diff --git a/.development/metadata_reader.py b/.development/metadata_reader.py
new file mode 100644
index 0000000..9b18ba6
--- /dev/null
+++ b/.development/metadata_reader.py
@@ -0,0 +1,138 @@
+""" Consumes a stream and returns a dict
+
+However, the dict won't contain "__doc__", "___version___" etc, but
+the shortened versions without underscores: "doc", "version".
+
+Currently not supported:
+* Dicts
+* Floating points
+* ints in list
+* Strings in any other format then "x" or 'y'
+* Docstrings with any other delimiter other than triple-" or triple='
+* Comments
+
+Feel free to expand if necessary
+"""
+
+class ParseException(Exception):
+    """Indicates a parsing exception"""
+    def __init__(self, message = ""):
+        super().__init__(message)
+
+def read_metadata(s):
+    result = {}
+
+    result["doc"] = _read_docstring(s)
+
+    while True:
+        key = _read_key(s)
+        if key:
+            result[key] = _read_value(s)
+        else:
+            break
+
+    return result
+
+def _read_docstring(s):
+    delimiter = _read_non_whitespace(s, 3)
+    if delimiter not in ["'''", '"""']:
+        raise ParseException("Docstring delimiter expected")
+    result = _read(s, 3);
+    while result[-3:] != delimiter:
+        result += _read(s)
+    return result[:-3]
+
+def _read_value(s):
+    char = _read_non_whitespace(s)
+    return _read_value_given_first_char(s, char)
+
+
+def _read_value_given_first_char(s, first_char):
+    if first_char in ["'", '"']:
+        return _read_string(s, first_char)
+    if first_char in "0123456789":
+        return _read_int(s, first_char)
+    if first_char in "TF":
+        return _read_bool(s, first_char)
+    if first_char == "[":
+        return _read_list(s)
+    raise ParseException("Invalid character %s found" % first_char)
+
+def _read_string(s, delimiter):
+    result = _read(s)
+    try:
+        while result[-1:] != delimiter:
+            result += _read(s)
+    except ParseException:
+        raise ParseException("Invalid string or not terminated: %s" % result)
+    return result[:-1]
+
+def _read_int(s, char):
+    result = char
+    while not char.isspace():
+        char = s.read(1)
+        if not char:
+            break
+        result += char
+        if not char in "0123456789":
+            raise ParseException("Invalid int: %s" % result)
+    return int(result)
+
+def _read_bool(s, char):
+    if char == "T":
+        _assert(char + _read(s, 3), "True", "Invalid boolean")
+        return True
+    else:
+        _assert(char + _read(s, 4), "False", "Invalid boolean")
+        return False
+
+def _read_list(s):
+    result = []
+    while True:
+        char = _read_non_whitespace(s)
+        if char == "]":
+            break
+        if result:
+            if char != ",":
+                raise ParseException("Expected comma, got '%s'" % char)
+            result.append(_read_value(s))
+        else:
+            result.append(_read_value_given_first_char(s, char))
+
+    return result
+
+def _read_key(s):
+    delimiter = _read_non_whitespace(s, 3)
+    if delimiter != "___":
+        return None
+    try:
+        result = _read(s, 3);
+        while result[-3:] != delimiter:
+            char = _read(s)
+            if char in [" ", "="]:
+                raise ParseException()
+            result += char
+    except ParseException:
+        raise ParseException("Invalid key: ___%s" % result)
+    _assert(_read_non_whitespace(s), "=", "Expected equals")
+    return result[:-3]
+
+def _read(s, l=1):
+    result = s.read(l)
+    if len(result)<l:
+        raise ParseException("Expected to read at least %s characters, got '%s'" % (l, result))
+    return result
+
+def _assert(input, expected, message):
+    if not input == expected:
+        raise ParseException(message + " ('%s' expected, '%s' found)" % (expected, input))
+
+def _read_non_whitespace(s, l=1):
+    result = s.read(1)
+    while result.isspace():
+        result = s.read(1)
+    if l == 1:
+        return result
+    else:
+        return result + s.read(l - 1)
+
diff --git a/.development/resources.py b/.development/resources.py
new file mode 100644
index 0000000..b8554db
--- /dev/null
+++ b/.development/resources.py
@@ -0,0 +1,208 @@
+import os, hashlib, binascii
+from metadata_reader import read_metadata, ParseException
+
+"""
+Resources are the basic components of any given library.
+
+supported the following types:
+* lib (files in lib/)
+* shared (file in shared/)
+* app (all other folders in root)
+* root (files in root)
+
+A resource has the following form:
+{
+    "app1": {"type": "app", dependencies: ["lib/lib1.py"], "files": {"app1/main.py": "abcdef1234", "app1/nested/text.txt": "abcdef1234"}},
+    "app2": {"type": "app", dependencies: ["lib/lib2.py"], "files": {"app2/main.py": "abcdef1234", "app2/other_content.txt": "abcdef1234", "app2/some_binary_file.gif": "abcdef1234"}},
+    "lib/lib1.py": {"type": "lib", , dependencies: ["lib/lib3.py"], "files": {"lib/lib1.py": "abcdef1234"}},
+    "lib/lib2.py": {"type": "lib", "files": {"lib/lib2.py": "abcdef1234"}},
+    "lib/lib3.py": {"type": "lib", "files": {"lib/lib3.py": "abcdef1234"}},
+    "lib/lib4.py": {"type": "lib", "files": {"lib/lib4.py": "abcdef1234"}},
+    "lib/lib5.py": {"type": "lib", "files": {"lib/lib5.py": "abcdef1234"}},
+    "shared/foo.txt": {"type": "shared", "files": {"shared/foo.txt": "abcdef1234"}},
+    "boot.py": {"type": "root", "files": {"boot.py": "abcdef1234"}}
+}
+
+Every resource can also contain other metadata fields which are extracted from the body
+of its main python class (in case of lib or app).
+
+This module has the following operations:
+resources = get_resources(path) # Gets resources for a given path
+add_hashes(path, resources)     # Adds hashes to the file dict - not needed for testing
+add_metadata(path, resources)   # Adds metadata
+resolve_dependencies(resources) # Merges all dependencies into each resource's file dict
+validate(resources)             # Runs basic validation
+
+This module encapsulates all the main operations the app library is expect to
+perform on a given checkout. It's intentionally kept in one file to make it easier
+to share between repositories. The only exception to this rule it metadata_reader
+(because it's rather complex and I didn't want to make this file impossible to read)
+
+Please make sure this file can be executes on any operating system running python3.
+Don't include any external dependencies. It forms part of the local toolchain.
+"""
+
+"""
+scan(path)
+
+A resource scanner for the Tilda filesystem. Returns a {path: {type:<type>, files:{...}}}
+
+ignored are the following:
+* dotfiles
+* __pycache__
+"""
+
+def _scan_files(path, rel_path = ""):
+    result = []
+    for element in os.listdir(path):
+        if element.startswith(".") or element == "__pycache__":
+            continue
+        element_path = os.path.join(path, element)
+        element_rel_path = os.path.join(rel_path, element)
+        if os.path.isdir(element_path):
+            result.extend(_scan_files(element_path, element_rel_path))
+        else:
+            result.append(element_rel_path)
+
+    return result
+
+def get_resources(path):
+    result = {}
+    for sub_path in os.listdir(path):
+        if sub_path.startswith(".") or sub_path == "__pycache__":
+            continue
+        full_path = os.path.join(path, sub_path)
+        if os.path.isfile(full_path):
+            result[sub_path] = {"type": "root", "files": {sub_path: None}}
+            continue
+        files = _scan_files(full_path, sub_path)
+        if sub_path in ["lib", "shared"]:
+            for rel_path in files:
+                result[rel_path] = {"type": sub_path, "files": {rel_path: None}}
+        else:
+            result[sub_path] = {"type": "app", "files": {}}
+            for rel_path in files:
+                result[sub_path]["files"][rel_path] = None
+    return result
+
+"""
+add_hashes(path, resource)
+
+Adds the first 10 characters of SHA256 hashes to all elements in "files".
+The hash is calcuated on the file content, not the file name.
+"""
+
+def add_hashes(path, resources):
+    for resource in resources.values():
+        for file_path in resource["files"]:
+            resource["files"][file_path] = _hash_file(os.path.join(path, file_path))
+
+def _hash_file(filename):
+    """Calculates the SHA256 hash of a file."""
+    with open(filename, "rb") as file:
+        sha256 = hashlib.sha256()
+        buf = file.read(128)
+        while len(buf) > 0:
+            sha256.update(buf)
+            buf = file.read(128)
+        return str(binascii.hexlify(sha256.digest()), "utf8")[:10]
+
+"""
+add_metadata(path, resource)
+
+Reads primary files for app and lib resources and extracts metadata information from its header
+"""
+
+def add_metadata(path, resources):
+    for resource in resources.values():
+        file = None
+        if resource['type'] == "app":
+            file = next(f for f in resource['files'] if "/main.py" in f)
+        elif resource['type'] == "lib":
+            file = next(iter(resource['files'].keys()))
+
+        if file:
+            try:
+                with open(os.path.join(path, file), "r") as stream:
+                    resource.update(_normalize_metadata(read_metadata(stream)))
+            except ParseException as e:
+                resource.setdefault("errors", []).append(file + ": " + str(e))
+
+def _normalize_metadata(metadata):
+    metadata['description'] = metadata.pop('doc')
+    if 'dependencies' in metadata:
+        metadata['dependencies'] = [_normalize_lib(l) for l in metadata.pop('dependencies')]
+
+    return metadata
+
+def _normalize_lib(lib):
+    """lib dependencies can be shortened to just their module name"""
+    if "." in lib or "/" in lib:
+        return lib
+    return "lib/%s.py" % lib
+
+"""
+resolve_dependencies(resources)
+
+merges files from dependent resources into the original files dict
+"""
+
+def resolve_dependencies(resources):
+    for file, resource in resources.items():
+        if 'dependencies' in resource:
+            already_added = [file]
+            to_add = resource['dependencies'].copy()
+            while len(to_add):
+                r = to_add.pop()
+                if r in already_added:
+                    continue
+                if r not in resources:
+                    resource.setdefault("errors", []).append("Dependency %s not found" % r)
+                    continue
+                already_added.append(r)
+                to_add.extend(resources[r].get("dependencies", []))
+                resource['files'].update(resources[r]['files'])
+
+
+"""
+validate(path, resources)
+
+does basic verification:
+* Is it valid python?
+* Are metadata fields missing
+* TBD: Does it have imports that are not dependencies?
+"""
+def validate(path, resources):
+    for resource in resources.values():
+        _validate_resource(path, resource)
+
+def _validate_resource(path, resource):
+    # Compile
+    for file in resource['files'].keys():
+        if file.endswith(".py"):
+            try:
+                filename = os.path.join(path, file)
+                with open(filename, 'r') as s:
+                    compile(s.read() + '\n', filename, 'exec')
+            except Exception as e:
+                resource.setdefault("errors", []).append(str(e))
+
+    # Metadata check
+    if resource['type'] in ["app", "lib"]:
+        pass #todo: what exactly are we're making required?
+
+
+"""
+helpers
+"""
+
+def get_error_summary(resources):
+    summary = ""
+    for key, resource in resources.items():
+        if "errors" in resource:
+            summary += "--- %s ---\n" % key
+            for error in resource['errors']:
+                summary += error + "\n"
+            summary += "\n"
+    return summary.strip()
+
diff --git a/.development/tilda_tools.py b/.development/tilda_tools.py
index a7c7966..0cf56a1 100755
--- a/.development/tilda_tools.py
+++ b/.development/tilda_tools.py
@@ -2,7 +2,7 @@
 
 """Toolchain for working with the TiLDA Mk4
 
-Usage (currently more of a wishlist)
+Usage
 ------------------------------------
 
 Reboot badge
@@ -27,7 +27,7 @@ $ tilda_toold.py sync --run some_other_file.py
 Executes a single file on the badge without copying anything (Using pyboard.py)
 $ tilda_tools run my_app/main.py
 
-Runs local validation against metadata (doesn't require a badge)
+Runs local validation (doesn't require a badge, but doesn't run unit tests)
 $ tilda_tools validate
 
 Runs local validation and badge-side tests
@@ -46,11 +46,12 @@ Common parameters
 
 import sys, glob
 import sync, pyboard_util
+from resources import *
 
 def main():
     import argparse
     cmd_parser = argparse.ArgumentParser(description='Toolchain for working with the TiLDA Mk4')
-    cmd_parser.add_argument('command', nargs=1, help='command')
+    cmd_parser.add_argument('command', nargs=1, help='command [test|reset|sync|run]')
     cmd_parser.add_argument('-d', '--device', help='the serial device of the badge')
     cmd_parser.add_argument('-s', '--storage', help='the usb mass storage path of the badge')
     cmd_parser.add_argument('-b', '--baudrate', default=115200, help='the baud rate of the serial device')
@@ -60,6 +61,23 @@ def main():
     cmd_parser.add_argument('paths', nargs='*', help='input files')
     args = cmd_parser.parse_args()
     command = args.command[0]
+    path = sync.get_root()
+
+    if command in ["test", "validate"]:
+        resources = get_resources(path)
+        add_metadata(path, resources)
+        resolve_dependencies(resources)
+        validate(path, resources)
+        errors = get_error_summary(resources)
+        if errors:
+            print("Problems found:\n")
+            print(errors)
+            sys.exit(1)
+        print("Local Test: PASS")
+        if command == "test":
+            command = "sync"
+            args.path = []
+            args.run = "test/main.py"
 
     if command in ["reset", "sync"]:
         pyboard_util.stop_badge(args)
diff --git a/.travis.yml b/.travis.yml
new file mode 100644
index 0000000..d733391
--- /dev/null
+++ b/.travis.yml
@@ -0,0 +1,2 @@
+script:
+    ./tilda_tools validate
diff --git a/lib/apps.py b/lib/apps.py
index 5d96408..c6d2b73 100644
--- a/lib/apps.py
+++ b/lib/apps.py
@@ -176,4 +176,4 @@ def empty_local_app_cache():
     global _public_apps_cache, _category_cache
     _public_apps_cache = None
     _category_cache = None
-        gc.collect()
+    gc.collect()
diff --git a/lib/unittest.py b/lib/unittest.py
index d755175..18b9698 100644
--- a/lib/unittest.py
+++ b/lib/unittest.py
@@ -1,3 +1,7 @@
+"""Base libarary for test cases"""
+
+___license___ = "MIT"
+
 import sys
 
 class SkipTest(Exception):