fix: replace deprecated gemini-2.5-flash-lite-preview-06-17 with gemini-2.5-flash-lite
Updated all references to the deprecated Gemini model in:
- graphiti_core/llm_client/gemini_client.py
- graphiti_core/cross_encoder/gemini_reranker_client.py
- tests/llm_client/test_gemini_client.py
- README.md
This resolves 404 errors when using Gemini clients.
Fixes#1075🤖 Generated with [Claude Code](https://claude.ai/code)
Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
Co-authored-by: Daniel Chalef <danielchalef@users.noreply.github.com>
* Update default Anthropic model to claude-haiku-4-5-latest
- Add Claude 4.5 models to AnthropicModel type (claude-sonnet-4-5-latest, claude-sonnet-4-5-20250929, claude-haiku-4-5-latest)
- Change DEFAULT_MODEL from claude-3-7-sonnet-latest to claude-haiku-4-5-latest
- Update test assertions to reflect new default model
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Add Claude 4.5 models to max_tokens mapping
- Add claude-sonnet-4-5-latest, claude-sonnet-4-5-20250929, and claude-haiku-4-5-latest to ANTHROPIC_MODEL_MAX_TOKENS
- All Claude 4.5 models support 64K (65536) max output tokens
- Based on official Anthropic documentation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Update uv.lock dependencies
---------
Co-authored-by: Claude <noreply@anthropic.com>
* Add dynamic max_tokens configuration for Anthropic models
Implements model-specific max output token limits for AnthropicClient,
following the same pattern as GeminiClient. This replaces the previous
hardcoded min() cap that was preventing models from using their full
output capacity.
Changes:
- Added ANTHROPIC_MODEL_MAX_TOKENS mapping with limits for all supported
Claude models (ranging from 4K to 65K tokens)
- Implemented _get_max_tokens_for_model() to lookup model-specific limits
- Implemented _resolve_max_tokens() with clear precedence rules:
1. Explicit max_tokens parameter
2. Instance max_tokens from initialization
3. Model-specific limit from mapping
4. Default fallback (8192 tokens)
This allows edge_operations.py to request 16384 tokens for edge extraction
without being artificially capped, while ensuring cheaper models with lower
limits are still properly handled.
Resolves TODO in anthropic_client.py:207-208.
* Clarify that max_tokens mapping represents standard limits
Updated comments to explicitly state that ANTHROPIC_MODEL_MAX_TOKENS
represents standard limits without beta headers. This prevents confusion
about extended limits (e.g., Claude 3.7's 128K with beta header) which
are not currently implemented in this mapping.
* Add Azure OpenAI example with Neo4j
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Convert Azure OpenAI example to use uv
- Remove requirements.txt (uv uses pyproject.toml)
- Update README to use 'uv sync' and 'uv run'
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Update Azure OpenAI example to use gpt-4.1
- Change default deployment from gpt-4 to gpt-4.1
- Update README recommendations to prioritize gpt-4.1 models
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Remove model recommendations from Azure OpenAI example
Model recommendations quickly become outdated.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Add default Neo4j credentials to docker-compose
Set sensible defaults (neo4j/password) to prevent NEO4J_AUTH error
when .env file is not present.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Update Azure OpenAI documentation to use v1 API
- Simplified Azure OpenAI setup using AsyncOpenAI with v1 endpoint
- Updated main README with clearer Quick Start example
- Removed outdated API version configuration
- Updated example deployment to gpt-5-mini
- Added note about v1 API endpoint format
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Update LLMConfig to include both model and small_model
Both parameters are needed for proper LLM configuration.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Address PR review feedback
- Remove flawed validation check in azure_openai_neo4j.py
- Remove unused azure-identity dependency
- Update docstrings to reflect dual client support (AsyncAzureOpenAI and AsyncOpenAI)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
* Use OpenAI structured output API for response validation
Replace prompt-based schema injection with native json_schema response format. This improves token efficiency and reliability by having OpenAI enforce the schema directly instead of embedding it in the prompt message.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Add type ignore for response_format to fix pyright error
* Increase OpenAIGenericClient max_tokens to 16K and update docs
- Set default max_tokens to 16384 (16K) for OpenAIGenericClient to better support local models
- Add documentation note clarifying OpenAIGenericClient should be used for Ollama and LM Studio
- Previous default was 8192 (8K)
* Refactor max_tokens override to use constructor parameter pattern
- Add max_tokens parameter to __init__ with 16K default
- Override self.max_tokens after super().__init__() instead of mutating config
- Consistent with OpenAIBaseClient and AnthropicClient patterns
- Avoids unintended config mutation side effects
---------
Co-authored-by: Claude <noreply@anthropic.com>
* feat: MCP Server v1.0.0rc0 - Complete refactoring with modular architecture
This is a major refactoring of the MCP Server to support multiple providers
through a YAML-based configuration system with factory pattern implementation.
## Key Changes
### Architecture Improvements
- Modular configuration system with YAML-based settings
- Factory pattern for LLM, Embedder, and Database providers
- Support for multiple database backends (Neo4j, FalkorDB, KuzuDB)
- Clean separation of concerns with dedicated service modules
### Provider Support
- **LLM**: OpenAI, Anthropic, Gemini, Groq
- **Embedders**: OpenAI, Voyage, Gemini, Anthropic, Sentence Transformers
- **Databases**: Neo4j, FalkorDB, KuzuDB (new default)
- Azure OpenAI support with AD authentication
### Configuration
- YAML configuration with environment variable expansion
- CLI argument overrides for runtime configuration
- Multiple pre-configured Docker Compose setups
- Proper boolean handling in environment variables
### Testing & CI
- Comprehensive test suite with unit and integration tests
- GitHub Actions workflows for linting and testing
- Multi-database testing support
### Docker Support
- Updated Docker images with multi-stage builds
- Database-specific docker-compose configurations
- Persistent volume support for all databases
### Bug Fixes
- Fixed KuzuDB connectivity checks
- Corrected Docker command paths
- Improved error handling and logging
- Fixed boolean environment variable expansion
Co-authored-by: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01PmXwij9S976CQk798DJ4PH
* fix: Improve MCP server configuration and initialization
- Fix API key detection: Remove hardcoded OpenAI checks, let factories handle provider-specific validation
- Fix .env file loading: Search for .env in mcp_server directory first
- Change default transport to SSE for broader compatibility (was stdio)
- Add proper error handling with warnings for failed client initialization
- Model already defaults to gpt-4o as requested
These changes ensure the MCP server properly loads API keys from .env files
and creates the appropriate LLM/embedder clients based on configuration.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01PCGAWzQUbmh7hAKBvadbYN
* chore: Update default transport from SSE to HTTP
- Changed default transport to 'http' as SSE is deprecated
- Updated all configuration files to use HTTP transport
- Updated Docker compose commands to use HTTP transport
- Updated comments to reflect HTTP transport usage
This change ensures the MCP server uses the recommended HTTP transport
instead of the deprecated SSE transport.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01FErZjFG5iWrvbdUD2acQwQ
* chore: Update default OpenAI model to gpt-4o-mini
Changed the default LLM model from gpt-4o to gpt-4o-mini across all
configuration files for better cost efficiency while maintaining quality.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01FETp6u9mWAMjJAeT6WFgAf
* conductor-checkpoint-msg_01AJJ48RbkPaZ99G2GmE6HUi
* fix: Correct default OpenAI model to gpt-4.1
Changed the default LLM model from gpt-4o-mini to gpt-4.1 as requested.
This is the latest GPT-4 series model.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_013HP1MYHKZ5wdTHHxrpBaT9
* fix: Update hardcoded default model to gpt-4.1 and fix config path
- Changed hardcoded default in schema.py from gpt-4o to gpt-4.1
- Fixed default config path to look in config/config.yaml relative to mcp_server directory
- This ensures the server uses gpt-4.1 as the default model everywhere
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01EaN8GZtehm8LV3a7CdWJ8u
* feat: Add detailed server URL logging and improve access information
- Added comprehensive logging showing exact URLs to access the MCP server
- Display localhost instead of 0.0.0.0 for better usability
- Show MCP endpoint, transport type, and status endpoint information
- Added visual separators to make server info stand out in logs
This helps users understand exactly how to connect to the MCP server
and troubleshoot connection issues.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01SNpbaZMdxWbefo2zsLprcW
* fix: Correct MCP HTTP endpoint path from / to /mcp/
- Remove incorrect /status endpoint reference
- Update logging to show correct MCP endpoint at /mcp/
- Align with FastMCP documentation standards
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01417YVh3s6afJadN5AM5Ahk
* fix: Configure consistent logging format between uvicorn and MCP server
- Use simplified format matching uvicorn's default (LEVEL message)
- Remove timestamps from custom logger format
- Suppress verbose MCP and uvicorn access logs
- Improve readability of server startup output
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_014BF6Kzdy7qXc5AgC7eeVa5
* conductor-checkpoint-msg_01TscHXmijzkqcTJX5sGTYP8
* conductor-checkpoint-msg_01Q7VLFTJrtmpkaB7hfUzZLP
* fix: Improve test runner to load API keys from .env file
- Add dotenv loading support in test runner
- Fix duplicate os import issue
- Improve prerequisite checking with helpful hints
- Update error messages to guide users
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01NfviLNeAhFDA1G5841YKCS
* conductor-checkpoint-msg_015EewQhKbqAGSQkasWtRQjp
* fix: Fix all linting errors in test suite
- Replace bare except with except Exception
- Remove unused imports and variables
- Fix type hints to use modern syntax
- Apply ruff formatting for line length
- Ensure all tests pass linting checks
* conductor-checkpoint-msg_01RedNheKT4yWyXcM83o3Nmv
* fix: Use contextlib.suppress instead of try-except-pass (SIM105)
- Replace try-except-pass with contextlib.suppress in test_async_operations.py
- Replace try-except-pass with contextlib.suppress in test_fixtures.py
- Fixes ruff SIM105 linting errors
* conductor-checkpoint-msg_01GuBj69k2CsqgojBsGJ2zFT
* fix: Move README back to mcp_server root folder
The main README for the MCP server should be in the root of the mcp_server folder for better discoverability
* conductor-checkpoint-msg_01VsJQ3MgDxPwyb4ynvswZfb
* docs: Update README with comprehensive features and database options
- Add comprehensive features list including all supported databases, LLM providers, and transports
- Document Kuzu as the default database with explanation of its benefits and archived status
- Add detailed instructions for running with different databases (Kuzu, Neo4j, FalkorDB)
- Update transport references from SSE to HTTP (default transport)
- Add database-specific Docker Compose instructions
- Update MCP client configurations to use /mcp/ endpoint
- Clarify prerequisites to reflect optional nature of external databases
- Add detailed database configuration examples for all supported backends
* conductor-checkpoint-msg_018Z7AjbTkuDfhqdjB9iGbbD
* docs: Address README review comments
- Shorten Kuzu database description to be more concise
- Update Ollama model example to use 'gpt-oss:120b'
- Restore Azure OpenAI environment variables documentation
- Remove implementation details from Docker section (irrelevant to container users)
- Clarify mcp-remote supports both HTTP and SSE transports
Addresses review comments #1-7 on the PR
* conductor-checkpoint-msg_01QMeMEMe9rTVDgd8Ce5hmXp
* docs: Remove SSE transport reference from Claude Desktop section
Since the MCP server no longer supports SSE transport, removed the
mention of SSE from the mcp-remote description. The server only
uses HTTP transport.
Addresses review comment on line 514
* conductor-checkpoint-msg_01DNn76rvpx7rmTBwsUQd1De
* docs: Remove telemetry from features list
Telemetry is not a feature but a notice about data collection,
so it shouldn't be listed as a feature.
Addresses review comment on line 29
* conductor-checkpoint-msg_01Jcb8sm9bpqB9Ksz1W6YrSz
* feat: Update default embedding model to text-embedding-3-small
Replace outdated text-embedding-ada-002 with the newer, more efficient
text-embedding-3-small model as the default embedder. The new model
offers better performance and is more cost-effective.
Updated:
- config/config.yaml: Changed default model
- README.md: Updated documentation to reflect new default
* conductor-checkpoint-msg_016AXAH98nYKTj5WCueBubmA
* fix: Resolve database connection and episode processing errors
Fixed two critical runtime errors:
1. Database connection check for KuzuDB
- KuzuDB session.run() returns None, causing async iteration error
- Added special handling for KuzuDB (in-memory, no query needed)
- Other databases (Neo4j, FalkorDB) still perform connection test
2. Episode processing parameter error
- Changed 'episode_type' parameter to 'source' to match Graphiti API
- Added required 'reference_time' parameter with current timestamp
- Added datetime imports (UTC, datetime)
Errors fixed:
- 'async for' requires an object with __aiter__ method, got NoneType
- Graphiti.add_episode() got an unexpected keyword argument 'episode_type'
* conductor-checkpoint-msg_01JvcW97a4s3icDWFkhF3kEJ
* fix: Use timezone.utc instead of UTC for Python 3.10 compatibility
The UTC constant was added in Python 3.11. Changed to use
timezone.utc which is available in Python 3.10+.
Fixed ImportError: cannot import name 'UTC' from 'datetime'
* conductor-checkpoint-msg_01Br69UnYf8QXvtAhJVTuDGD
* fix: Convert entity_types from list to dict for Graphiti API
The Graphiti add_episode() API expects entity_types as a
dict[str, type[BaseModel]], not a list. Changed entity type
building to create a dictionary mapping entity names to their
Pydantic model classes.
Fixed error: 'list' object has no attribute 'items'
Changes:
- Build entity_types as dict instead of list in config processing
- Add fallback to convert ENTITY_TYPES list to dict if needed
- Map entity type names to their model classes
* conductor-checkpoint-msg_0173SR9CxbBH9jVWp8tLooRp
* conductor-checkpoint-msg_0169v3hqZG1Sqb13Kp1Vijms
* fix: Remove protected 'name' attribute from entity type models
Pydantic BaseModel reserves 'name' as a protected attribute. Removed
the 'name' attribute from dynamically created entity type models as
it's not needed - the entity type name is already stored as the class
name and dict key.
Fixed error: name cannot be used as an attribute for Requirement as
it is a protected attribute name.
* conductor-checkpoint-msg_0118QJWvZLyoZfwb1UWZqrRa
* conductor-checkpoint-msg_01B78jtT59YDt1Xm5hJpoqQw
* conductor-checkpoint-msg_01MsqeFGoCEXpoNMiDRM3Gjh
* conductor-checkpoint-msg_01SwJkCDAScffk8116KPVpTd
* conductor-checkpoint-msg_01EBWwDRC8bZ7oLYxsrmVnLH
* conductor-checkpoint-msg_01SAcxuF3eqtP4exA47CBqAi
* conductor-checkpoint-msg_011dRKwJM31K3ob9Gy4JCmae
* conductor-checkpoint-msg_018d52yUXdPF48UBWPQdiB4W
* conductor-checkpoint-msg_01MGFAenMDnTX3H9HSZEbj2T
* conductor-checkpoint-msg_01MHw4g8TicrXegSK9phncfw
* conductor-checkpoint-msg_018YrqWa3c2ZpkxemiiaE9tA
* conductor-checkpoint-msg_01SNsax9AwiCBFrC7Fpo7BNe
* conductor-checkpoint-msg_01K7QC1X8iPiYaMdvbi7WtR5
* conductor-checkpoint-msg_01KgGgzpbiuM31KWKxQhNBfY
* conductor-checkpoint-msg_01KL3wzQUn3gekDmznXVgXne
* conductor-checkpoint-msg_016GKc3DYwYUjngGw8pArRJK
* conductor-checkpoint-msg_01QLbhPMGDeB5EHbMq5KT86U
* conductor-checkpoint-msg_01Qdskq96hJ6Q9DPg1h5Jjgg
* conductor-checkpoint-msg_01JhPXYdc6HGsoEW2f1USSyd
* conductor-checkpoint-msg_018NLrtFxs5zfcNwQnNCfvNg
* conductor-checkpoint-msg_01G1G9J7cbupmLkyiQufj335
* conductor-checkpoint-msg_01BHEPsv2EML14gFa6vkn1NP
* conductor-checkpoint-msg_0127MeSvxWk8BLXjB5k3wDJY
* conductor-checkpoint-msg_018dRGHW6fPNqJDN6eV6SpoH
* conductor-checkpoint-msg_01CPPZ9JKakjsmHpzzoFVhaM
* conductor-checkpoint-msg_014jJQ4FkGU4485gF41K2suG
* conductor-checkpoint-msg_01MS72hQDCrr1rB6GSd3zy4h
* conductor-checkpoint-msg_01P7ur6mQEusfHTYpBrBnpk3
* conductor-checkpoint-msg_01JiEiEuJN3sQXheqMzCa6hX
* conductor-checkpoint-msg_01D7XfEJqzTeKGyuE5EFmjND
* conductor-checkpoint-msg_01Gn6qZrD3DZd8c6a6fmMap7
* conductor-checkpoint-msg_01Ji7gxCG4jR145rBAupwU49
* conductor-checkpoint-msg_01CYzyiAtLo95iVLeqWSuYiR
* conductor-checkpoint-msg_017fAeUG21Ym1EeofanFzFGa
* conductor-checkpoint-msg_013rt24pyzMHbrmEQein2dJJ
* conductor-checkpoint-msg_016bN3uyAxN28Rh8uvDpExit
* conductor-checkpoint-msg_017QV6m73ShaMBdQi7L3kmhP
* conductor-checkpoint-msg_01LUZ9XS7C1LCG6A1VFNcRL2
* conductor-checkpoint-msg_0136b9tNU5ko18T3PmRkW3LJ
* conductor-checkpoint-msg_018FX6Mibr66cKLnpL84f2Js
* conductor-checkpoint-msg_01WRZxPMQYjNEjcFNTMzWYeL
* conductor-checkpoint-msg_015Tbxjxrj6dynf7TbZscFD3
* conductor-checkpoint-msg_01ELC9AyZZGry9tN4XKrwEM6
* conductor-checkpoint-msg_01Jk4ugkAqMs4iRYWwnaNAHR
* conductor-checkpoint-msg_01NLStrCDq7HZJy3pKyGSqxM
* conductor-checkpoint-msg_01BFZEVpXbdxuXJguFH3caek
* Remove User and Assistant exception from Preference prioritization
* conductor-checkpoint-msg_01JP4eGXZfEjoSXWUwTHNYoJ
* Add combined FalkorDB + MCP server Docker image
- Created Dockerfile.falkordb-combined extending official FalkorDB image
- Added startup script to run both FalkorDB daemon and MCP server
- Created docker-compose-falkordb-combined.yml for simplified deployment
- Added comprehensive README-falkordb-combined.md documentation
- Updated main README with Option 4 for combined image
- Single container solution for development and single-node deployments
* conductor-checkpoint-msg_01PRJ1fre9d6J4qgBmCBQhCu
* Fix Dockerfile syntax version and Python compatibility
- Set Dockerfile syntax to version 1 as requested
- Use Python 3.11 from Debian Bookworm instead of 3.12
- Add comment explaining Bookworm ships with Python 3.11
- Python 3.11 meets project requirement of >=3.10
- Build tested successfully
* conductor-checkpoint-msg_011Thrsv6CjZKRCXvordMWeb
* Fix combined FalkorDB image to run both services successfully
- Override FalkorDB ENTRYPOINT to use custom startup script
- Use correct FalkorDB module path: /var/lib/falkordb/bin/falkordb.so
- Create config-docker-falkordb-combined.yaml with localhost URI
- Create /var/lib/falkordb/data directory for persistence
- Both FalkorDB and MCP server now start successfully
- Tested: FalkorDB ready, MCP server running on port 8000
* conductor-checkpoint-msg_01FT3bsTuv7466EvCeRtgDsD
* Fix health check to eliminate 404 errors
- Changed health check to only verify FalkorDB (redis-cli ping)
- Removed non-existent /health endpoint check
- MCP server startup is visible in logs
- Container now runs without health check errors
* conductor-checkpoint-msg_01KWBc5S8vWzyovUTWLvPYNw
* Replace Kuzu with FalkorDB as default database
BREAKING CHANGE: Kuzu is no longer supported. FalkorDB is now the default.
- Renamed Dockerfile.falkordb-combined to Dockerfile (default)
- Renamed docker-compose-falkordb-combined.yml to docker-compose.yml (default)
- Updated config.yaml to use FalkorDB with localhost:6379 as default
- Removed Kuzu from pyproject.toml dependencies (now only falkordb extra)
- Updated Dockerfile to use graphiti-core[falkordb] instead of [kuzu,falkordb]
- Completely removed all Kuzu references from README
- Updated README to document FalkorDB combined container as default
- Docker Compose now starts single container with FalkorDB + MCP server
- Prerequisites now require Docker instead of Python for default setup
- Removed old Kuzu docker-compose files
Running from command line now requires external FalkorDB instance at localhost:6379
* conductor-checkpoint-msg_014wBY9WG9GRXP7cUZ2JiqGz
* Complete Kuzu removal from MCP server
Removed all remaining Kuzu references from:
- Test fixtures (test_fixtures.py): Changed default database to falkordb, removed kuzu configuration
- Test runner (run_tests.py): Removed kuzu from database choices, checks, and markers
- Integration tests (test_comprehensive_integration.py): Removed kuzu from parameterized tests and environment setup
- Test README: Updated all examples and documentation to reflect falkordb as default
- Docker README: Completely rewrote to remove KuzuDB section, updated with FalkorDB combined image as default
All Kuzu support has been completely removed from the MCP server codebase. FalkorDB (via combined container) is now the default database backend.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01FAgmoDFBPETezbBr18Bpir
* Fix Anthropic client temperature type error
Fixed pyright type error where temperature parameter (float | None) was being passed directly to Anthropic's messages.create() method which expects (float | Omit).
Changes:
- Build message creation parameters as a dictionary
- Conditionally include temperature only when not None
- Use dictionary unpacking to pass parameters
This allows temperature to be properly omitted when None, rather than passing None as a value.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_01KEuAQucnvsH94BwFgCAQXg
* Fix critical PR review issues
Fixed high-impact issues from PR #1024 code review:
1. **Boolean conversion bug (schema.py)**
- Fixed _expand_env_vars returning strings 'true'/'false' instead of booleans
- Now properly converts boolean-like strings (true/false/1/0/yes/no/on/off) to actual booleans
- Simplified logic by removing redundant string-to-string conversions
- Added support for common boolean string variations
2. **Dependency management (pyproject.toml)**
- Removed pytest from main dependencies (now only in dev dependencies)
- Moved azure-identity to optional dependencies under new [azure] group
- Prevents forcing Azure and testing dependencies on all users
3. **Conditional Azure imports (utils.py)**
- Made azure-identity import conditional in create_azure_credential_token_provider()
- Raises helpful ImportError with installation instructions if not available
- Follows lazy-import pattern for optional dependencies
4. **Documentation fix (graphiti_mcp_server.py)**
- Fixed confusing JSON escaping in add_memory docstring example
- Changed from triple-backslash escaping to standard JSON string
- Updated comment to clarify standard JSON escaping is used
Issues verified as already fixed:
- Docker build context (all docker-compose files use context: ..)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_013aEa7tUV8rEmfw38BzJatc
* Add comprehensive SEMAPHORE_LIMIT documentation
Added detailed documentation for SEMAPHORE_LIMIT configuration to help users optimize episode processing concurrency based on their LLM provider's rate limits.
Changes:
1. **graphiti_mcp_server.py**
- Expanded inline comments from 3 lines to 26 lines
- Added provider-specific tuning guidelines (OpenAI, Anthropic, Azure, Ollama)
- Documented symptoms of too-high/too-low settings
- Added monitoring recommendations
2. **README.md**
- Expanded "Concurrency and LLM Provider 429 Rate Limit Errors" section
- Added tier-specific recommendations for each provider
- Explained relationship between episode concurrency and LLM request rates
- Added troubleshooting symptoms and monitoring guidance
- Included example .env configuration
3. **config.yaml**
- Added header comment referencing detailed documentation
- Noted default value and suitable use case
4. **.env.example**
- Added SEMAPHORE_LIMIT with inline tuning guidelines
- Quick reference for all major LLM provider tiers
- Cross-reference to README for full details
Benefits:
- Users can now make informed decisions about concurrency settings
- Reduces likelihood of 429 rate limit errors from misconfiguration
- Helps users maximize throughput within their rate limits
- Provides clear troubleshooting guidance
Addresses PR #1024 review comment about magic number documentation.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_017netxNYzmam5Cu8PM2uQXW
* conductor-checkpoint-msg_012B8ESfBFcMeG3tFimpjbce
* conductor-checkpoint-msg_01Xe46bzgCGV4c8g4piPtSMQ
* conductor-checkpoint-start
* conductor-checkpoint-msg_01QPZK2pa2vUMpURRFmX93Jt
* conductor-checkpoint-msg_01UU5jQcfrW5btRJB3zy5KQZ
* conductor-checkpoint-msg_01884eN3wprtCkrEgEaRDzko
* conductor-checkpoint-msg_01GC2fQiu9gLGPGf8SvG5VW8
* conductor-checkpoint-msg_018ZD567wd4skoiAz7oML7WX
* conductor-checkpoint-msg_01C3AxzcQQSNZxJcuVxAMYpG
* conductor-checkpoint-msg_014w5iHAnv7mVkKfTroeNkuM
* docs: Add current LLM model reference to CLAUDE.md
Added comprehensive model reference section documenting valid model names for OpenAI, Anthropic, and Google Gemini as of January 2025.
OpenAI Models:
- GPT-5 family (reasoning models): gpt-5-mini, gpt-5-nano
- GPT-4.1 family (standard models): gpt-4.1, gpt-4.1-mini, gpt-4.1-nano
- Legacy models: gpt-4o, gpt-4o-mini
Anthropic Models:
- Claude 3.7 family (latest)
- Claude 3.5 family
- Legacy Claude 3 models
Google Gemini Models:
- Gemini 2.5 family (latest)
- Gemini 2.0 family (experimental)
- Gemini 1.5 family (stable)
This documents that model names like gpt-5-mini, gpt-4.1, and gpt-4.1-mini used throughout the codebase are valid OpenAI model identifiers, not errors.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* conductor-checkpoint-msg_014JsovjGyTM1mGwR1nVWLvX
* conductor-checkpoint-msg_013ooHLBEhPccaSY4cFse8vK
* conductor-checkpoint-msg_01WfmUCwXhWxEFtV7R3zJLwT
* conductor-checkpoint-msg_01SbjZ9mm9YwqeJHTDUDoKU8
* conductor-checkpoint-msg_01T2cR1aXUjNSegqzXQcW2jC
* conductor-checkpoint-msg_01EnQy5A9dMFD8F11hWKvzGo
* conductor-checkpoint-msg_01R1zsLmxvwjZ9SwKNhSnQAv
* refactor: Remove duplicate is_reasoning_model calculation in factories.py
* conductor-checkpoint-msg_015oLk8qck3TbfaCryY9gngJ
* conductor-checkpoint-msg_018YAxG5GsLq1dBMuGE6kwEJ
* conductor-checkpoint-msg_014fda5sUsvofb537BvqkuBY
* fix: Change default transport to http, mark SSE as deprecated
* conductor-checkpoint-msg_01S3x8oHkFTM2x4ZiT81QetV
* conductor-checkpoint-msg_01AVxUgejEA9piS6narw4omz
* conductor-checkpoint-msg_019W9KoNBmkobBguViYUj18s
* conductor-checkpoint-msg_01S2mYUmqLohxEmoZaNqsm2f
* conductor-checkpoint-msg_013ZGKfZjdDsqiCkAjAiuEk7
* fix: Handle default config path and empty env vars correctly
- Change default config path from 'config.yaml' to 'config/config.yaml'
- Fix env var expansion to return None for empty strings instead of False
- Prevents validation errors when optional string fields have unset env vars
* conductor-checkpoint-msg_01Bx1BqH3BaBxHMrnsbUQXww
* fix: Allow None for episode_id_prefix and convert to empty string
- Change episode_id_prefix type to str | None to accept None from YAML
- Add model_post_init to convert None to empty string for backward compatibility
* conductor-checkpoint-msg_01CXVkHJC8gp5i395MQMhp6D
* feat: Add helpful error message for database connection failures
- Catch Redis/database connection errors during initialization
- Provide clear, formatted error messages with startup instructions
- Include provider-specific guidance (FalkorDB vs Neo4j)
- Improves developer experience when database is not running
* conductor-checkpoint-msg_01Cd9u1z7pqmX1EG7vXXo4GA
* feat: Add specific Neo4j connection error message with startup instructions
* conductor-checkpoint-msg_01XgbmgFaUMPopni4Q8EhG23
* fix: Remove obsolete KuzuDB check from status endpoint
- Remove dead code checking for 'kuzu' provider (was removed)
- Simplify status check to use configured database provider directly
- Status now correctly reports neo4j or falkordb based on config
* conductor-checkpoint-msg_01WLjwygBwfvbJcVoUMDV3h6
* fix: Use service config instead of global config in status endpoint
- Changed status check to use graphiti_service.config.database.provider
- Ensures status reports the actual running database, not potentially stale global
- Fixes issue where status always reported falkordb regardless of config
* conductor-checkpoint-msg_01DoLD51xqrrdFvq3AgkYuQi
* conductor-checkpoint-msg_01EUW7ArnNM6kHCgFDrQZrro
* conductor-checkpoint-msg_01LqYK6nj1ZFfRNBRP15FMLo
* feat: Add standalone Dockerfile for external database deployments
- Create Dockerfile.standalone for MCP server without embedded FalkorDB
- Supports both Neo4j and FalkorDB via DATABASE_PROVIDER build arg
- Update docker-compose-neo4j.yml to use standalone Dockerfile
- Update docker-compose-falkordb.yml to use standalone Dockerfile
- Fixes issue where Neo4j compose was starting embedded FalkorDB
- Separate images: standalone-neo4j and standalone-falkordb
* conductor-checkpoint-msg_01QSHNgVZvF1id5UtLhpzuUa
* refactor: Unified standalone image with both Neo4j and FalkorDB drivers
- Modified Dockerfile.standalone to install both neo4j and falkordb extras
- Both compose files now use the same standalone image
- Config file determines which database to connect to at runtime
- Added build-standalone.sh script for building and pushing to DockerHub
- Image tags: standalone, {version}-standalone, {version}-graphiti-{core}-standalone
* conductor-checkpoint-msg_01H4isP3oHK25sGpVWzXq9kX
* fix: Correct config file paths in compose files
- Fix CONFIG_PATH env var: /app/config/config.yaml -> /app/mcp/config/config.yaml
- Fix volume mount path: /app/config/config.yaml -> /app/mcp/config/config.yaml
- Matches WORKDIR /app/mcp in Dockerfile.standalone
- Fixes issue where wrong config was being loaded
* conductor-checkpoint-msg_01Pv3Qj9UJJat288xZTsfCm3
* conductor-checkpoint-msg_01DkBq4kQA5Fdmxfikm8aBYG
* conductor-checkpoint-msg_01EBqphY68KNzRWei4QNpcYg
* feat: Add /health endpoint for Docker healthchecks
- Add @mcp.custom_route for /health endpoint using FastMCP
- Returns {status: 'healthy', service: 'graphiti-mcp'}
- Update Dockerfile.standalone healthcheck to use /health instead of /
- Eliminates 404 errors in logs from healthcheck pings
- Follows FastMCP best practices for operational monitoring
* conductor-checkpoint-msg_01UpNeurS45bREPEeGkV3uCx
* feat: Add logging to verify entity types are loaded from config
Added INFO level logging during GraphitiService initialization to confirm
that custom entity types from the configuration file are properly loaded.
This helps debug issues where the entity ontology may not be applied.
Logs the entity type names when custom types are present:
INFO - Using custom entity types: Preference, Requirement, Procedure, ...
* fix: Correct logging message for entity types and add embedder logging
Fixed copy-paste error where entity types else clause was logging about
embedder client. Also added missing else clause for embedder client logging
for consistency.
- Fixed: "No Embedder client configured" -> "Using default entity types"
- Added: Missing embedder client else clause logging
* conductor-checkpoint-msg_01WMuxAzUnkpsa5WSXKMyLLP
* fix: Return JSONResponse from health check endpoint
Fixed TypeError in health check endpoint by returning a proper Starlette
JSONResponse object instead of a plain dict. Starlette custom routes require
ASGI-compatible response objects.
Error was: TypeError: 'dict' object is not callable
* conductor-checkpoint-msg_01CSgKFQaLsKVrBJAYCFoSGa
* conductor-checkpoint-msg_01SFY9xCnHxeCFGf53FESncs
* feat: Return complete node properties and exclude all embeddings
Enhanced node search results to include all relevant properties:
- Added `attributes` dict for custom entity properties
- Changed from single `type` to full `labels` array
- Added `group_id` for partition information
- Added safety filter to strip any keys containing "embedding" from attributes
Added format_node_result() helper function for consistent node formatting
that excludes name_embedding vectors, matching the pattern used for edges.
Embeddings are now explicitly excluded in all data returns:
- EntityNode: name_embedding excluded + attributes filtered
- EntityEdge: fact_embedding excluded (existing)
- EpisodicNode: No embeddings to exclude
This ensures clients receive complete metadata while keeping payload
sizes manageable and avoiding exposure of internal vector representations.
---------
Co-authored-by: Claude <noreply@anthropic.com>
Changes to `to_prompt_json()` helper to default to minified JSON (no indentation) instead of 2-space indentation. This reduces token consumption in LLM prompts while maintaining all necessary information.
- Changed default `indent` parameter from `2` to `None` in `prompt_helpers.py`
- Updated all prompt modules to remove explicit `indent=2` arguments
- Minor code formatting fixes in LLM clients
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude <noreply@anthropic.com>
* Add OpenTelemetry distributed tracing support
- Add tracer abstraction with no-op and OpenTelemetry implementations
- Instrument add_episode and add_episode_bulk with tracing spans
- Instrument LLM client with cache-aware tracing
- Add configurable span name prefix support
- Refactor add_episode methods to improve code quality
- Add OTEL_TRACING.md documentation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Fix linting errors in tracing implementation
- Remove unused episodes_by_uuid variable
- Fix tracer type annotations for context manager support
- Replace isinstance tuple with union syntax
- Use contextlib.suppress for exception handling
- Fix import ordering and use AbstractContextManager
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Address PR review feedback on tracing implementation
Critical fixes:
- Remove flawed error span creation in graphiti.py that created orphaned spans
- Restructure LLM client tracing to create span once at start, eliminating code duplication
- Initialize LLM client tracer to NoOpTracer by default to fix type checking
Enhancements:
- Add comprehensive span attributes to add_episode: reference_time, entity/edge type counts, previous episodes count, invalidated edge count, community count
- Optimize isinstance check for better performance
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Add prompt name tracking to OpenTelemetry tracing spans
Add prompt_name parameter to all LLM client generate_response() methods
and set it as a span attribute in the llm.generate span. This enables
better observability by identifying which prompt template was used for
each LLM call.
Changes:
- Add prompt_name parameter to LLMClient.generate_response() base method
- Add prompt_name parameter and tracing to OpenAIBaseClient,
AnthropicClient, GeminiClient, and OpenAIGenericClient
- Update all 14 LLM call sites across maintenance operations to include
prompt_name:
- edge_operations.py: 4 calls
- node_operations.py: 6 calls (note: 7 listed but only 6 unique)
- temporal_operations.py: 2 calls
- community_operations.py: 2 calls
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Fix exception handling in add_episode to record errors in OpenTelemetry span
Moved try-except block inside the OpenTelemetry span context and added
proper error recording with span.set_status() and span.record_exception().
This ensures exceptions are captured in the distributed trace, matching
the pattern used in add_episode_bulk.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
* Add group_id parameter to get_extraction_language_instruction
Enable consumers to provide group-specific language extraction
instructions by passing group_id through the call chain.
Changes:
- Add optional group_id parameter to get_extraction_language_instruction()
- Add group_id parameter to all LLMClient.generate_response() methods
- Pass group_id through to language instruction function
- Maintain backward compatibility with default None value
Users can now customize extraction per group:
```python
def custom_instruction(group_id: str | None = None) -> str:
if group_id == 'spanish-users':
return '\n\nExtract in Spanish.'
return '\n\nExtract in original language.'
client.get_extraction_language_instruction = custom_instruction
```
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Pass group_id to generate_response in extraction operations
Thread group_id parameter through all extraction-related generate_response()
calls where it's naturally available (via episode.group_id or node.group_id).
This enables consumers to override get_extraction_language_instruction() with
group-specific language preferences.
Changes:
- edge_operations.py: Pass group_id in extract_edges()
- node_operations.py: Pass episode.group_id in extract_nodes() and
node.group_id in extract_attributes_from_node()
- node_operations.py: Add group_id parameter to extract_nodes_reflexion()
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Fix type inconsistency in extract_nodes_reflexion parameter
Change group_id parameter from str = '' to str | None = None to match
the pattern used throughout the codebase and align with the optional
nature of group_id in generate_response().
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Remove ensure_ascii parameter and uv.lock file
* Reset uv.lock to main branch version
---------
Co-authored-by: Claude <noreply@anthropic.com>
Replace MULTILINGUAL_EXTRACTION_RESPONSES constant with configurable
get_extraction_language_instruction() function to improve determinism
and allow customization.
Changes:
- Replace constant with function in client.py
- Update all LLM client implementations to use new function
- Maintain backward compatibility with same default behavior
- Enable users to override function for custom language requirements
Users can now customize extraction behavior by monkey-patching:
```python
import graphiti_core.llm_client.client as client
client.get_extraction_language_instruction = lambda: "Custom instruction"
```
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-authored-by: Claude <noreply@anthropic.com>
* gpt-5-mini and gpt-5-nano default
* bump version
* remove unused imports
* linter
* update
* disable neptune errors while we get a fixture in place
* update pyright
* revert non-structured completions
* fix typo
* feat: enhance GeminiClient with max tokens management
- Introduced a mapping for maximum output tokens for various Gemini models.
- Added methods to resolve max tokens based on precedence rules, allowing for more flexible token management.
- Updated tests to verify max tokens behavior, ensuring explicit parameters take precedence and fallback mechanisms work correctly.
This change improves the handling of token limits for different models, enhancing the client’s configurability and usability.
* refactor: streamline max tokens retrieval in GeminiClient
- Removed the fallback to DEFAULT_MAX_TOKENS in favor of directly using model-specific maximum tokens.
- Simplified the logic for determining max tokens, enhancing code clarity and maintainability.
This change improves the efficiency of token management within the GeminiClient.
* feat(gemini): embedding batch size & lite default
The new `gemini-embedding-001` model only allows one embedding input per batch
(instance), but has other impressive statistics:
https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/text-embeddings-api
The -DEFAULT_SMALL_MODEL must not have the 'models/' prefix.
* Refactor: Improve Gemini Client Error Handling and Reliability
This commit introduces several improvements to the Gemini client to enhance its robustness and reliability.
- Implemented more specific error handling for various Gemini API responses, including rate limits and safety blocks.
- Added a JSON salvaging mechanism to gracefully handle incomplete or malformed JSON responses from the API.
- Introduced detailed logging for failed LLM generations to simplify debugging and troubleshooting.
- Refined the Gemini embedder to better handle empty or invalid embedding responses.
- Updated and corrected tests to align with the improved error handling and reliability features.
* fix: cleanup in _log_failed_generation()
* fix: cleanup in _log_failed_generation()
* Fix ruff B904 error in gemini_client.py
* fix(gemini): correct retry logic and enhance error logging
Updated the retry mechanism in the GeminiClient to ensure it retries the maximum number of times specified. Improved error logging to provide clearer insights when all retries are exhausted, including detailed information about the last error encountered.
* fix(gemini): enhance error handling for safety blocks and update tests
Refined error handling in the GeminiClient to improve detection of safety block conditions. Updated test cases to reflect changes in exception messages and ensure proper retry logic is enforced. Enhanced mock responses in tests to better simulate real-world scenarios, including handling of invalid JSON responses.
* revert default gemini to text-embedding-001
---------
Co-authored-by: Daniel Chalef <131175+danielchalef@users.noreply.github.com>
The cross_encoder for Gemini already supported passing in a custom client.
I replicated the same input pattern to embedder and llm_client.
The value is, you can support custom API endpoints and other options like below:
cross_encoder=GeminiRerankerClient(
client=genai.Client(
api_key=os.environ.get('GOOGLE_GENAI_API_KEY'),
http_options=types.HttpOptions(api_version='v1alpha')),
config=LLMConfig(
model="gemini-2.5-flash-lite-preview-06-17"
)
))
* add support for Gemini 2.5 model thinking budget
* allow adding thinking config to support current and future gemini models
* merge
* improve client; add reranker
* refactor: change type hint for gemini_messages to Any for flexibility
* refactor: update GeminiRerankerClient to use direct relevance scoring and improve ranking logic. Add tests
* fix fixtures
---------
Co-authored-by: realugbun <github.disorder751@passmail.net>
* Refactor group_id handling and update dependencies
- Changed default behavior for `group_id` to 'default' instead of generating a UUID.
- Updated README to reflect the new default behavior for `--group-id`.
- Reformatted LLMConfig initialization for better readability.
- Bumped versions of several dependencies including `azure-core`, `azure-identity`, `certifi`, `charset-normalizer`, `sse-starlette`, and `typing-inspection`.
- Added `python-multipart` as a new dependency.
This update improves usability and ensures compatibility with the latest library versions.
* Update Graphiti MCP server instructions and refactor method names for clarity
- Revised the welcome message to enhance clarity about Graphiti's functionality.
- Renamed methods for better understanding: `add_episode` to `add_memory`, `search_nodes` to `search_memory_nodes`, `search_facts` to `search_memory_facts`, and updated related docstrings to reflect these changes.
- Updated references to "knowledge graph" to "graph memory" for consistency throughout the codebase.
* Update README for Graphiti MCP server configuration and integration with Claude Desktop
- Changed server name from "graphiti" to "graphiti-memory" in configuration examples for clarity.
- Added instructions for running the Graphiti MCP server using Docker.
- Included detailed steps for integrating Claude Desktop with the Graphiti MCP server, including optional installation of `mcp-remote`.
- Enhanced overall documentation to improve user experience and understanding of the setup process.
* Enhance error handling in GeminiEmbedder and GeminiClient
- Added checks to raise exceptions when no embeddings or response text are returned, improving robustness.
- Included type ignore comments for mypy compatibility in embed_content calls.
* Update graphiti_core/embedder/gemini.py
Co-authored-by: ellipsis-dev[bot] <65095814+ellipsis-dev[bot]@users.noreply.github.com>
* Update graphiti_core/llm_client/gemini_client.py
Co-authored-by: ellipsis-dev[bot] <65095814+ellipsis-dev[bot]@users.noreply.github.com>
---------
Co-authored-by: ellipsis-dev[bot] <65095814+ellipsis-dev[bot]@users.noreply.github.com>
* remove temporary debug logging
* add anthropic api to .env.example
* move anthropic int tests to llm_client dir to better match existing test structure
* update `TestLLMClient` to `MockLLMClient` to eliminate pytest warning
* Fix: use self.max_tokens when max_token isnt specified
* Fix: use self.max_tokens in OpenAI clients
* Fix: use self.max_tokens in Anthropic client
* Fix: use self.max_tokens in Gemini client
* update Anthropic client to use tool calling and add tests
* fix linting errors before creating pull request by making literal types for anthropic models
* Bump version from 0.9.0 to 0.9.1 in pyproject.toml and update google-genai dependency to >=0.1.0
* Bump version from 0.9.1 to 0.9.2 in pyproject.toml
* Update google-genai dependency version to >=0.8.0 in pyproject.toml
* loc file
* Update pyproject.toml to version 0.9.3, restructure dependencies, and modify author format. Remove outdated Google API key note from README.md.
* upgrade poetry and ruff
* first cut
* Update dependencies and enhance README for optional LLM providers
- Bump aiohttp version from 3.11.14 to 3.11.16
- Update yarl version from 1.18.3 to 1.19.0
- Modify pyproject.toml to include optional extras for Anthropic, Groq, and Google Gemini
- Revise README.md to reflect new optional LLM provider installation instructions and clarify API key requirements
* Remove deprecated packages from poetry.lock and update content hash
- Removed cachetools, google-auth, google-genai, pyasn1, pyasn1-modules, rsa, and websockets from the lock file.
- Added new extras for anthropic, google-genai, and groq.
- Updated content hash to reflect changes.
* Refactor import paths for GeminiClient in README and __init__.py
- Updated import statement in README.md to reflect the new module structure for GeminiClient.
- Removed GeminiClient from the __all__ list in __init__.py as it is no longer directly imported.
* Refactor import paths for GeminiEmbedder in README and __init__.py
- Updated import statement in README.md to reflect the new module structure for GeminiEmbedder.
- Removed GeminiEmbedder and GeminiEmbedderConfig from the __all__ list in __init__.py as they are no longer directly imported.
* implement so
* bug fixes and typing
* inject schema for non-openai clients
* correct datetime format
* remove List keyword
* Refactor node_operations.py to use updated prompt_library functions
* update example
* node distance and group_ids fixed
* get all with no group_id passed
* push
* push
* remove comments
* mypy
* mypy ids
* please mypy
* trust
* last one
* Override default max tokens for Anthropic and Groq clients
* Override default max tokens for Anthropic and Groq clients
* Override default max tokens for Anthropic and Groq clients
The code changes refactor the `OpenAIClient` initialization to accept an optional `client` parameter. This allows the client to be passed in from outside, providing more flexibility and enabling easier testing.
