Pyscript can be configured using the UI, or via yaml. To use the UI, go to the Configuration -> Integrations page and select “+” to add Pyscript Python scripting. After that, you can change the settings anytime by selecting “Options” under Pyscript in the Configuration -> Integrations page. You will need to select “reload” under Pyscript, or call the pyscript.reload service for the new settings to take effect.

Alternatively, for yaml configuration, add pyscript: to <config>/configuration.yaml. You can’t mix these two methods - your initial choice determines how you should update these settings later. If you want to switch configuration methods you will need to uninstall and reinstall pyscript.

Pyscript has two optional configuration parameters that allow any python package to be imported and exposes the hass variable as a global (both options default to false). Assuming you didn’t use the UI to configure pyscript, these can be set in <config>/configuration.yaml:

  allow_all_imports: true
  hass_is_global: true

It is recommended you put your pyscript configuration its own yaml file in the pyscript folder. That way changes to that file will be automatically detected and will trigger a reload, which includes rereading the configuration parameters:

pyscript: !include pyscript/config.yaml

The settings and behavior of your code can be controlled by additional user-defined yaml configuration settings. If you configured pyscript using the UI flow, you can still add additional configuration settings via yaml. Since they are free-form (no fixed schema) there is no UI configuration available for these additional settings.

Each application should have its configuration stored using the application’s name under an apps entry. For example, applications my_app1 and my_app2 would be configured as:

  allow_all_imports: true
      # any settings for my_app1 go here
      # any settings for my_app2 go here

If the include above is used, the pyscript/config.yaml would look like:

allow_all_imports: true
    # any settings for my_app1 go here
    # any settings for my_app2 go here

The apps section could also include other configuration files if you prefer to put each application’s configuration in its own file.

An application will not be loaded unless it has a configuration entry (even if it is empty). That provides an easy was to enable or disable an application.

An application’s configuration (eg, all the settings below my_app1 in the example above) are available in the variable pyscript.app_config in the global scope of the application’s main file (eg, apps/ or apps/my_app1/ Additionally, all the pyscript configuration settings are available via the variable pyscript.config, which also includes all the application configuration below the apps key. However, in a future release the apps entry will be removed so that apps do not have access to another app’s configuration (which might include token, passwords or keys). See this section for more information.

Note that if you used the UI flow to configure pyscript, the allow_all_imports and hass_is_global configuration settings will be ignored in the yaml file. In that case you should omit them from the yaml, and just use yaml for pyscript app configuration.

At startup, pyscript loads the following files. It also automatically unloads and reloads files when they are changed, renamed, created or deleted:

all files with a .py suffix in the top-level directory are autoloaded.
all files with a .py suffix below the scripts directory, recursively to any depth, are autoloaded. This is useful for organizing your scripts into subdirectories that are related in some way.
every file in a subdirectory in the apps subdirectory is autoloaded, provided app_name exists in the pyscript yaml configuration under apps. This package form is most convenient for sharing pyscript code, since all the files for one application are stored in their own directory.
all files in the apps subdirectory with a .py suffix are autoloaded, unless the package form above was loaded instead, and provided app_name exists in the pyscript yaml configuration under apps (that allows each app to be disabled by simply removing its configuration and reloading).

The idea is that the top-level directory pyscript and pyscript/scripts are for your own use (and are always loaded), and pyscript/apps/ is for things you’ve gotten from others or things you’ve written with the intent to reuse or share (and only are loaded if there is a configuration entry, which can be empty).

Any file name that starts with # is not loaded, and similarly scripts anywhere below a directory name that starts with #, are not loaded. That’s a convenient way to disable a specific script or entire directory below scripts - you can simply rename it with or without the leading # to disable or enable it, which automatically triggers a reload. Think of it as “commenting” the file name, rather than having to delete it or move it outside the pyscript directory.

Like regular Python, functions within one source file can call each other, and can share global variables (if necessary), but just within that one file. Each file has its own separate global context. Each Jupyter session also has its own separate global context, so functions, triggers, variables and services defined in each interactive session are isolated from the script files and other Jupyter sessions. Pyscript provides some utility functions to switch global contexts, which allows an interactive Jupyter session to interact directly with functions and global variables created by a script file, or even another Jupyter session.

The optional <config>/pyscript/modules subdirectory can contain modules (files with a .py extension) or packages (directories that contain at least a file) that can be imported by any other pyscript files, applications or modules. The module form is ignored if the package form is present. They are not autoloaded. Any modules or packages in <config>/pyscript/modules that are modified will be unloaded when they are modified, and any scripts or apps that depend on those modules will be reloaded. Importing modules and packages from <config>/pyscript/modules are not restricted if allow_all_imports is False. Typically common functions or features would be implemented in a module or package, and then imported and used by scripts in <config>/pyscript or applications in <config>/pyscript/apps.

Even if you can’t directly call one function from another script file, HASS state variables are global and services can be called from any script file.

Reloading Scripts

Manually reloading scripts via the pyscript.reload service is no longer necessary starting in 1.2.0, since changes are automatically detected and a reload is triggered.

By default, calling the reload service only reloads scripts, apps or modules that have changed since they were last loaded. A change means the script, app or module file contents or modification time has changed. Additionally, an app is considered changed if its yaml configuration has changed. When a change is detected, additional files are also reloaded if they depend on a changed file. Additional files include all other files in a particular module or app (meaning the entire module or app is reloaded if any of its files or imports are changed), and if a module has changed, any other module, app or script that imports that module directly or indirectly is reloaded too.

A file is also considered changed if it is newly created or deleted (or “commented” by renaming it or a parent directory to start with #). When reload detects a deleted file, the prior global context and all its triggers are deleted, just as when a file has changed.

If you want to reload a specific script or file, you can touch it to change its modification time. Alternatively, the pyscript.reload service takes an optional parameter global_ctx which specifies the name of a specific global context to reload. That is the one file considered to be changed by reload, and other changes are ignored. For example, specifying file.example will just reload, and apps.my_app1 will just reload apps/ See the Global Context section for the mapping of file or app names to global context names. Additional files might still be reloaded too (all other files in the module or app, and any modules, apps or scripts that import a module if global_ctx was set to a module).

If you want to manually reload all scripts, apps and modules, set global_ctx to * when you call pyscript.reload. Alternatively, the Reload option under pyscript in the in the Configuration -> Integrations page will always do a full reload.

Any function definitions, services and triggers in the reloaded files are re-created. Jupyter sessions are not affected. Any currently running functions (ie, functions that have been triggered and are actively executing Python code or waiting inside task.sleep() or task.wait_until()) are not stopped on reload - they continue to run until they finish (return). You can terminate these running functions too on reload if you wish by calling task.unique() in the script file preamble (ie, outside any function definition), so it is executed on load and reload, which will terminate any running functions that have previously called task.unique() with the same argument.

State Variables

These are typically called entities or entity_id in HASS.

State variables can be accessed in any Python code simply by name. State variables (also called entity_id) are of the form, where DOMAIN is typically the name of the component that sets that variable. You can set a state variable by assigning to it.

State variables only have string values, so you will need to convert them to int or float if you need their numeric value.

State variables have attributes that can be accessed by adding the name of the attribute, as in The attribute names and their meaning depend on the component that sets them, so you will need to look at the State tab in the Developer Tools to see the available attributes. You can set an attribute directly by assigning = value.

In cases where you need to compute the name of the state variable dynamically, or you need to set or get all the state attributes, you can use the built-in functions state.get(), state.getattr(), state.set() and state.setattr(); see State Functions.

The function state.names(domain=None) returns a list of all state variable names (ie, entity_ids) of a domain. If domain is not specified, it returns all HASS state variable (entity) names.

Also, service names (which are called as functions) take priority over state variable names, so if a component has a state variable name that collides with one of its services, you’ll need to use state.get(name) to access that state variable.

Accessing state variables that don’t exist will throw a NameError exception, and accessing an attribute that doesn’t exist will throw a AttributeError exception. One exception (!) to this is that in a @state_trigger expression, undefined state variables and attributes will evaluate to None instead of throwing an exception.

You can assign a state variable (or the return value of state.get()) to a normal Python variable, and that variable will capture the value and attributes of the state variable at the time of the assignment. So, for example, after this assignment:

test1_state = binary_sensor.test1

the variable test1_state captures both the value and attributes of binary_sensor.test1. Later, if binary_sensor.test1 changes, test1_state continues to represent the previous value and attributes at the time of the assignment.

State variables also support virtual methods that are service calls with that entity_id. For any state variable DOMAIN.ENTITY, any services registered by DOMAIN, eg: DOMAIN.SERVICE, that have an entity_id parameter can be called as a method DOMAIN.ENTITY.SERVICE(). Each of these statements are equivalent:"DOMAIN", "SERVICE", entity_id="DOMAIN.ENTITY", other_param=123)
DOMAIN.SERVICE(entity_id="DOMAIN.ENTITY", other_param=123)

In the case the service has only one other parameter in addition to entity_id, a further shorthand is that the method can be called with just a positional, rather than keyword, argument. So if the service only takes two parameters, entity_id and other_param, this additional form is equivalent to each of the above statements:


Here’s an example using input_number, assuming it has been configured to create an entity input_number.test. These statements all do the same thing (and the last one works because set_value only takes one other parameter):"input_number", "set_value", entity_id="input_number.test", value=13)
input_number.set_value(entity_id="input_number.test", value=13)

Two additional virtual attribute values are available when you use a variable directly as DOMAIN.entity.attr or call state.get("DOMAIN.entity.attr"):

  • last_changed is the last UTC time the state value was changed (not the attributes)
  • last_updated is the last UTC time the state entity was updated

If you need to compute how many seconds ago the binary_sensor.test1 state changed, you could do this:

from datetime import datetime as dt
from datetime import timezone as timezone

num_seconds_ago = ( - binary_sensor.test1.last_changed).total_seconds()

Note that these virtual attributes and methods take precedence over any entity attributes that have the same name. If an entity has attributes with those names and you need to access them, use state.getattr(name).

Calling services

Any service can be called by using the service name as a function, with keyword parameters to specify the service parameters. You’ll need to look up the service in the Service tab of Developer Tools to find the exact name and parameters. For example, inside any function you can call:

myservice.flash_light(light_name="front", light_color="red")

which calls the myservice.flash_light service with the indicated parameters. Obviously those parameter values could be any Python expression, and this call could be inside a loop, an if statement or any other Python code.

Starting in HASS 2023.7, services can return a response, which is a dictionary of values. You need to set the keyword parameter return_response=True to get the response. For example:

service_response = myservice.flash_light(light_name="front", light_color="red", return_response=True)

The function, name, **kwargs) can also be used to call a service when you need to compute the domain or service name dynamically. For example, the above service could also be called by:"myservice", "flash_light", light_name="front", light_color="red")

When making a service call, either using the function or the service name as the function, you can optionally pass the keyword argument blocking=True if you would like to wait for the service to finish execution before continuing execution in your function. If the service returns a response (new feature starting in HASS 2023.7), it will be returned by the call if you set the optional keyword parameter return_response=True.

Firing events

Any event can be triggered by calling, **kwargs). It takes the event_type as a first argument, and any keyword parameters as the event parameters. The event_type could be a user-defined string, or it could be one of the built-in events. You can access the names of those built-in events by importing from homeassistant.const, eg:

from homeassistant.const import EVENT_CALL_SERVICE

Function Trigger Decorators

There are four decorators for defining state, time, event and MQTT triggers, and two decorators for defining whether any trigger actually causes the function to run (i.e., is active), based on state-based expressions or one or more time-windows. The decorators should appear immediately before the function they refer to. A single function can have any or all of the decorator types specified. Multiple trigger decorators of the same type can be added to a single function, but only one @state_active, @time_active or @task_unique can be used per function.

A Python function with decorators is still a normal Python function that can be called by any other Python function. The decorators have no effect in the case where you call it directly from another function.


@state_trigger(str_expr, ..., state_hold=None, state_hold_false=None, state_check_now=False, kwargs=None, watch=None)

@state_trigger takes one or more string arguments that contain any expression based on one or more state variables, and evaluates to True or False (or non-zero or zero). Whenever any of the state variables or attribute values mentioned in the expression change (or specified via the watch argument), the expression is evaluated and the trigger occurs if it evaluates to True (or non-zero). For each state variable, eg:, the prior value is also available to the expression as in case you want to condition the trigger on the prior value too. Attribute values can be used in the expression too, using the forms and for the new and old attribute values respectively.

Multiple str_expr arguments are logically “or”ed together into a single expression. Any argument can alternatively be a list or set of strings, and they are treated the same as multiple arguments by “or”ing them together. Alternatively, independent state triggers can be specified by using multiple @state_trigger decorators.

Optional arguments are:

If set, the @state_trigger expression is evaluated immediately when the trigger function is defined (typically at startup), and the trigger occurs if the expression evaluates to True or non-zero. Normally the expression is only evaluated when a state variable changes, and not when the trigger function is first defined. This option is the same as in the task.wait_until function, except the default value is True in that case. Note that if you specify state_check_now=True, entries in state_trigger that are plain state variable names (which mean trigger on any change) are ignored during the initial check - only expressions are checked.
A numeric duration in seconds that delays executing the trigger function for this amount of time. If the state trigger expression changes back to False during that time, the trigger is canceled and a wait for a new trigger begins. If the state trigger expression changes, but is still True then the state_hold time is not restarted - the trigger will still occur that number of seconds after the first state trigger.

If set, the state trigger edge is triggered (triggers on a False to True transition), versus the default of level triggered (triggers when True). The @state_trigger expression must evaluate to False for this duration in seconds before the next trigger can occur. A value of 0 requires the expression be False before a trigger, but with no minimum time in that state. If the expression evaluates to True during the state_hold_false period, that trigger is ignored, and when the expression next is False the state_hold_false period starts over.

For example, by default the expression "int(sensor.temp_outside) >= 50" will trigger every time sensor.temp_outside changes to a value that is 50 or more. If instead state_hold_false=0, the trigger will only occur when sensor.temp_outside changes the first time to 50 or more. It has to go back below 50 for state_hold_false seconds before a new trigger can occur.

When state_hold_false is set, the state trigger expression is evaluated at startup. If False the state_hold_false period begins. If True, a wait for the next False value begins. If state_check_now is also set, the trigger will also occur at startup if the expression is True at startup, while the state_hold_false logic will continue to wait until the expression is False for that period before the next future trigger.

Additional keyword arguments can be passed to the trigger function by setting kwargs to a dict of additional keywords and values. These will override the standard keyword arguments such as value or var_name if you include those keywords in kwargs. A typical use for kwargs is if you have multiple @state_trigger decorators on a single trigger function and you want to pass additional parameters (based on the trigger, such as a setting or action) to the function. That could save several lines of code in the function determining which trigger occurred.
Specifies a list or set of string state variable names (entity names) or state attributes that are monitored for this trigger. When (and only when) a variable in this set changes, the trigger expression is evaluated. Normally this set of names is automatically extracted from the @state_trigger expression, but there could be cases where it doesn’t capture all the names (eg, if you have to use state.get() inside the trigger expression). You could also use watch to specify a subset of the names in the trigger expression, which has the effect of rendering those other variables as only conditions in the trigger expression that won’t cause a trigger themselves, since the expression won’t be evaluated when they change.

Here’s a summary of the trigger behavior with these parameter settings:

state_check_now state_hold_false trigger at start? trigger on state change?
default (False) default (None) no if expr True
default (False) 0 no if expr True, only after False
default (False) 10 no if expr True, only after False for ≥ 10 sec
True default (None) if expr True if expr True
True 0 if expr True if expr True, only after False
True 10 if expr True if expr True, only after False for ≥ 10 sec

If state_hold is also specified, all the entries in the table that say “if expr True” must remain True for that number of seconds for the trigger to occur.

All state variables in HASS have string values. So you’ll have to do comparisons against string values or cast the variable to an integer or float. These two examples are essentially equivalent (note the use of single quotes inside the outer double quotes):

@state_trigger("domain.light_level == '255' or domain.light2_level == '0'")
@state_trigger("int(domain.light_level) == 255 or int(domain.light2_level) == 0")

although the second will throw an exception if the variable string doesn’t represent a valid integer. If you want numerical inequalities you should use the second form, since string lexicographic ordering is not the same as numeric ordering.

You can also use state variable attributes in the trigger expression, with an idenfitier of the form Attributes maintain their original type, so there is no need to cast then to another type.

You can specify a state trigger on any change with a string that can take three forms:

  • "domain.entity": triggers on any change to the state variable value
  • "domain.entity.attr": triggers on any change to the state variable attribute attr value
  • "domain.entity.*": triggers on any change to any state variable attribute (but not its value)

For example:


will trigger any time the value of domain.light_level changes (not its attributes), which includes the cases when that variable is first created (ie, the old_value is None) and when it is deleted (ie, the value is None).

If you use the “any change” form, there’s no point in also specifying state_hold since the expression is always True whenever the state variable changes - there is no way for it to evaluate to False and to re-start the trigger process. If you do specify state_hold in this case it will simply delay the trigger by the specified time.

The trigger can include arguments with any mixture of string expressions (that are evaluated when any of the underlying state variables change) and string state variable or attribute names (that trigger whenever that variable or attribute changes).

Note that if a state variable and attributes are set to the same value, HASS doesn’t generate a state change event, so the @state_trigger condition will not be checked. It is only evaluated each time a state variable or any of its attributes change to a new value.

When the trigger occurs and the function is executed (meaning any active checks passed too), keyword arguments are passed to the function so it can tell which state variable caused it to succeed and run, in cases where the trigger condition involves multiple variables. These are:

kwargs = {
    "trigger_type": "state",
    "var_name": var_name,
    "value": new_value,
    "old_value": old_value

The value and old_value represent the current and old values of the state variable var_name whose change caused the trigger. Those variables include the state attributes too. If the trigger occurs when the state variable is newly created, old_value will be None, and if the trigger occurs when a state variable is deleted, value will be None.

Additional keyword parameters can be passed to the trigger function by setting the optional kwargs parameter to a dict with the keyword/value pairs.

If your function needs to know any of these values, you can list the keyword arguments you need, with defaults:

@state_trigger("domain.light_level == '255' or domain.light2_level == '0'")
def light_turned_on(trigger_type=None, var_name=None, value=None):

You don’t have to list all the default keyword parameters - just the ones your function needs. In contrast, if you specify additional keyword parameters via kwargs, you will get an excepton if the function doesn’t have matching keyword arguments (unless you use the **kwargs catch-all in the function definition).

Using trigger_type is helpful if you have multiple trigger decorators. The function can now tell which type of trigger, and which of the two variables changed to cause the trigger. You can also use the keyword catch-all declaration instead:

@state_trigger("domain.light_level == '255' or domain.light2_level == '0'")
def light_turned_on(**kwargs)"got arguments {kwargs}")

and all those values (including optional ones you specify with the kwargs argument to @state_trigger) will simply get passed via kwargs as a dict. That’s the most useful form to use if you have multiple decorators, since each one passes different variables into the function (although all of them set trigger_type).

If state_check_now is set to True and the trigger occurs during its immediate check, since there is no underlying state variable change, the trigger function is called with only this argument:

kwargs = {
    "trigger_type": "state",

If the trigger function uses var_name==None as a keyword argument, it can check if it is None to determine whether it was called immediately or not. Similarly, if it uses the kwargs form, if can check if var_name is in kwargs.

If state_hold is specified, the arguments to the trigger function reflect the variable change that cause the first trigger, not any subsequent ones during the state_hold period. Also, if the @time_active or @state_active decorators are used, they will be evaluated after the state_hold period, but with the initial trigger variable value (ie, the value that caused the initial trigger).

Inside str_expr, undefined state variables, undefined state attributes, and undefined .old variables evaluate to None, rather than throwing an exception. The .old variable will be None the first time the state variable is set (since it has no prior value), and when the str_expr is being evaluated because a different state variable changed (only the state variable change that caused str_expr to be evaluated gets its prior value in .old; any other .old variables will be None for that evaluation).


@time_trigger(time_spec, ..., kwargs=None)

@time_trigger takes one or more string specifications that specify time-based triggers. When multiple time triggers are specified, each are evaluated, and the earliest one is the next trigger. Then the process repeats. Alternatively, multiple time trigger specifications can be specified by using multiple @time_trigger decorators, although that is less efficient than passing multiple arguments to a single one.

Additional keyword parameters can be passed to the trigger function by setting the optional kwargs parameter to a dict with the keywords and values.

Several of the time specifications use a datetime format, which is ISO: yyyy/mm/dd hh:mm:ss, with the following features:

  • There is no time-zone (local is assumed).

  • The date is optional, and the year can be omitted with just mm/dd.

  • The date can also be replaced by a day of the week (either full like sunday or 3-letters like sun, in your local language based on the locale; however, on Windows and other platforms that lack locale.nl_langinfo, the days of week default to English).

  • The meaning of partial or missing dates depends on the trigger, as explained below.

  • The date and time can be replaced with now, which means the current date and time when the trigger was first evaluated (eg, at startup or when created as an inner function or closure), and remains fixed for the lifetime of the trigger.

  • The time can instead be sunrise, sunset, noon or midnight.

  • If the time is missing, midnight is assumed (so thursday is the same as thursday 00:00:00)

  • Seconds are optional, and can include a decimal (fractional) portion if you need finer resolution.

  • The datetime can be followed by an optional offset of the form [+-]number{seconds|minutes|hours|days|weeks} with abbreviations:

    • {s|sec|second|seconds} or empty for seconds,
    • {m|min|mins|minute|minutes} for minutes,
    • {h|hr|hour|hours} for hours,
    • {d|day|days} for days,
    • {w|week|weeks} for weeks.

    That allows things like sunrise + 30m to mean 30 minutes after sunrise, or sunday sunset - 1.5 hour to mean 1.5 hours before sunset on Sundays. The number can be floating point. (Note, there is no i18n support for those offset abbreviations - they are in English.)

In @time_trigger, each string specification time_spec can take one of four forms:

  • "startup" triggers on HASS start and reload (ie, on function definition), and is equivalent to "once(now)"

  • "shutdown" triggers on HASS shutdown and reload (ie, when the trigger function is no longer referenced)

  • "once(datetime)" triggers once on the date and time. If the year is omitted, it triggers once per year on the date and time (eg, birthday). If the date is just a day of week, it triggers once on that day of the week. If the date is omitted, it triggers once each day at the indicated time. once(now + 5 min) means trigger once 5 minutes after startup.

  • "period(datetime_start, interval, datetime_end)" or "period(datetime_start, interval)" triggers every interval starting at the starting datetime and finishing at the optional ending datetime. When there is no ending datetime, the periodic trigger runs forever. The interval has the form number{sec|min|hours|days|weeks} (the same as datetime offset without the leading sign), and the same abbreviations can be used. Period start and ends can also be based on now, for example period(now + 10m, 5min, now + 30min) will cause five triggers at 10, 15, 20, 25 and 30 minutes after startup.

  • "cron(min hr dom mon dow)" triggers according to Linux-style crontab. Each of the five entries are separated by spaces and correspond to minutes, hours, day-of-month, month, day-of-week (0 = sunday):

    field allowed values
    minute 0-59
    hour 0-23
    day of month 1-31
    month 1-12
    day of week 0-6 (0 is Sun)

    Each field can be a * (which means “all”), a single number, a range or comma-separated list of numbers or ranges (no spaces). Ranges are inclusive. For example, if you specify hours as 6,10-13 that means hours of 6,10,11,12,13. The trigger happens on the next minute, hour, day that matches the specification. See any Linux documentation for examples and more details (note: names for days of week and months are not supported; only their integer values are). The cron features use the croniter package, so check its documentation <> for additional specification formats that are supported (eg: */5 repeats every 5th unit, days of week can be specified with English abbreviations, and an optional 6th field allows seconds to be specified).

When the @time_trigger occurs and the function is called, the keyword argument trigger_type is set to "time", and trigger_time is the exact datetime of the time specification that caused the trigger (it will be slightly before the current time), or startup or shutdown in the case of a startup or shutdown trigger.

Additional optional keyword parameters can be specified in the kwargs parameter to @time_trigger.

A final special form of @time_trigger has no arguments, which causes the function to run once automatically on startup or reload, which is the same as providing a single "startup" time specification:

def run_on_startup_or_reload():
    """This function runs automatically once on startup or reload"""

The function is not re-started after it returns, unless a reload occurs. Startup occurs when the EVENT_HOMEASSISTANT_STARTED event is fired, which is after everything else is initialized and ready, so this function can call any services etc. A startup trigger can also occur after HASS is started when a new trigger function is defined (eg, on reload, defined at run-time using inner function/closure, or interactively in Jupyter).

Similarly, the shutdown trigger occurs when the EVENT_HOMEASSISTANT_STOP event is fired, meaning HASS is shutting down. It also occurs whenever a trigger function is no longer referenced (meaning it’s being deleted), which happens during reload or redefined interactively in Jupyter. It’s important that any trigger function based on shutdown runs as quickly as possible, since the shutdown of HASS (or reload completion) will be stalled until your function completes.


@event_trigger(event_type, str_expr=None, kwargs=None)

@event_trigger triggers on the given event_type. Multiple @event_trigger decorators can be applied to a single function if you want to trigger the same function with different event types.

An optional str_expr can be used to match the event data, and the trigger will only occur if that expression evaluates to True or non-zero. This expression has available all the event parameters sent with the event, together with these two variables:

  • trigger_type is set to “event”
  • event_type is the string event type, which will be the same as the first argument to @event_trigger

Note unlike state variables, the event data values are not forced to be strings, so typically that data has its native type.

When the @event_trigger occurs, those same variables are passed as keyword arguments to the function in case it needs them. Additional keyword parameters can be specified by setting the optional kwargs argument to a dict with the keywords and values.

The event_type could be a user-defined string, or it could be one of the built-in events. You can access the names of those events by importing from homeassistant.const, eg:

from homeassistant.const import EVENT_CALL_SERVICE

To figure out what parameters are sent with an event and what objects (eg: list, dict) are used to represent them, you can look at the HASS source code, or initially use the **kwargs argument to capture all the parameters and log them. For example, you might want to trigger on certain service calls (not ones directed to pyscript), but you are unsure which one and what parameters it has. So initially you trigger on all service calls just to see them:

from homeassistant.const import EVENT_CALL_SERVICE

def monitor_service_calls(**kwargs):"got EVENT_CALL_SERVICE with kwargs={kwargs}")

After running that, you see that you are interested in the service call lights.turn_on, and you see that the EVENT_CALL_SERVICE event has parameters domain set to lights and service set to turn_on, and the service parameters are passed as a dict in service_data. So then you can narrow down the event trigger to that particular service call:

from homeassistant.const import EVENT_CALL_SERVICE

@event_trigger(EVENT_CALL_SERVICE, "domain == 'lights' and service == 'turn_on'")
def monitor_light_turn_on_service(service_data=None):"lights.turn_on service called with service_data={service_data}")

This wiki page gives more examples of built-in and user events and how to create triggers for them.


@mqtt_trigger(topic, str_expr=None, kwargs=None)

@mqtt_trigger subscribes to the given MQTT topic and triggers whenever a message is received on that topic. Multiple @mqtt_trigger decorators can be applied to a single function if you want to trigger off different mqtt topics.

An optional str_expr can be used to match the MQTT message data, and the trigger will only occur if that expression evaluates to True or non-zero. This expression has available these four variables:

  • trigger_type is set to “mqtt”
  • topic is set to the topic the message was received on
  • payload is set to the string payload of the message
  • payload_obj if the payload was valid JSON, this will be set to the native python object representing that payload.
  • qos is set to the message QoS.

When the @mqtt_trigger occurs, those same variables are passed as keyword arguments to the function in case it needs them. Additional keyword parameters can be specified by setting the optional kwargs argument to a dict with the keywords and values.

Wildcards in topics are supported. The topic variables will be set to the full expanded topic the message arrived on.

NOTE: The MQTT Integration in Home Assistant must be set up to use @mqtt_trigger.



When any trigger occurs (whether time, state or event), the @state_active expression is evaluated. If it evaluates to False (or zero), the trigger is ignored and the trigger function is not called. Only a single @state_active decorator can be used per function - you can combine multiple conditions into str_expr using any, all, or logical operators like or or and.

This decorator is roughly equivalent to starting the trigger function with an if statement with the str_expr (the minor difference is that this decorator uses the @state_trigger variable value, if present, when evaluating str_expr, whereas an if statement at the start of the function uses its current value, which might be different if the state variable was changed immediately after the trigger, and the .old value is not available).

If the trigger was caused by @state_trigger, the prior value of the state variable that caused the trigger is available to str_expr with a .old suffix.

Inside the str_expr, undefined state variables, undefined state attributes, and undefined .old variables evaluate to None, rather than throwing an exception. Any .old variable will be None if the trigger is not a state trigger, if a different state variable change caused the state trigger, or if the state variable that caused the trigger was set for the first time (so there is no prior value).


@time_active(time_spec, ..., hold_off=None)

@time_active takes zero or more strings that specify time-based ranges. Only a single @time_active decorator can be used per function. When any trigger occurs (whether time, state or event), each time range specification is checked. If the current time doesn’t fall within any range specified, the trigger is ignored and the trigger function is not called. The optional numeric hold_off setting in seconds will ignore any triggers that are within that amount of time from the last successful one. Think of this as making the trigger inactive for that number of seconds immediately following each successful trigger. This can be used for rate-limiting trigger events or debouncing a noisy sensor.

Each string specification time_spec can take two forms:

  • "range(datetime_start, datetime_end)" is satisfied if the current time is in the indicated range, including the end points. As in @time_trigger, the year or date can be omitted to specify daily ranges. If the end is prior to the start, the range is satisfied if the current time is either greater than or equal to the start or less than or equal to the end. That allows a range like: @time_active("range(sunset - 20min, sunrise + 15min)") to mean at least 20 minutes before sunset, or at least 15 minutes after sunrise (note: at latitudes close to the polar circles, there can be cases where the sunset time is after midnight, so it is before the sunrise time, so this might not work correctly; at even greater latitudes sunset and sunrise will not be defined at all since there might not be daily sunrises or sunsets).
  • "cron(min hr dom mon dow)" is satisfied if the current time matches the range specified by the cron parameters. For example, if hr is 6-10 that means hours between 6 and 10 inclusive. If additionally min is * (i.e., any), then that would mean a time interval from 6:00 to immediately prior to 11:00.

Each argument specification can optionally start with not, which inverts the meaning of that range or cron specification. If you specify multiple arguments without not, they are logically or’ed together, meaning the active check is true if any of the (positive) time ranges are met. If you have several not arguments, they are logically and’ed together, so the active check will be true if the current time doesn’t match any of the “not” (negative) specifications. @time_active allows multiple arguments with and without not. The condition will be met if the current time matches any of the positive arguments, and none of the negative arguments.

Other Function Decorators


By default in pyscript all functions are async, so they cannot be used in task.executor, as callbacks or methods in python packages that expect regular functions, or used with built-in functions like filter, map or special class methods that are called by python internals (eg, __getattr__ or __del__).

The @pyscript_compile decorator causes the function to be treated as native Python and compiled, which results in a regular python function being defined, and it will run at full compiled speed. A lambda function is automatically compiled so it behaves like a regular python lambda function (which means the lambda function body cannot contain pyscript features).

For example:

def incr(x):
    return x + 1

x = list(map(incr, [0, 5, 10]))

sets x to [1, 6, 11].

One use for @pyscript_compile is to encapsulate functions that block (eg, doing I/O), so they can be called from task.executor. This might be a more convenient way to create native python functions called by task.executor, instead of moving them all to an external module or package outside of config/pyscript. This example reads a file using a native compiled function called by task.executor:

def read_file(file_name):
        with open(file_name, encoding="utf-8") as file_desc:
           return, None
    except Exception as exc:
        return None, exc

contents, exception = task.executor(read_file, "config/configuration.yaml")
if exception:
    raise exception"contents = {contents}")

This is an experimental feature and might change in the future. Restrictions include:

  • since it’s native python, the function cannot use any pyscript-specific features; but since it’s native python, all language features are available, including open, yield etc
  • if you use @pyscript_compile on an inner function (ie, defined inside a pyscript function), then binding of variables defined outside the scope of the inner function does not work.


The @pyscript_executor decorator does the same thing as @pyscript_compile and additionally wraps the compiled function with a call to task.executor. The resulting function is now a pyscript (async) function that can be called like any other pyscript function. This provides the cleanest way of defining a native python function that is executed in a new thread each time it is called, which is required for functions that does I/O or otherwise might block.

The file reading example above is simplified with the use of @pyscript_executor:

def read_file(file_name):
        with open(file_name, encoding="utf-8") as file_desc:
           return, None
    except Exception as exc:
        return None, exc

contents, exception = read_file("config/configuration.yaml")
if exception:
    raise exception"contents = {contents}")

Notice that read_file is called like a regular function, and it automatically calls task.executor, which runs the compiled native python function in a new thread, and then returns the result.

@service(service_name, …, supports_response=”none”)

The @service decorator causes the function to be registered as a service so it can be called externally. The string service_name argument is optional and defaults to "pyscript.FUNC_NAME", where FUNC_NAME is the name of the function. You can override that default by specifying a string with a single period of the form "DOMAIN.SERVICE". Multiple arguments and multiple @service decorators can be used to register multiple names (eg, aliases) for the same function.

The supports_response keyword argument can be set to one of three string values: "none" (the default), "only", or "optional", depending on whether the service provides no response, always provides a response, or sometimes provides a response. Services responses were added in HASS 2023.7. The service response is a dictionary returned by the function.

Other trigger decorators like @state_active and @time_active don’t affect the service. Those still allow state, time or other triggers to be specified in addition.

The function is called with keyword parameters set to the service call parameters, plus trigger_type is set to "service". The function definition should specify all the expected keyword arguments to match the service call parameters, or use the **kwargs argument declaration to capture all the keyword arguments.

The doc_string (the string immediately after the function declaration) is used as the service description that appears is in the Services tab of the Developer Tools page. The function argument names are used as the service parameter names, but there is no description.

Alternatively, if the doc_string starts with yaml, the rest of the string is used as a yaml service description. Here’s the first example above, with a more detailed doc_string (for a more complete example and explanation of the service description, check the Home Assistant developer documentation):

def hello_world(action=None, id=None):
name: Service example
description: hello_world service example using pyscript.
     description: turn_on turns on the light, fire fires an event
     example: turn_on
     required: true
           - turn_on
           - fire
     description: id of light, or name of event to fire
     example: kitchen.light
     required: true
""""hello world: got action {action}")
    if action == "turn_on" and id is not None:
        light.turn_on(entity_id=id, brightness=255)
    elif action == "fire" and id is not None:


@task_unique(task_name, kill_me=False)

This decorator is equivalent to calling task.unique() at the start of the function when that function is triggered. Like all the decorators, if the function is called directly from another Python function, this decorator has no effect. See this section for more details.


Most of these have been mentioned already, but here is the complete list of additional functions made available by pyscript.

Note that even though the function names contain a period, the left portion is not a class (e.g., state is not a class, and in fact isn’t even defined). These are simply functions whose name includes a period. This is one aspect where the interpreter behaves slightly differently from real Python.

However, if you set a variable like state, log or task to some value, then the functions defined with that prefix will no longer be available, since the portion after the period will now be interpreted as a method or class function acting on that variable. That’s the same behavior as Python - for example if you set bytes to some value, then the bytes.fromhex() class method is no longer available in the current scope.

State variables

State variables can be used and set just by using them as normal Python variables. However, there could be cases where you want to dynamically generate the variable name (eg, in a function or loop where the state variable name is computed dynamically). These functions allow you to get and set a variable using its string name. The set function also allows you to optionally set the attributes, which you can’t do if you are directly assigning to the variable:

Deletes the given state variable or attribute. The Python del statement can also be used to delete a state variable or attribute.
Returns the value of the state variable given its string name. A NameError exception is thrown if the name doesn’t exist. If name is a string of the form DOMAIN.entity.attr then the attribute attr of the state variable DOMAIN.entity is returned; an AttributeError exception is thrown if that attribute doesn’t exist.
Returns a dict of attribute values for the state variable name string, or None if it doesn’t exist. Alternatively, name can be a state variable. In pyscript prior to 1.0.0, this function was state.get_attr(). That deprecated name is still supported, but it logs a warning message and will be removed in a future version.
Returns a list of all state variable names (ie, entity_ids) of a domain. If domain is not specified, it returns all HASS state variable (entity_id) names.
state.persist(entity_id, default_value=None, default_attributes=None)
Indicates that the entity entity_id should be persisted. Optionally, a default value and default attributes (a dict) can be specified, which are applied to the entity if it doesn’t exist or doesn’t have any attributes respectively. “Persist” mean its value and attributes are preserved across HASS restarts. This only applies to entities in the pyscript domain (ie, name starts with pyscript.). See this section for more information
state.set(name, value=None, new_attributes=None, **kwargs)
Sets the state variable to the given value, with the optional attributes. The optional 3rd argument, new_attributes, should be a dict and it will overwrite all the existing attributes if specified. If instead attributes are specified using keyword arguments, then just those attributes will be set and other attributes will not be affected. If no optional arguments are provided, just the state variable value is set and the attributes are not changed. If no value is provided, just the state attributes are set and the value isn’t changed. To clear all the attributes, set new_attributes={}.
state.setattr(name, value)
Sets a state variable attribute to the given value. The name should fully specify the state variable and attribute as a string in the form DOMAIN.entity.attr.

Note that in HASS, all state variable values are coerced into strings. For example, if a state variable has a numeric value, you might want to convert it to a numeric type (eg, using int() or float()). Attributes keep their native type.

Service Calls, name, blocking=False, return_response=False, **kwargs)
calls the service with the given keyword arguments as parameters. If blocking is True, pyscript will wait for the service to finish executing before continuing the current routine. If the service returns a response, set return_response to True (which also causes blocking), and the response will be returned by the function call. Note that starting in HASS 2023.7, the blocking timeout parameter limit is no longer supported.
service.has_service(domain, name)
returns whether the service exists.

Event Firing, **kwargs)
sends an event with the given event_type string and the keyword parameters as the event data.


Five logging functions are provided, with increasing levels of severity:

log a message at debug level
log a message at info level
log a message at warning level
log a message at error level
same as log.debug(str); currently print doesn’t support other arguments.

The Logger component can be used to specify the logging level. Log messages below the configured level will not appear in the log. Each log message function uses a log name of the form:


where FUNCNAME is the name of the top-level Python function (e.g., the one called by a trigger or service), defined in the script file See the Global Context section for the logging paths for other cases.

That allows you to set the log level for each Python top-level script or function separately if necessary. That setting also applies to any other Python functions that the top-level Python function calls. For example, these settings:

  default: info
    custom_components.pyscript.file: info
    custom_components.pyscript.file.my_script.my_function: debug

will log all messages at info or higher (ie:, log.warning() and log.error()), and inside my_function defined in the script file (and any other functions it calls) will log all messages at debug or higher.

Note that in Jupyter, all the log functions will display output in your session, independent of the logger configuration settings.

Changing the log level via the main configuration yaml file will typically require a restart of HASS. To avoid that, you can set the log level for any specific logging path by calling the logger.set_level service from Jupyter, VSC, or in a pyscript script; for example:

logger.set_level(**{"custom_components.pyscript.file.my_script": "debug"})

will change the log level for the script to debug, while:

logger.set_level(**{"custom_components.pyscript.file.my_script.func1": "debug"})

will change the log level for the func1() function in script to debug. You can change multiple components in a single call just by adding them to the **kwargs dict.

Alternatively, you can use the Developer Tools to call the logger.set_level service with the logging path and desired level as parameters.


A task is a lightweight execution context that runs a function inside an event loop, effectively providing asynchronous (actually collaborative serial) execution. They are part of Python’s asyncio package and are central to how HASS and pyscript handle multiple activities. Tasks can run any Python code, provided it does not block (eg, for I/O) or run without giving up control for too long (which will prevent other tasks from running).

Task Management

These functions allow new tasks to be created and managed. Normally you won’t need to create your own tasks, since trigger functions automatically create a new task to run the function every time the trigger occurs.

task.create(func, *args, **kwargs)
Creates a new task and calls the given function func with the positional and keyword arguments. Returns the task id of the newly created task. The task id is an asyncio task object, which supports several methods. The function func runs asynchronously until it returns or the task is cancelled.
Cancels (kills) the task specified by the task_id returned by task.create. This is a simpler alternative to the task_id.cancel() method, which also requires waiting for the task to cancel. With no argument, task.cancel cancels the current task, which you might prefer to use in a trigger function on error, instead of a return or raising an exception.
Returns the task id of the current task.
Returns the task id given a name that task previously passed to task.unique. A NameError exception is raised if the task name is unknown. With no arguments it returns a dict mapping all names to task ids. The names are specific to the current global context.

Waits until the given set of tasks complete. This function calls asyncio.wait, so it takes the same arguments. task_set is a set or list of task ids, and it returns two sets of done and pending task ids. An example:

done, pending = task.wait({task_id1, task_id2})

This waits until both tasks complete, and sets done to {task_id1, task_id2} and pending to an empty set (pending might be non-empty if you specify a timeout or return_when).

You can check if a task is done with task_id.done(). After a task has finished, the function’s return value is available as task_id.result(), which raises an exception if the task is not finished or was cancelled.

task.add_done_callback(task_id, func, *args, **kwargs)
This is a more convenient alternative to the task_id.add_done_callback method that supports both pyscript function (coroutine) and regular function callbacks. The function func with the specified arguments is called when the task completes. Multiple callbacks (with different functions) can be added to one task. If you use the same func argument it replaces the prior callback for that function (ie, only one done callback per function is supported).
task.remove_done_callback(task_id, func)
This is a more convenient alternative to the task_id.remove_done_callback method that supports both pyscript function (coroutine) and regular function callbacks. This removes a previously added done callback, and the callback to that function will no longer occur when the task completes.

Task executor

If you call any Python functions that do I/O or otherwise block, they need to be run outside the main event loop using task.executor:

task.executor(func, *args, **kwargs)
Run the given function in a separate thread. The first argument is the function to be called, followed by each of the positional or keyword arguments that function expects. The func argument can only be a regular Python function (eg, defined in an imported module), not a function defined in pyscript. task.executor waits for the function to complete in the other thread, and it returns the return value from the function func.

See this section for more information.

Task sleep

sleeps for the indicated number of seconds, which can be floating point. Do not import time and use time.sleep() - that will block lots of other activity.

Task unique

task.unique(task_name, kill_me=False)
kills any currently running task that previously called task.unique with the same task_name. The name can be any string. If kill_me=True then the current task is killed if another task that is running previously called task.unique with the same task_name.

Note that task.unique is specific to the current global context, so names used in one global context will not affect another.

Once a task calls task.unique with a name, that name can be used to look up the task id by name using task.name2id. So even if task.unique is not used to kill a prior task, it can be used to associate that task with a name which might be helpful if you need to manage specific tasks (eg, cancel them, wait for them to complete, or get their return value when they finish).

A task can call task.unique multiple times with different names, which will kill any tasks that previously called task.unique with each of those names. The task continues to “own” all of those names, so any one of them could subsequently be used by a different task to kill the original task. Any of those names can be used to find the task id by name using task.name2id.

task.unique can also be called outside a function, for example in the preamble of a script file or interactively using Jupyter. That causes any currently running functions (ie, functions that have already been triggered and are running Python code) that previously called task.unique with the same name to be terminated. Since any currently running functions are not terminated on reload, this is the mechanism you can use should you wish to terminate specific functions on reload. If used outside a function or interactively with Jupyter, calling task.unique with kill_me=True causes task.unique to do nothing.

The task.unique functionality is also provided via a decorator @task_unique. If your function immediately and always calls task.unique, you could choose instead to use the equivalent function decorator form.

Task waiting

allows functions to wait for events, using identical syntax to the decorators. This can be helpful if at some point during execution of some logic you want to wait for some additional triggers.

It takes the following keyword arguments (all are optional):

  • state_trigger=None can be set to a string just like @state_trigger, or it can be a list of strings that are logically “or”ed together.
  • time_trigger=None can be set to a string or list of strings with datetime specifications, just like @time_trigger.
  • event_trigger=None can be set to a string or list of two strings, just like @event_trigger. The first string is the name of the event, and the second string (when the setting is a two-element list) is an expression based on the event parameters.
  • mqtt_trigger=None can be set to a string or list of two strings, just like @mqtt_trigger. The first string is the MQTT topic, and the second string (when the setting is a two-element list) is an expression based on the message variables.
  • timeout=None an overall timeout in seconds, which can be floating point.
  • state_check_now=True if set, task.wait_until() checks any state_trigger immediately to see if it is already True, and will return immediately if so. If state_check_now=False, task.wait_until() waits until a state variable change occurs, before checking the expression. Using True is safer to help avoid race conditions, although False makes task.wait_until() behave like @state_trigger, which by default doesn’t check at startup. However, if you use the default of True, and your function will call task.wait_until() again, it’s recommended you set that state variable to some other value immediately after task.wait_until() returns. Otherwise the next call will also return immediately. Note that entries in state_trigger that are plain state variable names (which mean trigger on any change) are ignored during this initial check; only expressions are evaluated.
  • state_hold=None is an optional numeric duration is seconds. If specified, any state_trigger delays returning for this amount of time. If the state trigger expression changes to False during that time, the trigger is canceled and a wait for a new trigger begins. If the state trigger expression changes, but is still True then the state_hold time is not restarted - task.wait_until() will return that number of seconds after the first state trigger (unless a different trigger type or a timeout occurs first). This setting also applies to the initial check when state_check_now=True.
  • state_hold_false=None requires the expression evaluate to False for this duration in seconds before a subsequent state trigger occurs. The default value of None means that the trigger can occur without the trigger expression having to be False. A value of 0 requires the expression become False before the trigger, but with no minimum time in that state. When state_hold_false is set, the state trigger expression is evaluated immediately. If False the state_hold_false period begins. If True, a wait for the next False value begins. If state_check_now is also set, task.wait_until() will still return immediately if the expression is initially True.

When a trigger occurs, the return value is a dict containing the same keyword values that are passed into the function when the corresponding decorator trigger occurs. There will always be a key trigger_type that will be set to:

  • "state", "time" or "event" when each of those triggers occur.
  • "timeout" if there is a timeout after timeout seconds (the dict has no other values)
  • "none" if you specify only time_trigger and no timeout, and there is no future next time that satisfies the trigger condition (e.g., a range or once is now in the past). Otherwise, task.wait_until() would never return.

In the special case that state_check_now=True and task.wait_until() returns immediately, the other return variables that capture the variable name and value that just caused the trigger are not included in the dict - it will just contain trigger_type="state".

Here’s an example. Whenever a door is opened, we want to do something if the door closes within 30 seconds. If a timeout of more than 30 seconds elapses (ie, the door is still open), we want to do some other action. We use a decorator trigger when the door is opened, and we use task.wait_until to wait for either the door to close, or a timeout of 30 seconds to elapse. The return value tells which of the two events happened:

@state_trigger("security.rear_door == 'open'")
def rear_door_open_too_long():
    """send alert if door is open for more than 30 seconds"""
    trig_info = task.wait_until(
                    state_trigger="security.rear_door == 'closed'",
    if trig_info["trigger_type"] == "timeout":
        # 30 seconds elapsed without the door closing; do some actions
        # the door closed within 30 seconds; do some other actions

task.wait_until() is logically equivalent to using the corresponding decorators, with some important differences. Consider these two alternatives, which each run some code whenever there is an event test_event3 with parameters args == 20 and arg2 == 30:

@event_trigger("test_event3", "arg1 == 20 and arg2 == 30")
def process_test_event3(**trig_info):
    # do some things, including waiting a while
    # do some more things


@time_trigger    # empty @time_trigger means run the function on startup
def wait_for_then_process_test_event3():
    while 1:
        trig_info = task.wait_until(
                        event_trigger=["test_event3", "arg1 == 20 and arg2 == 30"]
        # do some things, including waiting a while
        # do some more things

Logically they are the similar, but the important differences are:

  • task.wait_until() only looks for the trigger conditions when it is called, and it stops monitoring them as soon as it returns. That means the trigger (especially an event trigger) could occur before or after task.wait_until() is called, and you will miss the event. In contrast, the decorator triggers monitor the trigger conditions continuously, so they will not miss state changes or events once they are initialized. The reason for the state_check_now argument, and its default value of True is to help avoid this race condition for state triggers. Time triggers should generally be safe.
  • The decorators run each trigger function as a new independent task, and don’t wait for it to finish. So a function will be run for every matching event. In contrast, if your code runs for a while before calling task.wait_until() again (e.g., task.sleep() or any code), or even if there is no other code in the while loop, some events or state changes of interest will be potentially missed.

Summary: use trigger decorators whenever you can. Be especially cautious using task.wait_until() to wait for events; you must make sure your logic is robust to missing events that happen before or after task.wait_until() runs.

Global Context

Each pyscript file that is loaded, and each Jupyter session, runs inside its own global context, which means its global variables and functions are isolated from each other (unless they are a module or package that is explicitly imported). In normal use you don’t need to worry about global contexts. But for interactive debugging and development, you might want your Jupyter session to access variables and functions defined in a script file.

Here is the naming convention for each file’s global context (upper case mean any value; lower case are actual fixed names):

pyscript file path global context name
pyscript/ file.FILE
pyscript/modules/ modules.MODULE
pyscript/modules/MODULE/ modules.MODULE
pyscript/modules/MODULE/ modules.MODULE.FILE
pyscript/apps/ apps.APP
pyscript/apps/APP/ apps.APP
pyscript/apps/APP/ apps.APP.FILE
pyscript/scripts/ scripts.FILE
pyscript/scripts/DIR1/ scripts.DIR1.FILE
pyscript/scripts/DIR1/DIR2/ scripts.DIR1.DIR2.FILE

Note that if the package form of an app or module (pyscript/apps/APP/ or pyscript/modules/MODULE/ exists, the module form (pyscript/apps/ or pyscript/modules/ is ignored.

The logging path uses the global context name, so you can customize logging verbosity for each global context, to the granularity of specific functions eg:

  default: info
    custom_components.pyscript.file: info
    custom_components.pyscript.file.my_script.my_function: debug
    custom_components.pyscript.apps.my_app: debug
    custom_components.pyscript.apps.my_app.my_function: debug

Each Jupyter global context name is jupyter_NNN where NNN is a unique integer starting at 0.

You can set the log level for one or more logging paths by calling the logger.set_level service from Jupyter, VSC, or in a pyscript script; for example:

logger.set_level(**{"custom_components.pyscript.file.my_script": "debug"})

will change the log level for the script to debug.

When a script file has changed (or an app’s configuration has changed, provided the yaml file is below the pyscript directory), a reload is triggered, and the corresponding global context whose names starts with file., modules., apps. or scripts. is removed. As each file is reloaded, the corresponding global context is created.

Three functions are provided for getting, setting and listing the global contexts. That allows you to interactively change the global context during a Jupyter session. You could also use these functions in your script files, but that is strongly discouraged because it violates the name space isolation among the script files. Here are the functions:

returns the current global context name.
lists all the global contexts, with the current global context listed first.
sets the current global context to the given name.

When you exit a Jupyter session, its global context is deleted, which means any triggers, functions, services and variables you created are deleted (HASS state variables survive). If you switch to a script file’s context, then any triggers, functions, services or variables you interactively create there will persist after you exit the Jupyter session. However, if you don’t update the corresponding script file, whenever the script is modified and automatically reloaded, or upon HASS restart, those interactive changes will be lost, since reloading a script file recreates a new global context.

Advanced Topics


Without Jupyter, the pyscript workflow involves editing scripts in the <config>/pyscript folder. Each time a file is changed, it is automatically reloaded. You will need to look at the log file for error messages (eg, syntax errors), or log output from your code.

If a module or app file (or any yaml files in the pyscript directory) has changed, all files in that module or app get reloaded too, and also any other files that import that module. If you want to reload all the files, call the pyscript.reload service with the optional parameter global_ctx set to *.

An excellent alternative to repeatedly modifying a script file is to use Jupyter notebook to interactively develop and test functions, triggers and services.

Jupyter auto-completion (with <TAB>) is supported in Jupyter notebook, console and lab. It should work after you have typed at least the first character. After you hit <TAB> you should see a list of potential completions from which you can select. It’s a great way to easily see available state variables, functions or services.

In a Jupyter session, one or more functions can be defined in each code cell. Every time that cell is executed (eg, <Shift>Return), those functions are redefined, and any existing trigger decorators with the same function name are canceled and replaced by the new definition. You might have other function and trigger definitions in another cell - they won’t be affected (assuming those function names are different), and they will only be replaced when you re-execute that other cell.

When the Jupyter session is terminated, its global context is deleted, which means any trigger rules, functions, services and variables you created are deleted. The pyscript Jupyter kernel is intended as an interactive sandbox. As you finalize specific functions, triggers and automation logic, you should copy them to a pyscript script file, which will automatically reloaded once that file is written. That ensures they will be loaded and run each time you re-start HASS.

If a function you define has been triggered and is currently executing Python code, then re-running the cell in which the function is defined, or exiting the Jupyter session, will not stop or cancel the already running function. This is the same behavior as reload. In pyscript, each triggered function (ie, a trigger has occurred, the trigger conditions are met, and the function is actually executing Python code) runs as an independent task until it finishes. So if you are testing triggers of a long-running function (eg, one that uses task.sleep() or task.wait_until()) you could end up with many running instances. It’s strongly recommended that you use task.unique() to make sure old running function tasks are terminated when a new one is triggered. Then you can manually call task.unique() to terminate that last running function before exiting the Jupyter session.

If you switch global contexts to a script file’s context, and create some new variables, triggers, functions or services there, then those objects will survive the termination of your Jupyter session. However, if you reload that script, those newly-created objects will no longer exist. To make any additions or changes permanent (meaning they will be re-created on each reload or each time your restart HASS) then you should copy the changes or additions to one of your pyscript script files.


Pyscript supports importing two types of packages or modules:

  • Pyscript code can be put into modules or packages and stored in the <config>/pyscript/modules folder. Any pyscript code can import and use these modules or packages. These modules are not autoloaded on startup; they are only loaded when another script imports them. Any changes to a module’s files will cause all of the module files to be unloaded, and any scripts or apps that import that module will be reloaded. Imports of pyscript modules and packages are not affected by the allow_all_imports setting - if a file is in the <config>/pyscript/modules folder then it can be imported.

    Package-style layout is also supported where a PACKAGE is defined in <config>/pyscript/modules/PACKAGE/, and that file can, in turn, do relative imports of other files in that same directory. This form is most convenient for sharing useful pyscript libraries, since all the files for one package are stored in its own directory.

  • Installed Python packages can be imported. By default, pyscript only allows a short list of Python packages to be imported, for both security reasons and to reduce the risk that package functions that block doing I/O are called.

The rest of this section discusses the second style - importing installed Python modules and packages.

If you set the allow_all_imports configuration parameter, any available Python package can be imported. You should be cautious about setting this if you are going to install community pyscript code without inspecting it, since it could, for example, import os and call os.remove(). However, if you are developing your own code then there is no issue with enabling all imports.

Pyscript code is run using an asynchronous interpreter, which allows it to run in the HASS main event loop. That allows many of the “magic” features to be implemented without the user having to worry about the details. However, the performance will be much slower that regular Python code, which is typically compiled. Any Python packages you import will run at native, compiled speed.

So if you plan to run large chunks of code in pyscript without needing any of the pyscript-specific features, or you want access to native Python features that aren’t supported in pyscript (like yield, open, read or write), you might consider putting them in a package and importing it instead. That way it will run at native compiled speed and have full access to the native Python language.

One way to do that is in one of your pyscript script files, add this code:

import sys

if "/config/pyscript_modules" not in sys.path:

This adds the directory /config/pyscript_modules to Python’s module search path (you should use the correct full path specific to your installation). You will need to set the allow_all_imports configuration parameter to true to allow importing of sys. You can then add modules (files ending in .py) to that folder, which will contain native python that is compiled when imported (note that none of the pyscript-specific features are available in those modules).

Pyscript can install required Python packages if they are missing. Depending on how you run HASS (eg, using a minimal Docker container) it might not be convenient to manually install Python packages using pip. If your pyscript code requires particular Python packages that are not already installed by HASS, add a requirements.txt file the <config>/pyscript directory. This file lists each required package one per line, with an optional version if you require a specific version or minimum version of that package, eg:

# this is a comment

When a specific version of a package is required, the == specifier must be used. Unpinned packages (no version specified) are also accepted, but the highest pinned version will always take precedence when a package has been specified as a requirement multiple times.

Each app’s or module’s directory (assuming they use the directory-form of a package) can also contain an optional requirements.txt file:

  • <config>/pyscript/modules/my-package-name/requirements.txt
  • <config>/pyscript/apps/APP_NAME/requirements.txt

That allows you to specify the specific requirements for each pyscript module or app. If you release or share your module or app, all its code and requirements are self-contained, and any user can simply install the files in that directory and the requirements will be checked on the next start of HASS or reload.

If a required package version differs from the installed one, no change is made since it’s likely HASS has a requirement that pyscript should not change. In that case a warning message will be logged and the requirement will be skipped.

Trigger Closures

Pyscript supports trigger functions that are defined as closures, ie: functions defined inside another function. This allows you to easily create many similar trigger functions that might differ only in a couple of parameters (eg, a common function in different rooms or for each media setup). The trigger will be stopped when the function is no longer referenced in any scope. Typically the closure function is returned, and the return value is assigned to a variable. If that variable is re-assigned or deleted, the trigger function will be destroyed.

Here’s an example:

def state_trigger_factory(sensor_name, trig_value):

    @state_trigger(f"input_boolean.{sensor_name} == '{trig_value}'")
    def func_trig(value=None):"func_trig: {sensor_name} is {value}")

    return func_trig

f1 = state_trigger_factory("test1", "on")
f2 = state_trigger_factory("test2", "on")
f3 = state_trigger_factory("test3", "on")

This creates three trigger functions that fire when the given sensor input_boolean.testN is on. If you re-assign or delete f1 then that trigger will be destroyed, and the other two will not be affected. If you repeatedly re-run this block of code in Jupyter the right thing will happen - each time it runs the old triggers are destroyed when the variables are re-assigned.

Any data type could be used to maintain a reference to the trigger function. For example a list could be manually built:

input_boolean_test_triggers = [
    state_trigger_factory("test1", "on"),
    state_trigger_factory("test2", "on"),
    state_trigger_factory("test3", "on")

or dynamically in a loop:

input_boolean_test_triggers = []
for i in range(1, 4):
    input_boolean_test_triggers.append(state_trigger_factory(f"test{i}", "on"))

If you are writing a factory function and you prefer the caller not to bother with maintaining variables with the closure functions, you could move the appending into the function and use a global variable (a class could also be used):

input_boolean_test_triggers = []

def state_trigger_factory(sensor_name, trig_value):

    @state_trigger(f"input_boolean.{sensor_name} == '{trig_value}'")
    def func_trig(value=None):"func_trig: {sensor_name} is {value}")


state_trigger_factory("test1", "on")
state_trigger_factory("test2", "on")
state_trigger_factory("test3", "on")

Notice there is no return value from the factory function.

A dict could be used instead of a list, with a key that combines the unique parameters of the trigger. That way a new trigger with the same parameters will replace an old one when the dict entry is set, if that’s the behavior you want.

Accessing YAML Configuration

Pyscript binds all of its yaml configuration to the variable pyscript.config. That allows you to add configuration settings that can be processed by your pyscript code. Additionally, an application’s configuration (eg, for an application app_name, all the settings in app_name below apps) are available in the variable pyscript.app_config in the global scope of the application’s main file (eg, apps/ or apps/app_name/

One motivation is to allow pyscript apps to be developed and shared that can instantiate triggers and logic based on yaml configuration. That allows other users to use and configure your pyscript code without needing to edit or even understand it - they just need to add the corresponding yaml configuration.

The configuration for an application should be placed below that application’s name under the apps configuration key. For example, the settings for a pyscript application called auto_lights below an entry apps. That entry could contain a list of settings (eg, for handling multiple rooms or locations).

Here’s an example yaml configuration with settings for two applications, auto_lights and motion_light:

  allow_all_imports: true
      - room: living
        level: 60
         - 1
         - 20
      - room: dining
        level: 80
         - 1
         - 20
      - sensor: rear_left
        light: rear_flood
      - sensor: side_yard
        light: side_flood
      - sensor: front_patio
        light: front_porch
    setting1: 10
    setting2: true

For the auto_lights application, those settings are available to that application’s main source file (eg, apps/ or apps/auto_lights/ in the global variable pyscript.app_config, which will be set to:

    {"room": "living", "level": 60, "some_list": [1, 20]},
    {"room": "dining", "level": 80, "some_list": [1, 20]},

The corresponding global pyscript.config variable value will be:

    "allow_all_imports": True,
    "apps": {
        "auto_lights": [
            {"room": "living", "level": 60, "some_list": [1, 20]},
            {"room": "dining", "level": 80, "some_list": [1, 20]},
        "motion_light": [
            {"sensor": "rear_left", "light": "rear_flood"},
            {"sensor": "side_yard", "light": "side_flood"},
            {"sensor": "front_patio", "light": "front_porch"},
    "global": {
        "setting1": 10,
        "setting2": True,

Note that accessing pyscript.config["apps"] is deprecated. The apps entry will be removed in a future release so that apps do not have access to another app’s configuration (which might include token, passwords or keys).

Your application code for auto_lights would be in either

  • <config>/pyscript/apps/
  • <config>/pyscript/apps/auto_lights/

It can simply iterate over pyscript.app_config settings up the necessary triggers and application logic, eg:

def setup_triggers(room=None, level=None, some_list=None):
    # define some trigger functions etc

for inst in pyscript.app_config:

Validating the configuration can be done either manually or with the voluptuous package.

Secrets can also be bound to variables by substituting the value with !secret my_secret. This is useful for API keys or passwords.

Here is an example config for service_checker which needs a name (not a secret) and a URL and an API key (which are secret):

  allow_all_imports: true
      - service_name: my_service
        url: !secret my_secret_url
        api_key: !secret my_secret_api_key

Access to Hass

If the hass_is_global configuration setting is set (default is off), then the variable hass is available as a global variable in all pyscript contexts. That provides significant flexibility in accessing HASS internals for cases where pyscript doesn’t provide some binding or access.

Ideally you should only use hass for read-only access. However, you do need a good understanding of hass internals and objects if you try to call functions or update anything. With great power comes great responsibility!

For example, you can access configuration settings like hass.config.latitude or hass.config.time_zone.

You can use hass to compute sunrise and sunset times using the same method HASS does, eg:

import homeassistant.helpers.sun as sun
import datetime

location = sun.get_astral_location(hass)
sunrise = location[0].sunrise(
sunset = location[0].sunset(
print(f"today sunrise = {sunrise}, sunset = {sunset}")

(Note that the sun.sun attributes already provide times for the next sunrise and sunset, so this example is a bit contrived. Also note this only applies to HA 2021.5 and later; prior to that, sun.get_astral_location() doesn’t return the elevation, so replace location[0] with location in the two expressions if you have an older version of HA.)

Here’s another method that uses the installed version of astral directly, rather than the HASS helper function. HA 2021.5 upgraded to astral 2.2, and here’s one way of getting sunrise using this version (prior to this HA used a very old version of astral).

import astral
import astral.location
import datetime

here = astral.location.Location(astral.LocationInfo("City", "Country", str(hass.config.time_zone),
                                                    hass.config.latitude, hass.config.longitude))
sunrise = here.sunrise(
sunset = here.sunset(
print(f"today sunrise = {sunrise}, sunset = {sunset}")

If there are particular HASS internals that you think many pyscript users would find useful, consider making a feature request or PR so it becomes a built-in feature in pyscript, rather than requiring users to always have to delve into hass.

Avoiding Event Loop I/O

All pyscript code runs in the HASS main event loop. That means if you execute code that blocks, for example doing I/O like reading or writing files or fetching a URL, then the main loop in HASS will be blocked, which will delay all other tasks.

All the built-in functionality in pyscript is written using asynchronous code, which runs seamlessly together with all the other tasks in the main event loop. However, if you import Python packages and call functions that block (eg, file or network I/O) then you need to run those functions outside the main event loop. That can be accomplished wrapping those function calls with the task.executor function, which runs the function in a separate thread:

task.executor(func, *args, **kwargs)
Run the given function in a separate thread. The first argument is the function to be called, followed by each of the positional or keyword arguments that function expects. The func argument can only be a regular Python function, not a function defined in pyscript.

If you forget to use task.executor, you might get this warning from HASS:

WARNING (MainThread) [homeassistant.util.async_] Detected I/O inside the event loop. This is
causing stability issues. Please report issue to the custom component author for pyscript doing
I/O at custom_components/pyscript/, line 1583: return func(*args, **kwargs)

Currently the built-in functions that do I/O, such as open, read and write are not supported to avoid I/O in the main event loop, and also to avoid security issues if people share pyscripts. Also, the print function only logs a message, rather than implements the real print features, such as specifying an output file handle. If you want to do file I/O from pyscript, you have two choices:

  • put the code in a separate native Python module, so that functions like open, read and write are available, and call the function in that module from pyscript using task.executor. See Importing for how to set Python’s sys.path to import a local Python module.
  • you could use the os package (which can be imported by setting allow_all_imports) and calling the low-level functions like and using task.executor to wrap every function.

Here’s an example fetching a URL. Inside pyscript, this is the wrong way since it does I/O without using a separate thread:

import requests

url = ""
resp = requests.get(url)

The correct way is:

import requests

url = ""
resp = task.executor(requests.get, url)

An even better solution to fetch a URL is to use a Python package that uses asyncio, in which case there is no need for task.executor. In this case, aiohttp can be used (the await keyword is optional in pyscript):

import aiohttp

url = ""
async with aiohttp.ClientSession() as session:
    async with session.get(url) as resp:

Here’s another example that creates a client connection to a TCP server and exchanges messages in a manner that avoids event loop I/O by using asyncio.open_connection.

import asyncio

Reader, Writer = None, None

def do_client_connection():
    global Reader, Writer
    Reader, Writer = asyncio.open_connection('', 8956)

def client_send(message):
    if not Writer:
       raise("Client is not connected")
    return Reader.readline().decode()

This connects to the server (in this example at at startup, and then you can call client_send() and it will send the message and return the reply.

This assumes the protocol is line-oriented; you could call instead if you want to read bytes instead of expecting a newline with Reader.readline().

To test the code above, you can create a server by running nc -l 8956 before you run the code. When you call client_send("hello\n") you should see the hello printed by nc. Then whatever you type back at nc will be returned by client_send().

Persistent State

Pyscript has the ability to persist state variables in the pyscript. domain, meaning their values and attributes are preserved across HASS restarts. To specify that the value of a particular entity persists, you need to request persistence explicitly. This must be done in a code location that will be certain to run at startup.


@state_trigger('binary_sensor.motion == "on"')
def turn_on_lights():
  pyscript.last_light_on = "light.overhead"

With this in place, state.persist() will be called every time this script is parsed, ensuring the pyscript.last_light_on state variable state will persist between HASS restarts. If state.persist is not called on a particular state variable before HASS stops, then that state variable will not be preserved on the next start.

Language Limitations

Pyscript implements a Python interpreter in a fully-async manner, which means it can run safely in the main HASS event loop.

The language coverage is relatively complete, but it’s quite possible there are discrepancies with Python in certain cases. If so, please report them.

Here are some areas where pyscript differs from real Python:

  • The pyscript-specific function names and state names that contain a period are treated as plain identifiers that contain a period, rather than an attribute (to the right of the period) of an object (to the left of the period). For example, while pyscript.reload and state.get are functions, pyscript and state aren’t defined. However, if you set pyscript or state to some value (ie: assign them as a variable), then pyscript.reload and state.get are now treated as accessing those attributes in the pyscript or state object, rather than calls to the built-in functions, which are no longer available. That’s similar to regular Python, where if you set bytes to some value, the bytes.fromhex function is no longer available.
  • Since pyscript is async, it detects whether functions are real or async, and calls them in the correct manner. So it’s not necessary to use async and await in pyscript code - they are optional.
  • All pyscript functions are async. So if you call a Python module that takes a pyscript function as a callback argument, that argument is an async function, not a normal function. So a Python module won’t be able to call that pyscript function unless it uses await, which requires that function to be declared async. Unless the Python module is designed to support async callbacks, it is not currently possible to have Python modules and packages call pyscript functions. The workaround is to move your callbacks from pyscript and make them native Python functions; see Importing.
  • Continuing that point, special methods (eg, __eq__) in a class created in pyscript will not work since they are async functions and Python will not be able to call them. The two workarounds are to use the @pyscript_compile decorator so the method is compiled to a native (non-ascync) Python function, or write your class in native Python and import it into pyscript; see Importing.
  • The import function in pyscript fails to import certain complex packages. This is an open bug and it would be great if someone with some Python expertise could help fix it. In the meantime, the workaround is to import the module in a native Python file, and then import that shim module into pyscript. See Importing.
  • pyscript and the HASS primitives that it uses are not thread safe - the code should only be executed in the main event loop. The task.executor() function is one way that regular Python code (not pyscript code) can safely be executed in a separate thread. The threading package can potentially be used (although that is untested), so long as any threads created only run regular Python code (and not call any pyscript functions, which are all async). Bad things will happen if you call pyscript functions from a thread you create; currently there isn’t error checking for that case.

A handful of language features are not supported:

  • generators and the yield statement; these are difficult to implement in an interpreter.
  • built-in functions that do I/O, such as open, read and write are not supported to avoid I/O in the main event loop, and also to avoid security issues if people share pyscripts. The print function only logs a message, rather than implements the real print features, such as specifying an output file handle.
  • The built-in function decorators (eg, state_trigger) aren’t functions that can be called and used in-line. However, you can define your own function decorators that could include those decorators on the inner functions they define. Currently none of Python’s built-in decorators are supported.

Pyscript can call Python modules and packages, so you can always write your own native Python code (eg, if you need a generator or other unsupported feature) that can be called by pyscript (see Importing for how to create and import native Python modules in pyscript).