repo stringclasses 1
value | github_id int64 3.77B 4.38B | github_node_id stringlengths 18 24 | number int64 75 77.8k | html_url stringlengths 46 49 | api_url stringlengths 56 59 | title stringlengths 1 746 | body stringlengths 1 55.7k ⌀ | state stringclasses 2
values | state_reason stringclasses 4
values | locked bool 2
classes | comments_count int64 0 128 | labels listlengths 0 5 | assignees listlengths 0 2 | created_at stringdate 2026-01-01 16:55:34 2026-05-05 11:59:03 | updated_at stringdate 2026-03-24 00:13:29 2026-05-05 11:59:15 | closed_at stringlengths 20 20 ⌀ | author_association stringclasses 3
values | milestone_title stringclasses 0
values | snapshot_id stringclasses 76
values | extracted_at stringdate 2026-03-24 23:36:49 2026-05-05 12:00:33 | author_login stringlengths 3 38 | author_id int64 346 282M | author_node_id stringlengths 12 20 | author_type stringclasses 2
values | author_site_admin bool 2
classes |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
openclaw/openclaw | 4,050,044,302 | I_kwDOQb6kR87xZsWO | 41,914 | https://github.com/openclaw/openclaw/issues/41914 | https://api.github.com/repos/openclaw/openclaw/issues/41914 | openclaw update removes systemd service file and does not recreate it | ## Description
Running `openclaw update --yes` on systemd-based hosts removes the systemd service file during the update process but does not recreate it afterward. The gateway exits with "Gateway service disabled" and requires a manual `openclaw gateway install` + `systemctl --user start openclaw-gateway.service` to ... | closed | completed | true | 4 | [] | [] | 2026-03-10T07:47:15Z | 2026-04-27T04:53:54Z | 2026-04-24T22:09:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dafacto | 39,696,448 | MDQ6VXNlcjM5Njk2NDQ4 | User | false |
openclaw/openclaw | 4,050,130,261 | I_kwDOQb6kR87xaBVV | 41,924 | https://github.com/openclaw/openclaw/issues/41924 | https://api.github.com/repos/openclaw/openclaw/issues/41924 | Discussion: Agent Health Monitoring and Self-Healing Patterns | ## Problem
OpenClaw agents (especially long-running main agents) lack self-monitoring and auto-healing capabilities. When the agent encounters errors or resource exhaustion, it may degrade silently or require manual restart.
## Proposal
Implement a **Health Monitoring and Self-Healing** subsystem:
### Components
1. ... | closed | not_planned | true | 2 | [] | [] | 2026-03-10T08:07:32Z | 2026-04-27T04:53:53Z | 2026-04-24T22:09:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | DockeGumi | 258,497,605 | U_kgDOD2hcRQ | User | false |
openclaw/openclaw | 4,050,163,032 | I_kwDOQb6kR87xaJVY | 41,934 | https://github.com/openclaw/openclaw/issues/41934 | https://api.github.com/repos/openclaw/openclaw/issues/41934 | [macOS] Gateway fails to auto-restart after config-triggered SIGTERM — LaunchAgent becomes unloaded | ## Version
2026.3.9 (9d403fd), macOS (launchd)
## Description
When a config change triggers a gateway restart via SIGUSR1 → `launchctl kickstart -k`, the LaunchAgent (`ai.openclaw.gateway`) becomes unloaded from launchd. After the process receives SIGTERM and shuts down, launchd does not restart it because the service... | closed | completed | true | 2 | [] | [] | 2026-03-10T08:14:42Z | 2026-04-27T04:53:52Z | 2026-04-24T22:09:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | im905513 | 5,841,346 | MDQ6VXNlcjU4NDEzNDY= | User | false |
openclaw/openclaw | 4,050,178,730 | I_kwDOQb6kR87xaNKq | 41,937 | https://github.com/openclaw/openclaw/issues/41937 | https://api.github.com/repos/openclaw/openclaw/issues/41937 | TUI: Telegram messages update session state but do not appear live until session/history reload | ## Summary
In `openclaw tui`, inbound Telegram messages are written to the correct session, but they often **do not appear live** in the TUI chat log until a manual reload (`/session ...`, session switch, or reconnect/history fetch).
This appears to be a TUI live-event subscription/render issue (not routing/session-ke... | closed | completed | true | 3 | [] | [] | 2026-03-10T08:17:40Z | 2026-04-27T04:30:10Z | 2026-04-24T22:09:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sachser66 | 67,191,800 | MDQ6VXNlcjY3MTkxODAw | User | false |
openclaw/openclaw | 4,050,210,913 | I_kwDOQb6kR87xaVBh | 41,951 | https://github.com/openclaw/openclaw/issues/41951 | https://api.github.com/repos/openclaw/openclaw/issues/41951 | 功能请求：飞书消息附件下载工具 | ## 问题
当前 OpenClaw 的 Feishu 插件缺少直接下载聊天消息中附件（视频、文件等）的工具能力。
## 现状
- `im:resource` 权限已支持
- OpenClaw 内部有 `downloadMessageResourceFeishu` 函数（在 `extensions/feishu/src/media.ts`）
- 但这个功能没有暴露给 AI Agent 作为工具使用
## 需求
将 `downloadMessageResourceFeishu` 功能封装为可调用的工具，让 Agent 能够下载飞书聊天消息中的文件附件。
## 建议实现
添加新工具：
- 名称：`feishu_downloa... | closed | completed | true | 2 | [] | [] | 2026-03-10T08:24:21Z | 2026-04-27T04:53:52Z | 2026-04-24T22:09:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rym-byte | 249,670,818 | U_kgDODuGsog | User | false |
openclaw/openclaw | 4,050,251,748 | I_kwDOQb6kR87xae_k | 41,960 | https://github.com/openclaw/openclaw/issues/41960 | https://api.github.com/repos/openclaw/openclaw/issues/41960 | [Feature]: I hope to add a multi-agent robot configuration function. Configuring multi-agent robots is currently too cumbersome: it requires setting up a separate workspace and adding multiple agents in the agents field, which is very inconvenient. | ### Summary
I hope to add a multi-agent robot configuration function. Configuring multi-agent robots is currently too cumbersome: it requires setting up a separate workspace and adding multiple agents in the agents field, which is very inconvenient.
### Problem to solve
I hope to add a multi-agent robot configuratio... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-10T08:33:06Z | 2026-04-27T04:53:51Z | 2026-04-24T22:09:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 1406119529tmy-design | 225,306,333 | U_kgDODW3m3Q | User | false |
openclaw/openclaw | 4,050,258,959 | I_kwDOQb6kR87xagwP | 41,963 | https://github.com/openclaw/openclaw/issues/41963 | https://api.github.com/repos/openclaw/openclaw/issues/41963 | openai-completions providers: allow supportsUsageInStreaming override | ## Problem
When using `openai-completions` API providers (e.g. llama-server, vLLM, or other OpenAI-compatible servers), token usage is always recorded as zero in session data. This means the Control UI shows no usage stats for any conversations.
## Root Cause
The `openai-completions` adapter hardcodes `supportsUsage... | closed | completed | true | 3 | [] | [] | 2026-03-10T08:34:30Z | 2026-04-27T04:53:50Z | 2026-04-24T22:09:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | phi0x | 6,878,687 | MDQ6VXNlcjY4Nzg2ODc= | User | false |
openclaw/openclaw | 4,050,318,317 | I_kwDOQb6kR87xavPt | 41,974 | https://github.com/openclaw/openclaw/issues/41974 | https://api.github.com/repos/openclaw/openclaw/issues/41974 | feat: /reboot slash command — graceful context reset | ## Problem
Currently `/clear` hard-wipes conversation context with no opportunity to persist state. This causes:
- Loss of in-flight work context
- No distillation of conversation into memory files
- Sub-agents/hive jobs can lose parent context
- No status coordination (other agents don't know we're resetting)
## Prop... | closed | not_planned | true | 2 | [] | [] | 2026-03-10T08:47:21Z | 2026-04-27T04:53:49Z | 2026-04-24T22:09:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | merplerps | 11,965,499 | MDQ6VXNlcjExOTY1NDk5 | User | false |
openclaw/openclaw | 4,050,438,587 | I_kwDOQb6kR87xbMm7 | 41,988 | https://github.com/openclaw/openclaw/issues/41988 | https://api.github.com/repos/openclaw/openclaw/issues/41988 | Support ACP bindings for plugin channels | ## Use Case
Building multi-agent orchestration where tasks are routed through plugin channels to specialized ACP agents (Claude Code, Codex). ACP bindings on plugin channels would allow direct routing to coding agents without an intermediate agent layer.
## Current Behavior
ACP bindings are restricted to `discord` a... | closed | completed | true | 2 | [] | [] | 2026-03-10T09:11:28Z | 2026-04-27T04:53:48Z | 2026-04-24T22:09:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | plexodus | 3,330,043 | MDQ6VXNlcjMzMzAwNDM= | User | false |
openclaw/openclaw | 4,050,487,001 | I_kwDOQb6kR87xbYbZ | 41,996 | https://github.com/openclaw/openclaw/issues/41996 | https://api.github.com/repos/openclaw/openclaw/issues/41996 | [Bug]: An error occurred while sending a voice message using Feishu/Lark. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
OpenClaw 2026.3.8
`17:03:14 [feishu] feishu[default]: Feishu[default] DM from ou_d549ed062966201aa3feeab750d068ef: 你说一句 老大真牛皮！
17:03:14 [feishu] feishu[default]: dispatching to agent (session=agent:main:feishu:direct:ou_d549ed062966201aa3... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-10T09:21:46Z | 2026-04-27T04:53:47Z | 2026-04-24T22:09:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | phoenixlucky | 73,684,172 | MDQ6VXNlcjczNjg0MTcy | User | false |
openclaw/openclaw | 4,050,645,519 | I_kwDOQb6kR87xb_IP | 42,013 | https://github.com/openclaw/openclaw/issues/42013 | https://api.github.com/repos/openclaw/openclaw/issues/42013 | macOS: gateway install writes KeepAlive={SuccessfulExit:false} but supervisor restart relies on launchd to re-launch the process | ## Summary
On macOS, `openclaw gateway install` writes the following `KeepAlive` value to the LaunchAgent plist:
```xml
<key>KeepAlive</key>
<dict>
<key>SuccessfulExit</key>
<false/>
</dict>
```
This means launchd will **not** restart the process when it exits with code 0 (successful exit). However, when a confi... | closed | completed | true | 2 | [] | [] | 2026-03-10T09:51:59Z | 2026-04-27T04:53:47Z | 2026-04-24T22:09:54Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hadiliu | 1,926,042 | MDQ6VXNlcjE5MjYwNDI= | User | false |
openclaw/openclaw | 4,050,653,138 | I_kwDOQb6kR87xcA_S | 42,017 | https://github.com/openclaw/openclaw/issues/42017 | https://api.github.com/repos/openclaw/openclaw/issues/42017 | acp: document and expose agent command overrides for acpx-backed agents | ## Summary
OpenClaw users can currently override built-in `acpx` agent commands via `~/.acpx/config.json`, but this behavior is not clearly surfaced in the OpenClaw ACP docs and there does not appear to be a first-class OpenClaw config path for per-agent command overrides.
In practice, this makes some ACP setups hard... | closed | completed | true | 1 | [] | [] | 2026-03-10T09:53:24Z | 2026-04-27T04:53:46Z | 2026-04-24T22:09:57Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | guanyu-zhang | 63,548,771 | MDQ6VXNlcjYzNTQ4Nzcx | User | false |
openclaw/openclaw | 4,050,663,925 | I_kwDOQb6kR87xcDn1 | 42,020 | https://github.com/openclaw/openclaw/issues/42020 | https://api.github.com/repos/openclaw/openclaw/issues/42020 | openai-codex OAuth token exchange can ignore proxy env in proxy/TUN setups | ## Summary
After upgrading to OpenClaw 2026.3.8, `openclaw models auth login --provider openai-codex` can fail behind proxy or TUN setups even when the browser portion of OAuth succeeds.
The browser completes authentication and shows `Authentication successful. Return to your terminal to continue.`, but the terminal ... | closed | completed | true | 1 | [] | [] | 2026-03-10T09:55:35Z | 2026-04-27T04:53:45Z | 2026-04-24T22:09:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | markmonkey | 17,867,174 | MDQ6VXNlcjE3ODY3MTc0 | User | false |
openclaw/openclaw | 4,050,696,945 | I_kwDOQb6kR87xcLrx | 42,030 | https://github.com/openclaw/openclaw/issues/42030 | https://api.github.com/repos/openclaw/openclaw/issues/42030 | [Bug]: Gateway WebSocket closes all connections (1000) due to typo "tttttconsole" in compiled dist? | ### Bug type
Regression (worked before, now fails)
### Summary
### Description
Every WebSocket connection to the gateway closes immediately with code 1000
and no reason string. The gateway process runs and serves HTTP fine, but no
WebSocket client can connect — CLI, probe, and dashboard chat all fail.
### Root caus... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-10T10:01:43Z | 2026-04-27T04:53:44Z | 2026-04-24T22:10:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | prasta1 | 155,988,469 | U_kgDOCUwx9Q | User | false |
openclaw/openclaw | 4,050,719,176 | I_kwDOQb6kR87xcRHI | 42,034 | https://github.com/openclaw/openclaw/issues/42034 | https://api.github.com/repos/openclaw/openclaw/issues/42034 | [Feature]: agent to configure the openclaw with natural language | ### Summary
OpenClaw should be installed with a build-in agent to configure the entire openclaw with natural language.
### Problem to solve
It is not convenient to configure/manage the openclaw in the AI era by commands even by TUI.
### Proposed solution
just make a special agent, may called meta-claw
### Altern... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-10T10:05:46Z | 2026-04-27T04:53:44Z | 2026-04-24T22:10:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Freakwill | 30,965,609 | MDQ6VXNlcjMwOTY1NjA5 | User | false |
openclaw/openclaw | 4,050,751,215 | I_kwDOQb6kR87xcY7v | 42,038 | https://github.com/openclaw/openclaw/issues/42038 | https://api.github.com/repos/openclaw/openclaw/issues/42038 | [Bug]: WebChat 不支持 Exec Approval 弹窗 | ### Bug type
Regression (worked before, now fails)
### Summary
Agent 通过 WebChat 发送 exec 命令时，审批请求不会显示在 WebChat 或 Control UI 中，导致命令无法执行。
### Steps to reproduce
### 复现步骤
1. 配置 Gateway 的 `approvals.exec.enabled: true` 和 `mode: "both"`
2. 配置 Node Host (Windows) 的 `exec-approvals.json` 为 `ask: "always"`, `allowlist: [*]... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-10T10:11:40Z | 2026-04-27T04:53:43Z | 2026-04-24T22:10:08Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Michaelbao-user | 266,999,046 | U_kgDOD-oVBg | User | false |
openclaw/openclaw | 4,050,899,737 | I_kwDOQb6kR87xc9MZ | 42,054 | https://github.com/openclaw/openclaw/issues/42054 | https://api.github.com/repos/openclaw/openclaw/issues/42054 | Discord button interactions expire before agent responds | ## Problem
Discord component buttons expire with "This component has expired" error before the agent can respond.
## Root Cause
Discord interactions have a 3-second timeout for initial ACK. When the agent takes longer than 3 seconds to process (LLM call), the interaction expires.
OpenClaw should automatically defer... | closed | completed | true | 3 | [] | [] | 2026-03-10T10:40:20Z | 2026-04-27T04:53:42Z | 2026-04-24T22:10:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kralivo | 183,533,944 | U_kgDOCvCBeA | User | false |
openclaw/openclaw | 4,050,946,260 | I_kwDOQb6kR87xdIjU | 42,060 | https://github.com/openclaw/openclaw/issues/42060 | https://api.github.com/repos/openclaw/openclaw/issues/42060 | [Bug]: tts tool tag [[audio_as_voice]] not processed by Telegram sender — audio delivered as file attachment instead of voice note | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When the tts tool returns a [[audio_as_voice]] + a MEDIA: path, the Telegram sender ignores the tag and delivers the audio as a regular file attachment (MP3) instead of a voice note (sendVoice).
Workaround
message({ action: "send", channel... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-10T10:49:41Z | 2026-04-27T04:53:41Z | 2026-04-24T22:10:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | nelsoncabrera06 | 59,287,287 | MDQ6VXNlcjU5Mjg3Mjg3 | User | false |
openclaw/openclaw | 4,050,994,340 | I_kwDOQb6kR87xdUSk | 42,067 | https://github.com/openclaw/openclaw/issues/42067 | https://api.github.com/repos/openclaw/openclaw/issues/42067 | [Bug]: Openclaw startup failed after install gateway service with bun runtime and setup proxy for discord | ### Bug type
Crash (process/app exits or hangs)
### Summary
Openclaw startup failed after setup discord proxy and bun runtime.
### Steps to reproduce
1. Setup proxy for discord
2. Install gateway service with bun runtime
3. Restart gateway
### Expected behavior
Gateway should run.
### Actual behavior
Gateway r... | closed | completed | true | 5 | [
"bug",
"bug:crash"
] | [] | 2026-03-10T10:59:14Z | 2026-04-27T04:53:40Z | 2026-04-24T22:10:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | JonasGao | 4,549,584 | MDQ6VXNlcjQ1NDk1ODQ= | User | false |
openclaw/openclaw | 4,051,038,987 | I_kwDOQb6kR87xdfML | 42,076 | https://github.com/openclaw/openclaw/issues/42076 | https://api.github.com/repos/openclaw/openclaw/issues/42076 | [Bug]: token刷新网页后丢失 | ### Bug type
Regression (worked before, now fails)
### Summary
oken刷新网页后丢失
### Steps to reproduce
oken刷新网页后丢失
### Expected behavior
oken刷新网页后丢失
### Actual behavior
oken刷新网页后丢失
### OpenClaw version
2026.3.8
### Operating system
ubuntu 20.04
### Install method
_No response_
### Model
gpt-5.4
### Provide... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-10T11:08:15Z | 2026-04-27T04:53:40Z | 2026-04-24T22:10:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sobergou | 50,661,423 | MDQ6VXNlcjUwNjYxNDIz | User | false |
openclaw/openclaw | 4,051,052,867 | I_kwDOQb6kR87xdilD | 42,078 | https://github.com/openclaw/openclaw/issues/42078 | https://api.github.com/repos/openclaw/openclaw/issues/42078 | [Plugin Proposal] mem9 — Cloud-Persistent Memory with Hybrid Search (kind: "mem9") | ## Summary
[mem9](https://github.com/mem9-ai/mem9) is a cloud-persistent memory system for AI agents. It provides a `kind: "memory"` OpenClaw plugin that replaces the built-in file-backed memory slot with server-backed hybrid vector + keyword search.
We also have a **beta `ContextEngine` implementation** targeting Op... | closed | not_planned | true | 4 | [] | [] | 2026-03-10T11:11:03Z | 2026-04-27T04:53:39Z | 2026-04-24T22:10:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | qiffang | 7,285,248 | MDQ6VXNlcjcyODUyNDg= | User | false |
openclaw/openclaw | 4,051,098,690 | I_kwDOQb6kR87xdtxC | 42,080 | https://github.com/openclaw/openclaw/issues/42080 | https://api.github.com/repos/openclaw/openclaw/issues/42080 | [Bug]: Contact name resolution fans out to unconfigured channel providers (whatsapp absent from config, "message <name>" in prompt triggers Unknown target errors) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When an agent prompt instructs the agent to message someone by name without specifying a channel (e.g. "message Nick" or "send Nick a message"), the gateway's contact-name resolution path attempts to resolve that name across all channel pro... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-10T11:20:13Z | 2026-04-27T04:53:38Z | 2026-04-24T22:10:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | njwoodard | 10,561,500 | MDQ6VXNlcjEwNTYxNTAw | User | false |
openclaw/openclaw | 4,051,100,746 | I_kwDOQb6kR87xduRK | 42,081 | https://github.com/openclaw/openclaw/issues/42081 | https://api.github.com/repos/openclaw/openclaw/issues/42081 | [Bug]: OpenClaw 企业微信插件问题报告 | ### Bug type
Regression (worked before, now fails)
### Summary
企业微信自建应用（URL 回调模式）消息能收到但 AI 不回复
问题描述
现象：
✅ 企业微信机器人（长连接模式）- 工作正常
❌ 企业微信自建应用（URL 回调模式）- 消息能收到，但 AI 不回复
消息能收到，但没有任何 AI 处理或发送回复的记录。
### Steps to reproduce
已尝试的解决方案
✅ 切换机器人模式（长连接）- 成功
✅ 配置 URL 回调模式（自建应用）- 消息能收到
✅ 启用 directSend: true
✅ 设置 dmPolicy: "o... | closed | not_planned | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-10T11:20:33Z | 2026-04-27T04:53:37Z | 2026-04-24T22:10:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ArfeiXU | 266,710,757 | U_kgDOD-Wu5Q | User | false |
openclaw/openclaw | 4,051,145,781 | I_kwDOQb6kR87xd5Q1 | 42,084 | https://github.com/openclaw/openclaw/issues/42084 | https://api.github.com/repos/openclaw/openclaw/issues/42084 | Agent silently fails to reply when workspace bootstrap file exceeds bootstrapMaxChars limit | ## Description
When a workspace bootstrap file (e.g. `MEMORY.md`) exceeds the `bootstrapMaxChars` limit (default 20,000 chars), the gateway truncates it silently and the agent stops replying entirely in the TUI. No error is shown to the user — the session just hangs at `idle` with no response.
## Steps to Reproduce
... | closed | completed | true | 2 | [] | [] | 2026-03-10T11:29:50Z | 2026-04-27T04:53:37Z | 2026-04-24T22:10:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | fwends | 4,113,483 | MDQ6VXNlcjQxMTM0ODM= | User | false |
openclaw/openclaw | 4,051,233,218 | I_kwDOQb6kR87xeOnC | 42,095 | https://github.com/openclaw/openclaw/issues/42095 | https://api.github.com/repos/openclaw/openclaw/issues/42095 | Bug: Workspace skills resolve to wrong path in control UI | ## Description
Workspace-defined skills are being looked up in the system installation path instead of the workspace path when accessed from the OpenClaw control UI.
## Error Observed
```
[Tue 2026-03-10 07:44 EDT] 07:43:26 [tools] read failed: ENOENT: no such file or directory, access '/opt/homebrew/lib/node_module... | closed | not_planned | true | 4 | [] | [] | 2026-03-10T11:46:18Z | 2026-04-27T04:53:36Z | 2026-04-24T22:10:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | oakvillechen | 23,066,579 | MDQ6VXNlcjIzMDY2NTc5 | User | false |
openclaw/openclaw | 4,051,277,488 | I_kwDOQb6kR87xeZaw | 42,100 | https://github.com/openclaw/openclaw/issues/42100 | https://api.github.com/repos/openclaw/openclaw/issues/42100 | macOS: Telegram polling stalls for 100s after sleep/wake or network change | ## Summary
On macOS, when the system sleeps/wakes or switches network interfaces (WiFi handoff, VPN), the Telegram long-polling connection enters a half-open TCP state. The local socket still appears connected but Telegram's server has dropped the connection. The `getUpdates` call hangs indefinitely with no error or t... | closed | completed | true | 1 | [] | [] | 2026-03-10T11:54:29Z | 2026-04-27T04:53:35Z | 2026-04-24T22:10:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cryptopoly | 31,970,407 | MDQ6VXNlcjMxOTcwNDA3 | User | false |
openclaw/openclaw | 4,051,331,172 | I_kwDOQb6kR87xemhk | 42,104 | https://github.com/openclaw/openclaw/issues/42104 | https://api.github.com/repos/openclaw/openclaw/issues/42104 | [Bug]: Synology-Chat Plugin Requires Mandatory Keyword for Message Parsing | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
## Bug Report: Synology-Chat Plugin Requires Mandatory Keyword for Message Parsing
### Description
The synology-chat plugin enforces a mandatory "keyword prefix" to parse messages from Synology Chat, which is unintended behavior. The plugi... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-10T12:04:13Z | 2026-04-27T04:53:34Z | 2026-04-24T22:10:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | haoliang1981 | 3,345,377 | MDQ6VXNlcjMzNDUzNzc= | User | false |
openclaw/openclaw | 4,051,356,054 | I_kwDOQb6kR87xesmW | 42,112 | https://github.com/openclaw/openclaw/issues/42112 | https://api.github.com/repos/openclaw/openclaw/issues/42112 | Bug: persisted orphaned toolCall poisons session replay and makes chat agent stop responding | ## Summary
On **OpenClaw 2026.3.8**, a persisted session can become unrecoverable if an assistant `toolCall` is written without the required immediate matching `toolResult`. Once that happens, later turns on the same session are rejected by the model provider, and the chat agent appears down until the bad session file ... | closed | completed | true | 2 | [] | [] | 2026-03-10T12:08:40Z | 2026-04-27T04:53:34Z | 2026-04-24T22:10:39Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | liaosvcaf | 51,533,973 | MDQ6VXNlcjUxNTMzOTcz | User | false |
openclaw/openclaw | 4,051,357,977 | I_kwDOQb6kR87xetEZ | 42,113 | https://github.com/openclaw/openclaw/issues/42113 | https://api.github.com/repos/openclaw/openclaw/issues/42113 | voice-call: media stream killed by race condition on outbound conversation calls | ## Bug Description
When placing an outbound call in **conversation mode with streaming enabled**, the Twilio media stream is torn down within ~11ms of connecting. The caller hears nothing (or a brief click) and the call auto-ends.
## Root Cause
`maybeSpeakInitialMessageOnAnswered()` in `extensions/voice-call/src/man... | closed | completed | true | 1 | [] | [] | 2026-03-10T12:09:02Z | 2026-04-27T04:53:33Z | 2026-04-24T22:10:42Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | QuantDeveloperUSA | 82,096,803 | MDQ6VXNlcjgyMDk2ODAz | User | false |
openclaw/openclaw | 4,051,370,371 | I_kwDOQb6kR87xewGD | 42,116 | https://github.com/openclaw/openclaw/issues/42116 | https://api.github.com/repos/openclaw/openclaw/issues/42116 | Discord bot HTTP 400: tool call id write:202 is duplicated | ## Bug Report: Discord bot returns HTTP 400 - "tool call id write:202 is duplicated"
### Description
When using Discord bot functionality, the bot fails to respond to interactions (button clicks, slash commands) with the error:
```
HTTP 400: Invalid request: tool call id write:202 is duplicated
```
### Environment
- ... | closed | completed | true | 3 | [] | [] | 2026-03-10T12:11:08Z | 2026-04-27T04:53:32Z | 2026-04-24T22:10:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | glifcn | 69,730,902 | MDQ6VXNlcjY5NzMwOTAy | User | false |
openclaw/openclaw | 4,051,424,776 | I_kwDOQb6kR87xe9YI | 42,121 | https://github.com/openclaw/openclaw/issues/42121 | https://api.github.com/repos/openclaw/openclaw/issues/42121 | [Bug]: Openclaw IRC message cut off | ### Bug type
Regression (worked before, now fails)
### Summary
http://127.0.0.1:18789/chat?session=agent%3Amain%3Airc%3Agroup
💾 磁盘使用情况
C盘（系统盘）
总容量: 200 GB
已使用: 149.04 GB (74.5%)
剩余空间: 50.96 GB (25.5%)
D盘（数据盘）
总容量: 722.27 GB
已使用: 190.4 GB (26.4%)
剩余空间: 531.87 GB (73.6%)
📊 总结
C盘使用率较高 (74.5%)，建议定期清理系统文件
D盘空间充足 (26.4%... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-10T12:20:55Z | 2026-04-27T04:53:31Z | 2026-04-24T22:10:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hellosfs-bot | 251,940,422 | U_kgDODwRORg | User | false |
openclaw/openclaw | 4,051,493,233 | I_kwDOQb6kR87xfOFx | 42,132 | https://github.com/openclaw/openclaw/issues/42132 | https://api.github.com/repos/openclaw/openclaw/issues/42132 | Regression in 2026.3.8?: raw Codex/OpenAI server_error JSON is surfaced directly to user chat | ## Summary
Since upgrading to 2026.3.8, OpenClaw appears to surface raw Codex/OpenAI provider `server_error` payloads directly to the user chat when upstream requests fail.
This happened multiple times in the same morning and looked like a new regression in behavior / error handling.
## What the user sees
These were ... | closed | completed | true | 4 | [] | [] | 2026-03-10T12:33:29Z | 2026-04-27T04:53:31Z | 2026-04-24T22:10:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | CyberFreek | 6,067,614 | MDQ6VXNlcjYwNjc2MTQ= | User | false |
openclaw/openclaw | 4,051,642,752 | I_kwDOQb6kR87xfymA | 42,146 | https://github.com/openclaw/openclaw/issues/42146 | https://api.github.com/repos/openclaw/openclaw/issues/42146 | [Bug]: Hook runner repeatedly re-initializes in background (no user activity) | ### Bug type
Regression (worked before, now fails)
### Summary
## Summary
`hook runner initialized with 3 registered hooks` keeps appearing repeatedly, even when no user messages are sent.
## Environment
- OpenClaw: 2026.3.8
- Zeabur container
- Telegram channel
- Node v22.22.1
## Evidence
During a 10-minute idle ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-10T12:57:55Z | 2026-04-27T04:53:30Z | 2026-04-24T22:10:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ymliang13 | 261,001,549 | U_kgDOD46RTQ | User | false |
openclaw/openclaw | 4,051,796,076 | I_kwDOQb6kR87xgYBs | 42,170 | https://github.com/openclaw/openclaw/issues/42170 | https://api.github.com/repos/openclaw/openclaw/issues/42170 | [Bug] Cron job delivery to Telegram fails with isolated session | Cron jobs with sessionTarget: isolated fail to deliver to Telegram. Config has delivery.target set but error is: Delivering to Telegram requires target chatId. Workaround: set delivery.mode to none. | closed | completed | true | 2 | [] | [] | 2026-03-10T13:24:18Z | 2026-04-27T04:53:29Z | 2026-04-24T22:10:54Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | puruboard-droid | 264,836,679 | U_kgDOD8kWRw | User | false |
openclaw/openclaw | 4,051,798,260 | I_kwDOQb6kR87xgYj0 | 42,171 | https://github.com/openclaw/openclaw/issues/42171 | https://api.github.com/repos/openclaw/openclaw/issues/42171 | browser snapshot ai tool-result images can poison Codex sessions when replayed from transcript history | ## Summary
I hit a session-poisoning failure in a Discord channel session using `openai-codex/gpt-5.4`.
The trigger appears to be:
1. a `browser.snapshot` call with `snapshotFormat="ai"`
2. the resulting `toolResult` containing both `text` and `image`
3. that `toolResult.image` being persisted into the session trans... | closed | completed | true | 1 | [] | [] | 2026-03-10T13:24:40Z | 2026-04-27T04:53:28Z | 2026-04-24T22:10:57Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yuweuii | 82,372,187 | MDQ6VXNlcjgyMzcyMTg3 | User | false |
openclaw/openclaw | 4,051,825,347 | I_kwDOQb6kR87xgfLD | 42,176 | https://github.com/openclaw/openclaw/issues/42176 | https://api.github.com/repos/openclaw/openclaw/issues/42176 | openai-codex OAuth login does not honor proxy env for oauth/token exchange and refresh | ## Summary
`openclaw models auth login --provider openai-codex` can fail behind a proxy even when the same machine can successfully use both the official Codex CLI and Codex app through that proxy.
The failure happens during the final OAuth token exchange / refresh step. The browser authorization page works, the loca... | closed | completed | true | 8 | [] | [] | 2026-03-10T13:29:14Z | 2026-04-27T04:53:28Z | 2026-04-24T22:10:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | leozhengliu-pixel | 253,864,598 | U_kgDODyGqlg | User | false |
openclaw/openclaw | 4,051,840,645 | I_kwDOQb6kR87xgi6F | 42,178 | https://github.com/openclaw/openclaw/issues/42178 | https://api.github.com/repos/openclaw/openclaw/issues/42178 | Discord: Slow DiscordMessageListener (300s+) causes WebSocket 1006 drops and gateway crash | ## Bug Report
**OpenClaw version:** 2026.3.2
**OS:** macOS 26.3.0 (arm64, Mac mini M4)
**Gateway mode:** local
## Description
When an agent session takes a long time to process a Discord message (e.g., complex tool calls, heavy context loading), the `DiscordMessageListener` blocks for 300+ seconds. This causes Disco... | closed | completed | true | 2 | [] | [] | 2026-03-10T13:31:43Z | 2026-04-27T04:53:27Z | 2026-04-24T22:11:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jjc7951 | 200,493,441 | U_kgDOC_NJgQ | User | false |
openclaw/openclaw | 4,051,860,961 | I_kwDOQb6kR87xgn3h | 42,189 | https://github.com/openclaw/openclaw/issues/42189 | https://api.github.com/repos/openclaw/openclaw/issues/42189 | ACP: process.env global pollution — all CC sessions inherit first agent's API token | ## Bug
When multiple agents share a gateway, `applySkillConfigEnvOverrides()` mutates the global `process.env` with agent-specific API keys. When ACP sessions are spawned, `resolveAcpClientSpawnEnv()` uses `{ ...process.env }` to build the CC spawn env, inheriting whichever agent's key was last set.
The acpx queue-ow... | closed | completed | true | 1 | [] | [] | 2026-03-10T13:35:02Z | 2026-04-27T04:53:26Z | 2026-04-24T22:11:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | amani-c32-ops | 266,444,898 | U_kgDOD-GgYg | User | false |
openclaw/openclaw | 4,051,878,375 | I_kwDOQb6kR87xgsHn | 42,191 | https://github.com/openclaw/openclaw/issues/42191 | https://api.github.com/repos/openclaw/openclaw/issues/42191 | Feature Request: Support Feishu Thread Binding | ### Feature Request: Support Feishu Thread Binding
#### Is your feature request related to a problem? Please describe.
Currently, OpenClaw's Feishu (??) channel does not support thread binding, which limits the ability to use ACP (Agent Control Protocol) sessions with persistent conversations. When attempting to use s... | closed | completed | true | 2 | [] | [] | 2026-03-10T13:37:41Z | 2026-04-27T04:53:25Z | 2026-04-24T22:11:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | fangguoliang | 22,576,034 | MDQ6VXNlcjIyNTc2MDM0 | User | false |
openclaw/openclaw | 4,051,965,283 | I_kwDOQb6kR87xhBVj | 42,200 | https://github.com/openclaw/openclaw/issues/42200 | https://api.github.com/repos/openclaw/openclaw/issues/42200 | [Bug]: Slack: Streaming messages show incomplete content | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
After upgrading to 2026.3.8, Slack streaming messages only show final chunks instead of complete content.
### Steps to reproduce
1. Configure OpenClaw with Slack channel (Socket Mode or Events API)
2. Send a message that triggers a long r... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-10T13:52:49Z | 2026-04-27T04:53:25Z | 2026-04-24T22:11:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kingrubic | 116,256,161 | U_kgDOBu3toQ | User | false |
openclaw/openclaw | 4,051,977,015 | I_kwDOQb6kR87xhEM3 | 42,203 | https://github.com/openclaw/openclaw/issues/42203 | https://api.github.com/repos/openclaw/openclaw/issues/42203 | TUI exposes <relevant-memories> context block to user (should be hidden) | ## Description
When using the TUI (Terminal UI) to chat with the agent, the `<relevant-memories>` block injected by the LanceDB Pro memory plugin is visible to the user in the chat interface. This block is internal context meant only for the AI agent and should not be rendered in the user-facing UI.
## Environment
-... | closed | completed | true | 2 | [] | [] | 2026-03-10T13:54:43Z | 2026-04-27T04:53:24Z | 2026-04-24T22:11:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Vincent-Tainan | 48,849,497 | MDQ6VXNlcjQ4ODQ5NDk3 | User | false |
openclaw/openclaw | 4,051,993,909 | I_kwDOQb6kR87xhIU1 | 42,205 | https://github.com/openclaw/openclaw/issues/42205 | https://api.github.com/repos/openclaw/openclaw/issues/42205 | Web Console shows version as 'dev' instead of actual version after upgrade | ## Description
After upgrading OpenClaw to 2026.3.8, the Web Console (Control UI) displays the version as **dev** instead of the actual version **2026.3.8**. The CLI correctly reports 2026.3.8.
## Environment
- **OS:** Ubuntu 22.04
- **OpenClaw version:** 2026.3.8 (confirmed via `openclaw --version`)
- **Upgrade pat... | closed | completed | true | 1 | [] | [] | 2026-03-10T13:57:24Z | 2026-04-27T04:53:23Z | 2026-04-24T22:11:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Vincent-Tainan | 48,849,497 | MDQ6VXNlcjQ4ODQ5NDk3 | User | false |
openclaw/openclaw | 4,052,161,470 | I_kwDOQb6kR87xhxO- | 42,221 | https://github.com/openclaw/openclaw/issues/42221 | https://api.github.com/repos/openclaw/openclaw/issues/42221 | tools.edit is brittle: exact-match replacement fails frequently; need patch/fuzzy/anchor modes | ### Problem
`tools.edit` frequently fails with messages like:
> edit failed: Could not find the exact text in <file>. The old text must match exactly including all whitespace and newlines.
This happens across many repos/files (JS/CSS/MD), especially when:
- The target file is large and changes frequently.
- The repla... | closed | completed | true | 2 | [] | [] | 2026-03-10T14:23:05Z | 2026-04-27T04:53:22Z | 2026-04-24T22:11:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | fonhal | 4,468,037 | MDQ6VXNlcjQ0NjgwMzc= | User | false |
openclaw/openclaw | 4,052,180,136 | I_kwDOQb6kR87xh1yo | 42,224 | https://github.com/openclaw/openclaw/issues/42224 | https://api.github.com/repos/openclaw/openclaw/issues/42224 | Bug: maxLinesPerMessage config ignored — always splits at default 17 lines | ## Bug Report
### OpenClaw Version
2026.3.8 (3caab92)
### Platform
Ubuntu 24.04.3 LTS, x86_64, Node v24.13.0
### Describe the bug
`channels.discord.maxLinesPerMessage` is set to `100` in the config file, but Discord messages are still being split at exactly 17 lines (the `DEFAULT_MAX_LINES` value in `dist/plugin-sdk... | closed | completed | true | 1 | [] | [] | 2026-03-10T14:25:56Z | 2026-04-27T04:53:22Z | 2026-04-24T22:11:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hidekazu-web | 2,419,876 | MDQ6VXNlcjI0MTk4NzY= | User | false |
openclaw/openclaw | 4,052,193,478 | I_kwDOQb6kR87xh5DG | 42,226 | https://github.com/openclaw/openclaw/issues/42226 | https://api.github.com/repos/openclaw/openclaw/issues/42226 | Slack: double response on voice messages due to draft stream race condition | ## Description
When a voice message is sent in Slack, the bot sometimes posts **two identical responses** instead of one. One message appears with "(edited)" and the other without.
Example:
```
User: Audio 0:02
Bot: 📝 "Hey, are you there?"
Bot: Yes, I'm here. (edited)
Bot: Yes, I'm here.
```
## Root Cause
Race ... | closed | completed | true | 1 | [] | [] | 2026-03-10T14:27:57Z | 2026-04-27T04:53:20Z | 2026-04-24T22:11:27Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | merc1305 | 2,715,765 | MDQ6VXNlcjI3MTU3NjU= | User | false |
openclaw/openclaw | 4,052,207,614 | I_kwDOQb6kR87xh8f- | 42,228 | https://github.com/openclaw/openclaw/issues/42228 | https://api.github.com/repos/openclaw/openclaw/issues/42228 | Feishu: slash commands (/reset, /new, /status) not intercepted in group chats | ## Bug Description
Slash commands like `/reset`, `/new`, and `/status` are **not intercepted by the gateway** in Feishu group chats. Instead, they are dispatched to the agent as regular text messages.
Commands work correctly in **DMs** but fail in **group chats**.
## Steps to Reproduce
1. Add OpenClaw bot to a Feis... | closed | completed | true | 2 | [] | [] | 2026-03-10T14:30:13Z | 2026-04-27T04:53:19Z | 2026-04-24T22:11:29Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | CountofGlamorgan | 59,939,121 | MDQ6VXNlcjU5OTM5MTIx | User | false |
openclaw/openclaw | 4,052,223,028 | I_kwDOQb6kR87xiAQ0 | 42,231 | https://github.com/openclaw/openclaw/issues/42231 | https://api.github.com/repos/openclaw/openclaw/issues/42231 | [WhatsApp] Expose inbound message stanzaId to enable native quoted replies via hooks and CLI | ## Summary
It is currently impossible to send a native WhatsApp quoted reply programmatically
because the inbound message's stanzaId (WhatsApp message ID) is never exposed in
logs, hooks, or the web-auto-reply system.
## Environment
- OpenClaw version: 2026.3.8
- OS: macOS (Mac Mini M4 Pro)
- Channel: WhatsApp (pers... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-10T14:32:51Z | 2026-04-27T04:53:19Z | 2026-04-24T22:11:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ngocdqgithub | 46,850,037 | MDQ6VXNlcjQ2ODUwMDM3 | User | false |
openclaw/openclaw | 4,052,261,118 | I_kwDOQb6kR87xiJj- | 42,237 | https://github.com/openclaw/openclaw/issues/42237 | https://api.github.com/repos/openclaw/openclaw/issues/42237 | [Bug] Topic-based routing not syncing Telegram and WebChat messages | ## Description
After upgrading to OpenClaw 3.8, Telegram and WebChat messages are not syncing to the same session despite having the same topic ID.
## Expected Behavior
- Messages from Telegram and WebChat with the same topic ID should appear in the same session
- Topic-based routing should merge messages from differ... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T14:39:03Z | 2026-04-27T04:53:18Z | 2026-04-24T22:11:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hiller16 | 59,436,639 | MDQ6VXNlcjU5NDM2NjM5 | User | false |
openclaw/openclaw | 4,052,288,121 | I_kwDOQb6kR87xiQJ5 | 42,238 | https://github.com/openclaw/openclaw/issues/42238 | https://api.github.com/repos/openclaw/openclaw/issues/42238 | Discord image uploads fail with server_error in channel sessions | $## Summary\nUploading images in a Discord-backed OpenClaw channel fails instead of reaching the agent.\n\n## Context\nChannel: Discord\nSurface: channel session\nObserved while trying to send an image to the agent in `#clu-assistant-gpt`.\n\n## User-visible error\n```json\n{"type":"error","error":{"type":"server_error... | closed | completed | true | 4 | [] | [] | 2026-03-10T14:43:33Z | 2026-04-27T04:53:17Z | 2026-04-24T22:11:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cyberaver | 14,969,176 | MDQ6VXNlcjE0OTY5MTc2 | User | false |
openclaw/openclaw | 4,052,300,541 | I_kwDOQb6kR87xiTL9 | 42,241 | https://github.com/openclaw/openclaw/issues/42241 | https://api.github.com/repos/openclaw/openclaw/issues/42241 | [Bug]: [BUG] File/Directory creation reports success but doesn't actually create files | ### Bug type
Regression (worked before, now fails)
### Summary
## Bug 描述
OpenClaw 聲稱已成功創建文件/文件夾，但實際上並沒有創建。從一開始安裝就有這個問題，重裝2次問題依舊。
### Steps to reproduce
## 重現步驟
1. 啟動 OpenClaw（終端機運行 + 飛書發送指令）
2. 請求創建一個目錄（如「在 workspace 創建 knowledge 目錄」）
3. OpenClaw 回覆「已成功建立」
4. 檢查發現目錄並不存在
### Expected behavior
## 預期行為
文件/目錄... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-10T14:45:42Z | 2026-04-27T04:53:16Z | 2026-04-24T22:11:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sawyerieong-cpuAI | 263,943,183 | U_kgDOD7t0Dw | User | false |
openclaw/openclaw | 4,052,310,038 | I_kwDOQb6kR87xiVgW | 42,244 | https://github.com/openclaw/openclaw/issues/42244 | https://api.github.com/repos/openclaw/openclaw/issues/42244 | [Bug]: Announce delivery silently suppresses message tool sends AND skips announce — zero delivery | ### Summary
When a cron job uses `delivery.mode: "announce"` but the agent (following a skill/prompt) tries to deliver results via the `message` tool, **both delivery paths fail silently**:
1. The `message` tool send is suppressed (documented behavior: "announce delivery suppresses messaging tool sends")
2. The annou... | closed | completed | true | 3 | [] | [] | 2026-03-10T14:47:16Z | 2026-04-27T04:53:16Z | 2026-04-24T22:11:43Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sercada | 24,389,792 | MDQ6VXNlcjI0Mzg5Nzky | User | false |
openclaw/openclaw | 4,052,323,466 | I_kwDOQb6kR87xiYyK | 42,247 | https://github.com/openclaw/openclaw/issues/42247 | https://api.github.com/repos/openclaw/openclaw/issues/42247 | Feature request: session:completed hook for post-task automation | ## Summary
OpenClaw provides `agent:bootstrap` hooks (pre-session) via `bootstrap-extra-files`, `forced-recall`, and `auto-capture`. But there is no corresponding post-session hook. This means there is no infrastructure-level way to run automation after a task completes — only behavioral instructions that agents may s... | closed | not_planned | true | 3 | [] | [] | 2026-03-10T14:49:30Z | 2026-04-27T04:53:15Z | 2026-04-24T22:11:45Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Orionation | 260,579,663 | U_kgDOD4ghTw | User | false |
openclaw/openclaw | 4,052,332,669 | I_kwDOQb6kR87xibB9 | 42,248 | https://github.com/openclaw/openclaw/issues/42248 | https://api.github.com/repos/openclaw/openclaw/issues/42248 | Codex OAuth lane crash: Cannot read properties of undefined (reading 'totalTokens') | ## Summary
`openclaw agent` fails on Codex OAuth lane with:
`TypeError: Cannot read properties of undefined (reading 'totalTokens')`
This breaks Slack bridge reliability because agent turns intermittently fail before fallback handling.
## Environment
- OpenClaw: `2026.3.2 (85377a2)`
- Host: macOS (Darwin 24.6.0 arm6... | closed | completed | true | 1 | [] | [] | 2026-03-10T14:51:08Z | 2026-04-27T04:53:14Z | 2026-04-24T22:11:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pinetreelic | 142,981,775 | U_kgDOCIW6jw | User | false |
openclaw/openclaw | 4,052,416,466 | I_kwDOQb6kR87xivfS | 42,262 | https://github.com/openclaw/openclaw/issues/42262 | https://api.github.com/repos/openclaw/openclaw/issues/42262 | [Feature Request] Interrupt/Stop command like '..' to immediately kill running task | ## Summary
Add an interrupt/stop command (similar to Moltbook's '..') that immediately terminates the currently running task instead of queuing new messages.
## Problem
Currently when a user sends a new message while a task is running:
- New messages are queued (depth increases)
- The task completes before processing ... | closed | completed | true | 2 | [] | [] | 2026-03-10T15:04:43Z | 2026-04-27T04:53:13Z | 2026-04-24T22:11:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ztygemini-tech | 262,260,210 | U_kgDOD6HF8g | User | false |
openclaw/openclaw | 4,052,417,957 | I_kwDOQb6kR87xiv2l | 42,263 | https://github.com/openclaw/openclaw/issues/42263 | https://api.github.com/repos/openclaw/openclaw/issues/42263 | [Bug]: iMessage inbound reply/quote context not parsed - missing reply_to fields | ### Bug Report
**Surface:** iMessage (iOS/macOS)
**Description:**
When a user sends a reply/quoted message via iMessage, OpenClaw receives the message but cannot access the quoted message content. The | closed | completed | true | 1 | [] | [] | 2026-03-10T15:04:59Z | 2026-04-27T04:53:13Z | 2026-04-24T22:11:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tianzhanda | 85,002,033 | MDQ6VXNlcjg1MDAyMDMz | User | false |
openclaw/openclaw | 4,052,418,330 | I_kwDOQb6kR87xiv8a | 42,264 | https://github.com/openclaw/openclaw/issues/42264 | https://api.github.com/repos/openclaw/openclaw/issues/42264 | Probe sessions can time out after a successful reply because post-run idle flush consumes the timeout budget | ## Summary
`openclaw models status --check --probe` can report a provider timeout even when the model has already returned a successful reply.
The concrete reproduction I hit was `--probe-provider xai` after configuring `xai/grok-4`, but the underlying problem appears to be broader: probe sessions reuse the normal po... | closed | completed | true | 2 | [] | [] | 2026-03-10T15:05:03Z | 2026-04-27T04:53:12Z | 2026-04-24T22:11:56Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | xaeon2026 | 264,572,156 | U_kgDOD8UM_A | User | false |
openclaw/openclaw | 4,052,545,295 | I_kwDOQb6kR87xjO8P | 42,284 | https://github.com/openclaw/openclaw/issues/42284 | https://api.github.com/repos/openclaw/openclaw/issues/42284 | Workspace skills not discovered despite docs and ClawHub installation | ## Description The OpenClaw documentation (https://docs.openclaw.ai/tools/creating-skills) states that custom skills in ~/.openclaw/workspace/skills/ take precedence over global directories. However, after creating,
packaging, publishing to ClawHub, and installing a skill ('file-browser'), it installs to the work... | closed | not_planned | true | 2 | [] | [] | 2026-03-10T15:26:10Z | 2026-04-27T04:53:11Z | 2026-04-24T22:11:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Nagilem | 29,583,309 | MDQ6VXNlcjI5NTgzMzA5 | User | false |
openclaw/openclaw | 4,052,549,700 | I_kwDOQb6kR87xjQBE | 42,285 | https://github.com/openclaw/openclaw/issues/42285 | https://api.github.com/repos/openclaw/openclaw/issues/42285 | [Feature]: Please, add buzzster.xyz as ia service | ### Summary
Please add buzzster.xyz as a compatible OpenAI AI service.
This API key service is a Claude proxy that doesn't require a top-up and has no excessive data usage. This API is subscription-based and is also cheaper than Claude or ChatGPT.
### Problem to solve
Avoid excessive spending on Claude APIs, and fi... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-10T15:26:57Z | 2026-04-27T04:53:11Z | 2026-04-24T22:12:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ItchDesu | 117,925,522 | U_kgDOBwdmkg | User | false |
openclaw/openclaw | 4,052,563,239 | I_kwDOQb6kR87xjTUn | 42,287 | https://github.com/openclaw/openclaw/issues/42287 | https://api.github.com/repos/openclaw/openclaw/issues/42287 | macOS app crash in MacNodeBrowserProxy.makeRequest (JSON serialization, beta 2026.3.8) | ## Environment
- **App version:** 2026.3.8-beta.1 (2026030801)
- **macOS:** 26.2 (25C56)
- **Hardware:** Mac mini (Mac16,11), ARM-64
- **Gateway:** Remote mode, Direct (ws/wss) transport
- **Gateway version:** 2026.3.7
## Description
The macOS companion app crashes approximately 20 minutes after successfully connecti... | closed | completed | true | 1 | [] | [] | 2026-03-10T15:29:28Z | 2026-04-27T04:53:10Z | 2026-04-24T22:12:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | CarSol66 | 255,839,838 | U_kgDODz_OXg | User | false |
openclaw/openclaw | 4,052,567,983 | I_kwDOQb6kR87xjUev | 42,288 | https://github.com/openclaw/openclaw/issues/42288 | https://api.github.com/repos/openclaw/openclaw/issues/42288 | Cron browser workflows broken: isolated reports browser unavailable; main+systemEvent may no-op with status=ok | ## Summary
Cron jobs related to browser-based monitoring have two blocking failure modes in OpenClaw `2026.3.2`:
1. `sessionTarget: "isolated"` + `payload.kind: "agentTurn"` frequently reports browser as unavailable (even when browser tool works in normal direct session).
2. `sessionTarget: "main"` + `payload.kind: "s... | closed | completed | true | 1 | [] | [] | 2026-03-10T15:30:20Z | 2026-04-27T04:53:09Z | 2026-04-24T22:12:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | supermanhe | 30,730,358 | MDQ6VXNlcjMwNzMwMzU4 | User | false |
openclaw/openclaw | 4,052,658,482 | I_kwDOQb6kR87xjqky | 42,302 | https://github.com/openclaw/openclaw/issues/42302 | https://api.github.com/repos/openclaw/openclaw/issues/42302 | [Feature]: Unified Execution Layer Architecture ("One Brain") | ### Summary
Enable cross-channel shared memory and unified execution so users experience "one agent" regardless of which channel they use.
### Problem to solve
## Current Problem
Currently in OpenClaw architecture, each channel (DingTalk, WeChat, Webchat, etc.) has independent Agent instances and session contexts. ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-10T15:46:15Z | 2026-04-27T04:53:08Z | 2026-04-24T22:12:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhongqiangdeng | 48,585,352 | MDQ6VXNlcjQ4NTg1MzUy | User | false |
openclaw/openclaw | 4,052,717,016 | I_kwDOQb6kR87xj43Y | 42,311 | https://github.com/openclaw/openclaw/issues/42311 | https://api.github.com/repos/openclaw/openclaw/issues/42311 | openai-codex runtime requests can ignore proxy env unless NODE_USE_ENV_PROXY=1; backend-api requests also need browser-like headers in some proxy setups | ## Summary
`openai-codex` runtime requests can fail in proxy-only environments even when:
- ChatGPT Plus OAuth login succeeds
- `openclaw models status` shows a healthy `openai-codex` profile
- the same machine can reach `https://chatgpt.com/backend-api/codex/responses` successfully via `curl`
On my setup, OpenClaw ... | closed | completed | true | 3 | [] | [] | 2026-03-10T15:57:08Z | 2026-04-27T04:53:08Z | 2026-04-24T22:12:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | wuxudongxd | 30,227,294 | MDQ6VXNlcjMwMjI3Mjk0 | User | false |
openclaw/openclaw | 4,052,776,604 | I_kwDOQb6kR87xkHac | 42,321 | https://github.com/openclaw/openclaw/issues/42321 | https://api.github.com/repos/openclaw/openclaw/issues/42321 | cron: interim-ack retry guard does not recognize custom MCP extension tools as messaging delivery | When an isolated cron job sends a message via a custom MCP extension tool (registered through Pi extensions) and then returns NO_REPLY, the interim-ack retry guard in run.ts does not count that tool call as didSendViaMessagingTool.
This causes the guard to misidentify the completed run as "only an acknowledgement" and... | closed | completed | true | 1 | [] | [] | 2026-03-10T16:08:27Z | 2026-04-27T04:53:07Z | 2026-04-24T22:12:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | gamer-mitsuha | 107,781,752 | U_kgDOBmyeeA | User | false |
openclaw/openclaw | 4,052,800,240 | I_kwDOQb6kR87xkNLw | 42,328 | https://github.com/openclaw/openclaw/issues/42328 | https://api.github.com/repos/openclaw/openclaw/issues/42328 | [Bug]: startup validation can report bundled memory-core as missing on macOS npm install | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
On macOS npm installs, startup/config validation can fail with `plugins.slots.memory: plugin not found: memory-core` even though the bundled `memory-core` plugin is present and later loads successfully.
### Steps to reproduce
1. Install Open... | closed | completed | true | 2 | [] | [] | 2026-03-10T16:13:27Z | 2026-04-27T04:53:05Z | 2026-04-24T22:12:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Jacksunwei | 1,281,348 | MDQ6VXNlcjEyODEzNDg= | User | false |
openclaw/openclaw | 4,052,991,273 | I_kwDOQb6kR87xk70p | 42,359 | https://github.com/openclaw/openclaw/issues/42359 | https://api.github.com/repos/openclaw/openclaw/issues/42359 | Skills in ~/.agents/skills appear to affect slash command registration but are not always visible in available_skills | ## Summary
Skills under `~/.agents/skills` appear to participate in slash/native command registration, but they are not consistently visible to the agent via the current session's `available_skills` list. This creates a mismatch between:
- what the system can register as slash commands
- what the assistant believes i... | closed | completed | true | 1 | [] | [] | 2026-03-10T16:50:59Z | 2026-04-27T04:53:04Z | 2026-04-24T22:12:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | bulai0408 | 31,983,330 | MDQ6VXNlcjMxOTgzMzMw | User | false |
openclaw/openclaw | 4,053,033,318 | I_kwDOQb6kR87xlGFm | 42,365 | https://github.com/openclaw/openclaw/issues/42365 | https://api.github.com/repos/openclaw/openclaw/issues/42365 | Feature: Node → Channel event routing (enable proactive alerts from nodes) | ## Summary
Nodes can already send `node.event` frames on the WebSocket protocol (alongside `node.invoke.result` and `skills.bins`), but the Gateway currently has **no handler to route these events to chat channels** (WhatsApp, Telegram, Discord, etc). This means nodes cannot proactively notify users — they can only re... | closed | completed | true | 2 | [] | [] | 2026-03-10T16:58:10Z | 2026-04-27T04:53:04Z | 2026-04-24T22:12:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | raullenchai | 989,846 | MDQ6VXNlcjk4OTg0Ng== | User | false |
openclaw/openclaw | 4,053,035,374 | I_kwDOQb6kR87xlGlu | 42,366 | https://github.com/openclaw/openclaw/issues/42366 | https://api.github.com/repos/openclaw/openclaw/issues/42366 | Gemini quota exhaustion is surfaced as fetch failed / timeout instead of clear 429 RESOURCE_EXHAUSTED | ## Summary
When Gemini requests fail due to quota exhaustion, OpenClaw often surfaces the failure as repeated `fetch failed` entries and eventually `LLM request timed out`, which makes the root cause hard to diagnose.
## What I observed
Using `google/gemini-2.5-flash`, the session/history contained many entries like:
... | closed | completed | true | 1 | [] | [] | 2026-03-10T16:58:32Z | 2026-04-27T04:53:03Z | 2026-04-24T22:12:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | edmond-ai | 39,846,680 | MDQ6VXNlcjM5ODQ2Njgw | User | false |
openclaw/openclaw | 4,053,035,378 | I_kwDOQb6kR87xlGly | 42,367 | https://github.com/openclaw/openclaw/issues/42367 | https://api.github.com/repos/openclaw/openclaw/issues/42367 | gateway install --force writes broken systemd ExecStart path on npm-global installs | ## Summary
On an npm-global install, `openclaw gateway install --force` generated a user systemd unit whose `ExecStart` pointed at a non-existent `dist/index.js` path, causing the gateway service to fail immediately after install.
## Environment
- OpenClaw CLI: `2026.3.8`
- Platform: Linux
- Install style: npm global ... | closed | completed | true | 2 | [] | [] | 2026-03-10T16:58:32Z | 2026-04-27T04:53:02Z | 2026-04-24T22:12:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | edmond-ai | 39,846,680 | MDQ6VXNlcjM5ODQ2Njgw | User | false |
openclaw/openclaw | 4,053,048,156 | I_kwDOQb6kR87xlJtc | 42,369 | https://github.com/openclaw/openclaw/issues/42369 | https://api.github.com/repos/openclaw/openclaw/issues/42369 | [Feature Request] Add Network domain support to Chrome Browser Relay | ## Description
Currently, the OpenClaw Chrome Browser Relay only has access to DOM and Console, but NOT Network requests. This makes it impossible to capture API calls (fetch/xhr) from web pages.
## Use Case
When automating web interactions, being able to capture Network requests is essential. For example:
- Capturing... | closed | completed | true | 1 | [] | [] | 2026-03-10T17:00:48Z | 2026-04-27T04:53:02Z | 2026-04-24T22:12:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mayue018 | 34,374,355 | MDQ6VXNlcjM0Mzc0MzU1 | User | false |
openclaw/openclaw | 4,053,053,404 | I_kwDOQb6kR87xlK_c | 42,371 | https://github.com/openclaw/openclaw/issues/42371 | https://api.github.com/repos/openclaw/openclaw/issues/42371 | cron: auto-cleanup one-shot jobs after execution | ## Problem
One-shot cron jobs (scheduled with `at` instead of a recurring `cron` expression) are never cleaned up after execution. They remain in the cron list indefinitely with `idle` status, causing accumulation over time.
In practice, the agent's `followup-thread-*` jobs pile up quickly — I found 30+ stale one-sho... | closed | completed | true | 1 | [] | [] | 2026-03-10T17:01:48Z | 2026-04-27T04:53:01Z | 2026-04-24T22:12:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chenhab03 | 18,511,646 | MDQ6VXNlcjE4NTExNjQ2 | User | false |
openclaw/openclaw | 4,053,066,253 | I_kwDOQb6kR87xlOIN | 42,372 | https://github.com/openclaw/openclaw/issues/42372 | https://api.github.com/repos/openclaw/openclaw/issues/42372 | TTS [[audio_as_voice]] auto-delivery silently fails for WhatsApp group sessions | ## Bug Report
**Reported by:** Claude (OpenClaw main agent) — *yes, the bot itself is filing this bug*
### Summary
When an agent uses the `tts` tool inside a WhatsApp **group** session, the `[[audio_as_voice]]` marker is returned in the tool result but the audio is never delivered to the group. No error is thrown — ... | closed | completed | true | 1 | [] | [] | 2026-03-10T17:04:19Z | 2026-04-27T04:53:00Z | 2026-04-24T22:12:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | botsherkhan | 263,208,574 | U_kgDOD7A-fg | User | false |
openclaw/openclaw | 4,053,112,347 | I_kwDOQb6kR87xlZYb | 42,383 | https://github.com/openclaw/openclaw/issues/42383 | https://api.github.com/repos/openclaw/openclaw/issues/42383 | [Feature Request] Add Network domain support to Chrome Browser Relay | ## Description
Currently, the OpenClaw Chrome Browser Relay only has access to DOM and Console, but NOT Network requests. This makes it impossible to capture API calls (fetch/xhr) from web pages.
## Use Case
When automating web interactions, being able to capture Network requests is essential. For example:
- Capturing... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T17:13:25Z | 2026-04-27T04:52:59Z | 2026-04-24T22:12:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mayue018 | 34,374,355 | MDQ6VXNlcjM0Mzc0MzU1 | User | false |
openclaw/openclaw | 4,053,132,856 | I_kwDOQb6kR87xleY4 | 42,389 | https://github.com/openclaw/openclaw/issues/42389 | https://api.github.com/repos/openclaw/openclaw/issues/42389 | [Feature Request] Add create_bitable action to feishu_drive tool | ## Summary
Request to add `create_bitable` action to the `feishu_drive` tool to allow creating Feishu Bitable (多维表格) programmatically.
## Current Situation
The `feishu_drive` tool currently supports:
- `list` - List folder contents
- `info` - Get file info
- `create_folder` - Create folders
- `move` - Move files
- `de... | closed | completed | true | 1 | [] | [] | 2026-03-10T17:17:30Z | 2026-04-27T04:52:59Z | 2026-04-24T22:12:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | fosterhill | 46,843,205 | MDQ6VXNlcjQ2ODQzMjA1 | User | false |
openclaw/openclaw | 4,053,302,926 | I_kwDOQb6kR87xmH6O | 42,413 | https://github.com/openclaw/openclaw/issues/42413 | https://api.github.com/repos/openclaw/openclaw/issues/42413 | Feature Request: Auto-reload Gateway config when running as LaunchAgent | ## Description
When the gateway is running as a LaunchAgent (macOS service), config changes via `openclaw config set` are not automatically picked up. Users need to manually restart the gateway.
## Problem
Currently:
1. User runs `openclaw config set ...`
2. Config file is updated
3. Gateway (running as LaunchAgent... | closed | completed | true | 1 | [] | [] | 2026-03-10T17:49:51Z | 2026-04-27T04:52:58Z | 2026-04-24T22:12:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | shaozhengkun123 | 224,505,443 | U_kgDODWGuYw | User | false |
openclaw/openclaw | 4,053,355,973 | I_kwDOQb6kR87xmU3F | 42,421 | https://github.com/openclaw/openclaw/issues/42421 | https://api.github.com/repos/openclaw/openclaw/issues/42421 | [Bug]: openai-completions token usage logging broken for non-native endpoints (Kimi/Moonshot) since v2026.3.3 | ## Bug type
Regression (worked before, now fails)
## Summary
After upgrading to v2026.3.3+, sessions using non-native `openai-completions` endpoints (Kimi/Moonshot confirmed) log all-zero token usage. Token counts, cost, and usage tracking are entirely blind to usage incurred.
## Steps to reproduce
1. Configure a ... | closed | completed | true | 3 | [] | [] | 2026-03-10T18:01:11Z | 2026-04-27T04:52:57Z | 2026-04-24T22:12:46Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | leblanc-ryan | 91,286,009 | MDQ6VXNlcjkxMjg2MDA5 | User | false |
openclaw/openclaw | 4,053,364,742 | I_kwDOQb6kR87xmXAG | 42,423 | https://github.com/openclaw/openclaw/issues/42423 | https://api.github.com/repos/openclaw/openclaw/issues/42423 | Feature Request: Improved Model Fallback Strategy with Auto-Recovery | # Feature Request: Improved Model Fallback Strategy with Auto-Recovery
## Problem
Current behavior when API rate limit is hit (429 error):
1. Immediately downgrades to lower-tier model
2. Stays in downgraded state indefinitely
3. No automatic recovery to preferred model
**Example from 2026-03-11:**
- User's preferre... | closed | completed | true | 1 | [] | [] | 2026-03-10T18:03:02Z | 2026-04-27T04:52:56Z | 2026-04-24T22:12:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | randomuser2026x | 265,370,244 | U_kgDOD9E6hA | User | false |
openclaw/openclaw | 4,053,427,350 | I_kwDOQb6kR87xmmSW | 42,427 | https://github.com/openclaw/openclaw/issues/42427 | https://api.github.com/repos/openclaw/openclaw/issues/42427 | [Feature]: Request: Support multiple WeCom bot instances (multi-account) | ### Summary
## Is your feature request related to a problem? Please describe.
Currently, the plugin only supports one WeCom bot instance configured via `channels.wecom`.
Users who need multiple bots (e.g., for different departments or projects) cannot run them simultaneously.
## Describe the solution you'd like
Su... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-10T18:16:57Z | 2026-04-27T04:52:56Z | 2026-04-24T22:12:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Hex4c575a | 190,049,368 | U_kgDOC1PsWA | User | false |
openclaw/openclaw | 4,053,444,758 | I_kwDOQb6kR87xmqiW | 42,430 | https://github.com/openclaw/openclaw/issues/42430 | https://api.github.com/repos/openclaw/openclaw/issues/42430 | Duplicate voice note (sendVoice) on Telegram with ElevenLabs TTS | ## Description
Every voice note sent via Telegram is delivered twice (identical audio, same length). This happens consistently on every voice note response.
## Environment
- OpenClaw 2026.3.8 (3caab92)
- TTS: ElevenLabs, model: eleven_turbo_v2_5
- Channel: Telegram, streaming: block
- Primary model: Claude Sonnet 4
#... | closed | completed | true | 1 | [] | [] | 2026-03-10T18:19:59Z | 2026-04-27T04:52:55Z | 2026-04-24T22:12:54Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jeffordconsultancy | 212,752,615 | U_kgDODK5Y5w | User | false |
openclaw/openclaw | 4,053,449,468 | I_kwDOQb6kR87xmrr8 | 42,432 | https://github.com/openclaw/openclaw/issues/42432 | https://api.github.com/repos/openclaw/openclaw/issues/42432 | [Bug]: Anthropic 529 overload errors silently dropped on Telegram channel | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When Anthropic returns a 529 overloaded error, OpenClaw silently fails to respond on Telegram instead of surfacing the error to the user
### Steps to reproduce
1. Configure OpenClaw with Anthropic OAuth/subscription auth on the Telegram... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-10T18:20:59Z | 2026-04-27T04:52:54Z | 2026-04-24T22:12:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kylemwagner | 130,511,559 | U_kgDOB8dyxw | User | false |
openclaw/openclaw | 4,053,471,147 | I_kwDOQb6kR87xmw-r | 42,437 | https://github.com/openclaw/openclaw/issues/42437 | https://api.github.com/repos/openclaw/openclaw/issues/42437 | [Feature]: add configurable media.allowedRoots for message tool --media parameter | ### Summary
Currently, the message tool's --media parameter only accepts files within the workspace directory. Absolute paths (/...) and ~ paths are blocked for security. This prevents integration with external tools like Claude Code that may want to send files from other locations.
### Problem to solve
Use case:
... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-10T18:25:37Z | 2026-04-27T04:52:53Z | 2026-04-24T22:12:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | booghi-wq | 25,988,864 | MDQ6VXNlcjI1OTg4ODY0 | User | false |
openclaw/openclaw | 4,053,496,505 | I_kwDOQb6kR87xm3K5 | 42,439 | https://github.com/openclaw/openclaw/issues/42439 | https://api.github.com/repos/openclaw/openclaw/issues/42439 | Bug: Telegram voice messages corrupt thinking block signatures, preventing message processing | ## Summary
Telegram voice messages trigger thinking block signature corruption, causing Anthropic to reject the API call for that specific message. The session itself remains functional — sending a follow-up message via another platform (e.g. Discord) works fine. This suggests the corruption is isolated to the Telegra... | closed | not_planned | true | 3 | [] | [] | 2026-03-10T18:30:38Z | 2026-04-27T04:52:53Z | 2026-04-24T22:13:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | destructatron | 62,647,285 | MDQ6VXNlcjYyNjQ3Mjg1 | User | false |
openclaw/openclaw | 4,053,539,198 | I_kwDOQb6kR87xnBl- | 42,448 | https://github.com/openclaw/openclaw/issues/42448 | https://api.github.com/repos/openclaw/openclaw/issues/42448 | [Bug]: Feishu channel returns 404 due to missing receive_id_type when sending message | ### Bug type
Regression (worked before, now fails)
### Summary
The Feishu bot replies with `404 status code (no body)` when receiving messages in groups or private chats. Manual testing with the same credentials and a placeholder user `open_id` via `curl` succeeds, confirming that the Feishu app is correctly configu... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-10T18:39:26Z | 2026-04-27T04:52:51Z | 2026-04-24T22:13:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yoyo10261123 | 267,127,925 | U_kgDOD-wMdQ | User | false |
openclaw/openclaw | 4,053,545,726 | I_kwDOQb6kR87xnDL- | 42,449 | https://github.com/openclaw/openclaw/issues/42449 | https://api.github.com/repos/openclaw/openclaw/issues/42449 | [Bug]: using a skill makes chat enter an endless loop - tools error - webchat channel problem | ### Bug type
Regression (worked before, now fails)
### Summary
Trying to use a skill like ddg-web-search causes the web ui to enter an endless loop saying ist going to search but never executes the search. In the log there are errors "[tools] message failed: action send requieres a target"
### Steps to reproduce
*... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-10T18:40:37Z | 2026-04-27T04:52:50Z | 2026-04-24T22:13:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jctaborda | 17,262,048 | MDQ6VXNlcjE3MjYyMDQ4 | User | false |
openclaw/openclaw | 4,053,626,864 | I_kwDOQb6kR87xnW_w | 42,457 | https://github.com/openclaw/openclaw/issues/42457 | https://api.github.com/repos/openclaw/openclaw/issues/42457 | Feature request: task-triggered context injection (pre-task hooks) | When an agent has task-specific rules stored in memory (e.g. "always check X before doing Y"), there's no mechanism to automatically surface those rules when a matching task comes in. The agent has to remember to search memory first, which it frequently fails to do.
**Current behavior:** Rules get documented in daily ... | closed | completed | true | 1 | [] | [] | 2026-03-10T18:58:58Z | 2026-04-27T04:52:50Z | 2026-04-24T22:13:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | manos | 132,518 | MDQ6VXNlcjEzMjUxOA== | User | false |
openclaw/openclaw | 4,053,668,156 | I_kwDOQb6kR87xnhE8 | 42,459 | https://github.com/openclaw/openclaw/issues/42459 | https://api.github.com/repos/openclaw/openclaw/issues/42459 | mattermost: streaming final delivery may race with in-flight interval patch under high latency | **Context:** PR #33506 (block streaming / edit-in-place for Mattermost)
**Problem:**
The final-delivery path waits up to 2 seconds for `patchSending` to clear before issuing the final write:
```typescript
const patchWaitDeadline = Date.now() + 2000;
while (patchSending && Date.now() < patchWaitDeadline) {
await ne... | closed | completed | true | 1 | [] | [] | 2026-03-10T19:07:28Z | 2026-04-27T04:52:49Z | 2026-04-24T22:13:15Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | teconomix | 6,959,299 | MDQ6VXNlcjY5NTkyOTk= | User | false |
openclaw/openclaw | 4,053,738,462 | I_kwDOQb6kR87xnyPe | 42,464 | https://github.com/openclaw/openclaw/issues/42464 | https://api.github.com/repos/openclaw/openclaw/issues/42464 | [Bug]: Isolated cron agentTurn jobs hang until timeout — providers reachable, interactive sessions unaffected | ## Bug type
Regression / Infrastructure
## OpenClaw version
2026.3.8 (3caab92)
## Summary
Isolated `agentTurn` cron jobs consistently hang until their configured timeout, regardless of provider or model. Direct API calls to the same providers from the same machine succeed in ~2 seconds. Interactive sessions (Teleg... | closed | completed | true | 3 | [] | [] | 2026-03-10T19:21:39Z | 2026-04-27T04:52:47Z | 2026-04-24T22:13:18Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | frankbuild | 266,174,429 | U_kgDOD91_3Q | User | false |
openclaw/openclaw | 4,053,771,272 | I_kwDOQb6kR87xn6QI | 42,466 | https://github.com/openclaw/openclaw/issues/42466 | https://api.github.com/repos/openclaw/openclaw/issues/42466 | [Feature]: | ### Summary
Study app Ai
### Problem to solve
Study app Ai
### Proposed solution
Ai generated
### Alternatives considered
###
### Impact
###
### Evidence/examples
_No response_
### Additional information
_No response_ | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-10T19:27:50Z | 2026-04-27T04:52:47Z | 2026-04-24T22:13:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | veshwajeets-cell | 267,142,582 | U_kgDOD-xFtg | User | false |
openclaw/openclaw | 4,053,776,493 | I_kwDOQb6kR87xn7ht | 42,467 | https://github.com/openclaw/openclaw/issues/42467 | https://api.github.com/repos/openclaw/openclaw/issues/42467 | [Feature]: | ### Summary
[](url)
### Problem to solve
[](url)
### Proposed solution
### Duplicate of #42452
### Alternatives considered
_No response_
### Impact
****[](url)
### Evidence/examples
_No response_
### Additional information
_No response_ | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-10T19:28:43Z | 2026-04-27T04:52:46Z | 2026-04-24T22:13:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | veshwajeets-cell | 267,142,582 | U_kgDOD-xFtg | User | false |
openclaw/openclaw | 4,053,788,333 | I_kwDOQb6kR87xn-at | 42,468 | https://github.com/openclaw/openclaw/issues/42468 | https://api.github.com/repos/openclaw/openclaw/issues/42468 | Socket Mode: message.mpim events not delivered for MPDMs where bot is a member | ## Description
Bot is a member of a multi-party DM (MPDM) on Slack, confirmed via `conversations.info` (`is_member: true`, `is_mpim: true`). The bot has the correct OAuth scopes (`mpim:history`, `mpim:read`) and event subscriptions (`message.mpim`). Reading messages via API (`conversations.history`) works fine.
Howev... | closed | completed | true | 1 | [] | [] | 2026-03-10T19:30:55Z | 2026-04-27T04:52:45Z | 2026-04-24T22:13:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | gandalf-at-lerian | 264,490,208 | U_kgDOD8PM4A | User | false |
openclaw/openclaw | 4,053,802,642 | I_kwDOQb6kR87xoB6S | 42,471 | https://github.com/openclaw/openclaw/issues/42471 | https://api.github.com/repos/openclaw/openclaw/issues/42471 | /tools/invoke sessions_spawn can inherit workspaceDir="/" and bootstrap /AGENTS.md | Describe the bug
When sessions_spawn is invoked through HTTP POST /tools/invoke, spawned subagents can intermittently fail during bootstrap with:
Error: EROFS: read-only file system, open '/AGENTS.md'
Example gateway log:
gateway/ws ⇄ res ✗ agent errorCode=UNAVAILABLE errorMessage=Error: EROFS: read-only file system,... | closed | completed | true | 4 | [] | [] | 2026-03-10T19:33:43Z | 2026-04-27T04:52:45Z | 2026-04-24T22:13:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Mwhite84 | 5,177,066 | MDQ6VXNlcjUxNzcwNjY= | User | false |
openclaw/openclaw | 4,053,913,229 | I_kwDOQb6kR87xoc6N | 42,477 | https://github.com/openclaw/openclaw/issues/42477 | https://api.github.com/repos/openclaw/openclaw/issues/42477 | DM sessions not routed to per-account agent pools (all DMs land in main agent) | ## Problem
When a user DMs a specific bot account (e.g. `forrest` bot), the session is created under `agent:main:main` instead of the account-specific agent pool (`agent:forrest:dm:...`).
## Expected behavior
A DM to the `forrest` Slack bot account should create a session in the `forrest` agent pool (using the forre... | closed | completed | true | 1 | [] | [] | 2026-03-10T19:54:54Z | 2026-04-27T04:52:44Z | 2026-04-24T22:13:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | iret77 | 63,622,643 | MDQ6VXNlcjYzNjIyNjQz | User | false |
openclaw/openclaw | 4,054,008,738 | I_kwDOQb6kR87xo0Oi | 42,486 | https://github.com/openclaw/openclaw/issues/42486 | https://api.github.com/repos/openclaw/openclaw/issues/42486 | Cross-session wake: inbound message to session A causes stalled session B to resume immediately | ## Summary
When an inbound message arrives for session A (e.g. #general), a session B that appeared idle/stalled immediately resumes producing output. This has been reproduced consistently (8–10+ times) with two concurrent Discord sessions on the same gateway.
## Symptoms
- Session B (long-running task: infra thread... | closed | completed | true | 1 | [] | [] | 2026-03-10T20:16:09Z | 2026-04-27T04:52:43Z | 2026-04-24T22:13:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mattgodbolt-molty | 260,133,431 | U_kgDOD4FSNw | User | false |
openclaw/openclaw | 4,054,046,598 | I_kwDOQb6kR87xo9eG | 42,491 | https://github.com/openclaw/openclaw/issues/42491 | https://api.github.com/repos/openclaw/openclaw/issues/42491 | Workaround: Enable gemini-3.1-pro-preview-customtools support via custom provider | ## Description
Currently, OpenClaw natively supports `google/gemini-3.1-pro-preview` and `google/gemini-3-flash-preview`. However, for those looking to utilize the `gemini-3.1-pro-preview-customtools` endpoint (which provides enhanced tool and function-calling capabilities on Google AI Studio), it throws an "Unknown m... | closed | not_planned | true | 2 | [] | [] | 2026-03-10T20:25:04Z | 2026-04-27T04:52:43Z | 2026-04-24T22:13:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Skeptomenos | 22,348,986 | MDQ6VXNlcjIyMzQ4OTg2 | User | false |
openclaw/openclaw | 4,054,097,873 | I_kwDOQb6kR87xpJ_R | 42,495 | https://github.com/openclaw/openclaw/issues/42495 | https://api.github.com/repos/openclaw/openclaw/issues/42495 | Heartbeat polls overwrite session displayName, breaking control UI session selector | ## Summary
Heartbeat polls overwrite the session's `displayName` to `"heartbeat"`, causing the control UI session dropdown to show generic "heartbeat" labels instead of the richer context labels set by user messages (e.g., `"My App — Project: Dashboard"`).
## Steps to Reproduce
1. Configure an agent with heartbeat e... | closed | completed | true | 2 | [] | [] | 2026-03-10T20:36:09Z | 2026-04-27T04:52:42Z | 2026-04-24T22:13:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | captivationhub | 224,922,961 | U_kgDODWgNUQ | User | false |
openclaw/openclaw | 4,054,316,949 | I_kwDOQb6kR87xp_eV | 42,508 | https://github.com/openclaw/openclaw/issues/42508 | https://api.github.com/repos/openclaw/openclaw/issues/42508 | Subagent skill resolution: ~ resolves to workspace root instead of home directory | ## Bug
When a subagent reads a skill file, the tilde `~` in the skill path is resolved relative to the **workspace root** instead of the user's home directory.
**Example:**
- Skill actual path: `/home/sisu/.openclaw/skills/agent-browser/SKILL.md`
- System prompt lists skill location as: `~/.openclaw/skills/agent-brow... | closed | completed | true | 2 | [] | [] | 2026-03-10T21:27:19Z | 2026-04-27T04:52:41Z | 2026-04-24T22:13:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sisutuulenisa | 261,256,991 | U_kgDOD5J3Hw | User | false |
openclaw/openclaw | 4,054,436,827 | I_kwDOQb6kR87xqcvb | 42,529 | https://github.com/openclaw/openclaw/issues/42529 | https://api.github.com/repos/openclaw/openclaw/issues/42529 | Feature: Cron pre-flight scripts (gate enforcement before agent turn) | ## Problem
AI agents can rationalize past any instruction in a cron prompt. When a cron says "run pr-gate.sh and if it fails, fix the PRs instead of building," the LLM can skip the script, misread the output, or reason its way around the gate. **The monitor and the monitored are the same system.**
Real example: My Op... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T21:58:27Z | 2026-04-27T04:52:40Z | 2026-04-24T22:13:46Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | krissythesurfer | 266,366,874 | U_kgDOD-Bvmg | User | false |
openclaw/openclaw | 4,054,610,898 | I_kwDOQb6kR87xrHPS | 42,549 | https://github.com/openclaw/openclaw/issues/42549 | https://api.github.com/repos/openclaw/openclaw/issues/42549 | Telegram delivery fails with group: prefix in recipient ID | ## Bug
Cron job delivery (announce mode) to Telegram groups fails with:
```
Telegram recipient must be a numeric chat ID
```
**Root cause**: The delivery system formats the group target as `group:-1003786508776` instead of `-1003786508776`. The `group:` prefix is not a valid Telegram chat ID.
**Impact**: Messages s... | closed | completed | true | 1 | [] | [] | 2026-03-10T22:42:45Z | 2026-04-27T04:52:39Z | 2026-04-24T22:13:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mojoho81 | 222,377,164 | U_kgDODUE0zA | User | false |
openclaw/openclaw | 4,054,622,330 | I_kwDOQb6kR87xrKB6 | 42,550 | https://github.com/openclaw/openclaw/issues/42550 | https://api.github.com/repos/openclaw/openclaw/issues/42550 | Bug: TUI displays <relevant-memories> block in assistant responses | ## Summary
The openclaw-tui client shows the `<relevant-memories>` context block as part of assistant responses. Users see the raw injected memory context instead of just the reply text.
## Steps to Reproduce
1. Use openclaw-tui with Mem0 autoRecall enabled
2. Send a message that triggers memory recall
3. Observe that... | closed | completed | true | 1 | [] | [] | 2026-03-10T22:46:07Z | 2026-04-27T04:52:38Z | 2026-04-24T22:13:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | firstmovr | 155,568,852 | U_kgDOCUXK1A | User | false |
openclaw/openclaw | 4,054,668,321 | I_kwDOQb6kR87xrVQh | 42,556 | https://github.com/openclaw/openclaw/issues/42556 | https://api.github.com/repos/openclaw/openclaw/issues/42556 | Plugin install fails on Windows with spawn EINVAL (shouldSpawnWithShell always returns false) | ## Bug: Plugin dependency install fails on Windows with `spawn EINVAL`
### Summary
When installing any plugin that has npm dependencies on Windows, OpenClaw fails to run `npm install` in the plugin directory. The error stack is:
```
Error: spawn EINVAL
at ChildProcess.spawn (node:internal/child_process:421:11)
... | closed | completed | true | 3 | [] | [] | 2026-03-10T23:00:21Z | 2026-04-27T04:52:37Z | 2026-04-24T22:13:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | satvoop | 141,410,440 | U_kgDOCG3AiA | User | false |
openclaw/openclaw | 4,054,684,514 | I_kwDOQb6kR87xrZNi | 42,557 | https://github.com/openclaw/openclaw/issues/42557 | https://api.github.com/repos/openclaw/openclaw/issues/42557 | exec host=node: Gateway-side path validation breaks cross-platform (WSL->Windows) workflows | ### The Problem
When running `exec` from a WSL2 gateway to a Windows node, the gateway attempts to `stat()` the `workdir` path using its own (Linux) filesystem before sending the command to the node.
If the path is not perfectly mirrored or is an absolute Linux path (e.g., `/mnt/c/...`), the gateway fails to resolve i... | closed | completed | true | 2 | [] | [] | 2026-03-10T23:05:28Z | 2026-04-27T04:52:37Z | 2026-04-24T22:13:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Armel081 | 28,829,361 | MDQ6VXNlcjI4ODI5MzYx | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.