> ## Documentation Index > Fetch the complete documentation index at: https://docs.evermind.ai/llms.txt > Use this file to discover all available pages before exploring further. # Product Updates #### Unified v2 Memory API: one contract, Cloud or self-hosted EverOS Cloud now runs on the same v2 Memory API as open-source EverOS. Prototype locally, ship on managed Cloud, or switch between them for compliance, without rewriting your integration or getting locked into a black box. * **Write once, run either place**\ `add`, `flush`, `get`, and `search` behave identically on Cloud and self-hosted. * **Multi-party memory out of the box**\ Every message carries a `sender_id`, so group chats and human-plus-agent conversations attribute memory to the right participant automatically. #### Retrieve the right context * **Four retrieval strategies**\ Choose keyword, vector, hybrid, or agentic search per query, and scope any retrieval to one `user_id` or `agent_id` with time and field filters. #### Agent memory, not just user memory * **Cases and skills as first-class memory types**\ Alongside user profiles and episodes, retrieve what an agent has learned about itself, so your agents get better at a task the more they run it. #### Full lifecycle control Memory is a governable asset, not a write-once black box. * **Edit** a user's profile directly; your manual edits survive future background consolidation. * **Delete** by user, agent, or session to meet privacy and retention needs. #### Compatibility * The v1 API remains available for existing integrations. See the [Cloud Quickstart](/cloud/quickstart) and [API Reference](/api-reference/introduction). #### Request Log Improvements: * Find requests faster with request ID filtering, custom time-range search, and pagination. * Click any entry to view its status, latency, and error messages. #### API v0 Retirement: * The v0 API is now fully replaced by v1. Calls to v0 endpoints will return an error. * Existing data is migrated automatically, so users already on v1 are unaffected. * See the "API Reference (Cloud)" for the migration guide. #### Streamlined Onboarding: * **Simpler, friendlier setup**\ A more streamlined onboarding experience that's friendlier for users of any technical level. * **Helpful tips while you wait**\ While your memory space is being set up, we share helpful tips tailored to you to get you started. * **Refreshed look**\ A cleaner, fully branded onboarding experience. #### Console UI/UX Refresh: * **Zero-setup onboarding**\ **Copy a single prompt, paste it into any agent, and you're done.** No manual install, no API key creation, all automated. * **Redesigned Get Started page**\ Switch between Prompt (for agents) and SDK setup modes, with ready-to-use `add` / `get` / `search` code samples for both Agent and User memory. * **Clearer navigation**\ Reorganized sidebar tabs so you can find what you need faster. #### Performance Improvmenets: * System-level engineering optimizations for improved speed and stability #### New Features & Updates: * **API path set to v1**\ The EverOS API has been upgraded to v1 with improved structure and new capabilities. See the "API Reference" for full details and migration guide * **Multimodal storage and retrieval (mRAG)**\ Store and query images, documents, PDFs and web contents alongside text * **Self-evolving agent memory**\ Agents can now build and refine memory through Cases and Skills primitives * **Memory Bank**\ Visualize and manage user, group, and agent memory from a single console * **Interactive Playground**\ Explore and test memory APIs directly in the console, no setup required #### New Features & Updates: * Dashboard UI updates * Performance improvements * Bug fixes and reliability improvements #### New Features & Updates: * Message timestamp validation #### Bug Fixes: * Fixed dashboard API usage chart data anomaly #### New Features & Updates: * Refactored Get Memories and Search Memories API parameters * Dashboard UI interaction improvements * Added Google and GitHub third-party login support * Restricted available LLM providers and models on cloud platform * Bug fixes This is a non-backwards-compatible change. Update your API request parameters immediately. #### New Features & Updates: * **API path set to v0**\ All endpoints are served under `/api/v0/`. Requests to any other path are not supported. #### A new ergonomic client for the v2 API `everos-cloud` 1.0.0 wraps the v2 Memory API in a flat, ergonomic client: plain dicts in, response data out. * **Top-level methods**\ Call `client.add`, `client.flush`, `client.search`, `client.get`, `client.edit`, and `client.delete` directly, instead of the nested `client.v1.memories.*` namespace. * **Multimodal upload helper**\ `client.upload("photo.jpg")` presigns, uploads to storage, and returns the object key to reference in a message. * **Typed low-level client still available**\ The generated typed client is exposed as `client.memory` / `client.storage` when you need full control. * **Clearer exceptions**\ Errors raise `EverOSError`; HTTP failures raise `EverOSAPIError` with `.status` and `.body`. #### Breaking changes from 0.4.x * Targets the v2 Memory API: messages use per-message `sender_id`, timestamps are unix milliseconds, and `get` / `search` require a `user_id` or `agent_id`. * The client is synchronous. The previous `AsyncEverOS` client and task-polling APIs are removed. ```bash theme={null} pip install --upgrade everos-cloud ``` See the [Cloud Quickstart](/cloud/quickstart) and the [Cookbook](/cookbook/quickstart) for updated examples. #### Cloud SDK package renamed to `everos-cloud`: The Cloud Python SDK is now published as **`everos-cloud`**. Existing installations are unaffected; new installs should use the updated package name. ```bash theme={null} pip install everos-cloud ``` #### New Features & Updates: * **v1 API support**\ SDK updated to align with EverOS API v1; all endpoints are now accessible via the client.v1 namespace * **Multimodal storage and retrieval (mRAG)**\ Added multimodal support to memories.add(), enabling ingestion of images, PDFs, documents, and web content alongside text * **Self-evolving agent memory**\ Introduced Cases & Skills APIs to support structured storage and retrieval of self-evolving agent memory * **Async client**\ Added AsyncEverOS client with async/await support, with optional aiohttp backend for improved concurrency #### Bug Fixes * Fixed group\_ids parameter parsing error ```bash theme={null} pip install --upgrade evermemos ``` #### New Features & Updates: * **Core Memory Operation** * Updated memory types enum values * Search now accepts a list of group IDs * Get memories uses page-based pagination * **LLM Custom Settings** * Available provider: openrouter * Available models: qwen/qwen3-235b-a22b-2507 (recommended), openai/gpt-4.1-mini This is a non-backwards-compatible change. Update your SDK immediately. ```bash theme={null} pip install --upgrade evermemos ``` #### EverOS Python SDK v0.3.11 is now available: * Please upgrade to the latest version, as v0.3.10 and earlier are not compatible. #### Fresh Installs Work Again * The `openai` SDK is now pinned below 3.x. A fresh install on or after 2026-09-24 could pull in `openai` 3.x, which made every LLM and embedding call fail with a 500. Environments built from `uv.lock` were not affected. ```bash theme={null} pip install --upgrade everos ``` #### Native Windows Support * `pip install everos` now works on a stock Windows 11 machine. You don't need the Visual C++ Redistributable or WSL2, just an LLM key. #### Faster Vector Search * **Automatic vector indexes**\ Once a table holds enough vectors (2,000 by default), EverOS builds an ANN index on it in the background. Hybrid search no longer scans the whole column on every query. In our benchmark, query latency dropped from 245 ms to 84 ms with 0.98 recall\@10. #### Other Changes * **Safer index maintenance**\ `everos cascade sync`, `fix --apply` and `rebuild` now refuse to run while a server is running (exit code 3). Before, running them alongside the server could write every row twice. * **Reliability fixes under heavy write load**\ File changes are now indexed in the order they happen. Temporary storage errors are retried instead of marking the file as permanently failed. On first start, the server builds vector indexes for large tables. This takes about a second per 100k rows, and search keeps working meanwhile. If your store predates the `subject_vector` column, startup fails with `LanceDBSchemaMismatchError`. To fix it, stop the server and run `EVEROS_ROOT= everos cascade rebuild`. ```bash theme={null} pip install --upgrade everos ``` #### LLM-Guided Multi-Round Retrieval * **New `llm_multiround` search method**\ Set `method = "llm_multiround"` on `POST /api/v2/memory/search`. EverOS then searches in several rounds, and an LLM keeps the key evidence and writes follow-up queries to fill the gaps. You don't need a reranker. * **Separate decider model**\ Choose the model for this in the new `[decider]` section. Leave it empty and it uses your `[llm]` settings. #### Reproducible Benchmarks * One runner now covers LoCoMo, LongMemEval, EverMemBench and SubtleMemory. It shares staging, resumable runs and metrics across all four. #### Other Changes * **Bounded background jobs**\ Each OME strategy run now times out after 1,800 seconds by default. If a strategy legitimately runs longer, raise `EVEROS_OME_RUN_TIMEOUT_SECONDS` before you upgrade. * Embedding requests that fail permanently (for example, 401 or 422) are no longer retried. ```bash theme={null} pip install --upgrade everos ``` #### Milvus as an Optional Index Backend * **Scale out to Milvus or Zilliz Cloud**\ Install `everos[milvus]` and set `[index] backend = "milvus"`. LanceDB remains the default. Markdown is still the source of truth, so switching backends means a rebuild, not a data migration. * **Schema checked at startup**\ A collection that doesn't match the expected schema fails right away, not on the first write. Embedded Milvus Lite is not supported. #### Other Changes * **Consistent cosine distance**\ Vector search on LanceDB now uses cosine distance, like everywhere else. Rankings may shift slightly, but you don't need to rebuild. * **Fixed `memory/add` failing on fresh installs**\ In the default agent mode, `memory/add` returned a 500 caused by an incompatible dependency pair. The dependencies are now pinned. ```bash theme={null} pip install --upgrade everos ``` #### Agent Skill Extraction Fixes * **Skill extraction now works end to end**\ Fixed a race with background indexing that could prevent agent trajectories from producing skills, plus path-safety hardening for LLM-generated skill names. #### Other Changes * **Background maintenance is now supervised**\ Background loops restart automatically on crashes and report stalls in `GET /health`, instead of silently stopping. Keyword search stays available during index rebuilds. * **`extract_foresight` now ships disabled**\ Nothing consumes foresight output yet, so it is off by default to save tokens. Re-enable in `ome.toml` if you rely on it. ```bash theme={null} pip install --upgrade everos ``` #### Storage-Layer Reliability * **Fixed unbounded index disk growth**\ Index version cleanup no longer loses races against concurrent writes, and every storage operation runs under a deadline, so nothing can silently wedge a table. * **See and recover from index faults**\ `GET /health` gains a `cascade` readiness block you can alert on, and the new `everos cascade rebuild` command rebuilds a drifted or corrupt index from your markdown files. ```bash theme={null} pip install --upgrade everos ``` #### Run with Just an LLM * **`[embedding]` and `[rerank]` are now optional**\ EverOS boots with only `[llm]` configured and degrades gracefully: keyword search works out of the box, and adding `[embedding]` / `[rerank]` unlocks vector, hybrid, and agentic search. The new `everos cascade backfill` command fills in vectors for rows written before you added an embedding provider. #### API v2 Is Now Canonical * Docs and examples now use `/api/v2`. `/api/v1` keeps working as a legacy alias. This release fixes a path-traversal vulnerability in knowledge document upload. Versions before 1.1.4, and 1.2.0, are affected. Upgrade to 1.2.1 or later. ```bash theme={null} pip install --upgrade everos ``` #### OpenTelemetry Tracing * **See inside every memory operation**\ Native tracing exports to OTLP backends like Langfuse, covering add/flush, boundary processing, episode extraction, search, and reflection. * **Rich telemetry**\ Captures LLM and embedding token usage, per-request correlation, and recall-quality scores, so you can debug and measure memory quality in production. * **Off by default, privacy-aware**\ Enable it through the `[observability]` config with the optional `otel` extra. Content capture is opt-in and respects redaction settings. #### API v2 * **One codebase across OSS and Cloud**\ Business endpoints (`memory/*`, `ome/*`, `knowledge/*`) are now available under `/api/v2`, matching the EverOS Cloud API contract, so you can move between self-hosted and Cloud without rewriting integrations. * **Nothing breaks today**\ Existing `/api/v1` calls continue to work unchanged, so you can migrate to `/api/v2` at your own pace. ```bash theme={null} pip install --upgrade everos ``` #### Observability and Stability * **Langfuse tracing example**\ A new Langfuse OpenTelemetry integration example shows how to trace add, flush, extraction, search, and reflection operations end to end. #### Bug Fixes * **Reliable indexing when files are deleted**\ Deleting a file that was just edited no longer leaves stale search results or stalls the background processing queue. ```bash theme={null} pip install --upgrade everos ``` #### Bug Fixes * **Disk usage growth**\ Fixed an issue where index fragment merges could fail, causing disk usage to grow rapidly without being reclaimed. ```bash theme={null} pip install --upgrade everos ``` #### Bug Fixes * **Agent search reliability**\ Fixed an issue where agentic search could become unavailable in certain cases. ```bash theme={null} pip install --upgrade everos ``` #### Benchmarks and Search Quality EverOS OSS v1.1.1 is a quality release focused on reproducible benchmarks and improved search accuracy. * **LoCoMo benchmark runner**\ Reproduce EverOS memory benchmark results yourself with the new runner under [`benchmarks/`](https://github.com/EverMind-AI/EverOS/tree/main/benchmarks): TOML configuration, staged ingestion / search / answer / judge phases, and result artifacts. See [`benchmarks/README.md`](https://github.com/EverMind-AI/EverOS/blob/main/benchmarks/README.md) to get started. * **Improved hybrid search**\ Heap-driven lazy expansion and global top-N competition raise retrieval quality, with stable LoCoMo scores around 91% for hybrid search and 93% for agentic search. * **Python 3.12 and 3.13 support**\ CI coverage now includes Python 3.12 and 3.13. #### Bug Fixes * Concurrent Knowledge cascade upserts now use an atomic insert-or-update path * The OpenAPI application version stays in sync with the package version * Profile middleware re-raises inner handler exceptions instead of swallowing them * Reduced unnecessary LanceDB optimize I/O by throttling cascade optimize calls ```bash theme={null} pip install --upgrade everos ``` #### Knowledge Wiki and Reflection EverOS OSS v1.1.0 adds two major capabilities: a **Knowledge Wiki** that turns uploaded documents into searchable memory, and **Reflection** that consolidates scattered memories into clean, coherent narratives. #### Knowledge Wiki A third memory track alongside user and agent memory: upload documents (PDF, DOCX, Markdown, and more) and EverOS turns them into a searchable topic library. * **Progressive L0 → L1 → L2 structure**\ An LLM extracts each document into a category → document → topic hierarchy, so agents drill down on demand. * **Deterministic classification**\ Auto-sorted into an editable taxonomy of 20 built-in categories, predictable and overridable. * **Hybrid search, human-readable storage**\ Keyword, vector, or hybrid retrieval, with every document stored as editable Markdown and the original preserved. #### Reflection An offline background process that merges fragmented, overlapping memories into one coherent narrative. * **Cleaner memory over time**\ Related episodes merge into a single chronological narrative, resolving stale and contradictory fragments by keeping the latest state. * **Originals are never lost**\ Replaced memories are soft-archived rather than deleted, and stay fully traceable on disk. * **Off by default, you stay in control**\ Enable in `ome.toml`; runs on a weekly schedule and never touches the live conversation path. #### Getting Started ```bash theme={null} pip install --upgrade everos ``` See [Knowledge Wiki](/open-source/wiki) and [Reflection](/open-source/reflection) for full guides. #### Security release A path-traversal vulnerability in the memory-ingestion endpoint (`POST /api/v1/memory/add`) has been fixed. **All v1.0.0 users should upgrade.** * **`sender_id` path-safety validation** — `sender_id` now carries the same path-safety guard as `app_id` / `project_id`: a character whitelist plus rejection of `.` / `..`. The whitelist admits `@` and `+`, so email-style and plus-addressed identifiers still work. * **Defense-in-depth write containment** — the Markdown writer now rejects any write that resolves outside the configured memory root, before any filesystem access (covering both the write and the append read-modify-write paths). * **Severity:** High — unauthorized arbitrary `.md` file write on affected versions (≤ 1.0.0). No workaround other than upgrading. ```bash theme={null} pip install --upgrade everos ``` #### Initial Release — Agent-First Memory System EverOS OSS v1.0.0 is an agent-first, self-hosted memory system designed for full local deployment. Install with a single command — no Docker, no external databases, no infrastructure to manage. #### Markdown-First Storage * **Memory as plain Markdown files**\ Every memory record lives as a human-readable `.md` file on disk. Read, edit, version with Git, or browse in Obsidian — the data is fully yours. What you see is what is stored. * **Auditable and user-controlled**\ Memory files can be inspected and corrected directly without going through an API. Full data sovereignty. * **Lightweight embedded stack**\ Markdown (truth) + SQLite (state) + LanceDB (index). All retrieval — vector search, BM25 keyword, and scalar filtering — runs locally. Wipe `.index/` and the daemon rebuilds everything from the Markdown alone. #### Memory Classification * **Three memory tracks** — User Memory, Agent Memory, and Wiki Memory (coming soon) — organized by who owns the memory: the user, the agent, or shared world knowledge * **Multi-granularity coexistence** — episodes, atomic facts, foresight, and user profile exist side by side; the system selects the right granularity at retrieval time * **Independent memory scoping** — memory owner (`user_id` / `agent_id`), `app_id`, `project_id`, and `session_id` are orthogonal dimensions; add and search by any combination #### Self-Evolving Agent Memory * **Experience loop** — every agent execution is recorded as a case; cases are periodically clustered and distilled into reusable skills * **The system gets smarter with use** — skills generalize from individual traces into structured, retrievable knowledge about how to approach a class of task #### Online and Offline Memory Strategies * **Online strategy** — real-time memory ingestion and boundary detection during active sessions * **Offline strategy (OME)** — an in-process async engine that derives atomic facts, foresight, profile updates, and agent skills from accumulated episodes in the background * Each stage runs with configurable prompts and models #### Multimodal Support * Ingest and extract memories from: `text`, `image`, `audio`, `doc`, `pdf`, `html`, `email` * Multimodal content is processed alongside text through the same extraction pipeline #### Getting Started ```bash theme={null} pip install everos everos init # generate .env config everos server start ``` See the [OSS Quickstart](/open-source/quickstart) for a full walkthrough. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.