Version: 13 - TBD

Execution

class resources.execution.Execution

Base class: Activity

An automation (flow, script, or connection) which is or was running.

See the corresponding Flow Api class at Execution

Property Description Type Import/Export

activity_type Enum('EXECUTION', 'MESSAGE', 'PROCESS', name='activitytype') neither

break A flag to control if the execution should pause on the next line. Boolean() neither

breakpoints The breakpoints of the execution. ARRAY(SmallInteger()) import only

bundle_id Reference to the bundle this record is associated with. Allowed for BUNDLE_REPOSITORY, CONNECTOR, CUSTOM_OBJECT, DASHBOARD, DEVOLUTIONS_CONFIG, EXECUTION, FILE, FLOW, LDAP_CONFIG, MCP_USER_TOOL, MESSAGE, OBJECT_TEMPLATE, PLUGIN, PYTHON_MODULE, ROLE, SCHEDULE, SCHEDULER, SCHEMA, SECRET, SETTING, SYNC_CONFIG, TAG, VAULT_CONFIG, WEBHOOK, WRAPPER. Not allowed for BUNDLE, ORGANIZATION, PROCESS, PROJECT, USER, WORKSPACE. Mutually exclusive with project_id UUID(as_uuid=False) both

bundle_name The name of the bundle. Will look up the bundle and set bundle_id. String(length=128) import only

children Subrecords to create in the SAME transaction as the record, as a mapping of {subrecord_type: [ {column: value, ...}, ... ]}. A generic, schema-derived create-with-children: a key is accepted iff it names a subrecord type whose parent is one of this record's ancestor tables — e.g. an object_template with object_template_attribute rows, a role with role_permission rows, or a flow with resource_wrapper rows. The parent foreign key (<parent>_id) is auto-wired to the new record; each child is validated by its own subrecord create-path. Passing children inline makes the record and its subrecords atomic (any child failure rolls the whole create/update back), replacing a create-then-add-N sequence with one round-trip. Exposed identically on REST create/update bodies and the GraphQL create_<type>/update_<type> mutation input (as an arbitrary-JSON scalar), and via MCP create_record/update_record, since it is a single schema-derived column with no per-surface code. Accepted on both create and update. Nesting to arbitrary depth IS supported: a child entry may itself carry a children key, validated recursively against THAT child subrecord type's own ancestor tables and created depth-first in the same transaction. (Under the current subrecord model the only externally-creatable chains happen to be one level deep — subrecords are not record-backed, so e.g. a record_metadata cannot be nested under an object_template_attribute — but the recursion is applied generically so any deeper externally-creatable chain works with no per-type code.) On read the children are exposed via their subrecord lists, not this field. JSONEncodedData() neither

comment_count Denormalized count of record_comment rows attached to this record. Server-maintained on comment create/delete; read-only to clients. Integer() neither

commit_message The commit message for this change. Text() import only

connector_dispatched_at Set once, committed, immediately before a connector execution is handed to the connector thread. Because it is persisted it survives a hard process crash (e.g. an OOM SIGKILL that skips graceful shutdown), so after the resumer re-queues the still-AT_ENGINE execution into a fresh process the engine can tell "already dispatched once" apart from "never dispatched" and refuse to re-run a non-idempotent connector (at-most-once guard). See engine.run_execution_connection. Output only. DateTime(timezone=True) neither

connector_id The ID of the connector record the execution is using. UUID(as_uuid=False) both

connector_name The name of the connector. Only used for type CONNECTION String(length=128) neither

connector_type The type of the connector / the type of the remote system. Enum('AWS', 'AZURE', 'AZUREAI', 'CLAUDE', 'FTP', 'GIT', 'GOOGLE', 'HTTP', 'IMAP', 'JSONRPC', 'K8S', 'LDAP', 'MYSQL', 'OPENAI', 'OVH', 'PS', 'REDIS', 'REST', 'SCP', 'SFTP', 'SMB', 'SMTP', 'SOAP', 'SQLMSSQL', 'SQLORACLE', 'SQLPG', 'SSH', 'VAULT', 'WEBDAV', 'XMLRPC', name='connectortype') both

created_at DateTime(timezone=True) export only

created_by UUID(as_uuid=False) export only

custom_object_id If this execution was created by the on_create, on_update, or on_delete hook of an object template, this field contains the reference to the custom object. UUID(as_uuid=False) both

deleted_at When the record was moved to the trashbin. Null while the record is active. DateTime(timezone=True) import only

deleted_by UUID(as_uuid=False) import only

dependency_id The ID of the dependency record the execution is currently waiting for. UUID(as_uuid=False) export only

description A multiline description of what this record is and does. Text() both

end_status Enum('WAITING_DEPENDENCY', 'WAITING_LOCK', 'WAITING_RECORD', 'WAITING_ENGINE', 'WAITING_RESPONSE', 'AT_ENGINE', 'ENDED_SUCCESS', 'ENDED_ERROR', 'ENDED_CANCELLED', 'PAUSED', 'SCHEDULED', name='executionstatus') neither

ended_at The timestamp when the execution ended. DateTime(timezone=True) export only

expected_flow_version Optimistic-concurrency guard for approve-then-run flow integrity (type FLOW). Pass the flow record_version token you last read (e.g. the version you reviewed); if the flow was modified since, execution creation is refused with HTTP 409 so a silent post-review edit cannot slip through. Ignored when an inline script is supplied (that path runs the snapshot travelling with the request, so concurrent edits are deliberately irrelevant). Omit to keep the existing no-guard behaviour. Input only. Text() neither

flow_id The ID of the flow record the execution is processing. UUID(as_uuid=False) both

flow_name The name of the flow. Only used for type FLOW String(length=128) neither

id UUID(as_uuid=False) neither

identity_type Enum('EXECUTION', 'LDAP_CONFIG', 'SCHEDULE', 'SYNC_CONFIG', 'USER', 'WEBHOOK', 'WORKSPACE', name='identitytype') neither

input_schema The schema of the data the execution accepts in input_value. JSONEncodedData() both

input_value The data which was passed to the execution. JSONEncodedData() both

input_value_size_bytes The size of the input value in bytes. BigInteger() export only

is_bundle_readonly Boolean() neither

is_deleted Whether the record has been permanently deleted. Records in the trashbin are indicated by deleted_at instead. Boolean() export only

is_hidden Boolean() neither

is_paused A flag to control if the execution should be processed. Boolean() export only

is_productive A flag to control if the execution runs in productive mode. Boolean() both

is_project_readonly Boolean() neither

is_queued If set, the execution is waiting to be processed. Boolean() export only

is_readonly A flag to control if the record can be modified. Allowed for BUNDLE, BUNDLE_REPOSITORY, CONNECTOR, CUSTOM_OBJECT, DASHBOARD, DEVOLUTIONS_CONFIG, FILE, FLOW, MCP_USER_TOOL, OBJECT_TEMPLATE, PLUGIN, PROJECT, PYTHON_MODULE, ROLE, SCHEDULE, SCHEDULER, SCHEMA, SECRET, SETTING, SYNC_CONFIG, TAG, VAULT_CONFIG, WEBHOOK, WRAPPER. Not allowed for EXECUTION, LDAP_CONFIG, MESSAGE, ORGANIZATION, PROCESS, USER, WORKSPACE Boolean() both

is_remote_processed A flag to control if the execution is processed in a remote workspace. Boolean() export only

last_activity The last time the identity was active in Cloudomation. Output only. DateTime(timezone=True) neither

last_command The flow_api command which is currently executed. String(length=128) export only

last_sync Timestamp of when the execution was last synced with a workspace sync_config. DateTime(timezone=True) export only

lineno The line number of the script which is currently executed. SmallInteger() export only

location_inheritance

Controls the project/bundle association of a newly created record when no project_id/project_name/bundle_id/bundle_name is specified in the create request. Depending on the record type different options are available:

location_inheritance	record types	description
`created_by`	all	use the same location as the identity which creates the record
`default`	all	use the "Default project" where applicable or "Workspace" for record types which cannot be in a project
`wrapped_resource`	execution	use the same location as the innermost resource which is wrapped
`resource`	execution	use the same location as the resource on which the execution is based on

Notes:

If a project_id/project_name/bundle_id/bundle_name is specified in the request, it always takes precedence.
The options wrapped_resource and resource only work with executions which are based on a resource, not with ad-hoc connections or executions of type "SCRIPT". If used with ad-hoc connections or "SCRIPT" it will fall back to default.
If the record being created cannot be associated with the specified location, the fallback to default will be used.
If location_inheritance is unset, the default value used depends on the type of identity which creates the record as well as the type of record which is created. If an execution is created by any identity which is not an execution, it defaults to wrapped_resource. In all other cases it defaults to created_by.

String(length=128)

import only

lock_priority Optional acquire priority for a setting lock (operator-task-extend-lock-functionality, Phase 2). Stamped from the acquire() priority argument when the execution starts waiting; Engine.wait_lock grants a freed lock to the highest-priority waiter, tie-broken by lock_wait_seq (best-effort FIFO). Higher value = acquired first. NULL sorts lowest (default), so a lock with no priorities reduces exactly to best-effort FIFO. NULL when not waiting on a lock or acquired without a priority. Integer() neither

lock_wait_seq Monotonic arrival counter (nextval of the lock_wait_seq sequence) stamped when the execution starts waiting to acquire a setting lock; Engine.wait_lock grants a freed lock to the smallest-seq waiter for best-effort FIFO fairness. NULL when not waiting on a lock. BigInteger() neither

max_runtime_minutes The maximum runtime of the execution in minutes. If set to 0, there is no limit. Integer() both

mcp_user_tool_id The associated MCP user tool for this execution UUID(as_uuid=False) both

message A status message. Text() export only

message_id The ID of the message record the execution is currently awaiting a response from. UUID(as_uuid=False) export only

modified_at DateTime(timezone=True) neither

modified_by UUID(as_uuid=False) neither

name The name of this record. Must be unique across a workspace. String(length=255) both

operation_id UUID(as_uuid=False) neither

organization_id UUID(as_uuid=False) export only

origin_flow_id The id of the nearest ancestor flow that spawned this execution (the "calling flow"). Resolved once at creation from the parent execution's flow_id/origin_flow_id, so it is O(1). Used to correlate runtime statistics to the call site. NULL for top-level executions. UUID(as_uuid=False) export only

origin_lineno The source line in the calling flow at which the call chain that spawned this execution started (the this.connect/flow/script/child call site). Resolved O(1) at creation together with origin_flow_id: the parent lineno when the parent is a real flow, else the inherited origin_lineno of the parent so wrappers/scripts stay transparent to runtime-stats correlation. NULL for top-level executions. SmallInteger() export only

orphan If set, the child execution will be disconnected from the execution tree Boolean() neither

output_schema The schema of the data the execution produces in output_value. JSONEncodedData() both

output_value The data returned by the execution JSONEncodedData() export only

output_value_size_bytes The size of the output value in bytes. BigInteger() export only

parent_execution_id The ID of the execution which created this execution. UUID(as_uuid=False) export only

pending_delete_commit Boolean() neither

pending_delete_identity_id UUID(as_uuid=False) neither

pending_operation_id UUID(as_uuid=False) neither

pending_restore_commit Boolean() neither

pending_restore_identity_id UUID(as_uuid=False) neither

pending_trash_commit Boolean() neither

pending_trash_identity_id UUID(as_uuid=False) neither

plugin_action_id Reference to the plugin_action this execution was started by. UUID(as_uuid=False) both

priority The priority of the execution. Higher priority executions will be processed first. Executions with the same priority are processed in the order they were created. SmallInteger() both

process_id UUID(as_uuid=False) export only

project_id Reference to the project this record is associated with. Allowed for CONNECTOR, CUSTOM_OBJECT, DASHBOARD, DEVOLUTIONS_CONFIG, EXECUTION, FILE, FLOW, MCP_USER_TOOL, MESSAGE, OBJECT_TEMPLATE, PLUGIN, PYTHON_MODULE, SCHEDULE, SCHEDULER, SCHEMA, SECRET, SETTING, SYNC_CONFIG, TAG, VAULT_CONFIG, WEBHOOK, WRAPPER. Not allowed for BUNDLE, BUNDLE_REPOSITORY, LDAP_CONFIG, ORGANIZATION, PROCESS, PROJECT, ROLE, USER, WORKSPACE. Mutually exclusive with bundle_id. UUID(as_uuid=False) both

project_name The name of the project. Will look up the project and set project_id. String(length=128) import only

provisioning_type If this execution was created by the on_create, on_update, or on_delete hook of an object template. String(length=128) export only

record_metadata Record metadata to attach in the SAME transaction as the record create/update, as a mapping of {key: data} (key: string up to 128 chars, data: arbitrary JSON). Each entry is written as a record_metadata subrecord, upserted by (record_id, key). This removes the create-then-add race: without it a flow/API must first create the record and then add its metadata in a second call, leaving a window where the record exists un-marked (and may already be committed to git before the marker lands). Passing record_metadata inline makes the record and its metadata atomic. Accepted (and applied) on both create and update. On read the metadata is exposed via the record_metadata subrecord list, not this field. JSONEncodedData() neither

record_type Enum('EXECUTION', 'MESSAGE', 'PROCESS', 'BUNDLE', 'BUNDLE_REPOSITORY', 'CONNECTOR', 'CUSTOM_OBJECT', 'DASHBOARD', 'DEVOLUTIONS_CONFIG', 'FILE', 'FLOW', 'LDAP_CONFIG', 'MCP_USER_TOOL', 'OBJECT_TEMPLATE', 'ORGANIZATION', 'PLUGIN', 'PROJECT', 'PYTHON_MODULE', 'ROLE', 'SCHEDULE', 'SCHEDULER', 'SCHEMA', 'SECRET', 'SETTING', 'SYNC_CONFIG', 'TAG', 'USER', 'VAULT_CONFIG', 'WEBHOOK', 'WORKSPACE', 'WRAPPER', name='recordtype') neither

record_version Optimistic-concurrency version token. Read-only, computed on read from modified_at at full (microsecond) precision (request it by name; not in the default summary). Pass the value you last read back as expected_version on update to guard against lost updates: if the record changed meanwhile the write is rejected with HTTP 409 (StaleWrite) carrying the current record and token. Not stored; not exported. Text() neither

recurse_pending_delete Boolean() neither

recurse_pending_restore Boolean() neither

recurse_pending_trash Boolean() neither

remote_workspace_name The name of the workspace where this execution is remotely running String(length=128) neither

repository_path The path to the storing repository, relative to GIT_REPOSITORY_SAVE_PATH String(length=256) neither

resolved_icon The record's resolved display icon: the concrete subtype's own icon column (dashboard/project/bundle/object_template/workspace/user/plugin_action) when set, otherwise a per-record-type default glyph (a curated mdi:<key> reference) owned by the backend, so every consumer -- REST/MCP/agents, not only the Console -- gets a sensible icon for an iconless record. NULL only for record types that have no default. Read-only, computed on read (request it by name; not in the default summary); not stored, not exported. Lets the polymorphic base record expose a per-record icon (sidebar quick-access, quick-search, generic record lists) without shadowing the subtype icon columns. Named resolved_icon, not icon, for the same reason record_version is not version: a base virtual column named icon would override the subtype icon columns in their own all_columns (ancestors overwrite descendants), turning them read-only and non-exporting. Text() neither

roles The roles this identity has. JSONEncodedData() import only

runtime_key Optional caller-supplied runtime-statistics bucket key (e.g. this.flow(..., runtime_key="markdown_summary")). Groups this execution's runtime samples in a GLOBAL namespace shared across every flow that passes the same key, so a reusable flow called the same way from many call sites aggregates into one clean estimate. Overrides the automatic parent-flow/call-site tiers when its bucket is hot. Inherited transparently through wrapper/script parents; not inherited across a real-flow boundary. NULL means the automatic tiers apply (default). String(length=256) export only

savepoint_retention Controls which savepoints are retained during execution (e.g. LAST, ALL) Enum('ALL', 'LAST', name='savepointretention') both

schedule_id The associated schedule for this execution UUID(as_uuid=False) both

schedule_name The name of the schedule. Only used for type SCHEDULE String(length=128) neither

scheduled_for The timestamp when the execution will continue. This field is also used for timeouts. DateTime(timezone=True) export only

schema_version String(length=128) both

script The script the execution is processing. Text() both

secret_outputs An optional list of secret_outputs selector paths (see common.secret_paths) into this execution's output_value. At output finalization the engine wraps the selected leaf(s) into a Secret so they are masked everywhere the output is serialized (logs, output_value column, live view, records, git-sync). Set at child creation via connect()/script()/flow().run() secret_outputs=; NULL means no output masking. JSONEncodedData() both

setting_id The ID of the setting record the execution is currently trying to acquire a lock from. UUID(as_uuid=False) export only

started_at The timestamp when the execution started. DateTime(timezone=True) export only

status The status of the execution. Output only. Enum('WAITING_DEPENDENCY', 'WAITING_LOCK', 'WAITING_RECORD', 'WAITING_ENGINE', 'WAITING_RESPONSE', 'AT_ENGINE', 'ENDED_SUCCESS', 'ENDED_ERROR', 'ENDED_CANCELLED', 'PAUSED', 'SCHEDULED', name='executionstatus') export only

sync_config_id The ID of the workspace sync_config the execution is associated with. UUID(as_uuid=False) both

top_execution_id The ID of the top-most execution which created this execution. UUID(as_uuid=False) export only

track_in_git Should new records automatically be tracked in git. Boolean() both

type The type of the execution. Enum('FLOW', 'SCRIPT', 'CONNECTION', 'SCHEDULE', 'WRAPPER', name='executiontype') both

unwrapped If set, static wrappers which are registered on the flow are not used. Dynamically added wrappers and wrappers which are explicitly added using the wrappers field are still used. Boolean() neither

use_breakpoints A flag to control if the execution will stop at breakpoints. Boolean() both

use_only_explicit_breakpoints Should only explicitly defined breakpoints be used, this disables the step-through feature. Boolean() both

wait_record_id The ID of the record the execution is waiting for a trash, restore, or delete operation to complete. UUID(as_uuid=False) neither

wait_record_notify The notify channel prefix (record_deleted, record_restored, or record_trashed) for wait_record_id. Text() neither

webhook_id The associated webhook for this execution UUID(as_uuid=False) both

workspace_id UUID(as_uuid=False) export only

wrapper_id The ID of the wrapper record the execution is processing. UUID(as_uuid=False) both

wrapper_name The name of the wrapper. Only used for type WRAPPER String(length=128) neither

wrappers A list of dynamic wrappers to apply around the execution. Provide a list of objects in the form [{"wrapper_id": ..., "params": {...}}]. JSONEncodedData() neither