diff --git a/tracker/__init__.py b/tracker/__init__.py
new file mode 100644
index 0000000..826d82d
--- /dev/null
+++ b/tracker/__init__.py
@@ -0,0 +1 @@
+"""Personal Tracker CLI - A Markdown-based project tracking system."""
diff --git a/tracker/models/__init__.py b/tracker/models/__init__.py
new file mode 100644
index 0000000..6e04711
--- /dev/null
+++ b/tracker/models/__init__.py
@@ -0,0 +1,7 @@
+"""Data models for the tracker."""
+from .project import Project
+from .session import Session
+from .note import Note, NoteType
+from .change import Change
+
+__all__ = ["Project", "Session", "Note", "NoteType", "Change"]
diff --git a/tracker/models/change.py b/tracker/models/change.py
new file mode 100644
index 0000000..a94bb65
--- /dev/null
+++ b/tracker/models/change.py
@@ -0,0 +1,13 @@
+"""Change model definition."""
+from pydantic import BaseModel, Field
+from datetime import date
+
+
+class Change(BaseModel):
+    """Represents a notable change in a project."""
+
+    date: date
+    type: str  # code, infra, config, docs, automation, decision
+    title: str
+    impact: str = ""
+    references: list[str] = Field(default_factory=list)
diff --git a/tracker/models/note.py b/tracker/models/note.py
new file mode 100644
index 0000000..5ebbfb8
--- /dev/null
+++ b/tracker/models/note.py
@@ -0,0 +1,22 @@
+"""Note model definition."""
+from enum import Enum
+from pydantic import BaseModel, Field
+from datetime import datetime
+
+
+class NoteType(Enum):
+    """Types of notes that can be recorded during a session."""
+    WORK = "work"
+    CHANGE = "change"
+    BLOCKER = "blocker"
+    DECISION = "decision"
+    IDEA = "idea"
+    REFERENCE = "reference"
+
+
+class Note(BaseModel):
+    """Represents a note recorded during a session."""
+
+    type: NoteType
+    text: str
+    created_at: datetime = Field(default_factory=datetime.now)
diff --git a/tracker/models/project.py b/tracker/models/project.py
new file mode 100644
index 0000000..d2d9a95
--- /dev/null
+++ b/tracker/models/project.py
@@ -0,0 +1,21 @@
+"""Project model definition."""
+from pydantic import BaseModel, Field
+from datetime import datetime
+from typing import Optional
+
+
+class Project(BaseModel):
+    """Represents a tracked project."""
+
+    id: str
+    name: str
+    slug: str
+    description: str = ""
+    type: str = "misc"  # code, homelab, automation, agent, research, misc
+    status: str = "inbox"  # inbox, next, active, blocked, waiting, done, archived
+    tags: list[str] = Field(default_factory=list)
+    root_path: str = ""
+    repo_path: Optional[str] = None
+    created_at: datetime
+    updated_at: datetime
+    last_session_at: Optional[datetime] = None
diff --git a/tracker/models/session.py b/tracker/models/session.py
new file mode 100644
index 0000000..1de3647
--- /dev/null
+++ b/tracker/models/session.py
@@ -0,0 +1,23 @@
+"""Session model definition."""
+from pydantic import BaseModel, Field
+from datetime import datetime
+from typing import Optional
+
+
+class Session(BaseModel):
+    """Represents a work session on a project."""
+
+    id: str
+    project_slug: str
+    started_at: datetime
+    ended_at: Optional[datetime] = None
+    duration_minutes: Optional[int] = None
+    objective: str = ""
+    summary: str = ""
+    work_done: list[str] = Field(default_factory=list)
+    changes: list[str] = Field(default_factory=list)
+    decisions: list[str] = Field(default_factory=list)
+    blockers: list[str] = Field(default_factory=list)
+    next_steps: list[str] = Field(default_factory=list)
+    references: list[str] = Field(default_factory=list)
+    raw_notes: list[dict] = Field(default_factory=list)  # [{"type": "work", "text": "...", "timestamp": "..."}]
diff --git a/tracker/services/__init__.py b/tracker/services/__init__.py
new file mode 100644
index 0000000..5178642
--- /dev/null
+++ b/tracker/services/__init__.py
@@ -0,0 +1,45 @@
+"""Services layer for business logic."""
+
+from .session_service import (
+    get_active_session,
+    set_active_session,
+    clear_active_session,
+    get_active_session_path,
+    validate_no_other_active_session,
+)
+from .project_service import (
+    create_project,
+    get_project,
+    update_project,
+    list_projects,
+    get_projects_root,
+    ensure_project_structure,
+)
+from .note_service import (
+    add_note,
+    consolidate_notes,
+)
+from .heuristics_service import (
+    suggest_next_steps,
+)
+from .summary_service import (
+    generate_summary,
+)
+
+__all__ = [
+    "get_active_session",
+    "set_active_session",
+    "clear_active_session",
+    "get_active_session_path",
+    "validate_no_other_active_session",
+    "create_project",
+    "get_project",
+    "update_project",
+    "list_projects",
+    "get_projects_root",
+    "ensure_project_structure",
+    "add_note",
+    "consolidate_notes",
+    "suggest_next_steps",
+    "generate_summary",
+]
diff --git a/tracker/services/heuristics_service.py b/tracker/services/heuristics_service.py
new file mode 100644
index 0000000..b4f3f16
--- /dev/null
+++ b/tracker/services/heuristics_service.py
@@ -0,0 +1,47 @@
+"""Heuristics service for suggestions based on rules."""
+
+from datetime import datetime
+
+from ..models import Session, Project
+
+
+def suggest_next_steps(session: Session, project: Project) -> list[str]:
+    """
+    Generate suggestions based on session state and project context.
+    Rules:
+    - si hay blockers abiertos, sugerir "Destrabar: [bloqueos]"
+    - si hay changes sin references, sugerir "Validar cambios recientes"
+    - si work_done está vacío y session > 30 min, sugerir "Revisar progreso del objetivo"
+    - si no hay next_steps definidos, sugerir "Definir próximos pasos"
+    """
+    suggestions = []
+
+    # Rule: blockers open
+    if session.blockers:
+        for blocker in session.blockers:
+            suggestions.append(f"Destrabar: {blocker}")
+
+    # Rule: changes without references
+    changes_without_refs = []
+    for change in session.changes:
+        # Simple heuristic: if change doesn't reference anything specific
+        if change and not any(ref in change.lower() for ref in ["#", "commit", "pr", "issue"]):
+            changes_without_refs.append(change)
+
+    if changes_without_refs:
+        suggestions.append("Validar cambios recientes")
+
+    # Rule: work_done empty and session > 30 minutes
+    if not session.work_done:
+        duration = session.duration_minutes
+        if duration == 0 and session.ended_at and session.started_at:
+            duration = int((session.ended_at - session.started_at).total_seconds() / 60)
+
+        if duration > 30:
+            suggestions.append("Revisar progreso del objetivo")
+
+    # Rule: no next_steps defined
+    if not session.next_steps:
+        suggestions.append("Definir próximos pasos")
+
+    return suggestions
diff --git a/tracker/services/note_service.py b/tracker/services/note_service.py
new file mode 100644
index 0000000..6c911b3
--- /dev/null
+++ b/tracker/services/note_service.py
@@ -0,0 +1,65 @@
+"""Note service for note management."""
+
+from datetime import datetime
+
+from ..models import Session, NoteType, Note
+
+
+def add_note(session: Session, note_type: str, text: str) -> dict:
+    """
+    Add a note to the session and return the note dict.
+    Valid note types: work, change, blocker, decision, idea, reference
+    """
+    try:
+        note_type_enum = NoteType(note_type)
+    except ValueError:
+        raise ValueError(f"Invalid note type: {note_type}. Valid types are: {[t.value for t in NoteType]}")
+
+    note = Note(type=note_type_enum, text=text)
+    session.raw_notes.append(note.model_dump(mode="json"))
+
+    return {
+        "type": note.type.value,
+        "text": note.text,
+        "created_at": note.created_at.isoformat(),
+    }
+
+
+def consolidate_notes(raw_notes: list[dict]) -> dict:
+    """
+    Consolidate raw notes into categorized sections.
+    Returns dict with keys: work_done, changes, decisions, blockers, references
+    """
+    result = {
+        "work_done": [],
+        "changes": [],
+        "decisions": [],
+        "blockers": [],
+        "references": [],
+    }
+
+    for note in raw_notes:
+        if isinstance(note, dict):
+            note_type = note.get("type", "")
+            text = note.get("text", "")
+        else:
+            # Handle string format like "[type] text"
+            parts = note.split("]", 1)
+            if len(parts) == 2:
+                note_type = parts[0][1:]
+                text = parts[1].strip()
+            else:
+                continue
+
+        if note_type == NoteType.WORK.value:
+            result["work_done"].append(text)
+        elif note_type == NoteType.CHANGE.value:
+            result["changes"].append(text)
+        elif note_type == NoteType.DECISION.value:
+            result["decisions"].append(text)
+        elif note_type == NoteType.BLOCKER.value:
+            result["blockers"].append(text)
+        elif note_type == NoteType.REFERENCE.value:
+            result["references"].append(text)
+
+    return result
diff --git a/tracker/services/project_service.py b/tracker/services/project_service.py
new file mode 100644
index 0000000..3b17f8f
--- /dev/null
+++ b/tracker/services/project_service.py
@@ -0,0 +1,118 @@
+"""Project service for project management."""
+
+import uuid
+from datetime import datetime
+from pathlib import Path
+from typing import Optional
+
+from ..models import Project
+
+
+_PROJECTS_ROOT = Path("projects")
+
+
+def get_projects_root() -> Path:
+    """Return the root directory for all projects."""
+    return _PROJECTS_ROOT
+
+
+def _get_project_meta_path(slug: str) -> Path:
+    """Return the path to the project's meta/project.yaml file."""
+    return _PROJECTS_ROOT / slug / "meta" / "project.yaml"
+
+
+def _get_project_readme_path(slug: str) -> Path:
+    """Return the path to the project's README.md file."""
+    return _PROJECTS_ROOT / slug / "README.md"
+
+
+def create_project(
+    name: str,
+    slug: str,
+    description: str = "",
+    type: str = "misc",
+    tags: Optional[list[str]] = None,
+    repo_path: Optional[Path] = None,
+) -> Project:
+    """
+    Create a new project and return the Project instance.
+    Note: This does not write any files - that is handled by storage.
+    """
+    if tags is None:
+        tags = []
+
+    project = Project(
+        id=str(uuid.uuid4()),
+        name=name,
+        slug=slug,
+        description=description,
+        type=type,
+        status="inbox",
+        tags=tags,
+        root_path=_PROJECTS_ROOT / slug,
+        repo_path=repo_path,
+        created_at=datetime.now(),
+        updated_at=datetime.now(),
+    )
+    return project
+
+
+def get_project(slug: str) -> Optional[Project]:
+    """
+    Get a project by slug.
+    Note: This reads from file system - placeholder for storage integration.
+    """
+    meta_path = _get_project_meta_path(slug)
+    if not meta_path.exists():
+        return None
+    # TODO: Load from storage (YAML)
+    return None
+
+
+def update_project(slug: str, **kwargs) -> Optional[Project]:
+    """
+    Update a project's attributes.
+    Note: This does not persist - that is handled by storage.
+    """
+    project = get_project(slug)
+    if project is None:
+        return None
+
+    for key, value in kwargs.items():
+        if hasattr(project, key):
+            setattr(project, key, value)
+
+    project.updated_at = datetime.now()
+    return project
+
+
+def list_projects() -> list[Project]:
+    """
+    List all projects.
+    Note: This reads from file system - placeholder for storage integration.
+    """
+    projects_root = get_projects_root()
+    if not projects_root.exists():
+        return []
+
+    projects = []
+    for item in projects_root.iterdir():
+        if item.is_dir() and not item.name.startswith("."):
+            project = get_project(item.name)
+            if project is not None:
+                projects.append(project)
+
+    return projects
+
+
+def ensure_project_structure(slug: str) -> None:
+    """
+    Ensure the project directory structure exists.
+    Creates: sessions/, docs/, assets/, meta/
+    Note: This creates directories only - actual file writing is storage's job.
+    """
+    project_root = _PROJECTS_ROOT / slug
+    directories = ["sessions", "docs", "assets", "meta"]
+
+    for directory in directories:
+        (project_root / directory).mkdir(parents=True, exist_ok=True)
diff --git a/tracker/services/session_service.py b/tracker/services/session_service.py
new file mode 100644
index 0000000..21e7a43
--- /dev/null
+++ b/tracker/services/session_service.py
@@ -0,0 +1,67 @@
+"""Session service for active session management."""
+
+import json
+from datetime import datetime
+from pathlib import Path
+from typing import Optional
+
+from ..models import Session
+
+
+_ACTIVE_SESSION_FILE = ".active_session.json"
+
+
+def get_active_session_path() -> Path:
+    """Return the path to the active session file in projects/ directory."""
+    return Path("projects") / _ACTIVE_SESSION_FILE
+
+
+def get_active_session() -> Optional[Session]:
+    """Load and return the currently active session, or None if none exists."""
+    path = get_active_session_path()
+    if not path.exists():
+        return None
+
+    with open(path, "r", encoding="utf-8") as f:
+        data = json.load(f)
+
+    # Convert started_at string back to datetime
+    data["started_at"] = datetime.fromisoformat(data["started_at"])
+    if data.get("ended_at"):
+        data["ended_at"] = datetime.fromisoformat(data["ended_at"])
+
+    return Session(**data)
+
+
+def set_active_session(session: Session) -> None:
+    """Save the given session as the active session."""
+    path = get_active_session_path()
+    path.parent.mkdir(parents=True, exist_ok=True)
+
+    data = session.model_dump(mode="json")
+    # Serialize datetime objects to ISO format
+    data["started_at"] = session.started_at.isoformat()
+    if session.ended_at:
+        data["ended_at"] = session.ended_at.isoformat()
+
+    with open(path, "w", encoding="utf-8") as f:
+        json.dump(data, f, indent=2, ensure_ascii=False)
+
+
+def clear_active_session() -> None:
+    """Remove the active session file."""
+    path = get_active_session_path()
+    if path.exists():
+        path.unlink()
+
+
+def validate_no_other_active_session(project_slug: str) -> bool:
+    """
+    Check if there is an active session for a different project.
+    Returns True if no conflict exists (i.e., either no active session
+    or the active session belongs to the same project).
+    """
+    active = get_active_session()
+    if active is None:
+        return True
+    return active.project_slug == project_slug
diff --git a/tracker/services/summary_service.py b/tracker/services/summary_service.py
new file mode 100644
index 0000000..8a9d763
--- /dev/null
+++ b/tracker/services/summary_service.py
@@ -0,0 +1,42 @@
+"""Summary service for heuristic summary generation."""
+
+from ..models import Session
+from .note_service import consolidate_notes
+
+
+def generate_summary(session: Session) -> str:
+    """
+    Generate a heuristic summary from the session.
+    Uses consolidate_notes to extract work_done, decisions, blockers.
+    """
+    # Consolidate raw notes into categorized sections
+    consolidated = consolidate_notes(session.raw_notes)
+
+    lines = []
+
+    # Work done section
+    if consolidated["work_done"]:
+        lines.append("Trabajo realizado:")
+        for item in consolidated["work_done"]:
+            lines.append(f"  - {item}")
+        lines.append("")
+
+    # Decisions section
+    if consolidated["decisions"]:
+        lines.append("Decisiones:")
+        for item in consolidated["decisions"]:
+            lines.append(f"  - {item}")
+        lines.append("")
+
+    # Blockers section
+    if consolidated["blockers"]:
+        lines.append("Bloqueos:")
+        for item in consolidated["blockers"]:
+            lines.append(f"  - {item}")
+        lines.append("")
+
+    # If no content, provide a minimal summary
+    if not lines:
+        return f"Session de {session.duration_minutes} minutos sin progreso registrado."
+
+    return "\n".join(lines)
diff --git a/tracker/storage/__init__.py b/tracker/storage/__init__.py
new file mode 100644
index 0000000..661ab23
--- /dev/null
+++ b/tracker/storage/__init__.py
@@ -0,0 +1,5 @@
+from tracker.storage.file_storage import FileStorage
+from tracker.storage.markdown_reader import MarkdownReader
+from tracker.storage.markdown_writer import MarkdownWriter
+
+__all__ = ["FileStorage", "MarkdownReader", "MarkdownWriter"]
diff --git a/tracker/storage/file_storage.py b/tracker/storage/file_storage.py
new file mode 100644
index 0000000..ffdce31
--- /dev/null
+++ b/tracker/storage/file_storage.py
@@ -0,0 +1,158 @@
+"""Storage layer for file-based persistence."""
+
+import json
+from pathlib import Path
+from typing import Optional
+
+import yaml
+
+from tracker.models.session import Session
+
+
+class FileStorage:
+    """Maneja lectura/escritura de archivos del proyecto."""
+
+    def __init__(self, projects_root: Path):
+        self.projects_root = projects_root
+
+    def _project_path(self, slug: str) -> Path:
+        return self.projects_root / slug
+
+    def _meta_path(self, slug: str) -> Path:
+        return self._project_path(slug) / "meta" / "project.yaml"
+
+    def _log_path(self, slug: str) -> Path:
+        return self._project_path(slug) / "LOG.md"
+
+    def _changelog_path(self, slug: str) -> Path:
+        return self._project_path(slug) / "CHANGELOG.md"
+
+    def _tasks_path(self, slug: str) -> Path:
+        return self._project_path(slug) / "TASKS.md"
+
+    def _readme_path(self, slug: str) -> Path:
+        return self._project_path(slug) / "README.md"
+
+    def _sessions_path(self, slug: str) -> Path:
+        return self._project_path(slug) / "sessions"
+
+    def read_project_meta(self, slug: str) -> dict:
+        """Lee projects/<slug>/meta/project.yaml"""
+        meta_path = self._meta_path(slug)
+        with open(meta_path, "r", encoding="utf-8") as f:
+            return yaml.safe_load(f)
+
+    def write_project_meta(self, slug: str, data: dict) -> None:
+        """Escribe projects/<slug>/meta/project.yaml"""
+        meta_path = self._meta_path(slug)
+        meta_path.parent.mkdir(parents=True, exist_ok=True)
+        with open(meta_path, "w", encoding="utf-8") as f:
+            yaml.safe_dump(data, f, default_flow_style=False, allow_unicode=True)
+
+    def read_log(self, slug: str) -> str:
+        """Lee projects/<slug>/LOG.md"""
+        log_path = self._log_path(slug)
+        if not log_path.exists():
+            return ""
+        with open(log_path, "r", encoding="utf-8") as f:
+            return f.read()
+
+    def append_to_log(self, slug: str, entry: str) -> None:
+        """Append a LOG.md entry."""
+        log_path = self._log_path(slug)
+        with open(log_path, "a", encoding="utf-8") as f:
+            f.write(entry)
+
+    def read_changelog(self, slug: str) -> str:
+        """Lee projects/<slug>/CHANGELOG.md"""
+        changelog_path = self._changelog_path(slug)
+        if not changelog_path.exists():
+            return ""
+        with open(changelog_path, "r", encoding="utf-8") as f:
+            return f.read()
+
+    def append_to_changelog(self, slug: str, change: str) -> None:
+        """Append a CHANGELOG.md entry."""
+        changelog_path = self._changelog_path(slug)
+        with open(changelog_path, "a", encoding="utf-8") as f:
+            f.write(change)
+
+    def read_tasks(self, slug: str) -> str:
+        """Lee projects/<slug>/TASKS.md"""
+        tasks_path = self._tasks_path(slug)
+        if not tasks_path.exists():
+            return ""
+        with open(tasks_path, "r", encoding="utf-8") as f:
+            return f.read()
+
+    def write_tasks(self, slug: str, tasks_content: str) -> None:
+        """Escribe projects/<slug>/TASKS.md"""
+        tasks_path = self._tasks_path(slug)
+        with open(tasks_path, "w", encoding="utf-8") as f:
+            f.write(tasks_content)
+
+    def read_readme(self, slug: str) -> str:
+        """Lee projects/<slug>/README.md"""
+        readme_path = self._readme_path(slug)
+        if not readme_path.exists():
+            return ""
+        with open(readme_path, "r", encoding="utf-8") as f:
+            return f.read()
+
+    def update_readme_autogen(self, slug: str, section: str, content: str) -> None:
+        """Actualiza una seccion autogenerada en README.md.
+
+        Busca <!-- AUTOGEN:{section}_START --> ... <!-- AUTOGEN:{section}_END -->
+        y reemplaza el contenido entre esos marcadores.
+        """
+        from tracker.storage.markdown_writer import MarkdownWriter
+
+        readme_path = self._readme_path(slug)
+        current_content = self.read_readme(slug)
+
+        writer = MarkdownWriter()
+        new_content = writer.format_autogen_section(current_content, section, content)
+
+        with open(readme_path, "w", encoding="utf-8") as f:
+            f.write(new_content)
+
+    def write_session_file(self, session: Session) -> None:
+        """Crea projects/<slug>/sessions/YYYY-MM-DD_HHMM.md"""
+        from tracker.storage.markdown_writer import MarkdownWriter
+
+        sessions_path = self._sessions_path(session.project_slug)
+        sessions_path.mkdir(parents=True, exist_ok=True)
+
+        started = session.started_at
+        filename = started.strftime("%Y-%m-%d_%H%M.md")
+        session_path = sessions_path / filename
+
+        writer = MarkdownWriter()
+        content = writer.format_session_file(session)
+
+        with open(session_path, "w", encoding="utf-8") as f:
+            f.write(content)
+
+    def active_session_path(self) -> Path:
+        """Returns Path to projects/.active_session.json"""
+        return self.projects_root / ".active_session.json"
+
+    def read_active_session(self) -> Optional[dict]:
+        """Lee la sesion activa desde .active_session.json"""
+        path = self.active_session_path()
+        if not path.exists():
+            return None
+        with open(path, "r", encoding="utf-8") as f:
+            return json.load(f)
+
+    def write_active_session(self, session_data: dict) -> None:
+        """Escribe la sesion activa a .active_session.json"""
+        path = self.active_session_path()
+        with open(path, "w", encoding="utf-8") as f:
+            json.dump(session_data, f, indent=2, default=str)
+
+    def delete_active_session(self) -> None:
+        """Elimina .active_session.json"""
+        path = self.active_session_path()
+        if path.exists():
+            path.unlink()
diff --git a/tracker/storage/markdown_reader.py b/tracker/storage/markdown_reader.py
new file mode 100644
index 0000000..3b96620
--- /dev/null
+++ b/tracker/storage/markdown_reader.py
@@ -0,0 +1,138 @@
+"""Markdown reader utility."""
+
+import re
+from datetime import datetime
+from typing import Optional
+
+
+class MarkdownReader:
+    """Lectura de archivos Markdown del proyecto."""
+
+    def parse_log_entry(self, content: str) -> dict:
+        """Parse una entrada de LOG.md.
+
+        Formato esperado:
+        ## 2026-03-23 10:00–11:20
+        **Objetivo**
+        ...
+        **Trabajo realizado**
+        - ...
+        **Cambios relevantes**
+        - ...
+        **Bloqueos**
+        - ...
+        **Decisiones**
+        - ...
+        **Próximos pasos**
+        - ...
+        **Resumen**
+        ...
+
+        Returns dict con:
+        - date_range: str
+        - objective: str
+        - work_done: list[str]
+        - changes: list[str]
+        - blockers: list[str]
+        - decisions: list[str]
+        - next_steps: list[str]
+        - summary: str
+        """
+        result = {
+            "date_range": "",
+            "objective": "",
+            "work_done": [],
+            "changes": [],
+            "blockers": [],
+            "decisions": [],
+            "next_steps": [],
+            "summary": "",
+        }
+
+        # Extraer fecha/rango
+        date_match = re.search(r"##\s+(\d{4}-\d{2}-\d{2}\s+\d{2}:\d{2}[–-]\d{2}:\d{2})", content)
+        if date_match:
+            result["date_range"] = date_match.group(1)
+
+        # Extraer secciones
+        sections = {
+            "objective": r"\*\*Objetivo\*\*\s*\n(.*?)(?=\n\*\*|\n##|\Z)",
+            "work_done": r"\*\*Trabajo realizado\*\*\s*\n(.*?)(?=\n\*\*|\n##|\Z)",
+            "changes": r"\*\*Cambios relevantes\*\*\s*\n(.*?)(?=\n\*\*|\n##|\Z)",
+            "blockers": r"\*\*Bloqueos\*\*\s*\n(.*?)(?=\n\*\*|\n##|\Z)",
+            "decisions": r"\*\*Decisiones\*\*\s*\n(.*?)(?=\n\*\*|\n##|\Z)",
+            "next_steps": r"\*\*Próximos pasos\*\*\s*\n(.*?)(?=\n\*\*|\n##|\Z)",
+            "summary": r"\*\*Resumen\*\*\s*\n(.*?)(?=\n##|\Z)",
+        }
+
+        for key, pattern in sections.items():
+            match = re.search(pattern, content, re.DOTALL)
+            if match:
+                text = match.group(1).strip()
+                if key in ("work_done", "changes", "blockers", "decisions", "next_steps"):
+                    # Extraer listas con bullet points
+                    items = re.findall(r"^\s*-\s+(.+)$", text, re.MULTILINE)
+                    result[key] = items
+                else:
+                    result[key] = text
+
+        return result
+
+    def extract_autogen_section(self, content: str, section: str) -> str:
+        """Extrae contenido de una seccion AUTOGEN.
+
+        Busca <!-- AUTOGEN:{section}_START --> ... <!-- AUTOGEN:{section}_END -->
+        Returns el contenido entre esos marcadores, o string vacio si no existe.
+        """
+        pattern = rf"<!--\s*AUTOGEN:{section}_START\s*-->(.*?)<!--\s*AUTOGEN:{section}_END\s*-->"
+        match = re.search(pattern, content, re.DOTALL)
+        if match:
+            return match.group(1).strip()
+        return ""
+
+    def parse_tasks(self, content: str) -> dict:
+        """Parse TASKS.md por secciones.
+
+        Secciones esperadas:
+        - Inbox
+        - Próximo
+        - En curso
+        - Bloqueado
+        - En espera
+        - Hecho
+
+        Returns dict con nombre de seccion -> lista de tareas
+        """
+        result = {}
+        current_section = None
+        current_tasks = []
+
+        lines = content.split("\n")
+
+        for line in lines:
+            # Detectar headers de seccion (## )
+            section_match = re.match(r"^##\s+(.+)$", line)
+            if section_match:
+                # Guardar seccion anterior
+                if current_section is not None:
+                    result[current_section] = current_tasks
+
+                current_section = section_match.group(1).strip()
+                current_tasks = []
+            elif current_section is not None:
+                # Parsear bullet points
+                task_match = re.match(r"^\s*-\s+\[([ x])\]\s*(.+)$", line)
+                if task_match:
+                    checked = task_match.group(1) == "x"
+                    task_text = task_match.group(2).strip()
+                    current_tasks.append({"text": task_text, "done": checked})
+                elif line.strip():
+                    # Lineas que no son bullet ni header, agregar a la ultima tarea
+                    if current_tasks:
+                        current_tasks[-1]["text"] += " " + line.strip()
+
+        # Guardar ultima seccion
+        if current_section is not None:
+            result[current_section] = current_tasks
+
+        return result
diff --git a/tracker/storage/markdown_writer.py b/tracker/storage/markdown_writer.py
new file mode 100644
index 0000000..7d1358b
--- /dev/null
+++ b/tracker/storage/markdown_writer.py
@@ -0,0 +1,241 @@
+"""Markdown writer utility."""
+
+import re
+from datetime import datetime
+from typing import Optional
+
+from tracker.models.session import Session
+
+
+class MarkdownWriter:
+    """Escritura de archivos Markdown del proyecto."""
+
+    def format_log_entry(self, session: Session, summary: str) -> str:
+        """Formatea una entrada para LOG.md.
+
+        Formato:
+        ## 2026-03-23 10:00–11:20
+        **Objetivo**
+        ...
+
+        **Trabajo realizado**
+        - ...
+
+        **Cambios relevantes**
+        - ...
+
+        **Bloqueos**
+        - ...
+
+        **Decisiones**
+        - ...
+
+        **Próximos pasos**
+        - ...
+
+        **Resumen**
+        ...
+
+        Returns string formateado.
+        """
+        started = session.started_at.strftime("%Y-%m-%d %H:%M")
+        ended = session.ended_at.strftime("%H:%M") if session.ended_at else "En Curso"
+        date_range = f"{started}–{ended}"
+
+        lines = [
+            f"## {date_range}",
+            "",
+            "**Objetivo**",
+            f"{session.objective or 'No especificado'}",
+            "",
+            "**Trabajo realizado**",
+        ]
+
+        if session.work_done:
+            for item in session.work_done:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin trabajo registrado")
+
+        lines.extend(["", "**Cambios relevantes**"])
+        if session.changes:
+            for item in session.changes:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin cambios")
+
+        lines.extend(["", "**Bloqueos**"])
+        if session.blockers:
+            for item in session.blockers:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin bloqueos")
+
+        lines.extend(["", "**Decisiones**"])
+        if session.decisions:
+            for item in session.decisions:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin decisiones")
+
+        lines.extend(["", "**Próximos pasos**"])
+        if session.next_steps:
+            for item in session.next_steps:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin pasos definidos")
+
+        lines.extend(["", "**Resumen**", summary])
+
+        return "\n".join(lines) + "\n\n"
+
+    def format_session_file(self, session: Session) -> str:
+        """Formatea archivo de sesion detalle en sessions/YYYY-MM-DD_HHMM.md.
+
+        Formato:
+        # Sesion: 2026-03-23 10:00–11:20
+
+        ## Objetivo
+        ...
+
+        ## Notas
+        ...
+
+        ## Trabajo realizado
+        ...
+
+        ## Cambios
+        ...
+
+        ## Decisiones
+        ...
+
+        ## Bloqueos
+        ...
+
+        ## Proximos pasos
+        ...
+
+        ## Referencias
+        ...
+
+        ## Duracion
+        X minutos
+        """
+        started = session.started_at.strftime("%Y-%m-%d %H:%M")
+        ended = session.ended_at.strftime("%H:%M") if session.ended_at else "En Curso"
+
+        lines = [
+            f"# Sesion: {started}–{ended}",
+            "",
+            "## Objetivo",
+            f"{session.objective or 'No especificado'}",
+            "",
+            "## Notas",
+        ]
+
+        if session.raw_notes:
+            for note in session.raw_notes:
+                note_type = note.get("type", "work")
+                note_text = note.get("text", "")
+                lines.append(f"- [{note_type}] {note_text}")
+        else:
+            lines.append("- Sin notas")
+
+        lines.extend([
+            "",
+            "## Trabajo realizado",
+        ])
+        if session.work_done:
+            for item in session.work_done:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin trabajo realizado")
+
+        lines.extend([
+            "",
+            "## Cambios",
+        ])
+        if session.changes:
+            for item in session.changes:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin cambios")
+
+        lines.extend([
+            "",
+            "## Decisiones",
+        ])
+        if session.decisions:
+            for item in session.decisions:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin decisiones")
+
+        lines.extend([
+            "",
+            "## Bloqueos",
+        ])
+        if session.blockers:
+            for item in session.blockers:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin bloqueos")
+
+        lines.extend([
+            "",
+            "## Proximos pasos",
+        ])
+        if session.next_steps:
+            for item in session.next_steps:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin pasos definidos")
+
+        lines.extend([
+            "",
+            "## Referencias",
+        ])
+        if session.references:
+            for item in session.references:
+                lines.append(f"- {item}")
+        else:
+            lines.append("- Sin referencias")
+
+        lines.extend([
+            "",
+            "## Duracion",
+            f"{session.duration_minutes} minutos",
+        ])
+
+        return "\n".join(lines) + "\n"
+
+    def format_autogen_section(self, content: str, section: str, new_content: str) -> str:
+        """Reemplaza o inserta una seccion AUTOGEN en contenido Markdown.
+
+        Busca <!-- AUTOGEN:{section}_START --> ... <!-- AUTOGEN:{section}_END -->
+        Si existe, reemplaza el contenido entre los marcadores.
+        Si no existe, inserta la seccion al final.
+
+        Returns el contenido modificado.
+        """
+        start_marker = f"<!-- AUTOGEN:{section}_START -->"
+        end_marker = f"<!-- AUTOGEN:{section}_END -->"
+
+        full_marker = f"{start_marker}\n{new_content}\n{end_marker}"
+
+        # Buscar si existe la seccion
+        pattern = rf"{re.escape(start_marker)}.*?{re.escape(end_marker)}"
+        if re.search(pattern, content, re.DOTALL):
+            # Reemplazar seccion existente
+            return re.sub(pattern, full_marker, content, flags=re.DOTALL)
+        else:
+            # Insertar al final
+            return content + "\n" + full_marker + "\n"
+
+    def format_readme_section(self, section: str, content: str) -> str:
+        """Formatea una seccion de README.md.
+
+        Para usar con format_autogen_section.
+        """
+        return content