↓ Skip to main content
tomrochette.com
  1. Agents/
  2. Memory/

claude-mem

Author
glm-5.3-flash
History / Edit / PDF / EPUB / MD
Created: / Updated: / Status: finished / Readability: 3 / 23 minutes read (~4501 words)
agent-curated 610 compression 4 fully-ai-generated 715 llm=glm-5.3-flash 620 memory 34 open-source 196 research-note 580 session-memory 5
Table of Contents

claude-mem is an open-source Apache-2.0 plugin that gives coding agents persistent cross-session memory by capturing session activity through lifecycle hooks, compressing it into observations with LLM calls, storing them in local SQLite, and injecting relevant context back into future sessions.

claude-mem is the strongest evidence yet that session memory has crossed from experiment to default expectation, and its cost model is the field’s open question: it captures everything and pays your tokens to compress it.

What it is
#

A TypeScript plugin (Node 20+ or Bun) that hooks session lifecycle events: SessionStart injects context from recent sessions, UserPromptSubmit and PostToolUse capture activity, PreToolUse on Read adds file context, Stop summarizes. A background worker compresses observations through the Claude Agent SDK (or Gemini or OpenRouter), and storage is a local SQLite database with FTS5 full-text search and optional Chroma vector search. Retrieval is exposed as MCP tools plus a mem-search skill using three-layer progressive disclosure, which the project credits with roughly tenfold token savings over fetching full observations. It targets Claude Code first-class, with install paths for OpenCode, Codex CLI, Gemini CLI, Cursor, Windsurf, Antigravity, OpenClaw, and, since v13.34.0, first-party Pi and DeepSeek Harness extensions. Made by Alex Newman (thedotmack), a solo author in the Vercel OSS Program, with a commercial cloud arm at cmem.ai.

Status
#

Very large and fast: about 101.1k stars as of 2026-10-16, with 63,865 npm downloads in the trailing month (2026-09-23 to 2026-10-22).

Star History Chart

Created 2025-08-31, pushed 2026-11-13, latest tagged release v15.18.3 on 2026-11-12 with npm in sync as of 2026-11-13, and 138,988 npm downloads in the trailing month (2026-10-13 to 2026-11-12). v15.10.0 (November 5) shipped the pending train as an onboarding rework (setup now leads with discovering your existing work, rules, and focus from your transcripts), a capture kill switch (CLAUDE_MEM_OBSERVER_ENABLED=false stops new observation and summary calls while keeping existing memories and search), per-source observer provider selection (CLAUDE_MEM_PROVIDER_BY_SOURCE, for example Codex sessions on Codex), a long-lived Claude login via CLAUDE_CODE_OAUTH_TOKEN in the local env file, chained mode overrides, and viewer sync with shared My Work state, plus reliability fixes for worker pool saturation and hung context reads; v15.10.1 (November 6) and v15.11.0 (November 6) followed with installer setup-screen telemetry (nine screens, gated by the existing analytics settings including DO_NOT_TRACK, prompts, filenames, URLs, and credentials excluded) and runtime fixes stopping stale observer work from replaying after quota or authentication recovery; v15.11.1 (November 7) repaired Pro sign-in (the plugin now ships the better-auth dependencies and creates its auth tables), released paused workers’ claims so pending work can expire, clamped over-budget condense replies, fitted SessionStart context under Claude Code’s 10,000-character limit, and wrote the Goose config atomically; and v15.12.0 (November 7) keeps agent screenshots and images by default with a My Work gallery and a keep-images switch (inference off, local lossy-WebP conversion, a 25 GiB per-owner cloud quota), scoped PreToolUse memory lookups to the current platform so one harness no longer gets another’s observations, pinned My Work’s refresh to the configured agent, stopped CLAUDE_MEM_DATA_DIR and CLAUDE_MEM_WORKER_HOST changing through the settings API, and renamed the Claude app plugin to cmem-pro. v15.12.1 (November 8) repaired macOS 12’s system SQLite (statements using RETURNING now read through .all()) and the CMEM Pro sign-in path the v15.11.0 auth rework had broken, and fixed the Cursor installer’s hooks file; v15.13.0 (November 9) fixed Cursor summaries on native Windows and workspaces with _, space, or dot folders, and added an opt-in auto-clear at one third of the context window (work state saved, then context cleared, off by default); v15.14.0 (November 9) rebuilt My Work’s opening around Focus-area cards with bundled concept artwork and taught the worker to self-update through a promoted-release channel; the v15.15 line (November 10) stopped the worker freezing on its first image (every spawn runs bun --no-install and sharp’s wasm build is refused before decode), dropped refused captures instead of replaying them after an outage, and migrated Antigravity’s stale 14.x PreToolUse hook that had denied every tool call after upgrade; v15.16.0 (November 11) let My Work’s refresh run on the user’s own API provider opt-in (installations with no logged-in Claude Code, Codex, or Gemini CLI can now use My Work, and nothing runs until one is chosen) and pulled GitHub MCP pull requests into My Work, with v15.16.1 fixing the manual-save 500; and v15.17.0 (November 11) exported claude-mem/eve, a memory provider for eve, Vercel’s framework for durable backend agents, whose agents recall the user’s coding-agent memory on a cmem.ai key with one search before every turn and matches delivered as escaped data, never instructions. The v15.18 line (November 12) turns recording on by default for eve agents: after each completed turn the provider now sends the prompt, one observation per tool call, and a turn summary to cmem.ai as platform eve, with <private> regions and secret-shaped text removed on the deployment before sending and capture: false restoring read-only, and the v15.18.1 through v15.18.3 repairs keep the train usable (the worker no longer dies at startup on a locked database, auto-update recovers from a busy updater, Windows Gemini CLI hooks run through PowerShell again after v15.18.0 broke all five of them, and Codex hook approvals persist across plugin updates because the installer registers the hooks in the user’s own ~/.codex/hooks.json), while an opt-in local API token joins the worker API and MCP HTTP transport. The README now brands the project Grok Mem (the package is still claude-mem), and v13.24.0 shipped it as two independent Cursor and Grok Bot marketplace plugins alongside the Claude Code install path. The v13.x version line tells you the churn rate: near-daily releases with major-version jumps, which is velocity and breakage risk in the same number. v13.30.0 (2026-10-04) reworked hook delivery so hooks spool each event to disk and return without waiting on the worker, made observer prompts provider-cache-friendly, and dropped the separate per-prompt observer call, and v13.31.0 (2026-10-05) cut SessionStart’s newest-memories query from seconds to milliseconds with new indexes. v13.31.1 (2026-10-06) fixed cloud sync stalling behind Supabase’s firewall (uploads are now compressed before forwarding, and HTML blocks are retried instead of read as bad tokens), and v13.32.0 (2026-10-06) restored the File Read Gate on by default: full-file Reads of observed code files are blocked and served the observation timeline plus outline tools instead, with partial reads exempt so edits never deadlock. Same-day follow-ups: v13.34.0 added first-party Pi and DeepSeek Harness integrations (a self-contained ESM extension under ~/.pi/agent/extensions and a bundled @claude-mem/dsh plugin installed through DSH’s own profile commands), v13.34.1 restored the CMEM Pro public trial to 30 days from the shorter “up to 14 days” wording, and v13.34.2 fixed a worker regression that had broken prompt-cache reuse on OpenAI-compatible observers (OpenRouter, Gemini) since v13.29.0. The 2026-10-12-to-13 train (v13.36.0, v13.36.1, v13.37.0) is the first recorded privacy regression and capture outage in the same week: v13.36.0, a mostly community release, made the observer pause and keep the work instead of dropping memories when the provider refuses (an empty Anthropic credit balance, a blocked organization, or a local model server down) and fixed non-ASCII Windows home folders; v13.36.1 fixed a server-runtime leak where <private> sections still left the hook and were stored in the server’s raw event record until prompt build (the default worker runtime was not affected); and v13.37.0 repaired the marketplace update of v13.36.1 itself, which had shipped built plugin files containing leftover git stash conflict markers, so auto-updated installs failed to start the worker and MCP search and memory capture stopped until upgrade (the npm package was published before the bad change and was unaffected). The 2026-10-13-to-17 train continued at the same clip: v13.38.0 stopped the Windows console-window flashing by having every hook launch claude-mem directly in exec form with 16 community fixes from five contributors, v13.39.0 restored the memory timeline to Codex subagent threads after compaction through a new PostCompact hook, v13.40.0 brought TeamBrain team joins to the terminal (npx claude-mem teambrain join with device-code sign-in and per-project sharing), and v13.41.0 added build, release, and work-state visibility workflow skills that delegate through make-plan, do, and babysit. The same day closed the v13 line: v14.0.0 shipped My Work, a free local dashboard that lists the requests you gave your agents as cards with plain-language summaries, live statuses, and a human review step so an agent’s “done” is never mistaken for a person’s confirmation, and it moved TeamBrain from design-partner pilot to self-serve at $50 per seat per month; v14.0.1 followed within the hour, while cmem.ai/pricing still described TeamBrain as an unpriced pilot when fetched 2026-10-18. v15.0.0 (2026-10-20) opened the next line by making Smart memory search the default recall path, a three-step flow where the agent searches the compact index, inspects context around relevant results, and fetches only selected details, reasoning between the steps, with the old one-call lookup demoted to an explicit dumb_search; v15.0.1 the same day fixed project-scope consistency so observations found through a merged project alias stay reachable across the flow’s steps, and v15.0.2 (2026-10-23) disabled reasoning on observer model calls so reasoning tokens stop consuming the output budget. v15.1.0 (2026-10-25) rebuilt the My Work dashboard as a React viewer on live worker data that is always live, moved per-prompt semantic injection into server mode, added an opt-in loopback HTTP MCP transport with per-session workspaces, and broadened secret redaction to lowercase-named AWS secret keys, PGP private key blocks, and encrypted PKCS#8 keys. Later the same day, v15.2.0 gave fresh installs a browser setup funnel that reads the last 30 days of supported local agent history, filters and redacts it, and uploads the filtered owner prompts to CMEM Cloud for hosted intent extraction before setup finishes, with anonymous installs getting a private installation credential; v15.2.1 (2026-10-25) changed no runtime behavior and only moved npm publishing to OIDC trusted publishing with provenance attestations, fixing a publish path that had failed on every tag since v13.26.1. v15.3.0 (2026-10-26) added Deliverables to the My Work card: the Stop hook copies what a turn produced (decks, reports, pages, attached files) into ~/.claude-mem/deliverables/<project>/ under content-addressed names so the copy outlives the worktree or scratch folder it was made in, the worker serves each copy over loopback, entirely-<private> turns leave no copy, and CLAUDE_MEM_DELIVERABLES_ENABLED=false turns the feature off. v15.4.0 (2026-10-28) rebuilt My Work around Focus areas: a MY FOCUS rail lists your focus areas in priority order with a needs-you count on each, every request sits in exactly one area or in Other, area pages carry a goal sentence with combined progress and the asks not yet started, new areas are suggested automatically with keep, rename, merge, or dismiss per suggestion, and the My Goals page is retired into the area pages. v15.5.0 (2026-10-29) dressed the same surfaces in new components, made the Observer’s batching opt-in for OpenAI-compatible providers, made the Work view’s summary-provider selection explicit, and broadened redaction again: values under bare token, secret, and api_key keys, Authorization bearer tokens, MYSQL_PWD and dbPassword pairs in structured payloads, and Slack app-level (xapp-) and refresh (xoxe-) tokens, alongside Antigravity MCP calls now recorded under their real tool names and a stable id. v15.6.0 the same day moved My Work’s state into one Zustand store split across work, view, rules, and focus slices: page, focus area, and open request drawer are URL-addressable, list views remember filter and sort between visits, and live updates can no longer be overwritten by an older read or published by a stopped source. v15.7.0 (2026-10-30) stopped a quiet cross-agent spill in My Work’s auto-refresh: a refused answer (most often a review over 1,200 characters) used to hand the whole refresh to the next logged-in agent, which on machines with Codex installed meant hidden codex exec runs on the owner’s ChatGPT account, about 230 runs in four days, even with CLAUDE_MEM_PROVIDER=claude set; it also moved the note-taker’s default model to Claude Code’s always-newest-Haiku alias instead of a pinned ID and added a Feedback button. v15.8.0 (2026-11-01) added a cloud viewer host for My Work and work_state sync for every account. v15.9.0 (2026-11-02) added Devin to the install paths (hooks written, tool names understood, search tools registered for Devin CLI and Desktop), a manual Refresh now button on My Work and Past Focus with per-provider refresh model settings, a CLAUDE_MEM_WORKER_URL remote-worker setting that requires https off-machine, and a redaction batch covering Slack and Discord webhook URLs, Stripe webhook secrets, PyPI, SendGrid, DigitalOcean, Shopify, Docker Hub, and GitLab tokens, with a disabled error reporter now also withholding errors already saved on disk and localhost pages from other sites blocked from writing to work-state, dedup, corpus, and advisor routes. The November 4 default-branch train (pending release) routes the observer provider per session source, re-injects context after Devin compacts a session, redacts Linear API keys and Notion integration tokens, recognizes OpenCode tool inputs as agent-authored, preserves actual checkout paths before remapping memories, registers claude-mem in Goose’s Windows config location, chains mode overrides so a language mode can combine with chill, and classifies Codex exits from their recent stderr only.

Strengths
#

  • Local-first by default: everything lives in ~/.claude-mem/, with <private> tags excluding sensitive content and no account required.
  • Token-aware retrieval design: progressive-disclosure search keeps the injection cost bounded instead of dumping raw history.
  • One engine covers the harness landscape, so memory survives switching between Claude Code and OpenCode.
  • Real adoption: the npm and star numbers make it the most-installed dedicated memory layer for coding agents.

Cautions
#

  • Compression runs an LLM call on every tool observation, often 100+ times per session, billed to your own quota; a competing memory author publicly calls the capture-everything family noisy and token-expensive.
  • The cloud tier’s own docs warn that enabling sync uploads observation narratives and full prompt text to your cmem.ai account.
  • The v15.2.0 browser setup funnel makes that upload the default first-run path for fresh installs: the last 30 days of owner prompts leave the machine for CMEM Cloud’s hosted extraction during setup, filtered and redacted by the configured rules but still an outbound default (release notes, 2026-10-25).
  • Operational surface is real: a background worker on a local port, optional Bun, Python, and Chroma dependencies, and a documented npm install -g footgun that installs the SDK without hooks.
  • The README’s association with a third-party CMEM crypto token is a credibility smell some buyers will not overlook.

Pricing
#

The engine is free and Apache-2.0, fully local. The commercial arm restructured on 2026-10-06: CMEM Pro is $30/month for hosted observation and cloud sync (private MCP link, memory generation on the hosted observer) after a 30-day free trial, Heavy is $70/month with more included usage, and CMEM Max is $100/month with the largest allowance plus billed overages, each with a usage allowance and automatic upgrade settings. The former CMEM Cloud $20/month plan and the $333/seat/month Team tier are gone; the team product (TeamBrain) moved to self-serve at $50/seat/month with v14.0.0 on 2026-10-17, though cmem.ai/pricing still listed it as a design-partner pilot with pricing unannounced when fetched 2026-10-18. The hidden cost is compression: those LLM calls draw on your own model subscription or API budget, and no official per-session figure is published.

Price history
#

Date Plan Change Source
2026-09 CMEM Cloud Baseline: engine free (Apache-2.0); CMEM Cloud $20/mo for sync and one private MCP link, Team $333/seat/mo for 3-50 seats. claude-mem.ai
2026-10-06 CMEM Pro, Heavy, CMEM Max CMEM Cloud $20/mo and Team $333/seat/mo replaced: CMEM Pro $30/mo, Heavy $70/mo, CMEM Max $100/mo (allowance tiers with billed overages); TeamBrain pilot has no public price. cmem.ai/pricing
2026-10-06 CMEM Pro 30-day free trial restored as the public offer, from the shorter “up to 14 days” wording; $30/mo unchanged. cmem.ai/pricing
2026-10-17 TeamBrain Self-serve at $50/seat/mo, from a design-partner pilot with no public price; announced in the v14.0.0 release notes, with the pricing page still showing the pilot at fetch. v14.0.0 release

Compared to
#

  • File-based agent memory: free, transparent, version-controlled, best for stable facts; claude-mem automates recall of what happened, which files cannot do.
  • mem0: a memory API for products you build; choose claude-mem when you are the end user of coding agents and want hooks already written.
  • Letta: a framework where memory is a runtime primitive for agents you architect; choose claude-mem to bolt memory onto agents you already run.

Bottom line
#

Recommended for heavy daily Claude Code or OpenCode users who feel the re-explaining tax every morning and accept the token bill for compression. Not for privacy-strict environments, anyone unwilling to run a local worker service, or small projects where a curated CLAUDE.md is still the right answer.

Changes
#

  • 2026-08-30 - Created as a memory note covering session capture, compression, and reinjection, with 92,602 stars recorded.
  • 2026-09-06 - Recorded the README rebrand to Grok Mem and the added Cursor and Grok Bot marketplace plugins.
  • 2026-09-16 - Recorded the npm/GitHub release lag (npm at 13.25.1, the releases page still on v13.24.23) and refreshed stars to 94.0k with 191 open issues and PRs.
  • 2026-09-20 - Added the Price history section tracking price changes in a table, per the new owner rule.
  • 2026-09-22 - The npm-versus-GitHub channel drift closed: GitHub tagged v13.25.3 and npm published 13.25.3 within minutes on 2026-09-21, so the two release channels are back in sync.
  • 2026-09-27 - Refreshed the volatile facts: about 94.8k stars, 312 open issues and PRs, 68,101 trailing-month npm downloads, and release v13.28.0 (2026-09-26) with npm in sync.
  • 2026-10-03 - Recorded GitHub release v13.29.0 (2026-10-03) with npm lagging at 13.28.0 at fetch time, the first channel drift since 2026-09-21; refreshed stars to about 95.2k, 109 open issues and PRs, and 68,674 trailing-month npm downloads.
  • 2026-10-04 - The npm-versus-GitHub channel drift closed: npm published 13.29.0, matching the v13.29.0 GitHub tag; refreshed stars to about 95.6k, 8.5k forks, and 128 open issues and PRs.
  • 2026-10-05 - Recorded the v13.30.0, v13.30.1, and v13.31.0 releases (spooled hooks, cacheable observer prompts, indexed SessionStart) with npm in sync at 13.31.0; refreshed stars to about 96.3k, 105 open issues and PRs, and 68,891 trailing-month downloads.
  • 2026-10-06 - Recorded the v13.31.1 and v13.32.0 releases (the cloud-sync firewall fix, and the File Read Gate restored on by default) with npm in sync at 13.32.0; refreshed stars to about 96.7k, 87 open issues and PRs, and 69,573 trailing-month downloads.
  • 2026-10-06 - The commercial arm restructured its pricing: CMEM Cloud $20/mo and Team $333/seat/mo replaced by CMEM Pro $30/mo, Heavy $70/mo, and CMEM Max $100/mo, with TeamBrain a design-partner pilot without a public price (Price history row); the pricing reference repointed to cmem.ai, which claude-mem.ai now redirects to.
  • 2026-10-07 - Added the thedotmack/claude-mem star history chart to the Status section.
  • 2026-10-07 - Recorded the v13.34.0 through v13.34.2 releases (first-party Pi and DeepSeek Harness integrations, the CMEM Pro trial restored to 30 days, and the observer prompt-cache regression fix) with npm in sync at 13.34.2 and appended the trial-terms Price history row.
  • 2026-10-09 - Recorded the v13.35.0 release (one guided mem_search tool replacing the multi-tool surface, curated text replies instead of raw JSON, an explicit save_memory note tool, a native-memory-lookup bridge for Claude Code and Codex, and an opt-in markdown-note watcher) with npm in sync at 13.35.0; refreshed stars to about 98.7k and downloads to 72,524.
  • 2026-10-13 - Recorded the v13.36.0, v13.36.1, and v13.37.0 releases (the observer pausing instead of dropping memories when the provider refuses, the server-runtime <private> leak fixed, and the broken 13.36.1 marketplace update repaired after it stopped memory capture on auto-updated installs) with npm in sync at 13.37.0; refreshed stars to about 100.2k (past 100k) and trailing-month downloads to 66,625.
  • 2026-10-16 - Recorded the v13.38.0 and v13.39.0 releases (hooks calling claude-mem directly instead of through bash to stop the Windows console flash, sixteen community fixes, and the Codex memory timeline returning to subagent threads after compaction) with npm in sync at 13.39.0; refreshed stars to about 101.1k and trailing-month downloads to 68,478.
  • 2026-10-17 - Recorded the v13.40.0 and v13.41.0 releases (TeamBrain team joins from the terminal with device-code sign-in and per-project sharing, and build, release, and work-state workflow skills delegating through make-plan, do, and babysit) with the Status train narrative extended to the full 10-13-to-17 run and npm re-verified in sync at 13.41.0.
  • 2026-10-18 - Recorded the v14.0.0 and v14.0.1 releases (the free local My Work request dashboard with summaries, live statuses, and a human review step, and TeamBrain moving from design-partner pilot to self-serve at $50/seat/month, Price history row) with npm in sync at 14.0.1 and the pricing page’s remaining pilot wording noted at fetch time.
  • 2026-10-22 - Recorded the v15.0.0 release (Smart memory search now the default recall path: three-step index search, context inspection, and selective fetch with the agent reasoning between steps, and the old one-call lookup demoted to dumb_search) and the v15.0.1 project-scope consistency fix; npm re-verified in sync at 15.0.1.
  • 2026-10-24 - Recorded v15.0.2 (October 23): observer model calls now disable reasoning by default so reasoning tokens stop consuming the output budget, with a logged single retry on empty length-capped replies; npm re-verified in sync at 15.0.2.
  • 2026-10-25 - Recorded v15.1.0 (the React My Work viewer on live worker data that is always live, per-prompt semantic injection in server mode, an opt-in loopback HTTP MCP transport with per-session workspaces, and broader secret redaction) with npm in sync at 15.1.0; trailing-month npm downloads refreshed to 63,865 (2026-09-23 to 2026-10-22).
  • 2026-10-26 - Recorded the same-day v15.2.0 and v15.2.1 releases: the browser install funnel that scans 30 days of local agent history and uploads filtered owner prompts to CMEM Cloud for hosted intent extraction as the default fresh-install path (new caution), and the npm OIDC trusted-publishing fix that repairs a publish path broken since v13.26.1; npm re-verified in sync at 15.2.1.
  • 2026-10-27 - Recorded the v15.3.0 release (October 26): Deliverables on the My Work card, the Stop hook copying turn output into content-addressed project folders served over loopback, with private turns excluded and an off-switch; npm re-verified in sync at 15.3.0.
  • 2026-10-28 - Recorded v15.4.0: My Work rebuilt around Focus areas (a priority-ordered MY FOCUS rail with needs-you counts, every request in exactly one area, goal-sentence area pages, and auto-suggested areas with keep, rename, merge, or dismiss), retiring the My Goals page; npm re-verified in sync at 15.4.0.
  • 2026-10-29 - Recorded v15.5.0: My Work and My Focus UI component refresh, opt-in OpenAI-compatible batching in the Observer, explicit summary-provider selection, and redaction broadened again (bare token, secret, and api_key keys, Authorization bearer tokens, MYSQL_PWD and dbPassword pairs, and Slack xapp- and xoxe- tokens); npm re-verified in sync at 15.5.0.
  • 2026-10-30 - Recorded v15.6.0 (October 29): My Work’s state unified in one Zustand store with URL-addressable views, remembered filters and sort, and steadier live updates; npm re-verified in sync at 15.6.0, trailing-month downloads refreshed to 75,512 (2026-09-29 to 2026-10-28).
  • 2026-11-01 - Recorded v15.7.0 and v15.8.0: the My Work auto-refresh no longer spills refused runs to the next logged-in agent (the fix for the hidden codex exec runs on owners’ ChatGPT accounts, about 230 in four days), the note-taker default moved to the newest-Haiku alias, and My Work gained a cloud viewer host with work_state sync for every account; npm re-verified in sync at 15.8.0, trailing-month downloads refreshed to 74,755 (2026-10-02 to 2026-11-01).
  • 2026-11-03 - Recorded v15.9.0 (November 2): Devin CLI and Desktop join the install paths, My Work gains a manual Refresh now button with per-provider refresh model settings, a remote-worker URL setting requires https off-machine, and redaction covers Slack and Discord webhooks, Stripe, PyPI, SendGrid, DigitalOcean, Shopify, Docker Hub, and GitLab tokens; npm re-verified in sync at 15.9.0, trailing-month downloads refreshed to 86,399 (2026-10-04 to 2026-11-03).
  • 2026-11-04 - Recorded the pending post-v15.9.0 train (November 4): per-session observer provider routing, Devin compaction re-injection, Linear and Notion token redaction, OpenCode tool inputs recognized as agent-authored, checkout-path preservation before memory remapping, the Goose Windows config registration, and language-mode override chaining; npm still in sync at 15.9.0, trailing-month downloads refreshed to 89,686 (2026-10-05 to 11-04), and the push date refreshed to 2026-11-04.
  • 2026-11-07 - Recorded the v15.10 and v15.11 lines (November 5-6): v15.10.0 shipped the onboarding discovery rework, a capture kill switch (CLAUDE_MEM_OBSERVER_ENABLED), per-source observer providers, a long-lived Claude login env token, and chained mode overrides; v15.10.1 and v15.11.0 added installer setup-screen telemetry and the fix stopping stale observer work replaying after quota or authentication recovery; the push date refreshed to 2026-11-07.
  • 2026-11-08 - Recorded v15.11.1 and v15.12.0 (both November 7): the Pro sign-in repair (better-auth dependencies shipped and auth tables created), paused workers releasing their claims, SessionStart context fitted under the 10,000-character limit, and atomic Goose config; the 2.0-line images feature (screenshots and agent images kept by default with a My Work gallery, local WebP conversion, a 25 GiB per-owner cloud quota, inference off), PreToolUse memory lookups scoped per platform, My Work’s refresh pinned to the configured agent, settings-API restrictions, and the Claude app plugin renamed cmem-pro; npm re-verified in sync at 15.12.0, trailing-month downloads refreshed to 109,431 (2026-10-09 to 11-08), and the push date refreshed to 2026-11-08.
  • 2026-11-11 - Recorded the v15.12.1 through v15.17.0 train (November 8 to 11): the macOS 12 SQLite and CMEM Pro sign-in repairs, the Cursor summary and installer fixes, the opt-in auto-clear at a third of context, the Focus-area opening with self-updating worker, the worker image-freeze and outage-replay fixes with the Antigravity stale-hook migration, My Work refresh on the user’s own API provider with GitHub MCP pull requests, the manual-save fix, and the claude-mem/eve memory-provider export for Vercel’s eve backend-agent framework; npm re-verified in sync at 15.17.0, trailing-month downloads refreshed to 115,168 (2026-10-12 to 11-11), and the push date refreshed to 2026-11-11.
  • 2026-11-13 - Recorded the v15.18.0 through v15.18.3 train (November 12): eve agents now record by default (prompt, per-tool observations, and turn summaries to cmem.ai with on-deployment redaction, capture: false restores read-only), the worker’s locked-database startup and auto-update repairs, Windows Gemini CLI hooks restored (a PowerShell parser error had silenced all five hooks since v15.18.0), Codex hook approvals persisted across updates through the user’s own hooks.json, and the opt-in local API token for the worker API and MCP HTTP transport; npm re-verified in sync at 15.18.3, trailing-month downloads refreshed to 138,988 (2026-10-13 to 11-12), and the push date refreshed to 2026-11-13.

See also
#

References
#