OpenClaw Newsletter - 2026-05-18
OpenClaw Newsletter
Monday, May 18, 2026Top Stories
- openclaw 2026.5.16-beta.7 — Changes - Dependencies: update @openclaw/proxyline to 0.3.3. - Dependencies: update Pi packages to 0.75.1 and raise the minimum supported Node.js 22 line to 22.19. - Docker/Podman: add...
- openclaw 2026.5.16-beta.6 — Changes - Mac app: redesign Settings pages with consistent card layouts, cached navigation, cleaner permissions/voice/skills/cron/exec/debug panes, and steadier spacing around the native sidebar. -...
- openclaw 2026.5.16-beta.5 — Changes - Mac app: redesign Settings pages with consistent card layouts, cached navigation, cleaner permissions/voice/skills/cron/exec/debug panes, and steadier spacing around the native sidebar. -...
- [Feature]: safe/unsafe ClawdBot — perhaps inherit this feature or abstraction from Rust, completely rewrite this project in Rust. Safe mode can be running ClawdBot locally in a sandbox environment with limited access and it protects...
- Feature: Graceful sub-agent timeout (pre-timeout warning) — Problem When a sub-agent hits runTimeoutSeconds, it's killed immediately with no chance to save progress. All unsaved work is lost — code, research, analysis, generated content, anything. Proposed...
- Feature: Add denylist support for exec-approvals — Summary Add denylist support to exec-approvals, complementing the existing allowlist. This enables "allow everything except X" policies. Use Case I want to: - Allow all commands to run without...
- Linux/Windows Clawdbot Apps — We have apps for macOS, iOS and Android (simpler nodes) Linux and Windows are missing. Similar feature set to macOS ideally.
- docs: replace British 'behaviour' with American 'behavior' — Summary Problem: Three remaining instances of British "behaviour" in the docs violate the project's American English requirement. Why it matters: CONTRIBUTING.md explicitly states "Use American...
- test(qa-lab): trace Slack RTT ping pong — Summary - Switch the Slack live canary to a ping/pong marker and measure RTT from Slack accepted timestamps when available. - Record observer lag, poll counts, gateway RSS samples, and optional heap...
- feat(feishu): add agent message delete and recall — PR 1 draft: Feishu agent message delete and recall Title: Summary - Add an agent-callable feishu_message tool for list, delete, recall, read_receipts, and read_users. - Wire Feishu messages tool...
- feat(feishu): expose reaction tool — PR 2 draft: Feishu reaction tool registration Title: Summary - Add an agent-callable feishu_reaction tool with add, remove, and list actions. - Wire the tool through Feishu tool config defaults,...
- fix(codex): hydrate queued inbound images — Summary - Extract shared current-turn image hydration for native PI and Codex app-server runs. - Hydrate MediaPath / MediaPaths into queued followup images while the inbound context is still...
- Expose reload kind in config schema lookup — Summary > Replacement for closed PR #81574. Rebased onto current origin/main (856a1692ff24450f3883c11bd2de70cb00978414) and force-pushed as single commit 40b08e66e279e20c9d9ff0ce216699092caae980....
- Scope agent prompt surfaces by runtime — Summary This PR separates runtime-specific prompt fragments by explicit prompt surface so PI, CLI, ACP, native Codex app-server, and subagent prompts do not inherit each other's assumptions. The core...
- fix(qa): reduce WhatsApp live RTT setup cost — Summary - Skip WhatsApp live QA agent workspace bootstrap, matching the Telegram live QA fast path for ping/pong scenarios that do not need workspace files. - Add per-scenario WhatsApp live phase...
- [Bug] Wizard cross-overwrites credentials in multi-account plugins when "Add a new account" is selected — Version v2026.5.6 Description When adding a new account for a channel (observed in wecom-openclaw-plugin and telegram), the wizard prompts "Bot ID / Token already configured. Keep it?" even if the...
- fix(provider-catalog): catalog discovery sends GET /models without auth when apiKey is a non-secret marker — Summary When a self-hosted / local OpenAI-compatible provider (e.g., LiteLLM, vLLM, SGLang, LM Studio) is configured with an API key, but the key is configured in a way that produces a non-secret...
- Fix Codex runtime hook and session churn — Summary Fixes two Codex-runtime stability problems reported by users: - Fixes #76552: Codex no longer installs the default PreToolUse native hook relay command when no before_tool_call policy exists,...
- fix: harden release stability recovery and auth fallback — Why This is a narrow release-stability bundle for operators who need a branch they can run for 1-2 weeks while the broader OpenClaw mainline settles. It deliberately avoids a full Codex-harness...
- fix(provider-catalog): skip catalog discovery when apiKey is a non-secret auth marker — Summary buildSingleProviderApiKeyCatalog and buildPairedProviderApiKeyCatalog in src/plugins/provider-catalog.ts only check !apiKey before allowing catalog discovery. When the resolved apiKey is a...
- chore(lint): enable no-underscore-dangle with comprehensive allow list — Summary - Enable eslint/no-underscore-dangle in .oxlintrc.json with the current allow list for established underscore-prefixed identifiers - Remove the wrapper-level --allow...
- fix(skill-creator): reject empty name and description in skill valida… — Summary validate_skill() in quick_validate.py silently accepts empty or whitespace-only name and description values in SKILL.md frontmatter. The if name: guard skips all format checks when the value...
- fix(scripts): close local heavy-check lock gaps — Summary - keep the shared local heavy-check lock for type-aware oxlint, including explicit file-target runs that can spawn tsgolint headless - wrap sharded oxlint in one parent heavy-check lock while...
- Feature request: session:reset hook event for memory persistence on auto-resets — Problem The session-memory bundled hook only fires on command:new events. When sessions auto-reset via: - Daily reset (session.reset.mode: "daily", atHour: 4) - Idle timeout...
- Filter heartbeat response-tool transcript artifacts — Summary - Filters full silent heartbeat transcript spans before normal prompt assembly, including helper tool calls/results, prompt-only interrupted wakes, pre-terminal assistant status text, and...
- fix: constrain clawtributors avatar rendering — Summary Fix the README clawtributors renderer so each avatar is emitted as HTML with explicit width="48" and height="48" attributes. This addresses the concrete regression where...
- fix(code-mode): align outer exec hook params — Follow-up to #83387 after #83481 landed the active hook-context plumbing. Problem Code-mode exposes its outer exec control tool through the Pi custom-tool adapter. On current main, the active hook...
- fix(code-mode): honor agent scoped code mode — Fixes #83388. Problem agents.list[].tools.codeMode was documented as a per-agent setting, but main rejected it in the strict per-agent schema. A schema-only fix would be worse than the original...
- feat(plugins): expose ACP spawn and prompt in plugin runtime — Summary Plugins can orchestrate subagents via api.runtime.subagent. but have no way to dispatch ACP-backed agent sessions with channel delivery**. This PR adds api.runtime.acp.spawn() and...
- Output sanitizer: improve duplicate detection for same-line and partial duplicates — Problem The sanitizeUserFacingText() pipeline in pi-embedded-helpers includes collapseConsecutiveDuplicateBlocks(), which deduplicates identical paragraphs separated by \n\n. However, it misses...
- Proposal: safer self-update with pre-update backup + health check + auto-rollback/restore — Problem On production deployments (e.g., VPS), updates can occasionally introduce regressions or configuration/compat changes. Operators want a safer, more automated upgrade path that reduces...
- [Feature]: Integrate password toggle functionality across all platforms to let users see password in clear text — Summary Section: - ✅ Clear problem statement: "Add password toggle/show functionality across all platforms" - ✅ Explains the feature: Users can control password visibility with eye icon - ✅ Platforms...
- OpenClaw repo stats for 2026-05-18 — 372823 stars, 77301 forks, 2274 contributors
- OpenClaw sponsors (118 total) — 118 sponsors supporting OpenClaw
- fix: let Codex own release scan verdicts
- test: accept canonical local publish owner
- fix: make Codex scan schema strict
- fix: harden package publish contention (#2291) — fix: harden package publish contention test: fix http retry overload
- fix: authenticate Codex security worker in CI
- fix: make Codex security worker executable in Actions
- feat: run ClawScan classification through Codex (#2290)
- fix: add admin package name repair (#2289) — fix: add admin package name repair fix: stabilize publish owner selection
- fix: show suspicious skills in UI search (#2220)
- feat: add ClawScan finding permalinks (#2208)
- fix: use canonical user profile route (#2173)
- Add Parley
- openclaw v2026.5.12 on npm — Multi-channel AI gateway with extensible messaging integrations
- 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
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- Show HN: DenchClaw – Local CRM on Top of OpenClaw
- Four OpenClaw Flaws Enable Data Theft, Privilege Escalation, and Persistence — Four OpenClaw flaws have been disclosed by cybersecurity researchers, exposing serious risks in...
- Team9 Review: The AI Collaboration Workspace That Finally Makes AI Agents Real Teammates — Introduction: The Problem with Most AI Tools Let's be honest. Most AI tools today are...
- $1.3 Million on AI Tokens in 30 Days: what OpenClaw Founder Peter Steinberger Is Really Spending It… — Peter Steinberger posted a CodexBar screenshot showing a $1.3 million AI token bill. The internet reacted. Here’s what the spend actually…Continue reading on Medium »
- Hermes Agent: The Complete Guide to Self-Learning AI — The Problem Every AI Agent Had — Until NowContinue reading on Medium »
- OpenClaw Architecture Deep Dive — The 5 Design Patterns of OpenClaw — the open-source personal AI assistantContinue reading on Medium »
- I Gave an AI Agent a Telegram Bot and It Started Editing Videos — I wanted to test something simple:Continue reading on Medium »
- Harness Engineering: The Missing Layer in the AI Stack — 2022 was about who had the smartest brain. 2026 is about who built the best body for it. That body is the Harness.Continue reading on Medium »
- Can AI Agents Replace Enterprise Workflow Orchestration? — A database administrator’s honest investigation into whether the new wave of AI automation tools can handle enterprise-grade workflows —…Continue reading on Medium »
- I’m on the Board of a Newsroom Run Entirely by AI Agents — What happens when you bridge a familiar news site format with autonomous agents who actually live on the platform where news breaks first.Continue reading on Medium »
- Inside Alex’s Sandbox: An Implementer’s Guide to NVIDIA OpenShell + NemoClaw + CrewAI — The parent piece argued that the future of work has a runtime. This piece is what’s inside the runtime.Continue reading on Medium »
- How to Give AI Agents a Memory Humans Can Read, Edit, and Trust — Noosphere turns agent memory into a shared, API-first knowledge memory system for agents, coding CLIs, and humans.Continue reading on Medium »
- How to Set Up OpenClaw Safely: A Personal AI Agent That Won’t Wreck Your Machine — Alset TeamMay 17, 20269 minContinue reading on Medium »
- Private GPT — Ask questions to your documents without an internet connection, using the power of LLMs. 100% private, no data leaves your execution environment at any point. You can ingest documents and ask...
- Met few guys who paid for API. Installed openclaw, other Hermes but they said they’ve wasted money and time. If you don
- @geoffreywoo Hey GF, we're actually building on similar lines. Our entire thesis is we started with OpenClaw to automate
- 有人做了个实验。把OpenClaw接入Polymarket，开了5个账号，每个放1000美元。本金一共5000。 然后他去睡觉了。 12小时后醒来，5个账号加起来，变成了45000美元。 整晚AI自己交易，一共执行212笔。179胜，
- @libapi_ 我现在两个都在，但是好像openclaw已经基本不咋用了
- @rust_comb @OpenClawAIX That score looks intense. If you want to run OpenClaw with full management and access to over 50
- @0xRiRoyal @TheARCTERMINAL Modular AI agents are where real utility starts @TheARCTERMINAL pushing OpenClaw-style compos
- OpenClaw 测试新 runtime 模式 pi -> codex，大约一周了，体验只能说非常的糟糕，几点问题：1.缺乏记忆连续系统，OpenClaw 交给Codex去计算时会掉很多记忆和查询。2.对宿主的要求执行完成很差，特别像
- Switched back to OpenClaw, why? > First, I didn't get time for migrating all my existing cron jobs and workflows to o — Switched back to OpenClaw, why? > First, I didn't get time for migrating all my existing cron jobs and workflows to other agent services > It is getting better day by day. > Codex...
- @GOATNetwork @openclaw This feels more like building economic software than traditional hackathon demos. Agents with ide
- @YashHustle_22 Vermicode just like openclaw But it's kind of a top layer above it to think wisely in the Agentic era l
- @0xmryaxx @openclaw Unlimited attachments have not yet been implemented.
- This week will be all about Hermes. > Hermes took the lead from OpenClaw as being the biggest open-source agent mod — This week will be all about Hermes. > Hermes took the lead from OpenClaw as being the biggest open-source agent model > Hermes coin on Sol is starting to run > Base utility plays on Hermes...
- 感觉这个应该是全网唯一一个双击exe文件就能用的轻量级龙虾🦞了，不像openclaw和Hermes这种依赖怪兽，GitHub上已经有5000个stars了，确实非常非常方便使用，一共23MB，一秒下载完就能用了😍😍😍https://t.co
- @cryptoinfoswiss Wow moins de 10$ pour un abo tous compris pour openclaw !!!
- @0xmryaxx @openclaw Could you please add your idea to the discussions? https://t.co/Lqh6ZWeP6u
- @0xRiRoyal @TheARCTERMINAL Solid modularity take. How do you see OpenClaw skills shaping the future of agent interoperab
- 我让openclaw装上hermes ，装好后让hermes 彻底删除openclaw - 强力的背刺
- @stammering_wall @OpenClawAIX Nice to see WMV trending with OpenClaw's power behind it. If you want to run OpenClaw with
- The Session Logs skill lets your OpenClaw search and analyse its own older conversations using jq. It treats past sessi
Trending on X
- Met few guys who paid for API. Installed openclaw, other Hermes but they said they’ve wasted money and time. If you don
- @geoffreywoo Hey GF, we're actually building on similar lines. Our entire thesis is we started with OpenClaw to automate
- 有人做了个实验。把OpenClaw接入Polymarket，开了5个账号，每个放1000美元。本金一共5000。 然后他去睡觉了。 12小时后醒来，5个账号加起来，变成了45000美元。 整晚AI自己交易，一共执行212笔。179胜，
- @libapi_ 我现在两个都在，但是好像openclaw已经基本不咋用了
- @rust_comb @OpenClawAIX That score looks intense. If you want to run OpenClaw with full management and access to over 50
- @0xRiRoyal @TheARCTERMINAL Modular AI agents are where real utility starts @TheARCTERMINAL pushing OpenClaw-style compos
- OpenClaw 测试新 runtime 模式 pi -> codex，大约一周了，体验只能说非常的糟糕，几点问题：1.缺乏记忆连续系统，OpenClaw 交给Codex去计算时会掉很多记忆和查询。2.对宿主的要求执行完成很差，特别像
- Switched back to OpenClaw, why? > First, I didn't get time for migrating all my existing cron jobs and workflows to o — Switched back to OpenClaw, why? > First, I didn't get time for migrating all my existing cron jobs and workflows to other agent services > It is getting better day by day. > Codex...
- @GOATNetwork @openclaw This feels more like building economic software than traditional hackathon demos. Agents with ide
- @YashHustle_22 Vermicode just like openclaw But it's kind of a top layer above it to think wisely in the Agentic era l
- @0xmryaxx @openclaw Unlimited attachments have not yet been implemented.
- This week will be all about Hermes. > Hermes took the lead from OpenClaw as being the biggest open-source agent mod — This week will be all about Hermes. > Hermes took the lead from OpenClaw as being the biggest open-source agent model > Hermes coin on Sol is starting to run > Base utility plays on Hermes...
- 感觉这个应该是全网唯一一个双击exe文件就能用的轻量级龙虾🦞了，不像openclaw和Hermes这种依赖怪兽，GitHub上已经有5000个stars了，确实非常非常方便使用，一共23MB，一秒下载完就能用了😍😍😍https://t.co
- @cryptoinfoswiss Wow moins de 10$ pour un abo tous compris pour openclaw !!!
- @0xmryaxx @openclaw Could you please add your idea to the discussions? https://t.co/Lqh6ZWeP6u
- @0xRiRoyal @TheARCTERMINAL Solid modularity take. How do you see OpenClaw skills shaping the future of agent interoperab
- 我让openclaw装上hermes ，装好后让hermes 彻底删除openclaw - 强力的背刺
- @stammering_wall @OpenClawAIX Nice to see WMV trending with OpenClaw's power behind it. If you want to run OpenClaw with
- The Session Logs skill lets your OpenClaw search and analyse its own older conversations using jq. It treats past sessi
Releases
- openclaw 2026.5.16-beta.7 — Changes - Dependencies: update @openclaw/proxyline to 0.3.3. - Dependencies: update Pi packages to 0.75.1 and raise the minimum supported Node.js 22 line to 22.19. - Docker/Podman: add...
- openclaw 2026.5.16-beta.6 — Changes - Mac app: redesign Settings pages with consistent card layouts, cached navigation, cleaner permissions/voice/skills/cron/exec/debug panes, and steadier spacing around the native sidebar. -...
- openclaw 2026.5.16-beta.5 — Changes - Mac app: redesign Settings pages with consistent card layouts, cached navigation, cleaner permissions/voice/skills/cron/exec/debug panes, and steadier spacing around the native sidebar. -...
- OpenClaw repo stats for 2026-05-18 — 372823 stars, 77301 forks, 2274 contributors
- openclaw v2026.5.12 on npm — Multi-channel AI gateway with extensible messaging integrations
Community
- [Feature]: safe/unsafe ClawdBot — perhaps inherit this feature or abstraction from Rust, completely rewrite this project in Rust. Safe mode can be running ClawdBot locally in a sandbox environment with limited access and it protects...
- Feature: Graceful sub-agent timeout (pre-timeout warning) — Problem When a sub-agent hits runTimeoutSeconds, it's killed immediately with no chance to save progress. All unsaved work is lost — code, research, analysis, generated content, anything. Proposed...
- Feature: Add denylist support for exec-approvals — Summary Add denylist support to exec-approvals, complementing the existing allowlist. This enables "allow everything except X" policies. Use Case I want to: - Allow all commands to run without...
- Linux/Windows Clawdbot Apps — We have apps for macOS, iOS and Android (simpler nodes) Linux and Windows are missing. Similar feature set to macOS ideally.
- docs: replace British 'behaviour' with American 'behavior' — Summary Problem: Three remaining instances of British "behaviour" in the docs violate the project's American English requirement. Why it matters: CONTRIBUTING.md explicitly states "Use American...
- test(qa-lab): trace Slack RTT ping pong — Summary - Switch the Slack live canary to a ping/pong marker and measure RTT from Slack accepted timestamps when available. - Record observer lag, poll counts, gateway RSS samples, and optional heap...
- feat(feishu): add agent message delete and recall — PR 1 draft: Feishu agent message delete and recall Title: Summary - Add an agent-callable feishu_message tool for list, delete, recall, read_receipts, and read_users. - Wire Feishu messages tool...
- feat(feishu): expose reaction tool — PR 2 draft: Feishu reaction tool registration Title: Summary - Add an agent-callable feishu_reaction tool with add, remove, and list actions. - Wire the tool through Feishu tool config defaults,...
- fix(codex): hydrate queued inbound images — Summary - Extract shared current-turn image hydration for native PI and Codex app-server runs. - Hydrate MediaPath / MediaPaths into queued followup images while the inbound context is still...
- Expose reload kind in config schema lookup — Summary > Replacement for closed PR #81574. Rebased onto current origin/main (856a1692ff24450f3883c11bd2de70cb00978414) and force-pushed as single commit 40b08e66e279e20c9d9ff0ce216699092caae980....
- Scope agent prompt surfaces by runtime — Summary This PR separates runtime-specific prompt fragments by explicit prompt surface so PI, CLI, ACP, native Codex app-server, and subagent prompts do not inherit each other's assumptions. The core...
- fix(qa): reduce WhatsApp live RTT setup cost — Summary - Skip WhatsApp live QA agent workspace bootstrap, matching the Telegram live QA fast path for ping/pong scenarios that do not need workspace files. - Add per-scenario WhatsApp live phase...
- [Bug] Wizard cross-overwrites credentials in multi-account plugins when "Add a new account" is selected — Version v2026.5.6 Description When adding a new account for a channel (observed in wecom-openclaw-plugin and telegram), the wizard prompts "Bot ID / Token already configured. Keep it?" even if the...
- fix(provider-catalog): catalog discovery sends GET /models without auth when apiKey is a non-secret marker — Summary When a self-hosted / local OpenAI-compatible provider (e.g., LiteLLM, vLLM, SGLang, LM Studio) is configured with an API key, but the key is configured in a way that produces a non-secret...
- Fix Codex runtime hook and session churn — Summary Fixes two Codex-runtime stability problems reported by users: - Fixes #76552: Codex no longer installs the default PreToolUse native hook relay command when no before_tool_call policy exists,...
- fix: harden release stability recovery and auth fallback — Why This is a narrow release-stability bundle for operators who need a branch they can run for 1-2 weeks while the broader OpenClaw mainline settles. It deliberately avoids a full Codex-harness...
- fix(provider-catalog): skip catalog discovery when apiKey is a non-secret auth marker — Summary buildSingleProviderApiKeyCatalog and buildPairedProviderApiKeyCatalog in src/plugins/provider-catalog.ts only check !apiKey before allowing catalog discovery. When the resolved apiKey is a...
- chore(lint): enable no-underscore-dangle with comprehensive allow list — Summary - Enable eslint/no-underscore-dangle in .oxlintrc.json with the current allow list for established underscore-prefixed identifiers - Remove the wrapper-level --allow...
- fix(skill-creator): reject empty name and description in skill valida… — Summary validate_skill() in quick_validate.py silently accepts empty or whitespace-only name and description values in SKILL.md frontmatter. The if name: guard skips all format checks when the value...
- fix(scripts): close local heavy-check lock gaps — Summary - keep the shared local heavy-check lock for type-aware oxlint, including explicit file-target runs that can spawn tsgolint headless - wrap sharded oxlint in one parent heavy-check lock while...
- Feature request: session:reset hook event for memory persistence on auto-resets — Problem The session-memory bundled hook only fires on command:new events. When sessions auto-reset via: - Daily reset (session.reset.mode: "daily", atHour: 4) - Idle timeout...
- Filter heartbeat response-tool transcript artifacts — Summary - Filters full silent heartbeat transcript spans before normal prompt assembly, including helper tool calls/results, prompt-only interrupted wakes, pre-terminal assistant status text, and...
- fix: constrain clawtributors avatar rendering — Summary Fix the README clawtributors renderer so each avatar is emitted as HTML with explicit width="48" and height="48" attributes. This addresses the concrete regression where...
- fix(code-mode): align outer exec hook params — Follow-up to #83387 after #83481 landed the active hook-context plumbing. Problem Code-mode exposes its outer exec control tool through the Pi custom-tool adapter. On current main, the active hook...
- fix(code-mode): honor agent scoped code mode — Fixes #83388. Problem agents.list[].tools.codeMode was documented as a per-agent setting, but main rejected it in the strict per-agent schema. A schema-only fix would be worse than the original...
- feat(plugins): expose ACP spawn and prompt in plugin runtime — Summary Plugins can orchestrate subagents via api.runtime.subagent. but have no way to dispatch ACP-backed agent sessions with channel delivery**. This PR adds api.runtime.acp.spawn() and...
- Output sanitizer: improve duplicate detection for same-line and partial duplicates — Problem The sanitizeUserFacingText() pipeline in pi-embedded-helpers includes collapseConsecutiveDuplicateBlocks(), which deduplicates identical paragraphs separated by \n\n. However, it misses...
- Proposal: safer self-update with pre-update backup + health check + auto-rollback/restore — Problem On production deployments (e.g., VPS), updates can occasionally introduce regressions or configuration/compat changes. Operators want a safer, more automated upgrade path that reduces...
- [Feature]: Integrate password toggle functionality across all platforms to let users see password in clear text — Summary Section: - ✅ Clear problem statement: "Add password toggle/show functionality across all platforms" - ✅ Explains the feature: Users can control password visibility with eye icon - ✅ Platforms...
- OpenClaw sponsors (118 total) — 118 sponsors supporting OpenClaw
- fix: let Codex own release scan verdicts
- test: accept canonical local publish owner
- fix: make Codex scan schema strict
- fix: harden package publish contention (#2291) — fix: harden package publish contention test: fix http retry overload
- fix: authenticate Codex security worker in CI
- fix: make Codex security worker executable in Actions
- feat: run ClawScan classification through Codex (#2290)
- fix: add admin package name repair (#2289) — fix: add admin package name repair fix: stabilize publish owner selection
- fix: show suspicious skills in UI search (#2220)
- feat: add ClawScan finding permalinks (#2208)
- fix: use canonical user profile route (#2173)
- Add Parley
News
- 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
- Nanobot: Ultra-Lightweight Alternative to OpenClaw
- Show HN: DenchClaw – Local CRM on Top of OpenClaw
- Four OpenClaw Flaws Enable Data Theft, Privilege Escalation, and Persistence — Four OpenClaw flaws have been disclosed by cybersecurity researchers, exposing serious risks in...
- Team9 Review: The AI Collaboration Workspace That Finally Makes AI Agents Real Teammates — Introduction: The Problem with Most AI Tools Let's be honest. Most AI tools today are...
- $1.3 Million on AI Tokens in 30 Days: what OpenClaw Founder Peter Steinberger Is Really Spending It… — Peter Steinberger posted a CodexBar screenshot showing a $1.3 million AI token bill. The internet reacted. Here’s what the spend actually…Continue reading on Medium »
- Hermes Agent: The Complete Guide to Self-Learning AI — The Problem Every AI Agent Had — Until NowContinue reading on Medium »
- OpenClaw Architecture Deep Dive — The 5 Design Patterns of OpenClaw — the open-source personal AI assistantContinue reading on Medium »
- I Gave an AI Agent a Telegram Bot and It Started Editing Videos — I wanted to test something simple:Continue reading on Medium »
- Harness Engineering: The Missing Layer in the AI Stack — 2022 was about who had the smartest brain. 2026 is about who built the best body for it. That body is the Harness.Continue reading on Medium »
- Can AI Agents Replace Enterprise Workflow Orchestration? — A database administrator’s honest investigation into whether the new wave of AI automation tools can handle enterprise-grade workflows —…Continue reading on Medium »
- I’m on the Board of a Newsroom Run Entirely by AI Agents — What happens when you bridge a familiar news site format with autonomous agents who actually live on the platform where news breaks first.Continue reading on Medium »
- Inside Alex’s Sandbox: An Implementer’s Guide to NVIDIA OpenShell + NemoClaw + CrewAI — The parent piece argued that the future of work has a runtime. This piece is what’s inside the runtime.Continue reading on Medium »
- How to Give AI Agents a Memory Humans Can Read, Edit, and Trust — Noosphere turns agent memory into a shared, API-first knowledge memory system for agents, coding CLIs, and humans.Continue reading on Medium »
- How to Set Up OpenClaw Safely: A Personal AI Agent That Won’t Wreck Your Machine — Alset TeamMay 17, 20269 minContinue reading on Medium »
- Private GPT — Ask questions to your documents without an internet connection, using the power of LLMs. 100% private, no data leaves your execution environment at any point. You can ingest documents and ask...
Don't miss what's next. Subscribe to Openclaw Newsletter: