cognee

v1.6.6 — Conflict resolution & stability

Release Date: 2026-10-29 Changes: v1.6.6rc1 → v1.6.6

Pull Requests: #6266, #6337, #6339, #6340

Summary

This release focuses on making conflict handling more reliable and easier to manage, plus a handful of stability, vector, and data-layout fixes. You’ll find a new durable conflict-resolution API, safer retirement of conflicting choices, faster and safer cache/deletion paths, and improvements to vector handling that save re-embedding work.

Highlights

• New durable API for fact conflict resolution so you can resolve data disagreements reliably from the API.
• Conflict review is enabled by default and many fixes ensure 'choices' (user decisions about conflicting facts) are retired correctly on edits, deletes, and rebuilds.
• Copy stored vectors between vector stores without re-embedding to save time and cost when moving/replicating embeddings.
• Safer deletion and caching: Redis list/context deletions are now atomic or in-place, and permission/KMS errors are surfaced so failures won’t be silently ignored.

New Features

• Durable fact conflict resolution API — A new API endpoint that lets you resolve disagreements between conflicting facts in a durable (persistent) way. What it does: records and applies chosen resolutions so the system consistently uses the resolved fact. Why it matters: resolves data inconsistencies reliably across sessions and components, reducing incorrect answers or surprising behavior from stale/conflicting data.
• Default conflict review workflow — The system now reviews conflicts by default (previously optional). What it does: when the system detects conflicting facts (different sources or versions of the same information), it surfaces them for review automatically. Why it matters: makes conflict detection actionable out of the box, so fewer conflicts slip through and searches/agents return cleaner results.
• Copy stored vectors between stores without re-embedding — You can now move or copy vector embeddings from one vector store to another without re-running the embedding model. What it does: performs a direct copy of stored embeddings. Why it matters: speeds up migration and replication, and reduces compute and cost because models aren’t re-called.
• DLT column and dated-cell representation improvements — Every DLT column (a column used in change-tracking tables) is now treated as a value node by default, and dated cells are represented once as a Timestamp value. What it does: makes the platform treat change-tracking columns as concrete values and avoids duplicating date cells. Why it matters: fixes subtle correctness and duplication issues when reading dataset rows and keys, improving accuracy of reads and downstream processing.

Improvements

• Safer retirement of choices (user decisions about conflicting facts) — Choices are retired at the correct times: on chunk edits, connector deletes, dataset forget operations, and when rebuilds would otherwise lose them. This reduces stale decisions lingering after data changes.
• Fail-fast on invalidation and malformed keys — Editing will fail if invalidation of related state fails; malformed hints for DLT keys now raise an error instead of silently skipping them. This prevents silent data inconsistencies.
• Preserve retrieval context when resolving conflicts — The retrieval context used to fetch supporting evidence is now preserved while resolving conflicts so answers remain grounded in the same context.
• Merge and validate tags before writing vectors — Tags are now merged and validated before vector upserts, avoiding broken or incomplete metadata on vector records.
• Finish multi-session stamp rollout — Session stamps (used to track actions across sessions) are now rolled out consistently across components for improved session handling and correctness.
• Make classify() accept plaintext view of sealed objects — Useful when you need to classify contents of encrypted/sealed objects a

v1.6.6rc1 — History controls, connectors, and sync hardening

Release Date: 2026-10-23 Changes: v1.6.5.dev0 → v1.6.6rc1

Pull Requests: #5765, #5850, #5893, #5898, #5938, #5955, #5958, #5961, #6011, #6013, #6014, #6185

Summary

This release focuses on giving users clearer, safer controls over memory & search history, adding several new connectors, and hardening sync reliability. Key workflow changes include explicit history-purging behaviour, a new flag to remove search history from the CLI/API, and fixes that make Drive/Asana synchronisation and indexing more robust.

Highlights

• New forget(clean_history=True) option so you can remove search history without needing other inputs.
• Users can now delete their own search history via the API and server endpoints.
• New connectors: Obsidian (vaults), Asana (tasks) and OneNote are added and their syncs are hardened.
• Reliability improvements: Drive and Asana syncs are faster and resumable; graph and index operations avoid unnecessary work.

Breaking Changes

• Dataset delete now purges associated search and conversation history: When you delete a dataset, all associated history is removed. Action required: if you depend on preserved search or session traces for backups or audit, export or back them up before deleting datasets.
• run_info structure changed: run_info metadata no longer contains the full 'data' items that ingestion pipelines previously stored. Action required: any code or tooling that read run_info expecting embedded data must be updated to instead fetch original items via their data ids or use the new storage shapes. A migration to remove the old data shape is included in this release.
• Fail-on-history-purge: delete/forget operations that attempt to purge history will now fail if the purge cannot be completed. Action required: ensure the service has permission to delete search history and that long-running purge operations are allowed to complete; handle failures when calling delete APIs.
• Budget/quota errors: Budget exhaustion and payment-required errors are raised more consistently (LLMPaymentRequiredError). Action required: clients should catch and handle these exceptions if they interact with LLM billing paths.

New Features

• Forget(clean_history=True): A new clean_history flag for the forget command and API that lets you purge recorded search history while keeping other deletion behaviour. What it does: call forget(clean_history=True) to remove stored search history related to a dataset or user. Why it matters: you can now remove traces of user searches without performing other delete operations, making privacy and cleanup operations clearer and easier.
• Delete your own search history: The server and API now include a route that allows a user to delete their own search history. What it does: end-users (via API or UI) can remove their personal queries and search metadata. Why it matters: improves user privacy and compliance by giving individuals direct control over their search traces.
• Obsidian connector (vaults): A new connector that ingests Obsidian vaults (local notes and markdown vaults). What it does: lets you import notes stored in Obsidian vaults into Cognee so they become searchable and recallable. Why it matters: broadens supported sources so personal knowledge stored in Obsidian can be included in memory and search workflows.
• Asana connector: A first-class Asana connector for tasks and events. What it does: syncs Asana tasks and events into Cognee and makes them available for search and recall. Why it matters: teams using Asana can now include task context and history in their memory-driven workflows.
• OneNote connector: Add OneNote support for ingestion and sync. What it does: pulls OneNote content into Cognee so notebooks are searchable and linked into memory. Why it matters: integrates another major note-taking source so users' OneNote content participates in memory and search.
• Expose dataset and data id

v1.6.5 — Faster trace processing & richer code-graph docs

Release Date: 2026-10-16 Changes: v1.6.4 → v1.6.5

Pull Requests: #5287, #5534, #5620, #5622, #5720, #5721, #5722, #5725, #5726, #5751, #5798, #5898, #5938, #5955, #5958, #5961

Summary

This release speeds up trace and ingestion pipelines, improves session handling, and makes code documentation searchable inside the code graph. It also adds a new memify API, better embedding compatibility (base64), and several security and connector hardenings.

Highlights

• Faster trace processing: traces are persisted and processed in fewer, larger operations for lower latency and fewer database writes.
• Code-graph docstrings: function/class docstrings are stored as graph documents and linked to code symbols for better code recall and search.
• New Memify API endpoints: convert content into memory entries via the API to make ingestion and downstream recall simpler.
• Base64 embeddings support: request embeddings encoded as base64 where providers allow it to improve compatibility and reduce transport issues.
• Security & connectors: daily CVE Slack alert, Google Calendar connector hardening, and improved OAuth/token sharing.

Breaking Changes

• datamodel-code-generator requirement bumped: if you use a custom graph_model generation step, you must have datamodel-code-generator >= 0.56.0. What to do: upgrade that dependency when generating custom graph models to avoid generation errors.

New Features

• Store code docstrings in the code graph: Docstrings (the documentation strings attached to functions, classes and modules) are now saved as first-class graph documents and linked to the corresponding code symbols. What it does: when you ingest code, the human-readable docstrings are indexed and connected to the code symbol nodes. Why it matters: search and code-aware agents can return richer, context-aware answers and surface inline documentation alongside code references.
• New Memify API and endpoints: 'memify' is the process of converting content (files, messages, or other data) into memory entries the system can store and recall. What it does: we added server endpoints and CLI wiring to create memory entries (memify) directly via the API. Why it matters: it simplifies turning raw content into searchable, recallable memories for use in recall and improve flows.
• Request base64 embeddings where supported: some embedding providers accept embeddings encoded as base64 to avoid binary transport problems. What it does: Cognee will request base64-encoded embedding responses when the provider supports it. Why it matters: improves interoperability with certain providers and reduces problems when sending/receiving binary embedding payloads.
• Persist a session's trace backlog as one document: traces are the interaction records (events, steps, logs) from a user session. What it does: instead of writing many small trace documents, Cognee can now persist a session's backlog as one consolidated document. Why it matters: fewer database writes and improved throughput for high-traffic sessions, leading to faster ingestion and lower resource usage.
• Build trace pass windows in a single cognify run: 'cognify' is the pipeline run that converts raw traces into memory entries. What it does: trace windowing for a pass is computed in one cognify execution rather than multiple runs. Why it matters: reduces repeated pipeline runs and speeds up trace processing so results appear faster and more consistently.

Improvements

• Session/dataset binding reliability and speed: bound sessions (sessions tied to a specific dataset — a collection of documents you've added) now skip unnecessary re-reads and correctly route skill runs to the bound dataset. This reduces latency and fixes several edge cases where a stale or mismatched binding could cause errors.
• Curation and distillation fixes: distilled entries are sealed per curator batch and new entries are curated by default on ea

v1.6.5.dev0 — Distillation, Embeddings & Connector Hardening

Release Date: 2026-10-16 Changes: v1.6.4.dev0 → v1.6.5.dev0

Pull Requests: #5287, #5534, #5620, #5622, #5720, #5721, #5722, #5725, #5726, #5751, #5798, #5898

Summary

This release focuses on making memory distillation more reliable, improving how embeddings are produced and metered, and hardening connectors and LLM integrations. It includes several bug fixes, security and operational improvements, and a few breaking changes related to legacy embedding surfaces and local cache layout.

Highlights

• More reliable distillation: distillation now reads only the QA turns it needs, seals entries per curator batch, and curates every new entry by default so distilled memories are accurate and consistent.
• Embeddings improvements: support requesting base64-encoded embeddings where supported, meter embedding token usage, lower the default embedding batch size for stability, and keep local model cache under a predictable path.
• Connector and LLM fixes: Gmail and Google Calendar connectors were hardened (better quota handling and permissions), litellm and Mistral model handling were improved, and Azure token handling is shared to reduce auth complexity.
• Security & alerts: daily CVE alert delivery to Slack was added and repository security docs updated to help teams stay informed.

Breaking Changes

• Triplet embedding / triplet memify surfaces removed or deprecated — the older triplet embedding surfaces have been deprecated and parts removed. If you rely on 'triplet' embedding flows, migrate to the newer embedding surfaces and memify pipelines. Check your pipelines and replacements before upgrading.
• datamodel-code-generator minimum version raised — custom graph model generation requires datamodel-code-generator >= 0.56.0. Upgrade this dependency if you use custom graph models.
• Fastembed cache path change — FASTEMBED_CACHE_PATH was removed from the .env.template and fastembed now uses ~/.cognee/models by default. If you relied on a custom FASTEMBED_CACHE_PATH, update your environment or move your cache to the new location.

New Features

• Request Base64 embeddings where supported — Some embedding providers can return vectors encoded as base64. Cognee now requests base64 embeddings when the provider supports them. Why it matters: base64 can improve compatibility with certain providers and transports, reducing encoding errors and simplifying integrations.
• Daily CVE alert to Slack — A new automated alert posts daily security advisories (CVE summaries) to a configured Slack channel. Why it matters: keeps your team aware of new vulnerabilities affecting dependencies so you can act faster.
• Clone a chosen branch of a repo during ingestion — You can now select which branch to clone when pulling a repository. Why it matters: makes ingestion reproducible and lets you index branch-specific code or docs without manual repo handling.
• Embedding token usage metering — The system now records the number of embedding tokens used per activity. Why it matters: helps understand and audit embedding usage for cost tracking and debugging.

Improvements

• More focused distillation: distillation now reads only the required QA turns (instead of whole conversations), and curator runs seal distilled entries per batch so partial results don’t leak. This reduces work and makes distilled summaries more deterministic.
• Curate every new entry per distill run by default — every newly added memory entry is considered for distillation unless you change the setting, making sure new content is summarized promptly.
• Bind typed entries to the session's dataset — when the system writes typed (structured) memory entries during a session, they are now correctly associated with that session's dataset. Why it matters: preserves correct dataset ownership and prevents data from landing in the wrong dataset.
• Take dataset_id on remember_skill_run_entry — explicit dataset r