from org.opencontainers.image.source
 // Returns the raw markdown content, or empty string if not available
+// Uses the shared readme.Fetcher which validates Content-Type and rejects HTML content
 func (s *ManifestStore) fetchReadmeContent(ctx context.Context, annotations map[string]string) string {
 	if s.ctx.ReadmeFetcher == nil {
 		return ""
@@ -492,12 +507,12 @@ func (s *ManifestStore) fetchReadmeContent(ctx context.Context, annotations map[
 
 	// Priority 1: Direct README URL from io.atcr.readme annotation
 	if readmeURL := annotations["io.atcr.readme"]; readmeURL != "" {
-		content, err := s.fetchRawReadme(fetchCtx, readmeURL)
+		content, err := s.ctx.ReadmeFetcher.FetchRaw(fetchCtx, readmeURL)
 		if err != nil {
 			slog.Debug("Failed to fetch README from io.atcr.readme annotation", "url", readmeURL, "error", err)
-		} else if content != "" {
+		} else if len(content) > 0 {
 			slog.Info("Fetched README from io.atcr.readme annotation", "url", readmeURL, "length", len(content))
-			return content
+			return string(content)
 		}
 	}
 
@@ -510,7 +525,7 @@ func (s *ManifestStore) fetchReadmeContent(ctx context.Context, annotations map[
 				continue
 			}
 
-			content, err := s.fetchRawReadme(fetchCtx, readmeURL)
+			content, err := s.ctx.ReadmeFetcher.FetchRaw(fetchCtx, readmeURL)
 			if err != nil {
 				// Only log non-404 errors (404 is expected when trying main vs master)
 				if !readme.Is404(err) {
@@ -519,9 +534,9 @@ func (s *ManifestStore) fetchReadmeContent(ctx context.Context, annotations map[
 				continue
 			}
 
-			if content != "" {
+			if len(content) > 0 {
 				slog.Info("Fetched README from source URL", "sourceURL", sourceURL, "branch", branch, "length", len(content))
-				return content
+				return string(content)
 			}
 		}
 	}
@@ -529,48 +544,6 @@ func (s *ManifestStore) fetchReadmeContent(ctx context.Context, annotations map[
 	return ""
 }
 
-// fetchRawReadme fetches raw markdown content from a URL
-// Returns the raw markdown (not rendered HTML) for storage in the repo page record
-func (s *ManifestStore) fetchRawReadme(ctx context.Context, readmeURL string) (string, error) {
-	// Use a simple HTTP client to fetch raw content
-	// We want raw markdown, not rendered HTML (the Fetcher renders to HTML)
-	req, err := http.NewRequestWithContext(ctx, "GET", readmeURL, nil)
-	if err != nil {
-		return "", fmt.Errorf("failed to create request: %w", err)
-	}
-
-	req.Header.Set("User-Agent", "ATCR-README-Fetcher/1.0")
-
-	client := &http.Client{
-		Timeout: 10 * time.Second,
-		CheckRedirect: func(req *http.Request, via []*http.Request) error {
-			if len(via) >= 5 {
-				return fmt.Errorf("too many redirects")
-			}
-			return nil
-		},
-	}
-
-	resp, err := client.Do(req)
-	if err != nil {
-		return "", fmt.Errorf("failed to fetch URL: %w", err)
-	}
-	defer resp.Body.Close()
-
-	if resp.StatusCode != http.StatusOK {
-		return "", fmt.Errorf("unexpected status code: %d", resp.StatusCode)
-	}
-
-	// Limit content size to 100KB (repo page description has 100KB limit in lexicon)
-	limitedReader := io.LimitReader(resp.Body, 100*1024)
-	content, err := io.ReadAll(limitedReader)
-	if err != nil {
-		return "", fmt.Errorf("failed to read response body: %w", err)
-	}
-
-	return string(content), nil
-}
-
 // fetchAndUploadIcon fetches an image from a URL and uploads it as a blob to the user's PDS
 // Returns the blob reference for use in the repo page record, or nil on error
 func (s *ManifestStore) fetchAndUploadIcon(ctx context.Context, iconURL string) *atproto.ATProtoBlobRef {