Changelog

All notable changes to anchor are documented here. The format is based on Keep a Changelog, and the project adheres to Semantic Versioning.

[Unreleased]

Multi-provider LLM interface with unified API across Anthropic, OpenAI, Gemini, Grok, Ollama, OpenRouter, and LiteLLM
FallbackProvider for automatic failover across multiple LLM backends
Unified error hierarchy: ProviderError, AuthenticationError, RateLimitError, ContentFilterError, ModelNotFoundError, ServerError, TimeoutError, ProviderNotInstalledError
create_provider factory and register_provider for custom backends
Cost calculation via calculate_cost with MODEL_PRICING registry
Streaming support with StreamChunk and ToolCallDelta
Typed message/response models: Message, LLMResponse, ContentBlock, ToolCall, ToolResult, ToolSchema, Usage, StopReason

Bidirectional Model Context Protocol integration (requires astro-anchor[mcp])
FastMCPClientBridge for consuming external MCP servers as anchor tools
FastMCPServerBridge for exposing anchor pipelines and tools as MCP servers
MCPClientPool for managing connections to multiple MCP servers
MCPClient / MCPServer abstractions with typed config (MCPServerConfig)
MCP resource and prompt models: MCPResource, MCPPrompt, MCPPromptArgument
mcp_tool_to_agent_tool converter and parse_server_string utility
Error hierarchy: MCPError, MCPConfigError, MCPConnectionError, MCPTimeoutError, MCPToolError

Tool-calling agent with @tool decorator for defining agent tools
Agent class with async streaming via achat
Skill and SkillRegistry for organizing related tools
Built-in memory_skill / memory_tools and rag_skill / rag_tools for out-of-the-box capabilities

RoutedRetriever with pluggable routing: CallbackRouter, KeywordRouter, MetadataRouter
LateInteractionRetriever with LateInteractionScorer and MaxSimScorer for ColBERT-style retrieval
SharedSpaceRetriever for cross-modal retrieval in a unified embedding space
CrossModalEncoder and TokenLevelEncoder protocols for multi-modal embeddings
RoundRobinReranker and RerankerPipeline for composable reranking strategies
Async retriever/reranker variants: AsyncDenseRetriever, AsyncHybridRetriever, AsyncCrossEncoderReranker, AsyncCohereReranker

ContextualQueryTransformer for context-aware query rewriting
ConversationRewriter for resolving coreferences in multi-turn conversations
KeywordClassifier, EmbeddingClassifier, CallbackClassifier for query routing
QueryClassifier protocol and classified_retriever_step pipeline step

ProgressiveSummarizationMemory with multi-tier compaction
TierCompactor with CompactionStrategy and AsyncCompactionStrategy protocols

Unit tests for _math.py (cosine_similarity and clamp functions)
MemoryRetrieverAdapter tests verifying Retriever protocol compliance
PipelineExecutionError wrapping test with diagnostics verification
Golden path integration test mirroring README usage pattern
Example: examples/hybrid_rag.py -- hybrid RAG pipeline with dense retrieval
Example: examples/custom_retriever.py -- custom Retriever protocol implementation
Example: examples/budget_management.py -- token budget management and overflow handling
README sections for Priority System (1--10 scale) and Token Budgets

test_consolidator.py: eliminated shared mutable state (_orthogonal_index dict) by converting to factory function pattern (make_orthogonal_embed())
test_graph_memory.py: updated link_memory unknown entity test to expect KeyError instead of ValueError
README: fixed retrieval example with runnable embed_fn and ContextItem creation
README: updated test count from 961 to 1088

Core context pipeline with sync/async support (ContextPipeline)
Token-aware sliding window memory (SlidingWindowMemory)
Summary buffer memory with progressive compaction (SummaryBufferMemory)
Memory manager facade unifying conversation and persistent memory (MemoryManager)
Hybrid RAG retrieval: dense, sparse (BM25), and hybrid (RRF) retrievers
Multi-signal memory retrieval with recency/relevance/importance scoring (ScoredMemoryRetriever)
Provider-agnostic formatting: Anthropic, OpenAI, and generic text formatters
Anthropic multi-block system formatting with prompt caching support
Protocol-based extensibility (PEP 544) for all extension points
Token budget management with per-source allocations and overflow tracking
Pluggable eviction policies: FIFO, importance-based, and paired (user+assistant)
Memory decay: Ebbinghaus forgetting curve and linear decay
Recency scoring: exponential and linear strategies
Memory consolidation with content-hash dedup and cosine-similarity merging
Simple graph memory with BFS traversal for entity-relationship tracking
Memory garbage collection with two-phase expired+decayed pruning
Memory callback protocol for lifecycle observability
Pipeline query enrichment with memory context
Auto-promotion of evicted turns to long-term memory
In-memory reference implementations for all storage protocols
JSON file-backed persistent memory store
CLI with index and query commands (via typer+rich)
961 tests with 94% coverage