LangGraph Server changelog

LangGraph Server is an API platform for creating and managing agent-based applications. It provides built-in persistence, a task queue, and supports deploying, configuring, and running assistants (agentic workflows) at scale. This changelog documents all notable updates, features, and fixes to LangGraph Server releases.

v0.4.7

• Stopped the inclusion of x-api-key to enhance security.

v0.4.6

• Fixed a race condition when joining streams, preventing duplicate start events.

v0.4.5

• Ensured the checkpointer starts and stops correctly before and after the queue to improve shutdown and startup efficiency.
• Resolved an issue where workers were being prematurely cancelled when the queue was cancelled.
• Prevented queue termination by adding a fallback for cases when Redis fails to wake a worker.

v0.4.4

• Set the custom auth thread_id to None for stateless runs to prevent conflicts.
• Improved Redis signaling in the Go runtime by adding a wakeup worker and Redis lock implementation, and updated sweep logic.

v0.4.3

• Added stream mode to thread stream for improved data processing.
• Added a durability parameter to runs for improved data persistence.

v0.4.2

• Ensured pubsub is initialized before creating a run to prevent errors from missing messages.

v0.4.0

• Emitted attempt messages correctly within the thread stream.
• Reduced cluster conflicts by using only the thread ID for hashing in cluster mapping, prioritizing efficiency with stream_thread_cache.
• Introduced a stream endpoint for threads to track all outputs across sequentially executed runs.
• Made the filter query builder in PostgreSQL more robust against malformed expressions and improved validation to prevent potential security risks.

v0.3.4

• Added custom Prometheus metrics for Redis/PG connection pools and switched the queue server to Uvicorn/Starlette for improved monitoring.
• Restored Wolfi image build by correcting shell command formatting and added a Makefile target for testing with nginx.

v0.3.3

• Added timeouts to specific Redis calls to prevent workers from being left active.
• Updated the Golang runtime and added pytest skips for unsupported functionalities, including initial support for passing store to node and message streaming.
• Introduced a reverse proxy setup for serving combined Python and Node.js graphs, with nginx handling server routing, to facilitate a Postgres/Redis backend for the Node.js API server.

v0.3.1

• Added a statement timeout to the pool to prevent long-running queries.

v0.3.0

• Set a default 15-minute statement timeout and implemented monitoring for long-running queries to ensure system efficiency.
• Stop propagating run configurable values to the thread configuration, because this can cause issues on subsequent runs if you are specifying a checkpoint_id. This is a slight breaking change in behavior, since the thread value will no longer automatically reflect the unioned configuration of the most recent run. We believe this behavior is more intuitive, however.
• Enhanced compatibility with older worker versions by handling event data in channel names within ops.py.

v0.2.137

• Fixed an unbound local error and improved logging for thread interruptions or errors, along with type updates.

v0.2.136

• Added enhanced logging to aid in debugging metaview issues.
• Upgraded executor and runtime to the latest version for improved performance and stability.

v0.2.135

• Ensured async coroutines are properly awaited to prevent potential runtime errors.

v0.2.134

• Enhanced search functionality to improve performance by allowing users to select specific columns for query results.

v0.2.133

• Added count endpoints for crons, threads, and assistants to enhance data tracking (#1132).
• Improved SSH functionality for better reliability and stability.
• Updated @langchain/langgraph-api to version 0.0.59 to fix an invalid state schema issue.

v0.2.132

• Added Go language images to enhance project compatibility and functionality.
• Printed internal PIDs for JS workers to facilitate process inspection via SIGUSR1 signal.
• Resolved a run_pkey error that occurred when attempting to insert duplicate runs.
• Added ty run command and switched to using uuid7 for generating run IDs.
• Implemented the initial Golang runtime to expand language support.

v0.2.131

• Added support for object agent spec with descriptions in JS.

v0.2.130

• Added a feature flag (FF_RICH_THREADS=false) to disable thread updates on run creation, reducing lock contention and simplifying thread status handling.
• Utilized existing connections for aput and apwrite operations to improve performance.
• Improved error handling for decoding issues to enhance data processing reliability.
• Excluded headers from logs to improve security while maintaining runtime functionality.
• Fixed an error that prevented mapping slots to a single node.
• Added debug logs to track node execution in JS deployments for improved issue diagnosis.
• Changed the default multitask strategy to enqueue, improving throughput by eliminating the need to fetch inflight runs during new run insertions.
• Optimized database operations for Runs.next and Runs.sweep to reduce redundant queries and improve efficiency.
• Improved run creation speed by skipping unnecessary inflight runs queries.

v0.2.129

• Stopped passing internal LGP fields to context to prevent breaking type checks.
• Exposed content-location headers to ensure correct resumability behavior in the API.

v0.2.128

• Ensured synchronized updates between configurable and context in assistants, preventing setup errors and supporting smoother version transitions.

v0.2.127

• Excluded unrequested stream modes from the resumable stream to optimize functionality.

v0.2.126

• Made access logger headers configurable to enhance logging flexibility.
• Debounced the Runs.stats function to reduce the frequency of expensive calls and improve performance.
• Introduced debouncing for sweepers to enhance performance and efficiency (#1147).
• Acquired a lock for TTL sweeping to prevent database spamming during scale-out operations.

v0.2.125

• Updated tracing context replicas to use the new format, ensuring compatibility.

v0.2.123

• Added an entrypoint to the queue replica for improved deployment management.

v0.2.122

• Utilized persisted interrupt status in join to ensure correct handling of user’s interrupt state after completion.

v0.2.121

• Consolidated events to a single channel to prevent race conditions and optimize startup performance.
• Ensured custom lifespans are invoked on queue workers for proper setup, and added tests.

v0.2.120

• Restored the original streaming behavior of runs, ensuring consistent inclusion of interrupt events based on stream_mode settings.
• Optimized Runs.next query to reduce average execution time from ~14.43ms to ~2.42ms, improving performance.
• Added support for stream mode “tasks” and “checkpoints”, normalized the UI namespace, and upgraded @langchain/langgraph-api for enhanced functionality.

v0.2.117

• Added a composite index on threads for faster searches with owner-based authentication and updated the default sort order to updated_at for improved query performance.

v0.2.116

• Reduced the default number of history checkpoints from 10 to 1 to optimize performance.

v0.2.115

• Optimized cache re-use to enhance application performance and efficiency.

v0.2.113

• Improved thread search pagination by updating response headers with X-Pagination-Total and X-Pagination-Next for better navigation.

v0.2.112

• Ensured sync logging methods are awaited and added a linter to prevent future occurrences.
• Fixed an issue where JavaScript tasks were not being populated correctly for JS graphs.

v0.2.111

• Fixed JS graph streaming failure by starting the heartbeat as soon as the connection opens.

v0.2.110

• Added interrupts as default values for join operations while preserving stream behavior.

v0.2.109

• Fixed an issue where config schema was missing when config_type was not set, ensuring more reliable configurations.

v0.2.108

• Prepared for Langgraph v0.6 compatibility with new context API support and bug fixes.

v0.2.107

• Implemented caching for authentication processes to enhance performance and efficiency.
• Optimized database performance by merging count and select queries.

v0.2.106

• Made log streams resumable, enhancing reliability and improving user experience when reconnecting.

v0.2.105

• Added a heapdump endpoint to save memory heap information to a file.

v0.2.103

• Used the correct metadata endpoint to resolve issues with data retrieval.

v0.2.102

• Captured interrupt events in the wait method to preserve previous behavior from langgraph 0.5.0.
• Added support for SDK structlog in the JavaScript environment for enhanced logging capabilities.

v0.2.101

• Corrected the metadata endpoint for self-hosted deployments.

v0.2.99

• Improved license check by adding an in-memory cache and handling Redis connection errors more effectively.
• Reloaded assistants to preserve manually created ones while discarding those removed from the configuration file.
• Reverted changes to ensure the UI namespace for gen UI is a valid JavaScript property name.
• Ensured that the UI namespace for generated UI is a valid JavaScript property name, improving API compliance.
• Enhanced error handling to return a 422 status code for unprocessable entity requests.

v0.2.98

• Added context to langgraph nodes to improve log filtering and trace visibility.

v0.2.97

• Improved interoperability with the ckpt ingestion worker on the main loop to prevent task scheduling issues.
• Delayed queue worker startup until after migrations are completed to prevent premature execution.
• Enhanced thread state error handling by adding specific metadata and improved response codes for better clarity when state updates fail during creation.
• Exposed the interrupt ID when retrieving the thread state to improve API transparency.

v0.2.96

• Added a fallback mechanism for configurable header patterns to handle exclude/include settings more effectively.

v0.2.95

• Avoided setting the future if it is already done to prevent redundant operations.
• Resolved compatibility errors in CI by switching from typing.TypedDict to typing_extensions.TypedDict for Python versions below 3.12.

v0.2.94

• Improved performance by omitting pending sends for langgraph versions 0.5 and above.
• Improved server startup logs to provide clearer warnings when the DD_API_KEY environment variable is set.

v0.2.93

• Removed the GIN index for run metadata to improve performance.

v0.2.92

• Enabled copying functionality for blobs and checkpoints, improving data management flexibility.

v0.2.91

• Reduced writes to the checkpoint_blobs table by inlining small values (null, numeric, str, etc.). This means we don’t need to store extra values for channels that haven’t been updated.

v0.2.90

• Improve checkpoint writes via node-local background queueing.

v0.2.89

• Decoupled checkpoint writing from thread/run state by removing foreign keys and updated logger to prevent timeout-related failures.

v0.2.88

• Removed the foreign key constraint for thread in the run table to simplify database schema.

v0.2.87

• Added more detailed logs for Redis worker signaling to improve debugging.

v0.2.86

• Honored tool descriptions in the /mcp endpoint to align with expected functionality.

v0.2.85

• Added support for the on_disconnect field to runs/wait and included disconnect logs for better debugging.

v0.2.84

• Removed unnecessary status updates to streamline thread handling and updated version to 0.2.84.

v0.2.83

• Reduced the default time-to-live for resumable streams to 2 minutes.
• Enhanced data submission logic to send data to both Beacon and LangSmith instance based on license configuration.
• Enabled submission of self-hosted data to a Langsmith instance when the endpoint is configured.

v0.2.82

• Addressed a race condition in background runs by implementing a lock using join, ensuring reliable execution across CTEs.

v0.2.81

• Optimized run streams by reducing initial wait time to improve responsiveness for older or non-existent runs.

v0.2.80

• Corrected parameter passing in the logger.ainfo() API call to resolve a TypeError.

v0.2.79

• Fixed a JsonDecodeError in checkpointing with remote graph by correcting JSON serialization to handle trailing slashes properly.
• Introduced a configuration flag to disable webhooks globally across all routes.

v0.2.78

• Added timeout retries to webhook calls to improve reliability.
• Added HTTP request metrics, including a request count and latency histogram, for enhanced monitoring capabilities.

v0.2.77

• Added HTTP metrics to improve performance monitoring.
• Changed the Redis cache delimiter to reduce conflicts with subgraph message names and updated caching behavior.

v0.2.76

• Updated Redis cache delimiter to prevent conflicts with subgraph messages.

v0.2.74

• Scheduled webhooks in an isolated loop to ensure thread-safe operations and prevent errors with PYTHONASYNCIODEBUG=1.

v0.2.73

• Fixed an infinite frame loop issue and removed the dict_parser due to structlog’s unexpected behavior.
• Throw a 409 error on deadlock occurrence during run cancellations to handle lock conflicts gracefully.

v0.2.72

• Ensured compatibility with future langgraph versions.
• Implemented a 409 response status to handle deadlock issues during cancellation.

v0.2.71

• Improved logging for better clarity and detail regarding log types.

v0.2.70

• Improved error handling to better distinguish and log TimeoutErrors caused by users from internal run timeouts.

v0.2.69

• Added sorting and pagination to the crons API and updated schema definitions for improved accuracy.

v0.2.66

• Fixed a 404 error when creating multiple runs with the same thread_id using on_not_exist="create".

v0.2.65

• Ensured that only fields from assistant_versions are returned when necessary.
• Ensured consistent data types for in-memory and PostgreSQL users, improving internal authentication handling.

v0.2.64

• Added descriptions to version entries for better clarity.

v0.2.62

• Improved user handling for custom authentication in the JS Studio.
• Added Prometheus-format run statistics to the metrics endpoint for better monitoring.
• Added run statistics in Prometheus format to the metrics endpoint.

v0.2.61

• Set a maximum idle time for Redis connections to prevent unnecessary open connections.

v0.2.60

• Enhanced error logging to include traceback details for dictionary operations.
• Added a /metrics endpoint to expose queue worker metrics for monitoring.

v0.2.57

• Removed CancelledError from retriable exceptions to allow local interrupts while maintaining retriability for workers.
• Introduced middleware to gracefully shut down the server after completing in-flight requests upon receiving a SIGINT.
• Reduced metadata stored in checkpoint to only include necessary information.
• Improved error handling in join runs to return error details when present.

v0.2.56

• Improved application stability by adding a handler for SIGTERM signals.

v0.2.55

• Improved the handling of cancellations in the queue entrypoint.
• Improved cancellation handling in the queue entry point.

v0.2.54

• Enhanced error message for LuaLock timeout during license validation.
• Fixed the $contains filter in custom auth by requiring an explicit ::text cast and updated tests accordingly.
• Ensured project and tenant IDs are formatted as UUIDs for consistency.

v0.2.53

• Resolved a timing issue to ensure the queue starts only after the graph is registered.
• Improved performance by setting thread and run status in a single query and enhanced error handling during checkpoint writes.
• Reduced the default background grace period to 3 minutes.

v0.2.52

• Now logging expected graphs when one is omitted to improve traceability.
• Implemented a time-to-live (TTL) feature for resumable streams.
• Improved query efficiency and consistency by adding a unique index and optimizing row locking.

v0.2.51

• Handled CancelledError by marking tasks as ready to retry, improving error management in worker processes.
• Added LG API version and request ID to metadata and logs for better tracking.
• Added LG API version and request ID to metadata and logs to improve traceability.
• Improved database performance by creating indexes concurrently.
• Ensured postgres write is committed only after the Redis running marker is set to prevent race conditions.
• Enhanced query efficiency and reliability by adding a unique index on thread_id/running, optimizing row locks, and ensuring deterministic run selection.
• Resolved a race condition by ensuring Postgres updates only occur after the Redis running marker is set.

v0.2.46

• Introduced a new connection for each operation while preserving transaction characteristics in Threads state update() and bulk() commands.

v0.2.45

• Enhanced streaming feature by incorporating tracing contexts.
• Removed an unnecessary query from the Crons.search function.
• Resolved connection reuse issue when scheduling next run for multiple cron jobs.
• Removed an unnecessary query in the Crons.search function to improve efficiency.
• Resolved an issue with scheduling the next cron run by improving connection reuse.

v0.2.44

• Enhanced the worker logic to exit the pipeline before continuing when the Redis message limit is reached.
• Introduced a ceiling for Redis message size with an option to skip messages larger than 128 MB for improved performance.
• Ensured the pipeline always closes properly to prevent resource leaks.

v0.2.43

• Improved performance by omitting logs in metadata calls and ensuring output schema compliance in value streaming.
• Ensured the connection is properly closed after use.
• Aligned output format to strictly adhere to the specified schema.
• Stopped sending internal logs in metadata requests to improve privacy.

v0.2.42

• Added timestamps to track the start and end of a request’s run.
• Added tracer information to the configuration settings.
• Added support for streaming with tracing contexts.

v0.2.41

• Added locking mechanism to prevent errors in pipelined executions.