v0.1.11

[0.1.11] - 2025-02-05

Fixed

• Fixed timestamp handling in search_memories function to properly handle timezone conversions
• Added explicit timezone casting with AT TIME ZONE 'UTC' for all temporal comparisons
• Improved handling of _is_null operator for both created_at and expires_at filters

v0.1.10

[0.1.10] - 2025-02-05

Added

• Memory expiration support:
  • New expires_at field for memories to set expiration timestamps
  • Optional expires_at parameter in remember() function
  • Memories without expiration default to NULL (never expire)
• Enhanced memory search capabilities:
  • Added support for temporal filtering with standard Hasura operators (_gt, _gte, _lt, _lte, _eq)
  • Filter memories by created_at using created_at_filter
  • Filter memories by expires_at using expires_at_filter
  • All temporal filters are optional and can be combined with existing filters

Changed

• Updated search_memories function to support temporal filtering
• Extended memory schema with expires_at field
• Improved GraphQL queries to include expires_at in responses
• Updated Memory class to include optional expires_at field

v0.1.9

[0.1.9] - 2025-02-05

Added

• Automatic content chunking for large memories
  • Content exceeding 8192 tokens is automatically split into chunks
  • Chunks are linked with follows_up relationships
  • Each chunk includes metadata about its position and total chunks
  • Seamless handling of both single and multi-chunk memories
• Enhanced metadata filtering support in search_memories:
  • Exact value matching for numeric fields in metadata
  • Nested object matching in metadata
  • Array containment operations with _contains operator
  • Full support for deep nested metadata structures
• Improved IDE support and type hints:
  • Added py.typed marker for better type checking
  • Enhanced docstrings with examples and detailed descriptions
  • Added proper type hints for all public APIs
  • Better package metadata for development tools

Changed

• remember function now returns either a single Memory or List[Memory] depending on chunking
• Enhanced memory metadata to support chunk information and relationships
• Enhanced package structure for better IDE integration
• Improved import organization in core modules

v0.1.7

[0.1.7] - 2025-02-04

Added

• Human-readable slugs for agent identification
  • Slugs must match pattern ^[a-z][a-z0-9_-]*[a-z0-9]$ (e.g., "my-agent-1")
  • New get_agent_by_slug function to retrieve agents by slug
  • Optional slug parameter in register_agent function
  • CLI commands now support agent lookup by slug

Changed

• Agent registration now returns existing agent if slug is already in use
• CLI commands display agent slugs in output messages for easier reference
• Improved error messages for invalid slug formats

Fixed

• Better error handling when using invalid slugs in API calls

v0.1.6

Further improved semantic expansion for faster retrieval.

v0.1.5 Better Semantic Retrieval + Auto Expansion

Introducing automatic concept expansion for retrieval.

v0.1.4

Fixed an issue with semantic retrieval.

v0.1.3

Fixes an issue with similarity search.

v0.1.2

Fixed some issues regarding similarity function tracking.

v0.1.1

Contains a few bug fixes.