diff --git a/README.md b/README.md
index b412342..10acd60 100644
--- a/README.md
+++ b/README.md
@@ -229,13 +229,17 @@ Click **🎯 Tailor to JD**, paste any JD, and drag the intensity slider (0–10
 
 ### Persistent profile (My Profile)
 
-Click **🧠 My Profile** to store your identity, work history, projects, education, and target roles across five tabs. Profile data is automatically injected into every pipeline run.
+Click **🧠 My Profile** to store your identity, work history, projects, education, and target roles across five tabs. Profile data is injected into every pipeline run while the **🧠 Use memory** toggle is on — a global preference that persists across chats, sessions, and reloads (stored server-side). Each generated reply also notes whether your master resume and/or profile memory were used, so it's clear later what fed the result.
 
 **Import / Export** — download your profile as a portable JSON file and load it back on any install. Invalid files are rejected with a clear error.
 
 ### Master resumes
 
-Save any generated resume as a named **master resume** and set one as the default. The sidebar attach button starts a new chat with a specific master pre-loaded — great for keeping a polished base and forking variants per role.
+Save any generated resume as a named **master resume** and mark one as the default — great for keeping a polished base and forking variants per role. Three ways to put them to work:
+
+- **⭐ Use Master** (input bar) attaches a master as the starting point for your next message, so you can tailor it to a JD or refine it through the pipeline.
+- **⭐ New from Master** (sidebar) opens a fresh chat with the master loaded **verbatim — no AI changes**. Click the main button for your default, or the caret to pick any saved master. Ideal for forking a polished base, or just attaching the application tracker to an untouched copy.
+- **Compare to Master** on any resume version diffs it against a chosen master, so you can see exactly what tailoring changed relative to your base.
 
 ### Manual editor + Re-score
 
diff --git a/backend/app/api/_pipeline.py b/backend/app/api/_pipeline.py
index 7c15781..4da80d1 100644
--- a/backend/app/api/_pipeline.py
+++ b/backend/app/api/_pipeline.py
@@ -79,7 +79,8 @@ async def _save_progress(msg_id: str, events: list) -> None:
         pass
 
 
-async def _run_pipeline_background(msg_id: str, session_id: str, initial_state: dict):
+async def _run_pipeline_background(msg_id: str, session_id: str, initial_state: dict,
+                                   run_meta: dict | None = None):
     # Imported lazily so the test harness can stub the graph without forcing a
     # heavy import chain at app startup.
     from app.agents.graph import RESUME_GRAPH
@@ -159,7 +160,12 @@ def _sync_stream():
             )
         else:
             status_str = "complete"
-            summary = _build_summary(resume, has_cover=bool(cover_letter), n_emails=len(outreach_emails or []))
+            meta_info = run_meta or {}
+            summary = _build_summary(
+                resume, has_cover=bool(cover_letter), n_emails=len(outreach_emails or []),
+                used_memory=bool(meta_info.get("used_memory")),
+                from_master_name=meta_info.get("from_master_name"),
+            )
 
         async with AsyncSessionLocal() as db:
             result = await db.execute(select(Message).where(Message.id == msg_id))
@@ -301,7 +307,8 @@ def _safe_score(v) -> float:
         return 0.0
 
 
-def _build_summary(resume: dict | None, has_cover: bool = False, n_emails: int = 0) -> str:
+def _build_summary(resume: dict | None, has_cover: bool = False, n_emails: int = 0,
+                   used_memory: bool = False, from_master_name: str | None = None) -> str:
     if not resume:
         return "I encountered an issue generating your resume. Please try again."
     meta = resume.get("metadata", {})
@@ -337,6 +344,16 @@ def _build_summary(resume: dict | None, has_cover: bool = False, n_emails: int =
         for s in improvements[:3]:
             lines.append(f"• {s}")
 
+    # Provenance note — records what fed this generation so it's clear on a later
+    # read whether the master resume and/or profile memory were applied.
+    sources = []
+    if from_master_name:
+        sources.append(f'master resume **{from_master_name}**')
+    if used_memory:
+        sources.append("your **profile memory**")
+    if sources:
+        lines.append(f"\n📎 Generated using {' and '.join(sources)}.")
+
     lines.append("")
     lines.append("Tip: ask me to refine sections, raise seniority, target a company, add certifications, or paste a JD to tailor.")
     return "\n".join(lines)
diff --git a/backend/app/api/messages_routes.py b/backend/app/api/messages_routes.py
index 119f8e4..3ed2041 100644
--- a/backend/app/api/messages_routes.py
+++ b/backend/app/api/messages_routes.py
@@ -150,9 +150,15 @@ async def send_message(
         "section_preferences": section_prefs,
     }
 
+    # Provenance for the reply's "sources" note — what fed this generation.
+    run_meta = {
+        "used_memory": bool(memory_context),
+        "from_master_name": (req.from_master_name or "").strip() or None,
+    }
+
     background_tasks.add_task(
         _run_pipeline_background,
-        asst_msg.id, session_id, initial_state,
+        asst_msg.id, session_id, initial_state, run_meta,
     )
 
     return {"user_message_id": user_msg.id, "assistant_message_id": asst_msg.id, "status": "processing"}
diff --git a/backend/app/api/sessions_routes.py b/backend/app/api/sessions_routes.py
index 7aef522..28350af 100644
--- a/backend/app/api/sessions_routes.py
+++ b/backend/app/api/sessions_routes.py
@@ -1,4 +1,5 @@
 """Session CRUD endpoints — list/create/read/update/delete one or all."""
+import copy
 from datetime import datetime, timezone
 
 from fastapi import APIRouter, Depends, HTTPException
@@ -7,14 +8,15 @@
 from sqlalchemy.orm import selectinload
 
 from app.database import get_db
-from app.models import Session, Message
+from app.models import Session, Message, MasterResume
 from app.schemas import (
-    CreateSessionRequest, SessionOut, SessionSummary,
-    UpdateSessionRequest, SessionsPage,
+    CreateSessionRequest, CreateSessionFromMasterRequest, SessionOut,
+    SessionSummary, UpdateSessionRequest, SessionsPage,
 )
 from app.config import settings
 
 from ._pipeline import get_cancel_event
+from .memory_routes import _resolve_default
 
 
 router = APIRouter(prefix="/api", tags=["sessions"])
@@ -90,6 +92,81 @@ async def create_session(req: CreateSessionRequest, db: AsyncSession = Depends(g
                       updated_at=s.updated_at, messages=[])
 
 
+@router.post("/sessions/from-master", response_model=SessionOut, status_code=201)
+async def create_session_from_master(
+    req: CreateSessionFromMasterRequest, db: AsyncSession = Depends(get_db)
+):
+    """Create a new chat seeded with a master resume **verbatim** — no LLM /
+    pipeline run, so the resume is guaranteed identical to the master. Lets the
+    user start from their master (e.g. to track an application) and edit on top
+    of it without re-tailoring. The seeded resume lands as a completed assistant
+    message, exactly like a normal pipeline result, so editing/export/diff all
+    work unchanged.
+    """
+    # Resolve which master to load: explicit id, else the default (or earliest).
+    # `_resolve_default` is the single source of truth for "which master is the
+    # default", shared with the memory routes.
+    if req.master_id:
+        res = await db.execute(select(MasterResume).where(MasterResume.id == req.master_id))
+        master = res.scalar_one_or_none()
+        if not master:
+            raise HTTPException(404, "Master resume not found")
+    else:
+        master = await _resolve_default(db)
+        if not master:
+            raise HTTPException(404, "No master resume saved yet")
+
+    # Deep-copy so later edits to this chat never mutate the stored master.
+    resume = copy.deepcopy(master.resume) if isinstance(master.resume, dict) else (master.resume or {})
+    pi = resume.get("personal_info") or {}
+    meta = resume.get("metadata") or {}
+    full_name = (pi.get("full_name") or "").strip()
+    role = (pi.get("professional_title") or meta.get("jd_role") or "").strip()
+    title = (f"{full_name} — {role}" if full_name and role else full_name or master.name or "Master Resume")[:255]
+
+    try:
+        score = float(meta.get("overall_score"))
+    except (TypeError, ValueError):
+        score = None
+
+    s = Session(
+        title=title,
+        llm_provider=req.llm_provider or settings.llm_provider,
+        llm_model=req.llm_model or settings.llm_model,
+    )
+    db.add(s)
+    await db.flush()
+
+    ts = int(datetime.now(timezone.utc).timestamp() * 1000)
+    trace = [{
+        "agent": "Master Resume", "status": "complete",
+        "notes": f'Loaded "{master.name}" verbatim — no AI changes applied.',
+        "timestamp": ts,
+    }]
+    db.add(Message(
+        session_id=s.id, role="user", status="complete",
+        content=f'⭐ Start from master resume "{master.name}" (verbatim — no changes).',
+    ))
+    db.add(Message(
+        session_id=s.id, role="assistant", status="complete",
+        content=(
+            f"⭐ Loaded master resume **{master.name}** as-is — no AI changes were applied.\n\n"
+            "Edit any section directly, paste a job description to tailor it, or fill in "
+            "the application tracker for this chat."
+        ),
+        resume_json=resume,
+        agent_trace=trace,
+        progress_events=trace,
+        final_score=score,
+    ))
+    await db.commit()
+
+    result = await db.execute(
+        select(Session).options(selectinload(Session.messages)).where(Session.id == s.id)
+    )
+    return result.scalar_one()
+
+
 @router.get("/sessions/{session_id}", response_model=SessionOut)
 async def get_session(session_id: str, db: AsyncSession = Depends(get_db)):
     result = await db.execute(
diff --git a/backend/app/api/settings_routes.py b/backend/app/api/settings_routes.py
index 80981b5..6c59f7a 100644
--- a/backend/app/api/settings_routes.py
+++ b/backend/app/api/settings_routes.py
@@ -36,6 +36,8 @@ class AppSettingsUpsert(BaseModel):
     max_review_iterations: Optional[int] = None
     min_quality_score: Optional[float] = None
     default_jd_intensity: Optional[int] = None
+    # Global "Use memory" preference — persisted across chats/sessions.
+    memory_enabled: Optional[bool] = None
     # {"sections": {key: bool}, "fields": {"section.field": bool}} — missing = enabled
     section_preferences: Optional[dict] = None
 
diff --git a/backend/app/config.py b/backend/app/config.py
index 3e1dd8b..04b0074 100644
--- a/backend/app/config.py
+++ b/backend/app/config.py
@@ -44,6 +44,9 @@ class Settings(BaseSettings):
     # Remembered JD-tailoring slider value (0–100). DB overlay populates this
     # at runtime — the .env baseline is None so the UI defaults to 100%.
     default_jd_intensity: Optional[int] = None
+    # Global "Use memory" default. Baseline ON; the DB overlay flips it when the
+    # user toggles the pill, and the value persists across chats/sessions.
+    memory_enabled: bool = True
 
     # ── Optional Basic Auth (single admin user) ────────────────────────────
     # When AUTH_ENABLED=true, every /api/* request requires HTTP Basic auth
diff --git a/backend/app/database.py b/backend/app/database.py
index d453a98..9968425 100644
--- a/backend/app/database.py
+++ b/backend/app/database.py
@@ -56,6 +56,7 @@ async def get_db():
     ("app_settings", "default_jd_intensity", "ALTER TABLE app_settings ADD COLUMN default_jd_intensity INTEGER"),
     ("app_settings", "section_preferences", "ALTER TABLE app_settings ADD COLUMN section_preferences JSON"),
     ("app_settings", "openai_base_url", "ALTER TABLE app_settings ADD COLUMN openai_base_url VARCHAR(500)"),
+    ("app_settings", "memory_enabled", "ALTER TABLE app_settings ADD COLUMN memory_enabled BOOLEAN"),
 ]
 
 
diff --git a/backend/app/documents/docx_generator.py b/backend/app/documents/docx_generator.py
index 2b16e9d..f0bf41f 100644
--- a/backend/app/documents/docx_generator.py
+++ b/backend/app/documents/docx_generator.py
@@ -434,7 +434,7 @@ def _set_doc_margins(doc: Document, *, side_cm: float, vert_cm: float):
 _FA_ICONS = {
     "phone":    (_FA_SOLID, ""),
     "email":    (_FA_SOLID, ""),
-    "location": (_FA_SOLID, ""),
+    "location": (_FA_SOLID, ""),
     "linkedin": (_FA_BRANDS, ""),
     "github":   (_FA_BRANDS, ""),
     "web":      (_FA_SOLID, ""),
diff --git a/backend/app/documents/odt_generator.py b/backend/app/documents/odt_generator.py
index a397c76..d9d13e2 100644
--- a/backend/app/documents/odt_generator.py
+++ b/backend/app/documents/odt_generator.py
@@ -404,7 +404,7 @@ def _make_doc(*, side_cm: float, vert_cm: float, bg_color: str | None = None) ->
 _FA_ICONS = {
     "phone":    (_FA_SOLID, ""),   # faPhone
     "email":    (_FA_SOLID, ""),   # faEnvelope
-    "location": (_FA_SOLID, ""),   # faMapMarker
+    "location": (_FA_SOLID, ""),   # faMapMarker
     "linkedin": (_FA_BRANDS, ""),  # faLinkedin
     "github":   (_FA_BRANDS, ""),  # faGithub
     "web":      (_FA_SOLID, ""),   # faGlobe
diff --git a/backend/app/documents/pdf_generator.py b/backend/app/documents/pdf_generator.py
index 6ad1bb2..5af9490 100644
--- a/backend/app/documents/pdf_generator.py
+++ b/backend/app/documents/pdf_generator.py
@@ -817,7 +817,7 @@ def _ensure_fa_fonts() -> bool:
 _FA_ICONS = {
     "phone":    ("FA-Solid", ""),   # faPhone (handset)
     "email":    ("FA-Solid", ""),   # faEnvelope (solid)
-    "location": ("FA-Solid", ""),   # faMapMarker (solid teardrop)
+    "location": ("FA-Solid", ""),   # faMapMarkerAlt / location-dot (pin with cutout)
     "linkedin": ("FA-Brands", ""),  # faLinkedin
     "github":   ("FA-Brands", ""),  # faGithub
     "web":      ("FA-Solid", ""),   # faGlobe
diff --git a/backend/app/models.py b/backend/app/models.py
index 43ee85f..7f5d99f 100644
--- a/backend/app/models.py
+++ b/backend/app/models.py
@@ -106,6 +106,9 @@ class AppSettings(Base):
     # Last-used JD tailoring slider (0–100). Persisted so the UI starts on the
     # user's preferred intensity instead of resetting to 100 every reload.
     default_jd_intensity: Mapped[int | None] = mapped_column(Integer, nullable=True)
+    # Global "Use memory" preference. Persisted so the toggle is remembered
+    # across chats, sessions, and reloads. NULL → fall back to baseline (on).
+    memory_enabled: Mapped[bool | None] = mapped_column(Boolean, default=None, nullable=True)
     # Per-user section/field enable map. Shape:
     #   {"sections": {"languages": false, ...}, "fields": {"personal_info.website": false, ...}}
     # Missing keys default to enabled.
diff --git a/backend/app/runtime_settings.py b/backend/app/runtime_settings.py
index 85d1b8f..b53650d 100644
--- a/backend/app/runtime_settings.py
+++ b/backend/app/runtime_settings.py
@@ -27,7 +27,7 @@
     "ollama_base_url", "openai_base_url",
     "langsmith_api_key", "langsmith_project", "langsmith_tracing",
     "max_review_iterations", "min_quality_score",
-    "default_jd_intensity",
+    "default_jd_intensity", "memory_enabled",
 )
 
 
diff --git a/backend/app/schemas.py b/backend/app/schemas.py
index 301bffd..17c3aef 100644
--- a/backend/app/schemas.py
+++ b/backend/app/schemas.py
@@ -22,6 +22,14 @@ class CreateSessionRequest(BaseModel):
     llm_model: Optional[str] = None
 
 
+class CreateSessionFromMasterRequest(BaseModel):
+    # Which master resume to seed the new chat with. None → the default master
+    # (or the only/earliest one if none is explicitly marked default).
+    master_id: Optional[str] = None
+    llm_provider: Optional[str] = None
+    llm_model: Optional[str] = None
+
+
 class UpdateSessionRequest(BaseModel):
     title: Optional[str] = None
     llm_provider: Optional[str] = None
@@ -100,6 +108,7 @@ class SendMessageRequest(BaseModel):
     jd_text: Optional[str] = None             # job description for tailoring
     jd_intensity: Optional[int] = None        # 0–100; how aggressively to tailor to the JD (100 = full rewrite, 0 = barely touch the resume)
     attached_resume: Optional[dict] = None    # uploaded JSON resume to start from / refine
+    from_master_name: Optional[str] = None    # name of the master resume this turn was seeded from (for the "sources" note)
     use_memory: Optional[bool] = True         # apply persistent profile memory to this turn
     llm_provider: Optional[str] = None
     llm_model: Optional[str] = None
diff --git a/backend/tests/api/test_messages.py b/backend/tests/api/test_messages.py
index 5e3c5ca..65826cb 100644
--- a/backend/tests/api/test_messages.py
+++ b/backend/tests/api/test_messages.py
@@ -191,3 +191,34 @@ async def test_session_title_auto_updates_from_resume(async_client, fake_llm):
 
     detail = (await async_client.get(f"/api/sessions/{sid}")).json()
     assert "Test Person" in detail["title"]
+
+
+async def test_reply_notes_master_and_memory_sources(async_client, fake_llm):
+    """The generated reply records that the master resume + profile memory were
+    used, so it's clear on a later read what fed the generation."""
+    # Seed a profile so "use memory" actually loads something (else it's a no-op).
+    await async_client.put("/api/memory", json={"full_name": "Ada Lovelace"})
+    sid = await _create_session(async_client)
+    r = await async_client.post(
+        f"/api/sessions/{sid}/messages",
+        json={"content": "Senior Python dev", "from_master_name": "Backend Master",
+              "use_memory": True},
+    )
+    final = await _poll_until_done(async_client, r.json()["assistant_message_id"])
+    assert final["status"] == "complete"
+    assert final["resume_json"] is not None
+    content = final["content"]
+    assert "Backend Master" in content
+    assert "profile memory" in content.lower()
+
+
+async def test_reply_omits_sources_when_none_used(async_client, fake_llm):
+    """No master + memory off → no sources note clutters the reply."""
+    sid = await _create_session(async_client)
+    r = await async_client.post(
+        f"/api/sessions/{sid}/messages",
+        json={"content": "Senior Python dev", "use_memory": False},
+    )
+    final = await _poll_until_done(async_client, r.json()["assistant_message_id"])
+    assert final["resume_json"] is not None
+    assert "Generated using" not in final["content"]
diff --git a/backend/tests/api/test_sessions.py b/backend/tests/api/test_sessions.py
index 1608678..ca1ff42 100644
--- a/backend/tests/api/test_sessions.py
+++ b/backend/tests/api/test_sessions.py
@@ -120,3 +120,78 @@ async def test_session_app_status_fields(async_client):
     assert detail["app_status"] == "applied"
     assert detail["notes"] == "Followed up via LinkedIn."
     assert detail["apply_url"] == "https://example.com/careers/123"
+
+
+# ── New chat from master (verbatim seed, no pipeline) ─────────────────────────
+
+async def _create_master(async_client, resume, name=None, is_default=False) -> dict:
+    body = {"resume": resume, "is_default": is_default}
+    if name is not None:
+        body["name"] = name
+    r = await async_client.post("/api/memory/master-resumes", json=body)
+    assert r.status_code == 200, r.text
+    return r.json()
+
+
+async def test_from_master_no_master_404(async_client):
+    r = await async_client.post("/api/sessions/from-master", json={})
+    assert r.status_code == 404
+
+
+async def test_from_master_seeds_resume_verbatim(async_client, sample_resume):
+    await _create_master(async_client, sample_resume, name="Backend Master")
+
+    r = await async_client.post("/api/sessions/from-master", json={})
+    assert r.status_code == 201, r.text
+    sess = r.json()
+
+    # Title derived from the resume identity, not "New Resume".
+    assert sess["title"] == "Ada Lovelace — Senior Software Engineer"
+
+    # Seeded as a finished user + assistant pair, no processing.
+    roles = [m["role"] for m in sess["messages"]]
+    assert roles == ["user", "assistant"]
+    asst = sess["messages"][1]
+    assert asst["status"] == "complete"
+    # The resume is byte-for-byte the master — no AI edits.
+    assert asst["resume_json"] == sample_resume
+    assert asst["final_score"] == 88
+    # A synthetic trace marks it as a verbatim load (no LLM agents ran).
+    agents = [e["agent"] for e in (asst["progress_events"] or [])]
+    assert agents == ["Master Resume"]
+
+
+async def test_from_master_explicit_id_overrides_default(async_client, sample_resume, minimal_resume):
+    # First master becomes default automatically; second is explicit, non-default.
+    await _create_master(async_client, sample_resume, name="Default One")
+    second = await _create_master(async_client, minimal_resume, name="Picked One")
+
+    r = await async_client.post(
+        "/api/sessions/from-master", json={"master_id": second["id"]}
+    )
+    assert r.status_code == 201, r.text
+    asst = r.json()["messages"][1]
+    assert asst["resume_json"] == minimal_resume
+
+
+async def test_from_master_unknown_id_404(async_client, sample_resume):
+    await _create_master(async_client, sample_resume)
+    r = await async_client.post(
+        "/api/sessions/from-master", json={"master_id": "nope"}
+    )
+    assert r.status_code == 404
+
+
+async def test_from_master_deep_copies_resume(async_client, sample_resume):
+    """Editing the seeded chat's resume must not mutate the stored master."""
+    master = await _create_master(async_client, sample_resume, name="Immutable")
+    sess = (await async_client.post("/api/sessions/from-master", json={})).json()
+    msg_id = sess["messages"][1]["id"]
+
+    edited = dict(sample_resume)
+    edited["professional_summary"] = "MUTATED"
+    await async_client.patch(f"/api/messages/{msg_id}/resume", json={"resume_json": edited})
+
+    stored = (await async_client.get(f"/api/memory/master-resumes/{master['id']}")).json()
+    assert stored["resume"]["professional_summary"] == sample_resume["professional_summary"]
+    assert stored["resume"]["professional_summary"] != "MUTATED"
diff --git a/backend/tests/api/test_settings_routes.py b/backend/tests/api/test_settings_routes.py
index daa7d13..6430ec1 100644
--- a/backend/tests/api/test_settings_routes.py
+++ b/backend/tests/api/test_settings_routes.py
@@ -54,6 +54,26 @@ async def test_reset_settings_drops_overlay(async_client):
     assert body["max_review_iterations"] == 3
 
 
+async def test_memory_enabled_persists(async_client):
+    """The global 'Use memory' preference round-trips through GET/PUT so the
+    toggle is remembered across chats, sessions, and reloads."""
+    # Turn it OFF — a bool False must persist (not be treated as "unset/clear").
+    r = await async_client.put("/api/app-settings", json={"memory_enabled": False})
+    assert r.json()["memory_enabled"] is False
+    assert (await async_client.get("/api/app-settings")).json()["memory_enabled"] is False
+    # And back ON.
+    r2 = await async_client.put("/api/app-settings", json={"memory_enabled": True})
+    assert r2.json()["memory_enabled"] is True
+    assert (await async_client.get("/api/app-settings")).json()["memory_enabled"] is True
+
+
+async def test_memory_enabled_defaults_on_after_reset(async_client):
+    """With no overlay, memory defaults ON (baseline)."""
+    await async_client.put("/api/app-settings", json={"memory_enabled": False})
+    await async_client.delete("/api/app-settings")  # drop overlay → baseline
+    assert (await async_client.get("/api/app-settings")).json()["memory_enabled"] is True
+
+
 async def test_test_provider_uses_fake_llm(async_client, fake_llm):
     """/api/app-settings/test should hit the LLM and echo back a sample."""
     # Make the (fake) generator respond to a custom prompt by overriding
diff --git a/backend/tests/documents/test_contact_icons.py b/backend/tests/documents/test_contact_icons.py
new file mode 100644
index 0000000..7ec3040
--- /dev/null
+++ b/backend/tests/documents/test_contact_icons.py
@@ -0,0 +1,51 @@
+"""Contact-icon glyph contract — shared across the PDF / DOCX / ODT exporters.
+
+Regression guard for the location pin: it must be Font Awesome ``location-dot``
+(U+F3C5 — a pin with a hollow centre), NOT ``location-pin`` (U+F041 — a solid
+black teardrop that reads as an ink blob). All three exporters must agree so a
+résumé's contact line looks identical whichever format it's downloaded in.
+"""
+from __future__ import annotations
+
+import importlib
+
+import pytest
+
+pytestmark = pytest.mark.documents
+
+LOCATION_DOT = ""   # what we want — pin with a hollow centre
+LOCATION_PIN = ""   # the old solid blob — must never come back
+
+_GENERATORS = [
+    "app.documents.pdf_generator",
+    "app.documents.docx_generator",
+    "app.documents.odt_generator",
+]
+
+
+@pytest.mark.parametrize("module_path", _GENERATORS)
+def test_location_icon_is_location_dot_not_blob(module_path):
+    mod = importlib.import_module(module_path)
+    _face, glyph = mod._FA_ICONS["location"]
+    assert glyph == LOCATION_DOT, (
+        f"{module_path} location glyph is U+{ord(glyph):04X}, expected U+F3C5 (location-dot)"
+    )
+    assert glyph != LOCATION_PIN, f"{module_path} regressed to the solid blob U+F041"
+
+
+def test_all_exporters_agree_on_location_glyph():
+    glyphs = {importlib.import_module(p)._FA_ICONS["location"][1] for p in _GENERATORS}
+    assert glyphs == {LOCATION_DOT}, f"exporters disagree on the location glyph: {glyphs}"
+
+
+def test_location_dot_glyph_exists_in_bundled_font():
+    """The chosen glyph must actually exist in fa-solid-900.ttf, otherwise the
+    pin silently renders as blank/tofu instead of an icon."""
+    ttLib = pytest.importorskip("fontTools.ttLib")
+    import os
+    from app.documents.pdf_generator import _FONTS_DIR
+
+    font = ttLib.TTFont(os.path.join(_FONTS_DIR, "fa-solid-900.ttf"))
+    cmap = font.getBestCmap()
+    assert ord(LOCATION_DOT) in cmap, "location-dot (U+F3C5) missing from fa-solid-900.ttf"
+    assert cmap[ord(LOCATION_DOT)] == "location-dot"
diff --git a/backend/tests/unit/test_migrations.py b/backend/tests/unit/test_migrations.py
new file mode 100644
index 0000000..a193018
--- /dev/null
+++ b/backend/tests/unit/test_migrations.py
@@ -0,0 +1,67 @@
+"""Lightweight SQLite migrations (app/database.py).
+
+Regression guard for the *"no such column: app_settings.memory_enabled"* startup
+crash: when a column is added to a model, an existing DB created by an older
+build must be upgraded in place by ``_apply_lightweight_migrations`` —
+``create_all`` only creates missing *tables*, it never alters existing ones.
+This path isn't exercised by the normal test suite (which recreates every table
+fresh per test), so it gets its own test.
+"""
+from __future__ import annotations
+
+import pytest
+from sqlalchemy import text
+from sqlalchemy.ext.asyncio import create_async_engine
+
+pytestmark = pytest.mark.unit
+
+
+async def _columns(conn, table: str) -> set[str]:
+    res = await conn.execute(text(f"PRAGMA table_info({table})"))
+    return {row[1] for row in res.fetchall()}
+
+
+async def test_migration_adds_memory_enabled_to_existing_db(tmp_path):
+    from app.database import _apply_lightweight_migrations
+
+    eng = create_async_engine(f"sqlite+aiosqlite:///{tmp_path / 'old.db'}")
+    try:
+        # Simulate an app_settings table from an older build — no memory_enabled.
+        async with eng.begin() as conn:
+            await conn.execute(text(
+                "CREATE TABLE app_settings ("
+                "id VARCHAR(36) PRIMARY KEY, llm_provider VARCHAR(50), "
+                "llm_model VARCHAR(100), default_jd_intensity INTEGER)"
+            ))
+            await conn.execute(text(
+                "INSERT INTO app_settings (id, llm_provider) VALUES ('1', 'google')"
+            ))
+            assert "memory_enabled" not in await _columns(conn, "app_settings")
+
+        # This is what init_db() runs right after create_all.
+        async with eng.begin() as conn:
+            await _apply_lightweight_migrations(conn)
+            cols = await _columns(conn, "app_settings")
+
+        assert "memory_enabled" in cols  # the column the crash was about
+    finally:
+        await eng.dispose()
+
+
+async def test_migrations_are_idempotent(tmp_path):
+    """Re-running migrations must be a no-op, not a 'duplicate column' error —
+    every ALTER is guarded by a PRAGMA column check."""
+    from app.database import _apply_lightweight_migrations
+
+    eng = create_async_engine(f"sqlite+aiosqlite:///{tmp_path / 'old.db'}")
+    try:
+        async with eng.begin() as conn:
+            await conn.execute(text("CREATE TABLE app_settings (id VARCHAR(36) PRIMARY KEY)"))
+        async with eng.begin() as conn:
+            await _apply_lightweight_migrations(conn)
+        async with eng.begin() as conn:
+            await _apply_lightweight_migrations(conn)  # second pass — no error
+            cols = await _columns(conn, "app_settings")
+        assert "memory_enabled" in cols
+    finally:
+        await eng.dispose()
diff --git a/backend/tests/unit/test_pipeline_helpers.py b/backend/tests/unit/test_pipeline_helpers.py
index 2d43d03..e51b65f 100644
--- a/backend/tests/unit/test_pipeline_helpers.py
+++ b/backend/tests/unit/test_pipeline_helpers.py
@@ -66,6 +66,20 @@ def test_cover_letter_and_emails_called_out(self):
         assert "cover letter" in out
         assert "3 outreach email templates" in out
 
+    def test_notes_master_and_memory_sources(self):
+        """The reply records which inputs fed the generation, for later reference."""
+        from app.api._pipeline import _build_summary
+        resume = {"metadata": {"overall_score": 90}, "personal_info": {"full_name": "Ada"}}
+        out = _build_summary(resume, used_memory=True, from_master_name="Backend Master")
+        assert "Backend Master" in out
+        assert "profile memory" in out.lower()
+
+    def test_omits_sources_line_when_nothing_used(self):
+        from app.api._pipeline import _build_summary
+        resume = {"metadata": {"overall_score": 90}, "personal_info": {"full_name": "Ada"}}
+        out = _build_summary(resume)  # no master, memory off
+        assert "Generated using" not in out
+
 
 # ── _build_no_resume_error ────────────────────────────────────────────────────
 
diff --git a/frontend/css/sidebar.css b/frontend/css/sidebar.css
index 1e8b0f9..320250b 100644
--- a/frontend/css/sidebar.css
+++ b/frontend/css/sidebar.css
@@ -40,6 +40,14 @@ body.sb-resizing{cursor:col-resize;user-select:none}
 .logo-sub{font-size:10px;color:var(--text3);text-transform:uppercase;letter-spacing:.6px;margin-top:-2px}
 #new-btn{width:100%;padding:9px 13px;background:var(--abg);border:1px solid var(--ab);border-radius:8px;color:var(--accent2);font-family:var(--font);font-size:13.5px;font-weight:500;cursor:pointer;display:flex;align-items:center;gap:8px;transition:all .15s}
 #new-btn:hover{background:rgba(124,131,255,.22)}
+.nfm-split{display:flex;width:100%;margin-top:6px}
+#new-from-master-btn{flex:1;min-width:0;padding:7px 13px;background:transparent;border:1px solid var(--bdr);border-radius:8px 0 0 8px;border-right:none;color:var(--text2);font-family:var(--font);font-size:12.5px;font-weight:500;cursor:pointer;display:flex;align-items:center;gap:8px;transition:all .15s}
+#new-from-master-btn:hover{background:var(--bg3);border-color:var(--amber);color:var(--amber)}
+#new-from-master-btn .ic{color:var(--amber)}
+.nfm-caret{flex-shrink:0;width:34px;display:flex;align-items:center;justify-content:center;background:transparent;border:1px solid var(--bdr);border-radius:0 8px 8px 0;color:var(--text2);cursor:pointer;transition:all .15s}
+.nfm-caret:hover,.nfm-caret.on{background:var(--bg3);border-color:var(--amber);color:var(--amber)}
+.nfm-caret .ic{color:var(--amber);transition:transform .15s}
+.nfm-caret.on .ic{transform:rotate(180deg)}
 
 .ss{flex:1;overflow-y:auto;padding:6px 8px}
 .ss::-webkit-scrollbar{width:3px}.ss::-webkit-scrollbar-thumb{background:var(--border2);border-radius:2px}
diff --git a/frontend/index.html b/frontend/index.html
index 7003223..b7ec74a 100644
--- a/frontend/index.html
+++ b/frontend/index.html
@@ -71,6 +71,8 @@
   <symbol id="ic-zap" viewBox="0 0 24 24"><path d="M13 2 3 14h9l-1 8 10-12h-9l1-8Z"/></symbol>
   <symbol id="ic-link" viewBox="0 0 24 24"><path d="M10 13a5 5 0 0 0 7.54.54l3-3a5 5 0 0 0-7.07-7.07l-1.72 1.71"/><path d="M14 11a5 5 0 0 0-7.54-.54l-3 3a5 5 0 0 0 7.07 7.07l1.71-1.71"/></symbol>
   <symbol id="ic-star" viewBox="0 0 24 24"><path d="m12 2 3.09 6.26 6.91 1-5 4.87 1.18 6.88L12 17.77l-6.18 3.24L7 14.13 2 9.26l6.91-1L12 2Z"/></symbol>
+  <symbol id="ic-compare" viewBox="0 0 24 24"><rect x="3" y="4" width="7" height="16" rx="1"/><rect x="14" y="4" width="7" height="16" rx="1"/></symbol>
+  <symbol id="ic-chevron-down" viewBox="0 0 24 24"><path d="m6 9 6 6 6-6"/></symbol>
   <symbol id="ic-star-fill" viewBox="0 0 24 24"><path d="m12 2 3.09 6.26 6.91 1-5 4.87 1.18 6.88L12 17.77l-6.18 3.24L7 14.13 2 9.26l6.91-1L12 2Z" fill="currentColor"/></symbol>
   <symbol id="ic-message" viewBox="0 0 24 24"><path d="M21 15a2 2 0 0 1-2 2H7l-4 4V5a2 2 0 0 1 2-2h14a2 2 0 0 1 2 2Z"/></symbol>
   <symbol id="ic-bookmark" viewBox="0 0 24 24"><path d="m19 21-7-4-7 4V5a2 2 0 0 1 2-2h10a2 2 0 0 1 2 2Z"/></symbol>
@@ -102,6 +104,15 @@
       <svg class="ic sm"><use href="#ic-plus"/></svg>
       <span class="sb-label">New Resume</span>
     </button>
+    <div id="nfm-wrap" class="nfm-split" style="display:none">
+      <button id="new-from-master-btn" class="nfm-main" onclick="newChatFromMaster()" title="New chat from your default master resume — loaded as-is, no AI changes">
+        <svg class="ic sm"><use href="#ic-star"/></svg>
+        <span class="sb-label">New from Master</span>
+      </button>
+      <button id="nfm-caret" class="nfm-caret" onclick="toggleNewFromMasterMenu(event)" title="Pick which master resume to start from" aria-label="Pick which master resume to start from">
+        <svg class="ic sm"><use href="#ic-chevron-down"/></svg>
+      </button>
+    </div>
     <div class="sb-search">
       <svg class="ic sm sb-search-ic"><use href="#ic-search"/></svg>
       <input type="text" id="sb-search" placeholder="Search chats…" oninput="filterSessions()">
@@ -734,6 +745,7 @@
             <div id="vm-role" style="font-size:12px;color:var(--text3);overflow:hidden;text-overflow:ellipsis;white-space:nowrap"></div>
           </div>
           <div style="display:flex;gap:6px;flex-wrap:wrap;justify-content:flex-end">
+            <button class="bsec" onclick="newChatFromMaster(_selectedMasterId)" title="Open a new chat seeded with this master resume, unchanged — edit on top of it"><svg class="ic" style="width:13px;height:13px"><use href="#ic-plus"/></svg> New chat</button>
             <button class="bsec" onclick="replaceSelectedMaster()" title="Replace this resume's JSON from a file"><svg class="ic" style="width:13px;height:13px"><use href="#ic-attach"/></svg> Replace</button>
             <button class="bsec" onclick="copyMasterJson()"><svg class="ic" style="width:13px;height:13px"><use href="#ic-copy"/></svg> Copy JSON</button>
             <button class="bsec" onclick="downloadMasterJson()"><svg class="ic" style="width:13px;height:13px"><use href="#ic-download"/></svg> JSON</button>
@@ -802,6 +814,10 @@
     <button class="mx" onclick="closeDiff()">×</button>
     <div class="modal-title" style="margin-bottom:4px"><svg class="ic" style="margin-right:6px"><use href="#ic-refresh"/></svg>Resume Changes <span id="diff-ver" style="font-size:13px;color:var(--text3);font-weight:400;margin-left:4px"></span></div>
     <div style="font-size:11px;color:var(--text3);margin-bottom:16px;flex-shrink:0"><span class="df-del">removed</span> &nbsp;·&nbsp; <span class="df-add">added</span></div>
+    <div id="diff-master-row" style="display:none;margin-bottom:16px;flex-shrink:0;align-items:center;gap:8px">
+      <span style="font-size:12px;color:var(--text3);white-space:nowrap">Compare against master:</span>
+      <select id="diff-master-pick" class="fi" onchange="_renderMasterDiff()" style="font-size:12px;padding:4px 8px;min-width:0;flex:1;max-width:320px"></select>
+    </div>
     <div id="diff-body" style="overflow-y:auto;flex:1;padding-right:4px"></div>
   </div>
 </div>
diff --git a/frontend/js/diff.js b/frontend/js/diff.js
index 80c1e51..2eac5a1 100644
--- a/frontend/js/diff.js
+++ b/frontend/js/diff.js
@@ -1,14 +1,58 @@
 // ── Resume Diff ───────────────────────────────────────────────────────────────
+// The resume currently being compared against a master (master-mode only), so
+// the picker's onchange can re-render without re-resolving the message.
+let _diffMasterTargetResume=null;
+
+function _hideDiffMasterRow(){
+  const row=document.getElementById('diff-master-row');
+  if(row)row.style.display='none';
+}
+
 function openDiff(msgId,sessionId){
   const versions=_resumeVersions[sessionId]||[];
   const myIdx=versions.findIndex(v=>v.id===msgId);
   if(myIdx<1)return;
   const prev=versions[myIdx-1].resume;
   const curr=versions[myIdx].resume;
+  _diffMasterTargetResume=null;
+  _hideDiffMasterRow();
   document.getElementById('diff-ver').textContent=`(v${myIdx} → v${myIdx+1})`;
   document.getElementById('diff-body').innerHTML=_buildDiffHtml(prev,curr);
   document.getElementById('diff-modal').classList.add('open');
 }
+
+// Compare any resume version against a saved master resume (the reference).
+// Direction is master → this version, so tailoring/edits read as additions.
+function openDiffVsMaster(msgId){
+  const curr=window['__r_'+msgId];
+  if(!curr){showToast('Resume not loaded yet');return;}
+  if(typeof _masterResumes==='undefined'||!_masterResumes.length){
+    showToast('No master resume saved to compare against');return;
+  }
+  _diffMasterTargetResume=curr;
+  const sel=document.getElementById('diff-master-pick');
+  if(sel){
+    sel.innerHTML=_masterResumes.map(it=>
+      `<option value="${it.id}"${it.is_default?' selected':''}>${esc(it.name||'Untitled')}${it.is_default?' · default':''}</option>`
+    ).join('');
+  }
+  const row=document.getElementById('diff-master-row');
+  if(row)row.style.display='flex';
+  _renderMasterDiff();
+  document.getElementById('diff-modal').classList.add('open');
+}
+
+function _renderMasterDiff(){
+  if(!_diffMasterTargetResume)return;
+  const sel=document.getElementById('diff-master-pick');
+  const mid=sel?sel.value:null;
+  const m=(_masterResumes.find(x=>x.id===mid))||_defaultMaster();
+  const body=document.getElementById('diff-body');
+  if(!m){body.innerHTML='<div style="color:var(--text3);font-size:13px;text-align:center;padding:32px 0">No master resume to compare against.</div>';return;}
+  document.getElementById('diff-ver').textContent=`(Master “${m.name}” → this version)`;
+  body.innerHTML=_buildDiffHtml(m.resume,_diffMasterTargetResume);
+}
+
 function closeDiff(){document.getElementById('diff-modal').classList.remove('open');}
 document.getElementById('diff-modal').addEventListener('click',e=>{if(e.target===e.currentTarget)closeDiff();});
 
diff --git a/frontend/js/init.js b/frontend/js/init.js
index c45eb61..43b03c0 100644
--- a/frontend/js/init.js
+++ b/frontend/js/init.js
@@ -39,9 +39,8 @@ async function init(){
     document.getElementById('sb').classList.add('collapsed');
     document.getElementById('sb-expand').style.display='flex';
   }
-  // Pull live server defaults so the pill shows what'll actually run
+  // Pull live server defaults (provider/model, JD slider, memory pref) in one go
   await loadServerDefaults();
-  loadJdIntensityDefault();
   updProvPill();
   await loadSessions();
   await loadMemoryChip();
@@ -64,27 +63,24 @@ function _scheduleSessionRefreshIfNeeded(){
 }
 
 async function loadServerDefaults(){
-  // Only override localStorage if user hasn't picked anything yet
-  if(localStorage.getItem('sr_p')&&localStorage.getItem('sr_m'))return;
+  // One fetch for every server-side default applied on bootstrap.
   try{
     const r=await fetch(`${API}/api/app-settings`);
     if(!r.ok)return;
     const s=await r.json();
-    if(s.llm_provider)curProv=s.llm_provider;
-    if(s.llm_model)curModel=s.llm_model;
-  }catch(e){}
-}
-
-async function loadJdIntensityDefault(){
-  try{
-    const r=await fetch(`${API}/api/app-settings`);
-    if(!r.ok)return;
-    const s=await r.json();
-    if(s.default_jd_intensity==null)return;
-    const v=Math.max(0,Math.min(100,parseInt(s.default_jd_intensity,10)));
-    if(isNaN(v))return;
-    const sl=document.getElementById('jd-intensity');
-    if(sl){sl.value=v;updateJdIntensityUI();}
+    // Adopt the server provider/model only if the user hasn't picked one yet.
+    if(!(localStorage.getItem('sr_p')&&localStorage.getItem('sr_m'))){
+      if(s.llm_provider)curProv=s.llm_provider;
+      if(s.llm_model)curModel=s.llm_model;
+    }
+    // Remembered JD-tailoring slider
+    if(s.default_jd_intensity!=null){
+      const v=Math.max(0,Math.min(100,parseInt(s.default_jd_intensity,10)));
+      const sl=document.getElementById('jd-intensity');
+      if(sl&&!isNaN(v)){sl.value=v;updateJdIntensityUI();}
+    }
+    // Global "Use memory" preference (persisted across chats/sessions)
+    applyMemoryPref(s.memory_enabled);
   }catch(e){}
 }
 
diff --git a/frontend/js/master-resumes.js b/frontend/js/master-resumes.js
index b627352..0d67b5c 100644
--- a/frontend/js/master-resumes.js
+++ b/frontend/js/master-resumes.js
@@ -34,6 +34,10 @@ function _updateMasterUI(){
   // Sidebar footer badge
   const badge=document.getElementById('master-sb-badge');
   if(badge)badge.style.display=has?'':'none';
+  // Sidebar "New from Master" split-button (only useful once a master exists)
+  const nfm=document.getElementById('nfm-wrap');
+  if(nfm)nfm.style.display=has?'':'none';
+  if(!has)_hideNfmMenu();
   // Data Management section
   const dmTitle=document.getElementById('dm-master-title');
   if(dmTitle){
@@ -367,7 +371,7 @@ function attachMasterById(id){
   _hideMasterAttachMenu();
   const it=_masterResumes.find(x=>x.id===id);
   if(!it){showToast('Master resume not found');return;}
-  attachedResume={kind:'json',resume:it.resume,filename:`master_${(it.name||'resume').replace(/\s+/g,'_')}.json`,fromMaster:true};
+  attachedResume={kind:'json',resume:it.resume,filename:`master_${(it.name||'resume').replace(/\s+/g,'_')}.json`,fromMaster:true,masterName:it.name||'Master Resume'};
   document.getElementById('att-ic').innerHTML='<svg class="ic"><use href="#ic-star"/></svg>';
   document.getElementById('att-label').textContent=it.name||'Master Resume';
   document.getElementById('att-meta').textContent='JSON';
@@ -382,6 +386,87 @@ function closeMasterAttachAndOpenModal(){
   openMasterModal();
 }
 
+// ── New chat seeded from a master (verbatim — no AI edits) ───────────────────
+// Creates a fresh chat whose first resume version IS the master, untouched, so
+// you can edit on top of it (or just fill the application tracker) without
+// re-running the pipeline. With no id: uses the only master, else the default.
+async function newChatFromMaster(id){
+  _hideMasterAttachMenu();
+  _hideNfmMenu();
+  let masterId=id||null;
+  if(!masterId){
+    if(_masterResumes.length===0){showToast('No master resumes saved yet');return;}
+    if(_masterResumes.length===1)masterId=_masterResumes[0].id;
+    else masterId=(_defaultMaster()||{}).id||null;
+  }
+  try{
+    const r=await fetch(`${API}/api/sessions/from-master`,{
+      method:'POST',headers:{'Content-Type':'application/json'},
+      body:JSON.stringify({master_id:masterId,llm_provider:curProv,llm_model:curModel}),
+    });
+    if(!r.ok){const e=await r.json().catch(()=>({detail:'Failed to create chat'}));throw new Error(e.detail||'Failed to create chat');}
+    const sess=await r.json();
+    closeViewMaster();
+    await _refreshSessions();   // surface the new chat at the top of the sidebar
+    await loadSession(sess.id); // …and open it
+    showToast('⭐ New chat from master — edit away');
+  }catch(e){showToast('⚠ '+e.message);}
+}
+
+// ── "New from Master" caret dropdown — pick a specific master (≥1 saved) ──────
+// The split button's main half uses the default master; this menu lists them
+// all (rendered into <body> with position:fixed so the sidebar can't clip it).
+function _ensureNfmMenu(){
+  let menu=document.getElementById('nfm-menu');
+  if(!menu){
+    menu=document.createElement('div');
+    menu.id='nfm-menu';
+    menu.style.cssText='display:none;position:fixed;min-width:240px;max-width:340px;background:var(--bg2);border:1px solid var(--border2);border-radius:9px;box-shadow:var(--shadow-md);padding:4px;z-index:1500;max-height:320px;overflow-y:auto';
+    document.body.appendChild(menu);
+  }
+  return menu;
+}
+
+function toggleNewFromMasterMenu(ev){
+  if(ev)ev.stopPropagation();
+  if(_masterResumes.length===0){showToast('No master resumes saved yet');return;}
+  const caret=document.getElementById('nfm-caret');
+  const menu=_ensureNfmMenu();
+  if(menu.dataset.open==='1'){_hideNfmMenu();return;}
+  menu.innerHTML=`<div style="padding:6px 10px;font-size:11px;color:var(--text3);text-transform:uppercase;letter-spacing:.06em">New chat from…</div>`+
+    _masterResumes.map(it=>`<button class="vm-menu-item" onclick="newChatFromMaster('${it.id}')" data-id="${it.id}">
+      <span style="opacity:${it.is_default?1:0.35}">${it.is_default?'⭐':'☆'}</span>
+      <span style="flex:1;min-width:0;overflow:hidden;text-overflow:ellipsis;white-space:nowrap">${esc(it.name||'Untitled')}</span>
+    </button>`).join('');
+  // Anchor to the caret; open downward if there's room, else upward.
+  const r=caret.getBoundingClientRect();
+  menu.style.visibility='hidden';menu.style.display='block';menu.style.top='0px';menu.style.left='0px';
+  const mh=menu.offsetHeight,mw=menu.offsetWidth;
+  const spaceBelow=window.innerHeight-r.bottom,spaceAbove=r.top;
+  const top=(spaceBelow>=mh+8||spaceBelow>spaceAbove)?Math.min(window.innerHeight-mh-8,r.bottom+6):Math.max(8,r.top-mh-6);
+  const left=Math.max(8,Math.min(window.innerWidth-mw-8,r.right-mw));
+  menu.style.top=`${top}px`;menu.style.left=`${left}px`;menu.style.visibility='';
+  menu.dataset.open='1';
+  if(caret)caret.classList.add('on');
+  document.addEventListener('click',_onNfmClickAway);
+}
+
+function _onNfmClickAway(ev){
+  const wrap=document.getElementById('nfm-wrap');
+  const menu=document.getElementById('nfm-menu');
+  if(wrap&&wrap.contains(ev.target))return;
+  if(menu&&menu.contains(ev.target))return;
+  _hideNfmMenu();
+}
+
+function _hideNfmMenu(){
+  const menu=document.getElementById('nfm-menu');
+  if(menu){menu.style.display='none';menu.dataset.open='0';}
+  const caret=document.getElementById('nfm-caret');
+  if(caret)caret.classList.remove('on');
+  document.removeEventListener('click',_onNfmClickAway);
+}
+
 // Stable entry points called from sidebar and data-mgmt.
 function attachMasterResume(){toggleMasterAttachMenu();}
 async function deleteMasterResume(){return deleteSelectedMaster();}
diff --git a/frontend/js/render.js b/frontend/js/render.js
index 932e6dd..68c472e 100644
--- a/frontend/js/render.js
+++ b/frontend/js/render.js
@@ -142,11 +142,13 @@ function renderDone(c,msg){
     return;
   }
   const agMap={};trace.forEach(e=>{agMap[e.agent]=e;});
+  let _nSteps=0;
   let stepsHtml=`<div class="prog-panel" style="margin-bottom:14px">
     <div class="prog-hdr" style="color:var(--green)">${ico('ic-check')} Pipeline complete</div>
     <div class="step-list">`;
   STEPS.forEach(step=>{
     const ev=agMap[step.agent];if(!ev)return;
+    _nSteps++;
     const cls=ev.status==='complete'?'done':ev.status==='error'?'err':'pend';
     const ic=cls==='done'?ico('ic-check'):ico('ic-x');
     let llmHtml='';
@@ -163,6 +165,9 @@ function renderDone(c,msg){
         <div class="snotes">${esc(ev.notes||step.desc)}</div>${llmHtml}</div></div>`;
   });
   stepsHtml+='</div></div>';
+  // A verbatim "from master" seed has no pipeline agents to show — skip the
+  // (otherwise empty) "Pipeline complete" panel for it.
+  if(_nSteps===0)stepsHtml='';
 
   let scHtml='';
   if(meta.overall_score!==undefined){
@@ -186,9 +191,19 @@ function renderDone(c,msg){
         _resumeVersions[sid].push({id:msg.id,resume});
       const versions=_resumeVersions[sid];
       const myIdx=versions.findIndex(v=>v.id===msg.id);
-      if(myIdx>0){
-        const jvr=c.querySelector('.jv-r');
-        if(jvr){
+      const jvr=c.querySelector('.jv-r');
+      if(jvr){
+        // Compare this resume version against a saved master reference.
+        if(typeof _masterResumes!=='undefined' && _masterResumes.length){
+          const mbtn=document.createElement('button');
+          mbtn.className='ibtn cmp';
+          mbtn.innerHTML=ico('ic-compare')+` Compare to Master`;
+          mbtn.title='Compare this resume against your master resume — see what changed';
+          mbtn.onclick=()=>openDiffVsMaster(msg.id);
+          jvr.prepend(mbtn);
+        }
+        // Version-to-version diff (only once a previous version exists).
+        if(myIdx>0){
           const btn=document.createElement('button');
           btn.className='ibtn cmp';
           btn.innerHTML=ico('ic-refresh')+` v${myIdx}→v${myIdx+1}`;
diff --git a/frontend/js/send.js b/frontend/js/send.js
index 6bd4229..ef6a572 100644
--- a/frontend/js/send.js
+++ b/frontend/js/send.js
@@ -38,10 +38,12 @@ async function sendMsg(){
   let finalContent=content;
   let attachedJson=null;
   let attachLabel=null;
+  let fromMasterName=null;
   if(attachedResume){
     if(attachedResume.kind==='json'&&attachedResume.resume){
       attachedJson=attachedResume.resume;
       attachLabel=`JSON · ${attachedResume.filename||'resume.json'}`;
+      if(attachedResume.fromMaster)fromMasterName=attachedResume.masterName||'Master Resume';
     }else if(attachedResume.kind==='text'&&attachedResume.text){
       const trimmed=attachedResume.text.length>14000?attachedResume.text.slice(0,14000)+'\n…[truncated]':attachedResume.text;
       finalContent=`[ATTACHED RESUME — extracted from ${attachedResume.filename||'upload'}]\n${trimmed}\n[END ATTACHED RESUME]\n\n${content}`;
@@ -73,7 +75,7 @@ async function sendMsg(){
   try{
     const r=await fetch(`${API}/api/sessions/${sessionId}/messages`,{
       method:'POST',headers:{'Content-Type':'application/json'},
-      body:JSON.stringify({content:finalContent,jd_text:jdText||null,jd_intensity:jdIntensity,attached_resume:submitAttached,use_memory:isMemoryOn(),llm_provider:curProv,llm_model:curModel})});
+      body:JSON.stringify({content:finalContent,jd_text:jdText||null,jd_intensity:jdIntensity,attached_resume:submitAttached,from_master_name:fromMasterName,use_memory:isMemoryOn(),llm_provider:curProv,llm_model:curModel})});
     if(!r.ok)throw new Error(`HTTP ${r.status}`);
     const d=await r.json();
 
diff --git a/frontend/js/sessions.js b/frontend/js/sessions.js
index 1926740..78bd222 100644
--- a/frontend/js/sessions.js
+++ b/frontend/js/sessions.js
@@ -141,18 +141,16 @@ async function newChat(){
   document.getElementById('tb-title').textContent='New Resume';
   syncToolbar();
   document.querySelectorAll('.si').forEach(e=>e.classList.remove('on'));
-  // Memory toggle is per-chat; new chats always start with it enabled.
-  const mt=document.getElementById('mem-toggle');
-  mt.classList.add('on');mt.classList.remove('off');
+  // "Use memory" is a global preference (loaded on bootstrap, persisted to the
+  // backend) — leave the toggle as the user set it, don't reset per chat.
   document.getElementById('mi').focus();
 }
 
 async function loadSession(id){
   closeMobileSidebar();
   curSession=id;
-  // Memory toggle resets to on for each chat (per-chat, not global).
-  const mt=document.getElementById('mem-toggle');
-  mt.classList.add('on');mt.classList.remove('off');
+  // "Use memory" is global (not per-chat) — the toggle keeps the user's
+  // persisted preference across chats; nothing to reset here.
   let session;
   try{
     const r=await fetch(`${API}/api/sessions/${id}`);
diff --git a/frontend/js/toggles.js b/frontend/js/toggles.js
index 62f2395..65f3d25 100644
--- a/frontend/js/toggles.js
+++ b/frontend/js/toggles.js
@@ -1,13 +1,30 @@
-// ── Memory toggle ───────────────────────────────────────────────────────────
+// ── Memory toggle (global, persisted across chats/sessions) ──────────────────
 function toggleMemPill(){
   const el=document.getElementById('mem-toggle');
   const willBeOn=el.classList.contains('off');
   el.classList.toggle('on',willBeOn);
   el.classList.toggle('off',!willBeOn);
+  _saveMemoryPref(willBeOn);
 }
 function isMemoryOn(){
   return document.getElementById('mem-toggle').classList.contains('on');
 }
+// Reflect the persisted global preference (default ON when unset/null).
+function applyMemoryPref(enabled){
+  const el=document.getElementById('mem-toggle');
+  if(!el)return;
+  const on=enabled!==false;
+  el.classList.toggle('on',on);
+  el.classList.toggle('off',!on);
+}
+// Persist the preference to the backend so it survives reloads and applies to
+// every chat. Fire-and-forget — the toggle is already updated in the DOM.
+function _saveMemoryPref(on){
+  fetch(`${API}/api/app-settings`,{
+    method:'PUT',headers:{'Content-Type':'application/json'},
+    body:JSON.stringify({memory_enabled:!!on}),
+  }).catch(()=>{});
+}
 
 // ── JD toggle ───────────────────────────────────────────────────────────────
 function toggleJD(){