diff --git a/backend/openedx_ai_extensions/processors/component_extractors.py b/backend/openedx_ai_extensions/processors/component_extractors.py
new file mode 100644
index 00000000..8e3b682c
--- /dev/null
+++ b/backend/openedx_ai_extensions/processors/component_extractors.py
@@ -0,0 +1,338 @@
+"""
+Clean, LLM-friendly formatting for HTML, Video, Problem, and all other XBlocks.
+"""
+
+import json
+import logging
+from typing import Optional
+
+from bs4 import BeautifulSoup  # pylint: disable=import-error
+from django.conf import settings
+
+logger = logging.getLogger(__name__)
+
+filters = settings.AI_EXTENSIONS_FIELD_FILTERS
+ALLOWED_FIELDS = filters.get("allowed_fields", [])
+ALLOWED_FIELD_SUBSTRINGS = filters.get("allowed_field_substrings", [])
+# -----------------------------
+# Embedded content helpers
+# -----------------------------
+
+
+def _extract_iframes(soup: BeautifulSoup) -> list[str]:
+    """
+    Extract all <iframe> tags from the HTML and return
+    a list of LLM-friendly strings describing the embedded iframes.
+
+    Example output:
+    '[Embedded iframe: title="Video Player", src="/path/to/video"]'
+    """
+    embeddings = []
+    for iframe in soup.find_all("iframe"):
+        src = iframe.get("src")
+        title = iframe.get("title", "").strip() or "iframe"
+        if src:
+            embeddings.append(f"[Embedded iframe: title=\"{title}\", src=\"{src}\"]")
+    return embeddings
+
+
+def _extract_objects(soup: BeautifulSoup) -> list[str]:
+    """
+    Extract <object> tags, identifying PDFs and other objects.
+    Returns a list of strings describing each embedded object.
+
+    Example output:
+    '[Embedded PDF: title="Manual", file="/static/manual.pdf"]'
+    '[Embedded object: title="3D Model", type="application/x-3d", file="/assets/model.obj"]'
+    """
+    embeddings = []
+    for obj in soup.find_all("object"):
+        data = obj.get("data")
+        mime = obj.get("type", "")
+        title = obj.get("title", "").strip() or "Object"
+        if data:
+            if "pdf" in mime.lower() or data.lower().endswith(".pdf"):
+                embeddings.append(f"[Embedded PDF: title=\"{title}\", file=\"{data}\"]")
+            else:
+                embeddings.append(f"[Embedded object: title=\"{title}\", type=\"{mime}\", file=\"{data}\"]")
+    return embeddings
+
+
+def _extract_images(soup: BeautifulSoup) -> list[str]:
+    """
+    Extract <img> tags and return a list of strings describing the images.
+
+    Example output:
+    '[Embedded image: alt="Logo", src="/static/logo.png"]'
+    """
+    embeddings = []
+    for img in soup.find_all("img"):
+        src = img.get("src")
+        alt = img.get("alt", "").strip() or "image"
+        if src:
+            embeddings.append(f"[Embedded image: alt=\"{alt}\", src=\"{src}\"]")
+    return embeddings
+
+
+def _extract_media(soup: BeautifulSoup) -> list[str]:
+    """
+    Extract <video> and <audio> tags, including <source> elements.
+    Returns a list of descriptive strings.
+
+    Example output:
+    '[Embedded video: title="Lecture", sources="/videos/lec.mp4, /videos/lec.webm"]'
+    """
+    embeddings = []
+    for media in soup.find_all(["video", "audio"]):
+        sources = [s.get("src") for s in media.find_all("source") if s.get("src")]
+        direct_src = media.get("src")
+        if direct_src and direct_src not in sources:
+            sources.append(direct_src)
+
+        title = media.get("title", "").strip() or media.name
+        if sources:
+            sources_str = ", ".join(sources)
+            embeddings.append(f"[Embedded {media.name}: title=\"{title}\", sources=\"{sources_str}\"]")
+    return embeddings
+
+
+def _extract_embeds(soup: BeautifulSoup) -> list[str]:
+    """
+    Extract <embed> tags and return descriptive strings.
+
+    Example output:
+    '[Embedded content: tag=embed, src="/assets/interactive.swf"]'
+    """
+    embeddings = []
+    for embed in soup.find_all("embed"):
+        src = embed.get("src")
+        if src:
+            embeddings.append(f"[Embedded content: tag=embed, src=\"{src}\"]")
+    return embeddings
+
+
+def _extract_xblocks(soup: BeautifulSoup) -> list[str]:
+    """
+    Extract custom XBlock placeholders from <div> tags with data-type or data-block-id.
+    Returns descriptive strings for LLM consumption.
+
+    Example output:
+    '[Embedded XBlock: type="problem", id="block-v1:course+unit+block"]'
+    """
+    embeddings = []
+    for div in soup.find_all("div"):
+        xblock_type = div.get("data-type")
+        block_id = div.get("data-block-id")
+        if xblock_type or block_id:
+            desc = "[Embedded XBlock"
+            if xblock_type:
+                desc += f": type=\"{xblock_type}\""
+            if block_id:
+                desc += f", id=\"{block_id}\""
+            desc += "]"
+            embeddings.append(desc)
+    return embeddings
+
+
+# ---------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------
+
+def html_to_text(raw_html: str) -> str:
+    """
+    Convert HTML into clean, LLM-friendly text, preserving embedded content info.
+    """
+    logger.debug("html_to_text: started cleaning HTML (%d chars)", len(raw_html) if raw_html else 0)
+    try:
+        soup = BeautifulSoup(raw_html, "html.parser")
+
+        embeddings = []
+        embeddings.extend(_extract_iframes(soup))
+        embeddings.extend(_extract_objects(soup))
+        embeddings.extend(_extract_images(soup))
+        embeddings.extend(_extract_media(soup))
+        embeddings.extend(_extract_embeds(soup))
+        embeddings.extend(_extract_xblocks(soup))
+
+        # Remove embedded tags after extraction
+        for tag in soup.find_all(["iframe", "object", "img", "video", "audio", "embed"]):
+            tag.extract()
+
+        # Remove scripts & styles
+        for tag in soup(["script", "style"]):
+            tag.extract()
+
+        clean = "\n".join(line.strip() for line in soup.get_text(separator="\n").splitlines() if line.strip())
+
+        if embeddings:
+            clean += "\n\n" + "\n".join(embeddings)
+
+        logger.debug("html_to_text: finished cleaning HTML (%d chars)", len(clean))
+        return clean.strip()
+    except Exception as exc:  # pylint: disable=broad-exception-caught
+        logger.exception("html_to_text: failed to clean HTML: %s", exc)
+        return raw_html  # fallback
+
+
+def _load_transcript_content(block) -> Optional[dict]:
+    """
+    Returns transcript content as clean text.
+    """
+    logger.debug("_load_transcript_content: loading transcript for block %s", block)
+
+    transcripts = getattr(block, "transcripts", {}) or {}
+    if not transcripts:
+        logger.info("_load_transcript_content: no transcripts found")
+        return None
+
+    # prefer english, fallback to next available
+    language_code = "en" if "en" in transcripts else next(iter(transcripts.keys()))
+    logger.debug("_load_transcript_content: selected language '%s'", language_code)
+
+    # use edxval API
+    try:
+        # pylint: disable=import-error,import-outside-toplevel
+        from edxval.api import get_video_transcript_data
+
+        edx_video_id = getattr(block, "edx_video_id", None)
+        logger.debug("_load_transcript_content: fetching transcript via edxval for video %s", edx_video_id)
+
+        transcript = get_video_transcript_data(video_id=edx_video_id, language_code=language_code)
+
+        content = transcript["content"]
+        if isinstance(content, bytes):
+            content = content.decode("utf-8")
+            logger.debug("_load_transcript_content: decoded bytes content")
+
+        try:
+            content = json.loads(content)
+        except json.JSONDecodeError:
+            logger.warning("_load_transcript_content: transcript content is not valid JSON, returning raw content")
+
+        logger.debug("_load_transcript_content: transcript content successfully loaded")
+        return content.get("text") if isinstance(content, dict) else content
+
+    except Exception as exc:  # pylint: disable=broad-exception-caught
+        logger.exception("_load_transcript_content: failed to load transcript: %s", exc)
+        return transcripts
+
+
+# ---------------------------------------------------------
+# Block Extractors
+# ---------------------------------------------------------
+
+def extract_video_info(block) -> dict:
+    """Extract rich metadata for video blocks, including transcripts."""
+    logger.debug("extract_video_info: extracting video block %s", block.location)
+
+    info = {
+        "type": "video",
+        "block_id": str(block.location),
+        "title": block.display_name,
+        "edx_video_id": getattr(block, "edx_video_id", None),
+        "youtube_id": getattr(block, "youtube_id_1_0", None),
+    }
+
+    transcript_text = _load_transcript_content(block)
+    if transcript_text:
+        logger.debug("extract_video_info: transcript found (%d chars)", len(str(transcript_text)))
+        info["transcript_text"] = transcript_text
+    else:
+        logger.info("extract_video_info: no transcript available for block %s", block.location)
+
+    return info
+
+
+def extract_html_info(block) -> dict:
+    logger.debug("extract_html_info: processing block %s", block.location)
+    raw_html = getattr(block, "data", "") or ""
+    text = html_to_text(raw_html)
+    return {
+        "type": "html",
+        "block_id": str(block.location),
+        "title": block.display_name,
+        "text": text,
+    }
+
+
+def extract_problem_info(block) -> dict:
+    logger.debug("extract_problem_info: processing %s", block.location)
+    raw = getattr(block, "data", "") or ""
+    text = html_to_text(raw)
+    return {
+        "type": "problem",
+        "block_id": str(block.location),
+        "title": block.display_name,
+        "text": text,
+    }
+
+
+def extract_discussion_info(block) -> dict:
+    logger.debug("extract_discussion_info: processing %s", block.location)
+    return {
+        "type": "discussion",
+        "block_id": str(block.location),
+        "title": block.display_name,
+        "discussion_id": getattr(block, "discussion_id", None),
+        "category": getattr(block, "discussion_category", None),
+        "target": getattr(block, "discussion_target", None),
+    }
+
+
+def _is_field_allowed(field_name: str) -> bool:
+    """
+    Returns True if field_name is explicitly allowed or matches an allowed substring.
+    """
+    fname = field_name.lower()
+
+    # 1. Exact allow-list
+    if fname in ALLOWED_FIELDS:
+        return True
+
+    # 2. Substring allow-list
+    for substring in ALLOWED_FIELD_SUBSTRINGS:
+        if substring in fname:
+            return True
+
+    return False
+
+
+def extract_generic_info(block) -> dict:
+    """
+    Catch-all extractor for unknown block types with allow-list filtering.
+    Only fields explicitly allowed or matching allowed substrings are included.
+    """
+    logger.debug("extract_generic_info: processing %s", block.location)
+
+    safe_fields = {}
+    for field in getattr(block, "fields", []):
+        if not _is_field_allowed(field):
+            logger.debug("extract_generic_info: excluded field '%s' (not allowed)", field)
+            continue
+
+        value = getattr(block, field, None)
+
+        # Only safe primitive-like values are included
+        if isinstance(value, (str, int, float, bool, list, dict, type(None))):
+            safe_fields[field] = value
+        else:
+            logger.debug(
+                "extract_generic_info: excluded field '%s' (unsupported type %s)",
+                field,
+                type(value)
+            )
+
+    return {
+        "type": getattr(block, "category", "unknown"),
+        "block_id": str(block.location),
+        "title": block.display_name,
+        "fields": safe_fields,
+    }
+
+
+COMPONENT_EXTRACTORS = {
+    "html": extract_html_info,
+    "problem": extract_problem_info,
+    "video": extract_video_info,
+    "discussion": extract_discussion_info,
+}
diff --git a/backend/openedx_ai_extensions/processors/openedx/openedx_processor.py b/backend/openedx_ai_extensions/processors/openedx/openedx_processor.py
index 075408cc..9754d75d 100644
--- a/backend/openedx_ai_extensions/processors/openedx/openedx_processor.py
+++ b/backend/openedx_ai_extensions/processors/openedx/openedx_processor.py
@@ -6,6 +6,8 @@
 
 from opaque_keys.edx.keys import UsageKey
 
+from openedx_ai_extensions.processors.component_extractors import COMPONENT_EXTRACTORS, extract_generic_info
+
 logger = logging.getLogger(__name__)
 
 
@@ -53,30 +55,49 @@ def get_unit_content(self, context):
                 "blocks": [],
             }
 
-            if hasattr(unit, "children") and unit.children:
-                for child_key in unit.children:
-                    try:
-                        child = store.get_item(child_key)
-                        block_info = {
-                            "block_id": str(child.location),
-                            "display_name": child.display_name,
-                            "category": child.category,
-                        }
-
-                        if child.category == "html":
-                            block_info["content"] = getattr(child, "data", "")
-                        elif child.category == "problem":
-                            block_info["content"] = getattr(child, "data", "")
-
-                        unit_info["blocks"].append(block_info)
-                    except Exception as e:  # pylint: disable=broad-exception-caught
-                        logger.warning(f"Could not load block {child_key}: {e}")
-
-            # limit for dev
-            if char_limit and len(unit_info) > char_limit:
-                unit_info = unit_info[:char_limit]
+            for child_key in getattr(unit, "children", []):
+                block_info = self._extract_block(store, child_key)
+                if block_info:
+                    unit_info["blocks"].append(block_info)
+
+            if char_limit:
+                self._truncate_unit_text(unit_info, char_limit)
 
             return unit_info
 
-        except Exception as e:  # pylint: disable=broad-exception-caught
-            return {"error": f"Error accessing content: {str(e)}"}
+        except Exception as exc:  # pylint: disable=broad-exception-caught
+            return {"error": f"Error accessing content: {str(exc)}"}
+
+    def _extract_block(self, store, block_key):
+        """Helper to extract block info safely"""
+        try:
+            block = store.get_item(block_key)
+            block_type = block.category.lower()
+            extractor = COMPONENT_EXTRACTORS.get(block_type, extract_generic_info)
+            return extractor(block)
+        except Exception as exc:  # pylint: disable=broad-exception-caught
+            logger.warning(f"Could not load block {block_key}: {exc}")
+            return None
+
+    def _truncate_unit_text(self, unit_info, char_limit):
+        """Helper to safely truncate text fields in unit blocks (dev-only)"""
+        try:
+            total_text = "".join(
+                b.get("text", "") for b in unit_info["blocks"] if isinstance(b.get("text"), str)
+            )
+
+            if len(total_text) <= char_limit:
+                return
+
+            logger.debug("char_limit=%s triggered, truncating text fields safely", char_limit)
+
+            blocks_with_text = [b for b in unit_info["blocks"] if isinstance(b.get("text"), str)]
+            per_block = char_limit // len(blocks_with_text)
+
+            for block in blocks_with_text:
+                block["text"] = block["text"][:per_block]
+
+            unit_info["truncated"] = True
+
+        except Exception as trunc_err:  # pylint: disable=broad-exception-caught
+            logger.debug("char_limit truncation skipped due to error: %s", trunc_err)
diff --git a/backend/openedx_ai_extensions/settings/common.py b/backend/openedx_ai_extensions/settings/common.py
index 2ac33866..301665c6 100644
--- a/backend/openedx_ai_extensions/settings/common.py
+++ b/backend/openedx_ai_extensions/settings/common.py
@@ -7,6 +7,27 @@
 logger = logging.getLogger(__name__)
 
 
+DEFAULT_FIELD_FILTERS = {
+    "allowed_fields": [
+        "name",
+        "display_name",
+        "tags",
+        "title",
+        "format",
+        "text",
+        "type",
+        "due",
+        "source_file",
+        "data",
+        "graded",
+    ],
+    "allowed_field_substrings": [
+        "description",
+        "name",
+    ],
+}
+
+
 def plugin_settings(settings):
     """
     Add plugin settings to main settings object.
@@ -23,3 +44,9 @@ def plugin_settings(settings):
     # This prevents context window from growing too large while maintaining conversation continuity
     if not hasattr(settings, "AI_EXTENSIONS_MAX_CONTEXT_MESSAGES"):
         settings.AI_EXTENSIONS_MAX_CONTEXT_MESSAGES = 3
+
+    # -------------------------
+    # Default field filters
+    # -------------------------
+    if not hasattr(settings, "AI_EXTENSIONS_FIELD_FILTERS"):
+        settings.AI_EXTENSIONS_FIELD_FILTERS = DEFAULT_FIELD_FILTERS.copy()
diff --git a/backend/requirements/base.in b/backend/requirements/base.in
index 640c6b10..71e5d48b 100644
--- a/backend/requirements/base.in
+++ b/backend/requirements/base.in
@@ -6,6 +6,7 @@ djangorestframework  # REST API framework
 django-filter      # Filtering for DRF
 edx-opaque-keys        # Open edX CourseKeyField
 litellm            # LLM Gateway to provide model access
+beautifulsoup4     # HTML parsing for component extractors
 
 openedx-atlas
 edx-submissions
diff --git a/backend/requirements/base.txt b/backend/requirements/base.txt
index 13c02d78..d60377a6 100644
--- a/backend/requirements/base.txt
+++ b/backend/requirements/base.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --output-file=requirements/base.txt requirements/base.in
@@ -18,11 +18,15 @@ anyio==4.11.0
     #   openai
 asgiref==3.8.1
     # via django
+async-timeout==5.0.1
+    # via aiohttp
 attrs==25.4.0
     # via
     #   aiohttp
     #   jsonschema
     #   referencing
+beautifulsoup4==4.14.2
+    # via -r requirements/base.in
 certifi==2025.10.5
     # via
     #   httpcore
@@ -60,6 +64,8 @@ edx-opaque-keys==2.12.0
     # via -r requirements/base.in
 edx-submissions==3.12.1
     # via -r requirements/base.in
+exceptiongroup==1.3.1
+    # via anyio
 fastuuid==0.13.5
     # via litellm
 filelock==3.20.0
@@ -156,6 +162,8 @@ sniffio==1.3.1
     # via
     #   anyio
     #   openai
+soupsieve==2.8
+    # via beautifulsoup4
 sqlparse==0.5.3
     # via django
 stevedore==5.4.1
@@ -170,10 +178,15 @@ tqdm==4.67.1
     #   openai
 typing-extensions==4.13.2
     # via
+    #   -c requirements/constraints.txt
     #   aiosignal
     #   anyio
+    #   asgiref
+    #   beautifulsoup4
     #   edx-opaque-keys
+    #   exceptiongroup
     #   huggingface-hub
+    #   multidict
     #   openai
     #   pydantic
     #   pydantic-core
diff --git a/backend/requirements/ci.txt b/backend/requirements/ci.txt
index 0c53594d..2bf88dcb 100644
--- a/backend/requirements/ci.txt
+++ b/backend/requirements/ci.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --output-file=requirements/ci.txt requirements/ci.in
@@ -28,7 +28,15 @@ pluggy==1.5.0
     # via tox
 pyproject-api==1.9.0
     # via tox
+tomli==2.3.0
+    # via
+    #   pyproject-api
+    #   tox
 tox==4.25.0
     # via -r requirements/ci.in
+typing-extensions==4.13.2
+    # via
+    #   -c requirements/constraints.txt
+    #   tox
 virtualenv==20.30.0
     # via tox
diff --git a/backend/requirements/dev.txt b/backend/requirements/dev.txt
index fdd7693a..49b3d5a7 100644
--- a/backend/requirements/dev.txt
+++ b/backend/requirements/dev.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --output-file=requirements/dev.txt requirements/dev.in
@@ -34,12 +34,18 @@ astroid==3.3.9
     #   -r requirements/quality.txt
     #   pylint
     #   pylint-celery
+async-timeout==5.0.1
+    # via
+    #   -r requirements/quality.txt
+    #   aiohttp
 attrs==25.4.0
     # via
     #   -r requirements/quality.txt
     #   aiohttp
     #   jsonschema
     #   referencing
+beautifulsoup4==4.14.2
+    # via -r requirements/quality.txt
 build==1.3.0
     # via
     #   -r requirements/pip-tools.txt
@@ -160,6 +166,11 @@ edx-opaque-keys==2.12.0
     # via -r requirements/quality.txt
 edx-submissions==3.12.1
     # via -r requirements/quality.txt
+exceptiongroup==1.3.1
+    # via
+    #   -r requirements/quality.txt
+    #   anyio
+    #   pytest
 fastuuid==0.13.5
     # via
     #   -r requirements/quality.txt
@@ -433,6 +444,10 @@ snowballstemmer==2.2.0
     # via
     #   -r requirements/quality.txt
     #   pydocstyle
+soupsieve==2.8
+    # via
+    #   -r requirements/quality.txt
+    #   beautifulsoup4
 sqlparse==0.5.3
     # via
     #   -r requirements/quality.txt
@@ -455,6 +470,18 @@ tokenizers==0.22.1
     # via
     #   -r requirements/quality.txt
     #   litellm
+tomli==2.3.0
+    # via
+    #   -r requirements/ci.txt
+    #   -r requirements/pip-tools.txt
+    #   -r requirements/quality.txt
+    #   build
+    #   coverage
+    #   pip-tools
+    #   pylint
+    #   pyproject-api
+    #   pytest
+    #   tox
 tomlkit==0.13.2
     # via
     #   -r requirements/quality.txt
@@ -468,15 +495,23 @@ tqdm==4.67.1
     #   openai
 typing-extensions==4.13.2
     # via
+    #   -c requirements/constraints.txt
+    #   -r requirements/ci.txt
     #   -r requirements/quality.txt
     #   aiosignal
     #   anyio
+    #   asgiref
+    #   astroid
+    #   beautifulsoup4
     #   edx-opaque-keys
+    #   exceptiongroup
     #   huggingface-hub
+    #   multidict
     #   openai
     #   pydantic
     #   pydantic-core
     #   referencing
+    #   tox
     #   typing-inspection
 typing-inspection==0.4.2
     # via
diff --git a/backend/requirements/doc.txt b/backend/requirements/doc.txt
index 0cb06ded..bcb18fdc 100644
--- a/backend/requirements/doc.txt
+++ b/backend/requirements/doc.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --output-file=requirements/doc.txt requirements/doc.in
@@ -33,6 +33,10 @@ asgiref==3.8.1
     # via
     #   -r requirements/test.txt
     #   django
+async-timeout==5.0.1
+    # via
+    #   -r requirements/test.txt
+    #   aiohttp
 attrs==25.4.0
     # via
     #   -r requirements/test.txt
@@ -43,8 +47,12 @@ babel==2.17.0
     # via
     #   pydata-sphinx-theme
     #   sphinx
-beautifulsoup4==4.13.3
-    # via pydata-sphinx-theme
+backports-tarfile==1.2.0
+    # via jaraco-context
+beautifulsoup4==4.14.2
+    # via
+    #   -r requirements/test.txt
+    #   pydata-sphinx-theme
 build==1.2.2.post1
     # via -r requirements/doc.in
 certifi==2025.10.5
@@ -137,6 +145,11 @@ edx-opaque-keys==2.12.0
     # via -r requirements/test.txt
 edx-submissions==3.12.1
     # via -r requirements/test.txt
+exceptiongroup==1.3.1
+    # via
+    #   -r requirements/test.txt
+    #   anyio
+    #   pytest
 fastuuid==0.13.5
     # via
     #   -r requirements/test.txt
@@ -189,6 +202,7 @@ imagesize==1.4.1
 importlib-metadata==8.7.0
     # via
     #   -r requirements/test.txt
+    #   keyring
     #   litellm
 iniconfig==2.1.0
     # via
@@ -391,8 +405,10 @@ sniffio==1.3.1
     #   openai
 snowballstemmer==2.2.0
     # via sphinx
-soupsieve==2.6
-    # via beautifulsoup4
+soupsieve==2.8
+    # via
+    #   -r requirements/test.txt
+    #   beautifulsoup4
 sphinx==8.1.3
     # via
     #   -r requirements/doc.in
@@ -435,6 +451,14 @@ tokenizers==0.22.1
     # via
     #   -r requirements/test.txt
     #   litellm
+tomli==2.3.0
+    # via
+    #   -r requirements/test.txt
+    #   build
+    #   coverage
+    #   doc8
+    #   pytest
+    #   sphinx
 tqdm==4.67.1
     # via
     #   -r requirements/test.txt
@@ -444,17 +468,22 @@ twine==6.1.0
     # via -r requirements/doc.in
 typing-extensions==4.13.2
     # via
+    #   -c requirements/constraints.txt
     #   -r requirements/test.txt
     #   aiosignal
     #   anyio
+    #   asgiref
     #   beautifulsoup4
     #   edx-opaque-keys
+    #   exceptiongroup
     #   huggingface-hub
+    #   multidict
     #   openai
     #   pydantic
     #   pydantic-core
     #   pydata-sphinx-theme
     #   referencing
+    #   rich
     #   typing-inspection
 typing-inspection==0.4.2
     # via
diff --git a/backend/requirements/pip-tools.txt b/backend/requirements/pip-tools.txt
index 7bfbffab..2eb10351 100644
--- a/backend/requirements/pip-tools.txt
+++ b/backend/requirements/pip-tools.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --output-file=requirements/pip-tools.txt requirements/pip-tools.in
@@ -16,6 +16,10 @@ pyproject-hooks==1.2.0
     # via
     #   build
     #   pip-tools
+tomli==2.3.0
+    # via
+    #   build
+    #   pip-tools
 wheel==0.45.1
     # via pip-tools
 
diff --git a/backend/requirements/pip.txt b/backend/requirements/pip.txt
index 1fe96604..b6425cd4 100644
--- a/backend/requirements/pip.txt
+++ b/backend/requirements/pip.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --allow-unsafe --output-file=requirements/pip.txt requirements/pip.in
diff --git a/backend/requirements/quality.txt b/backend/requirements/quality.txt
index 66bfa71c..d6e9b6b2 100644
--- a/backend/requirements/quality.txt
+++ b/backend/requirements/quality.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --output-file=requirements/quality.txt requirements/quality.in
@@ -33,12 +33,18 @@ astroid==3.3.9
     # via
     #   pylint
     #   pylint-celery
+async-timeout==5.0.1
+    # via
+    #   -r requirements/test.txt
+    #   aiohttp
 attrs==25.4.0
     # via
     #   -r requirements/test.txt
     #   aiohttp
     #   jsonschema
     #   referencing
+beautifulsoup4==4.14.2
+    # via -r requirements/test.txt
 certifi==2025.10.5
     # via
     #   -r requirements/test.txt
@@ -127,6 +133,11 @@ edx-opaque-keys==2.12.0
     # via -r requirements/test.txt
 edx-submissions==3.12.1
     # via -r requirements/test.txt
+exceptiongroup==1.3.1
+    # via
+    #   -r requirements/test.txt
+    #   anyio
+    #   pytest
 fastuuid==0.13.5
     # via
     #   -r requirements/test.txt
@@ -349,6 +360,10 @@ sniffio==1.3.1
     #   openai
 snowballstemmer==2.2.0
     # via pydocstyle
+soupsieve==2.8
+    # via
+    #   -r requirements/test.txt
+    #   beautifulsoup4
 sqlparse==0.5.3
     # via
     #   -r requirements/test.txt
@@ -371,6 +386,12 @@ tokenizers==0.22.1
     # via
     #   -r requirements/test.txt
     #   litellm
+tomli==2.3.0
+    # via
+    #   -r requirements/test.txt
+    #   coverage
+    #   pylint
+    #   pytest
 tomlkit==0.13.2
     # via pylint
 tqdm==4.67.1
@@ -380,11 +401,17 @@ tqdm==4.67.1
     #   openai
 typing-extensions==4.13.2
     # via
+    #   -c requirements/constraints.txt
     #   -r requirements/test.txt
     #   aiosignal
     #   anyio
+    #   asgiref
+    #   astroid
+    #   beautifulsoup4
     #   edx-opaque-keys
+    #   exceptiongroup
     #   huggingface-hub
+    #   multidict
     #   openai
     #   pydantic
     #   pydantic-core
diff --git a/backend/requirements/test.txt b/backend/requirements/test.txt
index 74475cdf..16679116 100644
--- a/backend/requirements/test.txt
+++ b/backend/requirements/test.txt
@@ -1,5 +1,5 @@
 #
-# This file is autogenerated by pip-compile with Python 3.12
+# This file is autogenerated by pip-compile with Python 3.10
 # by the following command:
 #
 #    pip-compile --output-file=requirements/test.txt requirements/test.in
@@ -29,12 +29,18 @@ asgiref==3.8.1
     # via
     #   -r requirements/base.txt
     #   django
+async-timeout==5.0.1
+    # via
+    #   -r requirements/base.txt
+    #   aiohttp
 attrs==25.4.0
     # via
     #   -r requirements/base.txt
     #   aiohttp
     #   jsonschema
     #   referencing
+beautifulsoup4==4.14.2
+    # via -r requirements/base.txt
 certifi==2025.10.5
     # via
     #   -r requirements/base.txt
@@ -104,6 +110,11 @@ edx-opaque-keys==2.12.0
     # via -r requirements/base.txt
 edx-submissions==3.12.1
     # via -r requirements/base.txt
+exceptiongroup==1.3.1
+    # via
+    #   -r requirements/base.txt
+    #   anyio
+    #   pytest
 fastuuid==0.13.5
     # via
     #   -r requirements/base.txt
@@ -282,6 +293,10 @@ sniffio==1.3.1
     #   -r requirements/base.txt
     #   anyio
     #   openai
+soupsieve==2.8
+    # via
+    #   -r requirements/base.txt
+    #   beautifulsoup4
 sqlparse==0.5.3
     # via
     #   -r requirements/base.txt
@@ -302,6 +317,10 @@ tokenizers==0.22.1
     # via
     #   -r requirements/base.txt
     #   litellm
+tomli==2.3.0
+    # via
+    #   coverage
+    #   pytest
 tqdm==4.67.1
     # via
     #   -r requirements/base.txt
@@ -309,11 +328,16 @@ tqdm==4.67.1
     #   openai
 typing-extensions==4.13.2
     # via
+    #   -c requirements/constraints.txt
     #   -r requirements/base.txt
     #   aiosignal
     #   anyio
+    #   asgiref
+    #   beautifulsoup4
     #   edx-opaque-keys
+    #   exceptiongroup
     #   huggingface-hub
+    #   multidict
     #   openai
     #   pydantic
     #   pydantic-core
diff --git a/tutor/openedx_ai_extensions/patches/openedx-auth b/tutor/openedx_ai_extensions/patches/openedx-auth
index b0d89177..eccd7729 100644
--- a/tutor/openedx_ai_extensions/patches/openedx-auth
+++ b/tutor/openedx_ai_extensions/patches/openedx-auth
@@ -26,3 +26,19 @@ AI_EXTENSIONS_MODEL_PROXY:
     file: "{{ proxy.file }}"
   {%- endfor %}
 {%- endif %}
+{%- if AI_EXTENSIONS_FIELD_FILTERS is defined %}
+AI_EXTENSIONS_FIELD_FILTERS:
+  {%- if AI_EXTENSIONS_FIELD_FILTERS.allowed_fields is defined %}
+  allowed_fields:
+    {%- for f in AI_EXTENSIONS_FIELD_FILTERS.allowed_fields %}
+    - "{{ f }}"
+    {%- endfor %}
+  {%- endif %}
+
+  {%- if AI_EXTENSIONS_FIELD_FILTERS.allowed_field_substrings is defined %}
+  allowed_field_substrings:
+    {%- for s in AI_EXTENSIONS_FIELD_FILTERS.allowed_field_substrings %}
+    - "{{ s }}"
+    {%- endfor %}
+  {%- endif %}
+{%- endif %}