-
released this
2026-02-02 20:31:33 -05:00 | 197 commits to main since this release📅 Originally published on GitHub: Tue, 03 Feb 2026 01:31:40 GMT
🏷️ Git tag created: Tue, 03 Feb 2026 01:31:33 GMTFixed
- Experiment Run Migration: Fixed
trace_id does not match first part of dotted_ordervalidation errors- Root runs now correctly set
trace_id = run_id(API requirement) - Child runs properly look up
trace_idfrom mapping set by root run - Added
_regenerate_dotted_order()to remap UUIDs in dotted_order field - Runs sorted by dotted_order to ensure parents processed before children
- Generate new UUIDs for destination runs to avoid conflicts
- Remove None values from run payloads to avoid 422 errors
- Root runs now correctly set
Improved
- API Error Messages: Enhanced error detail extraction to handle validation errors and various error response formats
Added
- Dev Dependencies: Added pytest as dev dependency
Downloads
- Experiment Run Migration: Fixed
-
released this
2026-02-02 18:33:33 -05:00 | 200 commits to main since this release📅 Originally published on GitHub: Mon, 02 Feb 2026 23:33:50 GMT
🏷️ Git tag created: Mon, 02 Feb 2026 23:33:33 GMT[0.0.41] - 2026-02-02
Added
- Production Readiness Audit Fixes: Comprehensive error handling and robustness improvements
- New Exception Classes:
AuthenticationError(401/403) andConflictError(409) for explicit HTTP status handling - Batch Error Tracking:
BatchResultandBatchItemResultclasses to track individual item success/failure in batch operations - Thread Safety: Thread locks for shared state access during parallel migrations
- Attachment Validation: Size limits (100MB max) and content-type validation for attachment downloads
- URL Validation: Configuration validates URL format (must include scheme)
- CLI Input Validation:
--batch-size(1-1000) and--workers(1-10) now use IntRange constraints - Environment Variable Documentation: CLI help text now shows environment variable names
Fixed
- Silent Data Loss: Batch operations now track which items failed instead of returning
[None] - Authentication Errors: 401/403 errors now raise
AuthenticationErrorwith clear guidance - Conflict Handling: 409 errors now raise
ConflictErrorfor proper duplicate handling - Bare
except:Clauses: Replaced 3 bare except clauses with specific exception types - Rate Limiting: Added
Retry-Afterheader support and maximum backoff cap (60 seconds) - Network Errors: Now handles
Timeout,ReadTimeout, andsocket.timeoutin addition toConnectionError - Response Validation: All API responses now validated before accessing fields
- Environment Parsing: Safe parsing of numeric environment variables with try-catch
Changed
- API Client:
test_connection()now returnsTuple[bool, Optional[str]](success, error_message) - Orchestrator:
test_connections_detailed()now returns 4-tuple with error messages - Logging: Filtered field warnings in rules migration changed from "info" to "warning" level
- Chart Migration: Added verbose logging when charts are filtered out
Security
- Attachment Downloads: Enforces size limits during streaming to prevent memory exhaustion
- Content-Type Validation: Validates attachment MIME types before download
Downloads
-
released this
2026-02-02 12:25:08 -05:00 | 204 commits to main since this release📅 Originally published on GitHub: Mon, 02 Feb 2026 17:25:31 GMT
🏷️ Git tag created: Mon, 02 Feb 2026 17:25:08 GMT[0.0.4] - 2025-02-02
Added
- Feedback Migration: New
FeedbackMigratorclass to migrate feedback records for experiment runs- Automatically migrates feedback when using
--include-experimentswith datasets - Maps run IDs correctly between source and destination
- Automatically migrates feedback when using
- CLAUDE.md: Development guidance file for Claude Code
Fixed
- Multi-experiment Run Migration: Fixed bug where only the first experiment's runs were migrated
- Now queries runs per-experiment instead of all at once (LangSmith API only processes first session ID in list)
- Added detailed logging for run migration progress
- SSL Option: Added
@ssl_optiondecorator to all CLI commands for consistent SSL handling
Changed
- README.md: Simplified from ~430 lines to ~170 lines for better readability
- Kept full Limitations section for transparency
- Condensed installation, configuration, and usage sections
- Experiment Prompt: CLI now prompts whether to include experiments when running
datasetscommand
Downloads
- Feedback Migration: New
-
released this
2025-12-10 17:37:18 -05:00 | 206 commits to main since this release📅 Originally published on GitHub: Wed, 10 Dec 2025 22:37:44 GMT
🏷️ Git tag created: Wed, 10 Dec 2025 22:37:18 GMTRelease 0.0.3
Downloads
-
released this
2025-12-08 20:47:17 -05:00 | 207 commits to main since this release📅 Originally published on GitHub: Tue, 09 Dec 2025 01:47:36 GMT
🏷️ Git tag created: Tue, 09 Dec 2025 01:47:17 GMT[0.0.2] - 2025-12-08
Added
- Rules Migration: Full support for v3+ evaluator rules
- Automatic project and dataset ID mapping
- Support for LLM evaluators with hub prompts
- Support for code evaluators
--create-enabledflag to control rule state on creation- Rules created as disabled by default to bypass secrets validation
- Prompt Migration: Enhanced prompt migration with model configuration
- Proper parent commit handling for existing prompts
include_model=truesupport for evaluator prompts- 409 conflict handling for up-to-date prompts
- Enhanced Error Handling:
- Detailed error messages for evaluator validation failures
- Better guidance for missing secrets/API keys
- Improved 409 conflict resolution
Changed
- Rules are now created as disabled by default to bypass API key/secrets validation
- Use
--create-enabledflag to create rules as enabled - Helpful post-migration message explaining how to enable rules
- Use
- Improved prompt commit handling with automatic parent commit detection
- Better verbose logging for debugging migration issues
Fixed
- Fixed "RunnableSequence must have at least 2 steps" error for v3+ evaluators
- Fixed parent commit validation failures when updating existing prompts
- Fixed 409 errors being incorrectly reported as failures when prompts are up-to-date
Downloads
- Rules Migration: Full support for v3+ evaluator rules
-
released this
2025-12-08 11:42:44 -05:00 | 208 commits to main since this release📅 Originally published on GitHub: Mon, 08 Dec 2025 16:43:05 GMT
🏷️ Git tag created: Mon, 08 Dec 2025 16:42:44 GMT[0.0.1] - 2025-12-08
Added
- Initial release of LangSmith Data Migration Tool
- Dataset migration with examples and attachments support
- Experiment migration alongside datasets
- Annotation queue migration with configurations
- Project rules migration with automatic ID mapping
- Prompt migration with version history support
- Chart and dashboard migration with automatic section creation
- Interactive CLI with user-friendly selection UI
- Connection testing functionality
- All-in-one migration wizard (
migrate-allcommand) - SSL certificate verification options for self-hosted instances
- Batch processing and concurrent workers for performance
- Dry-run mode for testing migrations
- Comprehensive error handling and retry logic
- Progress tracking and verbose output options
- Session management with resume capability
Features
- Dataset Migration: Migrate datasets with examples and file attachments
- Experiment Migration: Transfer experiments and their runs
- Annotation Queues: Migrate queue configurations and settings
- Project Rules: Copy automation rules between instances with automatic project mapping
- Prompts: Migrate prompts and their version history
- Charts: Transfer monitoring charts and dashboards
- Interactive TUI: User-friendly command-line interface with search and selection
- Automatic Mapping: Smart ID mapping for projects, datasets, and resources
Documentation
- Comprehensive README with installation instructions
- Usage examples for all commands
- Troubleshooting guide
- Configuration documentation
- API reference for core classes
Downloads
mirror of
https://github.com/langchain-ai/langsmith-data-migration-tool.git
synced 2026-07-19 13:44:53 -04:00