Python API

Name: OMEGA
Author: OMEGA

omega.bridge public API -- 36 functions.

This is the main programmatic interface used internally by the MCP handlers. All functions are importable from omega.bridge.

from omega.bridge import store, query, remember, auto_capture

Core

Function	Signature	Description
`store`	`store(content, event_type="memory", metadata=None, session_id=None, entity_id=None) -> str`	Store a memory with metadata. Wraps `auto_capture` with a default event type.
`remember`	`remember(text, session_id=None) -> str`	Store a permanent memory as `user_preference` type.
`auto_capture`	`auto_capture(content, event_type, metadata=None, session_id=None, project=None, ttl_override=None, entity_id=None) -> str`	Primary ingestion function. Handles dedup (SHA256 + Jaccard), evolution (appends to similar memories), auto-tagging, auto-relating, and blocklist filtering.
`delete_memory`	`delete_memory(memory_id) -> dict`	Delete a memory by its node ID. Returns `{"success": bool}`.
`edit_memory`	`edit_memory(memory_id, new_content) -> dict`	Update a memory's content and regenerate its embedding.

Function	Signature	Description
`query`	`query(text, limit=10, event_type=None, filter_tags=None, temporal_range=None, context_file=None, context_tags=None, entity_id=None, project=None, session_id=None) -> str`	Semantic search with blended ranking (70% vector, 30% FTS5) and contextual re-ranking. Returns markdown-formatted results.
`query_structured`	`query_structured(text, limit=10, event_type=None, filter_tags=None, ...) -> list[dict]`	Same search pipeline as `query` but returns structured dicts instead of markdown.
`phrase_search`	`phrase_search(phrase, limit=10, event_type=None, project=None, case_sensitive=False) -> str`	Exact substring match via FTS5 full-text search.
`find_similar_memories`	`find_similar_memories(memory_id, limit=5) -> str`	Find memories similar to a given memory by embedding distance.

Function	Signature	Description
`welcome`	`welcome(session_id=None, project=None) -> dict`	Session briefing: recent memories, user profile, project context.
`get_session_context`	`get_session_context(session_id, project=None) -> dict`	Retrieve session context for handoff or resume.
`clear_session`	`clear_session(session_id) -> dict`	Delete all memories associated with a session.
`batch_store`	`batch_store(items) -> dict`	Store multiple memories in a single call. Each item is a dict with `content` and optional `event_type`, `metadata`, etc.

Function	Signature	Description
`check_health`	`check_health(warn_mb=350, critical_mb=500, max_nodes=10000) -> dict`	System health check: memory usage, node counts, cache stats, warnings.
`status`	`status() -> dict`	Memory count, database size, model status, edge count.
`get_dedup_stats`	`get_dedup_stats() -> dict`	Deduplication statistics (duplicates found, evolved, blocked).
`type_stats`	`type_stats() -> dict`	Memory counts grouped by event type.
`session_stats`	`session_stats() -> dict`	Memory counts grouped by session (top 20).
`get_activity_summary`	`get_activity_summary(days=7) -> dict`	Recent session activity overview.

Function	Signature	Description
`get_profile`	`get_profile() -> dict`	User profile built from memory patterns.
`save_profile`	`save_profile(profile) -> bool`	Save or update user profile fields.
`extract_preferences`	`extract_preferences(text) -> dict`	Extract preference signals from text content.
`list_preferences`	`list_preferences() -> list[dict]`	List all stored user preferences.

Function	Signature	Description
`get_cross_session_lessons`	`get_cross_session_lessons(task=None, limit=5, project_path=None, exclude_project=None, exclude_session=None) -> list`	Lessons across sessions, ranked by verification count and access frequency.
`get_cross_project_lessons`	`get_cross_project_lessons(task=None, limit=5, exclude_project=None, exclude_session=None) -> list`	Lessons across all projects.

Function	Signature	Description
`consolidate`	`consolidate(prune_days=30, max_summaries=50) -> str`	Prune stale low-value memories, cap session summaries, clean orphaned edges. Auto-backs up before running.
`compact`	`compact(event_type="lesson_learned", threshold=0.6, min_cluster=3, dry_run=False) -> str`	Cluster similar memories (Jaccard similarity) and create summary nodes. Originals marked as superseded.
`deduplicate`	`deduplicate() -> dict`	Remove exact content duplicates (SHA256 hash match).
`timeline`	`timeline(days=7, limit_per_day=10) -> str`	Memory timeline grouped by day.
`traverse`	`traverse(memory_id, max_hops=2, min_weight=0.0) -> str`	BFS traversal of the memory relationship graph (max 5 hops).

Function	Signature	Description
`export_memories`	`export_memories(filepath) -> str`	Export all memories to a JSON file.
`import_memories`	`import_memories(filepath, clear_existing=True) -> str`	Import memories from a JSON file. Optionally clears existing data first.
`reingest`	`reingest() -> dict`	Reload entries from legacy store.jsonl into the graph system.

Function	Signature	Description
`check_constraints`	`check_constraints(file_path, project=None) -> list[dict]`	Check a file against stored project constraints.
`list_constraints`	`list_constraints(project=None) -> dict`	List all constraints for a project.
`save_constraints`	`save_constraints(constraints, project=None) -> dict`	Save or update project constraints.

Function	Signature	Description
`record_feedback`	`record_feedback(memory_id, rating, reason=None) -> dict`	Rate a memory as helpful, unhelpful, or outdated. Affects future search ranking via feedback dampening.

Function	Signature	Description
`reset_memory`	`reset_memory() -> None`	Reset the memory store. For testing only -- destroys all data.

All functions are importable from omega.bridge.
The MCP handlers in server/handlers.py call these functions directly.
auto_capture is the primary ingestion path -- store and remember are convenience wrappers around it.
Functions returning str typically return markdown-formatted output suitable for display in agent conversations.
Functions returning dict or list return structured data for programmatic use.