agentgit.checkpoints package

Submodules

agentgit.checkpoints.checkpoint module

Checkpoint model for the LangGraph rollback agent system.

Represents saved states that can be restored with full execution history preservation.

class agentgit.checkpoints.checkpoint.Checkpoint(id=None, internal_session_id=0, checkpoint_name=None, session_state=<factory>, conversation_history=<factory>, is_auto=False, created_at=None, metadata=<factory>, user_id=None, tool_invocations=<factory>)

Bases: object

Checkpoint model for saving and restoring LangGraph agent states.

A checkpoint captures the complete state of an internal session at a specific point in time, allowing non-destructive rollback to that state by creating a new internal session branch.

id

Unique identifier for the checkpoint.

internal_session_id

ID of the internal session this checkpoint was created from.

checkpoint_name

Optional user-friendly name for the checkpoint.

session_state

The agent’s session state at checkpoint time.

conversation_history

The conversation history at checkpoint time.

is_auto

Whether this checkpoint was created automatically.

created_at

When this checkpoint was created.

metadata

Additional metadata about the checkpoint.

user_id

Optional ID of the user who owns this checkpoint.

tool_invocations

History of tool invocations up to this checkpoint.

Example

>>> checkpoint = Checkpoint(
...     internal_session_id=1,
...     checkpoint_name="Before tool call",
...     session_state={"counter": 5},
...     is_auto=True
... )

checkpoint_name: Optional[str] = None

conversation_history: List[Dict[str, Any]]

created_at: Optional[datetime] = None

classmethod from_dict(data)

Create a Checkpoint from dictionary data.

Parameters:

data (dict) – Dictionary containing checkpoint data.

Return type:

Checkpoint

Returns:

Checkpoint instance.

classmethod from_internal_session(internal_session, checkpoint_name=None, is_auto=False, user_id=None, tool_invocations=None)

Create a checkpoint from an internal session.

Parameters:

• internal_session – The internal session to checkpoint.

• checkpoint_name (Optional[str]) – Optional name for the checkpoint.

• is_auto (bool) – Whether this is an automatic checkpoint.

• user_id (Optional[int]) – Optional user ID who owns this checkpoint.

• tool_invocations (Optional[List[Dict[str, Any]]]) – Optional tool invocation history.

Return type:

Checkpoint

Returns:

New Checkpoint instance with session data.

get_summary()

Get a summary of the checkpoint for display.

Return type:

str

Returns:

Human-readable summary of the checkpoint.

get_tool_track_position()

Get the tool track position from metadata.

Return type:

int

Returns:

The tool track position, or 0 if not set.

has_tool_invocations()

Check if this checkpoint has any tool invocations.

Return type:

bool

Returns:

True if there are tool invocations, False otherwise.

id: Optional[int] = None

internal_session_id: int = 0

is_auto: bool = False

metadata: Dict[str, Any]

session_state: Dict[str, Any]

to_dict()

Convert checkpoint to dictionary representation.

Return type:

dict

Returns:

Dictionary with checkpoint data.

tool_invocations: List[Dict[str, Any]]

user_id: Optional[int] = None

Module contents