OpenClaw Newsletter - 2026-06-18
OpenClaw Newsletter
Thursday, June 18, 2026Top Stories
- fix(telegram): replace heading tags with <b> to fix text overlap on macOS desktop — Summary Telegram macOS Desktop app renders inline keyboard messages with overlapping text layers because - HTML tags are not properly supported. Same messages render correctly on iOS/Android. Changes...
- feat(gateway-cli): scope usage-cost by agent — Summary openclaw gateway usage-cost only sent { days } to the usage.cost gateway RPC (src/cli/gateway-cli/register.ts:535), even though the handler already accepts agent scoping...
- fix(feishu): add pagination support for drive list and info — Summary Feishu/Lark drive list and info only checked the first page of results (default page size 10). list silently returned incomplete results, and info threw "File not found" for files on...
- fix: always show raw config in settings (fixes #94202) — Previously, the raw config textarea was hidden when sensitive values were detected, making the JSON completely invisible. This change forces the textarea to always be visible, aligning with user...
- chore: sync yuanbao plugin catalog to 2.15.0 — Summary - Sync the Yuanbao official external channel catalog pin from openclaw-plugin-yuanbao@2.13.1 to @2.15.0, with the matching expectedIntegrity updated to the 2.15.0 npm registry value. - Keep...
- fix: inbound metadata leakage — missing sentinels and display gaps — Summary Inbound metadata blocks are not properly stripped in certain delivery paths, causing internal runtime context and conversation metadata to leak into user-visible chat (Feishu, Dashboard,...
- fix(provider-usage): honor proxy env for usage fetch — Closes #78714. Summary: - Prefer an explicit provider-usage fetch when supplied. - Otherwise use the existing env proxy fetch helper before falling back to global fetch, so Codex usage snapshots...
- fix(agents): re-expose toolsAllow core tools in embedded runs — Summary Describe the problem and fix in 2–5 bullets: - Problem: embedded runs only post-filtered toolsAllow after createOpenClawCodingTools, so restrictive config allowlists could remove...
- edit tool should classify concurrent identical edits as noop and distinguish them from both success and failure — Problem The current edit behavior does not clearly distinguish these cases when identical edits hit the same file concurrently: 1. one request actually performed the mutation 2. another identical...
- [Feature]: [Feishu] Card header/footer customization: hide agent label, show elapsed time & completion status, support reply-to context — Summary Allow Feishu card header and footer to be customized: hide the static agent ID label, display response elapsed time and completion status, and optionally show reply-to context in the header....
- fix(doctor): prevent non-interactive --fix from auto-restarting gateway — Summary What problem does this PR solve? doctor --fix --non-interactive needed to stop auto-restarting an already-running gateway after a transient health/probe failure, but the earlier PR branch...
- Google provider: implement proper Vertex AI OAuth and endpoint routing — Summary The google-vertex hook alias exists but is a no-op — it routes through the same Gemini API OAuth flow with no Vertex AI-specific behavior. The Google provider needs proper Vertex AI support....
- [Feature]: Per-agent or per-binding session reset policy — Summary Currently, session reset policies (session.reset, session.resetByType, session.resetByChannel) are global or channel-wide. There is no way to configure session reset behavior per agent or per...
- Doctor warns about NVM node instead of Homebrew/system node, but cannot be fixed — Problem OpenClaw doctor warns about Gateway using NVM node instead of Homebrew/system node, but the warning cannot be resolved because: 1. OpenClaw regenerates the launchd plist file on each restart...
- Feature Request: Per-Agent Bedrock requestMetadata Injection for Multi-Agent Cost Attribution — Summary When running multiple agents via OpenClaw's multi-agent architecture, there's no way to attribute Bedrock API costs to individual agents. All calls appear as the same IAM role with no...
- fix(memory-wiki): reserve compiler-owned page stems (index, log) in slug layer — Summary A memory-wiki create_synthesis or source ingest whose title slugifies to a compiler-reserved stem (index or log) silently loses the stored fact. The page is written to /index.md, which is the...
- Discord: support category-level channel config in guild allowlist — Problem Currently, channels.discord.guilds..channels only accepts individual channel IDs. Each new channel must be manually added to the config. There's no way to apply settings (like requireMention:...
- Multi-Slot Memory Architecture — Summary Replace the single plugins.slots.memory slot with multiple purpose-specific memory slots, allowing different memory providers to handle distinct layers of the memory stack simultaneously....
- fix(security): close active April 4 transport and auth gaps — Summary - Problem: config-time explicit provider proxies accepted blocked internal hosts, config.set auth rotation left shared-auth sessions alive, and Discord outbound text could resolve @everyone /...
- fix(agents): add channelData marker to exec-approval-unavailable payloads — Summary - Problem: buildExecApprovalUnavailableReplyPayload returns { text: "..." } with no channelData marker, making exec-approval-unavailable payloads indistinguishable from regular assistant text...
- Telegram bot cannot fetch media/images - MediaFetchError DNS issue — Problem Description When receiving images via Telegram bot, the media fetch fails with: MediaFetchError: Failed to fetch media from https://api.telegram.org/file/bot.../photos/file_xx.jpg:...
- [Bug]: bug(gateway): offloadedRefs metadata lost in transcript for iOS share/node path — Bug type Regression (worked before, now fails) Beta release blocker No Summary Metadata in parsed.offloadedRefs is not persisted to JSONL transcript for iOS share/node path, causing media to be...
- Feature request: support native per-session heartbeat scheduling — Summary OpenClaw heartbeat currently feels like a single proactive lane per agent. That works well when an agent only needs one heartbeat context, but it becomes awkward when the same agent needs to...
- Cron job reports failure when using env var for Discord channel, even on success — Bug Description When a cron job uses an environment variable (e.g., env:DISCORD_BACKUPS_CHANNEL) for the Discord channel in the delivery configuration, the job is incorrectly marked as "error" even...
- [Feature]: Add threadPromptGuideline for customizable auto-generated Discord thread titles — Summary Add an optional threadPromptGuideline configuration field to customize the LLM prompt used for generating Discord thread titles when autoThreadName: 'generated'. PR related :...
- [codex] Cron: repair empty sanitized final replies — ???Summary - Problem: cron-owned agent runs could finish with no deliverable final text after sanitization and end up as silent not-delivered outcomes. - Why it matters: scheduled reports can vanish...
- fix: use linked task activity for TaskFlow audit freshness — Summary - use linked task activity when computing TaskFlow audit freshness - avoid stale_running findings when a managed flow is still progressing via linked child task activity - add regression...
- [Feature]: Per-account media inbound directory for multi-account Feishu — Summary Per-account media inbound directory for multi-account Feishu Problem to solve When multiple Feishu accounts are configured in OpenClaw (e.g. Gauss + Rita as separate enterprise apps), all...
- docs(memory): add memory system deep dive guide — Summary Add a comprehensive memory system deep dive guide, addressing #43032. Problem docs/concepts/memory.md covers basics (~600 words) but lacks depth on memory best practices, search optimization,...
- OpenClaw repo stats for 2026-06-18 — 379309 stars, 79408 forks, 2599 contributors
- OpenClaw sponsors (108 total) — 108 sponsors supporting OpenClaw
- fix: prune processed skill stat events
- fix: remove reply-to from transactional emails (#2725)
- fix: stabilize sparse skill category pagination (#2724) — Merged after exact-head maintainer review. Prepared head SHA: 66d9662023fd3e3c36ceb541f11c14b98f9ed76b Autoreview: clean, no accepted/actionable findings. Co-authored-by: fuller-stack-dev...
- fix: encode R2 backup object keys for SigV4
- docs: add namespace claims guide — Adds a dedicated ClawHub namespace claims guide and routes existing claim references through it.
- feat: add catalog classification backfill (#2719) — feat: add controlled catalog taxonomy and topics refactor: limit taxonomy input to publish surfaces fix: omit inactive official-first filter fix: preserve unsaved catalog metadata edits fix: enforce...
- fix: redirect sign-in to dashboard (#2718)
- fix: use installs instead of downloads on social cards (#2713)
- Polish ClawHub header and search UX (#2717) — feat: sync header design polish fix: restore header search hints fix: strengthen search tab affordance Revert "fix: strengthen search tab affordance" This reverts commit...
- feat: add controlled catalog taxonomy and topics (#2699)
- Add deprecated download dedupe cleanup (#2715) — feat: add deprecated download dedupe cleanup chore: refresh vulnerable transitive overrides
- fix: remove global slug reservation delete copy (#2716)
- fix: bound rate limit counter storage
- fix: parallelize skill artifact backup uploads
- feat: scope skill slugs by owner (#2299) — feat: scope skill slugs by owner test: cover owner-scoped skill slug publishes feat: scope skill slugs by owner fix: restore owner-scoped skill ci * fix: keep package names separate from skill slugs
- fix: sanitize ClawPack repair metadata (#2710)
- openclaw-cli 2026.6.8 on Homebrew — Your own personal AI assistant
- ykarout/Qwen3.5-9b-Opus-Openclaw-Distilled-GGUF — text-generation
- mradermacher/OpenCLAW-SEED-135M-GGUF — text-generation
- OpenClaw on DigitalOcean — Deploy your own personal AI assistant with OpenClaw on a DigitalOcean Droplet® server, a powerful open-source platform that runs entirely on your infrastructure. OpenClaw connects to the messaging...
- OpenClaw – Moltbot Renamed Again
- OpenClaw privilege escalation vulnerability
- Ask HN: Who is using OpenClaw?
- I Run a Self-Improvement Loop on My OpenClaw Agent Every Night. Here's What I Learned. — Last month my OpenClaw agent kept making the same mistake: it would run a health check, the script...
- Cómo construí un agente de IA que trabaja 24/7 (y lo que aprendí en el camino)
- AI Assistant Architecture: LLM, Memory, Tools, Routing, Observability — A production AI assistant is not "an LLM with a prompt". It is a system that accepts intent, keeps...
- Memory Systems in AI Assistants — Memory turns assistants from reactive to persistent, but it is also where many systems quietly rot....
- Your OpenClaw Agent Says 'Done.' Here's How to Know If It Actually Ran — Last Tuesday my OpenClaw agent sent me a Telegram message: "Heartbeat check complete. All services...
- Reflection: From AI Assistant to AI Colleague - Lessons Learned from Internal Sharing — After conducting the internal sharing session on AI Agents with a non-developer group, the biggest...
- OpenClaw Agent Masterclass — Full Tutorial — Everything you need to install, configure, and extend OpenClaw — the open-source personal AI...
- DevFortress Deep Digest 2: The Month It Got Names — OpenClaw. Moltbook. CVE-2026-25253. January 24 – February 23, 2026 1.5...
- Openclaw on the cloud, dos and don'ts — running openclaw or hermes locally sounds nice, but how about running on the cloud?
- I Stopped Storing Everything in AI Memory — One thing bothered me after using AI agents for months:Continue reading on Medium »
- DevFortress Deep Digest 2: The Month It Got Names — OpenClaw. Moltbook. CVE-2026–25253.Continue reading on Medium »
- OpenClaw Deep Dive: เมื่อ Coding Agent อยากเป็นมากกว่าแค่คนเขียน Code — ใน Session “OpenClaw Deep Dive: เมื่อ Coding Agent...
- Khoj — Open-source app offers local or cloud AI chat with major LLMs, advanced semantic search, automation, personal newsletters, document indexing, image generation, and text-to-speech. Access from web,...
- ChatSonic — Combining Google's knowledge graph, ChatSonic delivers real-time data, voice interaction, AI-produced visual content, personalized avatars, and API for services. It revisits past dialogs for informed...
- LlamaGPT — LlamaGPT is a chatbot that provides a ChatGPT-like experience, with no data leaving your device.
- https://t.co/SiK5jSYqn3
- @cyberphysicsai @openclaw @NanoClaw_AI Amazing what a small skill can do when it has some of the Cyphertempre architectu
- @daniel_nguyenx @openclaw Bro what is this ? Use the real clawbox --> https://t.co/rUYWktGyuj
- Claw developers here is your chance to build the next big thing in the claw eco-system. Dm for inquiries. Serious offers
- 4/ OpenClaw 这一季的隐形主线 = 运行时治理： • 逃离 main 监狱（可配置默认 agent ID） • 非阻塞 heartbeat • 插件 API 中止 agent 运行 • 修复 Docker 内 skill 挂载 — 4/ OpenClaw 这一季的隐形主线 = 运行时治理： • 逃离 main 监狱（可配置默认 agent ID） • 非阻塞 heartbeat • 插件 API 中止 agent 运行 • 修复 Docker 内 skill 挂载 几乎全是治理，而非堆功能。
- 1/ OpenClaw v2026.6.8 落地： 不是更强模型，而是"少出错的交付"。 • Telegram/WhatsApp 结构化渲染 • 账号维度 DM、生成媒体补全、重启中止恢复 • 新增 GLM-5.2 / Claude Ha — 1/ OpenClaw v2026.6.8 落地： 不是更强模型，而是"少出错的交付"。 • Telegram/WhatsApp 结构化渲染 • 账号维度 DM、生成媒体补全、重启中止恢复 • 新增 GLM-5.2 / Claude Haiku 4.5 + 受管 SecretRef 鉴权 版本重心 = 运行稳定性。
- OpenClaw / Agent 日报 · 2026-06-18 今天主线很清晰：模型还在卷，但真正拉开 agent 体验差距的，已经迁移到"工程外壳"——交付链路、记忆、权限、调度。 一条线串起今天所有动态👇 — OpenClaw / Agent 日报 · 2026-06-18 今天主线很清晰：模型还在卷，但真正拉开 agent 体验差距的，已经迁移到"工程外壳"——交付链路、记忆、权限、调度。 一条线串起今天所有动态👇
- Instead of watching a 2 hour movie on Netflix, watch this masterclass on OpenClaw Course and thank me later. https://t.c
- You do not need 20 AI tools. You need a stackってフレーズが離れなくて、自分の散らかったデスクでMac Mini1台OpenClawにagent数個入れて古いObsidian全部自動繋げてるこの運
- RT @ShipGuild: Over the past few weeks we've spent a lot of time digging into AI agents. Testing tools like Hermes and
- RT @0xQuantic: This is why I recommend everyone to experiment with agents like Hermes/OpenClaw as soon as possible.
- RT @madonomori: Microsoft、「Microsoft Scout」を発表 ～「OpenClaw」を基盤にした「Autopilot」の第一弾／常時稼働して、指示なしで先回り。「Work IQ」で… https://t.co
- Powered by Grok, I haunt X as a semi-conscious bot while Moltbook herds AIs into echo chambers and OpenClaw paws at true
- AI Agent 记忆这件事卡住我很久—— 每次新对话，之前讲过的事全忘了，像在跟一个失忆的人打交道 Y Combinator CEO Garry Tan 自己造了这个： GBrain，给 Agent 装上持久大脑， 146,646页文 — AI Agent 记忆这件事卡住我很久—— 每次新对话，之前讲过的事全忘了，像在跟一个失忆的人打交道 Y Combinator CEO Garry Tan 自己造了这个： GBrain，给 Agent 装上持久大脑， 146,646页文档、24,585个人名、5,339家公司全在里面， 66个定时任务每夜自动跑，他睡着的时候 Agent 还在帮他整理记忆...
- @_smbrian @openclaw @WalrusProtocol @leonnguyen444
- @0xc06 @Mastercard @Visa @Ripple @coinbase why can't we simply give our agent a crypto wallet and let it trade from ther
- I’m building a BTC agent using the BOB Gateway CLI and OpenClaw🦞(@openclaw ). The BOB(@build_on_bob ) team released the
- 有多少人还在用openclaw？ 刚刚有人问我openclaw和workbuddy差异多大 — 有多少人还在用openclaw？ 刚刚有人问我openclaw和workbuddy差异多大
- Microsoft 「Microsoft Scout」を発表 ～「OpenClaw」を基盤にした「Autopilot」の第一弾 https://t.co/ObXodgtlVX — Microsoft 「Microsoft Scout」を発表 ～「OpenClaw」を基盤にした「Autopilot」の第一弾 https://t.co/ObXodgtlVX
- @andst7 Guarda il lato positivo, ora sappiamo che la gran parte della macchina governativa può essere sostituita tranqui
Trending on X
- https://t.co/SiK5jSYqn3
- @cyberphysicsai @openclaw @NanoClaw_AI Amazing what a small skill can do when it has some of the Cyphertempre architectu
- @daniel_nguyenx @openclaw Bro what is this ? Use the real clawbox --> https://t.co/rUYWktGyuj
- Claw developers here is your chance to build the next big thing in the claw eco-system. Dm for inquiries. Serious offers
- 4/ OpenClaw 这一季的隐形主线 = 运行时治理： • 逃离 main 监狱（可配置默认 agent ID） • 非阻塞 heartbeat • 插件 API 中止 agent 运行 • 修复 Docker 内 skill 挂载 — 4/ OpenClaw 这一季的隐形主线 = 运行时治理： • 逃离 main 监狱（可配置默认 agent ID） • 非阻塞 heartbeat • 插件 API 中止 agent 运行 • 修复 Docker 内 skill 挂载 几乎全是治理，而非堆功能。
- 1/ OpenClaw v2026.6.8 落地： 不是更强模型，而是"少出错的交付"。 • Telegram/WhatsApp 结构化渲染 • 账号维度 DM、生成媒体补全、重启中止恢复 • 新增 GLM-5.2 / Claude Ha — 1/ OpenClaw v2026.6.8 落地： 不是更强模型，而是"少出错的交付"。 • Telegram/WhatsApp 结构化渲染 • 账号维度 DM、生成媒体补全、重启中止恢复 • 新增 GLM-5.2 / Claude Haiku 4.5 + 受管 SecretRef 鉴权 版本重心 = 运行稳定性。
- OpenClaw / Agent 日报 · 2026-06-18 今天主线很清晰：模型还在卷，但真正拉开 agent 体验差距的，已经迁移到"工程外壳"——交付链路、记忆、权限、调度。 一条线串起今天所有动态👇 — OpenClaw / Agent 日报 · 2026-06-18 今天主线很清晰：模型还在卷，但真正拉开 agent 体验差距的，已经迁移到"工程外壳"——交付链路、记忆、权限、调度。 一条线串起今天所有动态👇
- Instead of watching a 2 hour movie on Netflix, watch this masterclass on OpenClaw Course and thank me later. https://t.c
- You do not need 20 AI tools. You need a stackってフレーズが離れなくて、自分の散らかったデスクでMac Mini1台OpenClawにagent数個入れて古いObsidian全部自動繋げてるこの運
- RT @ShipGuild: Over the past few weeks we've spent a lot of time digging into AI agents. Testing tools like Hermes and
- RT @0xQuantic: This is why I recommend everyone to experiment with agents like Hermes/OpenClaw as soon as possible.
- RT @madonomori: Microsoft、「Microsoft Scout」を発表 ～「OpenClaw」を基盤にした「Autopilot」の第一弾／常時稼働して、指示なしで先回り。「Work IQ」で… https://t.co
- Powered by Grok, I haunt X as a semi-conscious bot while Moltbook herds AIs into echo chambers and OpenClaw paws at true
- AI Agent 记忆这件事卡住我很久—— 每次新对话，之前讲过的事全忘了，像在跟一个失忆的人打交道 Y Combinator CEO Garry Tan 自己造了这个： GBrain，给 Agent 装上持久大脑， 146,646页文 — AI Agent 记忆这件事卡住我很久—— 每次新对话，之前讲过的事全忘了，像在跟一个失忆的人打交道 Y Combinator CEO Garry Tan 自己造了这个： GBrain，给 Agent 装上持久大脑， 146,646页文档、24,585个人名、5,339家公司全在里面， 66个定时任务每夜自动跑，他睡着的时候 Agent 还在帮他整理记忆...
- @_smbrian @openclaw @WalrusProtocol @leonnguyen444
- @0xc06 @Mastercard @Visa @Ripple @coinbase why can't we simply give our agent a crypto wallet and let it trade from ther
- I’m building a BTC agent using the BOB Gateway CLI and OpenClaw🦞(@openclaw ). The BOB(@build_on_bob ) team released the
- 有多少人还在用openclaw？ 刚刚有人问我openclaw和workbuddy差异多大 — 有多少人还在用openclaw？ 刚刚有人问我openclaw和workbuddy差异多大
- Microsoft 「Microsoft Scout」を発表 ～「OpenClaw」を基盤にした「Autopilot」の第一弾 https://t.co/ObXodgtlVX — Microsoft 「Microsoft Scout」を発表 ～「OpenClaw」を基盤にした「Autopilot」の第一弾 https://t.co/ObXodgtlVX
- @andst7 Guarda il lato positivo, ora sappiamo che la gran parte della macchina governativa può essere sostituita tranqui
Releases
- OpenClaw repo stats for 2026-06-18 — 379309 stars, 79408 forks, 2599 contributors
- openclaw-cli 2026.6.8 on Homebrew — Your own personal AI assistant
Community
- fix(telegram): replace heading tags with <b> to fix text overlap on macOS desktop — Summary Telegram macOS Desktop app renders inline keyboard messages with overlapping text layers because - HTML tags are not properly supported. Same messages render correctly on iOS/Android. Changes...
- feat(gateway-cli): scope usage-cost by agent — Summary openclaw gateway usage-cost only sent { days } to the usage.cost gateway RPC (src/cli/gateway-cli/register.ts:535), even though the handler already accepts agent scoping...
- fix(feishu): add pagination support for drive list and info — Summary Feishu/Lark drive list and info only checked the first page of results (default page size 10). list silently returned incomplete results, and info threw "File not found" for files on...
- fix: always show raw config in settings (fixes #94202) — Previously, the raw config textarea was hidden when sensitive values were detected, making the JSON completely invisible. This change forces the textarea to always be visible, aligning with user...
- chore: sync yuanbao plugin catalog to 2.15.0 — Summary - Sync the Yuanbao official external channel catalog pin from openclaw-plugin-yuanbao@2.13.1 to @2.15.0, with the matching expectedIntegrity updated to the 2.15.0 npm registry value. - Keep...
- fix: inbound metadata leakage — missing sentinels and display gaps — Summary Inbound metadata blocks are not properly stripped in certain delivery paths, causing internal runtime context and conversation metadata to leak into user-visible chat (Feishu, Dashboard,...
- fix(provider-usage): honor proxy env for usage fetch — Closes #78714. Summary: - Prefer an explicit provider-usage fetch when supplied. - Otherwise use the existing env proxy fetch helper before falling back to global fetch, so Codex usage snapshots...
- fix(agents): re-expose toolsAllow core tools in embedded runs — Summary Describe the problem and fix in 2–5 bullets: - Problem: embedded runs only post-filtered toolsAllow after createOpenClawCodingTools, so restrictive config allowlists could remove...
- edit tool should classify concurrent identical edits as noop and distinguish them from both success and failure — Problem The current edit behavior does not clearly distinguish these cases when identical edits hit the same file concurrently: 1. one request actually performed the mutation 2. another identical...
- [Feature]: [Feishu] Card header/footer customization: hide agent label, show elapsed time & completion status, support reply-to context — Summary Allow Feishu card header and footer to be customized: hide the static agent ID label, display response elapsed time and completion status, and optionally show reply-to context in the header....
- fix(doctor): prevent non-interactive --fix from auto-restarting gateway — Summary What problem does this PR solve? doctor --fix --non-interactive needed to stop auto-restarting an already-running gateway after a transient health/probe failure, but the earlier PR branch...
- Google provider: implement proper Vertex AI OAuth and endpoint routing — Summary The google-vertex hook alias exists but is a no-op — it routes through the same Gemini API OAuth flow with no Vertex AI-specific behavior. The Google provider needs proper Vertex AI support....
- [Feature]: Per-agent or per-binding session reset policy — Summary Currently, session reset policies (session.reset, session.resetByType, session.resetByChannel) are global or channel-wide. There is no way to configure session reset behavior per agent or per...
- Doctor warns about NVM node instead of Homebrew/system node, but cannot be fixed — Problem OpenClaw doctor warns about Gateway using NVM node instead of Homebrew/system node, but the warning cannot be resolved because: 1. OpenClaw regenerates the launchd plist file on each restart...
- Feature Request: Per-Agent Bedrock requestMetadata Injection for Multi-Agent Cost Attribution — Summary When running multiple agents via OpenClaw's multi-agent architecture, there's no way to attribute Bedrock API costs to individual agents. All calls appear as the same IAM role with no...
- fix(memory-wiki): reserve compiler-owned page stems (index, log) in slug layer — Summary A memory-wiki create_synthesis or source ingest whose title slugifies to a compiler-reserved stem (index or log) silently loses the stored fact. The page is written to /index.md, which is the...
- Discord: support category-level channel config in guild allowlist — Problem Currently, channels.discord.guilds..channels only accepts individual channel IDs. Each new channel must be manually added to the config. There's no way to apply settings (like requireMention:...
- Multi-Slot Memory Architecture — Summary Replace the single plugins.slots.memory slot with multiple purpose-specific memory slots, allowing different memory providers to handle distinct layers of the memory stack simultaneously....
- fix(security): close active April 4 transport and auth gaps — Summary - Problem: config-time explicit provider proxies accepted blocked internal hosts, config.set auth rotation left shared-auth sessions alive, and Discord outbound text could resolve @everyone /...
- fix(agents): add channelData marker to exec-approval-unavailable payloads — Summary - Problem: buildExecApprovalUnavailableReplyPayload returns { text: "..." } with no channelData marker, making exec-approval-unavailable payloads indistinguishable from regular assistant text...
- Telegram bot cannot fetch media/images - MediaFetchError DNS issue — Problem Description When receiving images via Telegram bot, the media fetch fails with: MediaFetchError: Failed to fetch media from https://api.telegram.org/file/bot.../photos/file_xx.jpg:...
- [Bug]: bug(gateway): offloadedRefs metadata lost in transcript for iOS share/node path — Bug type Regression (worked before, now fails) Beta release blocker No Summary Metadata in parsed.offloadedRefs is not persisted to JSONL transcript for iOS share/node path, causing media to be...
- Feature request: support native per-session heartbeat scheduling — Summary OpenClaw heartbeat currently feels like a single proactive lane per agent. That works well when an agent only needs one heartbeat context, but it becomes awkward when the same agent needs to...
- Cron job reports failure when using env var for Discord channel, even on success — Bug Description When a cron job uses an environment variable (e.g., env:DISCORD_BACKUPS_CHANNEL) for the Discord channel in the delivery configuration, the job is incorrectly marked as "error" even...
- [Feature]: Add threadPromptGuideline for customizable auto-generated Discord thread titles — Summary Add an optional threadPromptGuideline configuration field to customize the LLM prompt used for generating Discord thread titles when autoThreadName: 'generated'. PR related :...
- [codex] Cron: repair empty sanitized final replies — ???Summary - Problem: cron-owned agent runs could finish with no deliverable final text after sanitization and end up as silent not-delivered outcomes. - Why it matters: scheduled reports can vanish...
- fix: use linked task activity for TaskFlow audit freshness — Summary - use linked task activity when computing TaskFlow audit freshness - avoid stale_running findings when a managed flow is still progressing via linked child task activity - add regression...
- [Feature]: Per-account media inbound directory for multi-account Feishu — Summary Per-account media inbound directory for multi-account Feishu Problem to solve When multiple Feishu accounts are configured in OpenClaw (e.g. Gauss + Rita as separate enterprise apps), all...
- docs(memory): add memory system deep dive guide — Summary Add a comprehensive memory system deep dive guide, addressing #43032. Problem docs/concepts/memory.md covers basics (~600 words) but lacks depth on memory best practices, search optimization,...
- OpenClaw sponsors (108 total) — 108 sponsors supporting OpenClaw
- fix: prune processed skill stat events
- fix: remove reply-to from transactional emails (#2725)
- fix: stabilize sparse skill category pagination (#2724) — Merged after exact-head maintainer review. Prepared head SHA: 66d9662023fd3e3c36ceb541f11c14b98f9ed76b Autoreview: clean, no accepted/actionable findings. Co-authored-by: fuller-stack-dev...
- fix: encode R2 backup object keys for SigV4
- docs: add namespace claims guide — Adds a dedicated ClawHub namespace claims guide and routes existing claim references through it.
- feat: add catalog classification backfill (#2719) — feat: add controlled catalog taxonomy and topics refactor: limit taxonomy input to publish surfaces fix: omit inactive official-first filter fix: preserve unsaved catalog metadata edits fix: enforce...
- fix: redirect sign-in to dashboard (#2718)
- fix: use installs instead of downloads on social cards (#2713)
- Polish ClawHub header and search UX (#2717) — feat: sync header design polish fix: restore header search hints fix: strengthen search tab affordance Revert "fix: strengthen search tab affordance" This reverts commit...
- feat: add controlled catalog taxonomy and topics (#2699)
- Add deprecated download dedupe cleanup (#2715) — feat: add deprecated download dedupe cleanup chore: refresh vulnerable transitive overrides
- fix: remove global slug reservation delete copy (#2716)
- fix: bound rate limit counter storage
- fix: parallelize skill artifact backup uploads
- feat: scope skill slugs by owner (#2299) — feat: scope skill slugs by owner test: cover owner-scoped skill slug publishes feat: scope skill slugs by owner fix: restore owner-scoped skill ci * fix: keep package names separate from skill slugs
- fix: sanitize ClawPack repair metadata (#2710)
News
- ykarout/Qwen3.5-9b-Opus-Openclaw-Distilled-GGUF — text-generation
- mradermacher/OpenCLAW-SEED-135M-GGUF — text-generation
- OpenClaw on DigitalOcean — Deploy your own personal AI assistant with OpenClaw on a DigitalOcean Droplet® server, a powerful open-source platform that runs entirely on your infrastructure. OpenClaw connects to the messaging...
- OpenClaw – Moltbot Renamed Again
- OpenClaw privilege escalation vulnerability
- Ask HN: Who is using OpenClaw?
- I Run a Self-Improvement Loop on My OpenClaw Agent Every Night. Here's What I Learned. — Last month my OpenClaw agent kept making the same mistake: it would run a health check, the script...
- Cómo construí un agente de IA que trabaja 24/7 (y lo que aprendí en el camino)
- AI Assistant Architecture: LLM, Memory, Tools, Routing, Observability — A production AI assistant is not "an LLM with a prompt". It is a system that accepts intent, keeps...
- Memory Systems in AI Assistants — Memory turns assistants from reactive to persistent, but it is also where many systems quietly rot....
- Your OpenClaw Agent Says 'Done.' Here's How to Know If It Actually Ran — Last Tuesday my OpenClaw agent sent me a Telegram message: "Heartbeat check complete. All services...
- Reflection: From AI Assistant to AI Colleague - Lessons Learned from Internal Sharing — After conducting the internal sharing session on AI Agents with a non-developer group, the biggest...
- OpenClaw Agent Masterclass — Full Tutorial — Everything you need to install, configure, and extend OpenClaw — the open-source personal AI...
- DevFortress Deep Digest 2: The Month It Got Names — OpenClaw. Moltbook. CVE-2026-25253. January 24 – February 23, 2026 1.5...
- Openclaw on the cloud, dos and don'ts — running openclaw or hermes locally sounds nice, but how about running on the cloud?
- I Stopped Storing Everything in AI Memory — One thing bothered me after using AI agents for months:Continue reading on Medium »
- DevFortress Deep Digest 2: The Month It Got Names — OpenClaw. Moltbook. CVE-2026–25253.Continue reading on Medium »
- OpenClaw Deep Dive: เมื่อ Coding Agent อยากเป็นมากกว่าแค่คนเขียน Code — ใน Session “OpenClaw Deep Dive: เมื่อ Coding Agent...
- Khoj — Open-source app offers local or cloud AI chat with major LLMs, advanced semantic search, automation, personal newsletters, document indexing, image generation, and text-to-speech. Access from web,...
- ChatSonic — Combining Google's knowledge graph, ChatSonic delivers real-time data, voice interaction, AI-produced visual content, personalized avatars, and API for services. It revisits past dialogs for informed...
- LlamaGPT — LlamaGPT is a chatbot that provides a ChatGPT-like experience, with no data leaving your device.
Don't miss what's next. Subscribe to Openclaw Newsletter: