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 | 4,345,351,052 | I_kwDOQb6kR88AAAABAwDLjA | 73,728 | https://github.com/openclaw/openclaw/issues/73728 | https://api.github.com/repos/openclaw/openclaw/issues/73728 | Default DEFAULT_PLUGIN_DISCOVERY_CACHE_MS / DEFAULT_PLUGIN_MANIFEST_CACHE_MS of 1 second is too short for gateway use | ## Environment
- openclaw \2026.4.26\
- File: \src/plugins/cache-controls.ts\
## Summary
Both \DEFAULT_PLUGIN_DISCOVERY_CACHE_MS\ and \DEFAULT_PLUGIN_MANIFEST_CACHE_MS\ default to \1e3\ (1 second). This makes the \capabilityProviderPluginIdCache\ and related snapshot caches effectively useless for any conversation-pac... | closed | completed | true | 2 | [] | [] | 2026-04-28T18:25:24Z | 2026-05-03T04:36:34Z | 2026-04-30T14:58:35Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | sg1416-zg | 259,938,361 | U_kgDOD35YOQ | User | false |
openclaw/openclaw | 4,345,357,678 | I_kwDOQb6kR88AAAABAwDlbg | 73,729 | https://github.com/openclaw/openclaw/issues/73729 | https://api.github.com/repos/openclaw/openclaw/issues/73729 | resolvePluginCapabilityProviders triggers redundant full plugin loads per message (image/video/music generation) | ## Environment
- openclaw `2026.4.26`
- Windows 11, Node.js (bundled)
- feishu channel plugin, ~7 user plugins + bundled providers
## Summary
Every message dispatch triggers three independent full-registry loads (image / video / music generation providers), each re-registering all ~119 plugins. Together they cost **~1... | closed | completed | true | 1 | [] | [] | 2026-04-28T18:26:52Z | 2026-05-02T04:29:03Z | 2026-04-29T06:52:34Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | sg1416-zg | 259,938,361 | U_kgDOD35YOQ | User | false |
openclaw/openclaw | 4,338,530,888 | I_kwDOQb6kR88AAAABApi6SA | 73,013 | https://github.com/openclaw/openclaw/issues/73013 | https://api.github.com/repos/openclaw/openclaw/issues/73013 | [Bug]: NVIDIA provider plugin omits apiKey field, silently breaks pi-coding-agent's models.json validation | ## Symptom
Any custom provider entry in `~/.openclaw/agents/<id>/agent/models.json`
without an `apiKey` field causes pi-coding-agent's `validateConfig()` to
throw, which **silently drops the entire `models.json`**. The runtime
falls through to built-in models only — every other custom provider
(codex, xai, github-copi... | closed | completed | true | 1 | [] | [
"vincentkoc"
] | 2026-04-27T20:00:37Z | 2026-05-01T04:42:19Z | 2026-04-28T18:30:59Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | bautrey | 5,750,806 | MDQ6VXNlcjU3NTA4MDY= | User | false |
openclaw/openclaw | 4,098,604,940 | I_kwDOQb6kR870S7-M | 50,107 | https://github.com/openclaw/openclaw/issues/50107 | https://api.github.com/repos/openclaw/openclaw/issues/50107 | [Bug]: NVIDIA Provider Sends Anthropic-Style Message Format to OpenAI-Compatible API | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When spawning a subagent with `sessions_spawn` using an NVIDIA provider configured as `"api": "openai-completions"`, OpenClaw sends Anthropic-style message content (arrays with `type` objects) instead of OpenAI-style simple strings. This ca... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [
"vincentkoc"
] | 2026-03-19T01:13:45Z | 2026-05-01T04:42:20Z | 2026-04-28T18:30:59Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | ifearghal | 249,224,915 | U_kgDODtre0w | User | false |
openclaw/openclaw | 4,338,532,925 | I_kwDOQb6kR88AAAABApjCPQ | 73,014 | https://github.com/openclaw/openclaw/issues/73014 | https://api.github.com/repos/openclaw/openclaw/issues/73014 | [Bug]: parseStaticModelRef strips provider prefix, breaks NVIDIA NIM (and OpenRouter-style ids) on direct invocation | ## Symptom
```
openclaw infer model run --model nvidia/nemotron-3-super-120b-a12b ...
```
fails with `404 page not found` from NVIDIA NIM.
NIM requires the **full** `nvidia/<model>` form in the request body,
but `parseStaticModelRef` (line 28-52 of `model-ref-shared-DcLkozLZ.js`)
splits on the first `/` and yields:
... | closed | completed | true | 1 | [] | [
"vincentkoc"
] | 2026-04-27T20:00:53Z | 2026-05-01T04:42:18Z | 2026-04-28T18:32:35Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | bautrey | 5,750,806 | MDQ6VXNlcjU3NTA4MDY= | User | false |
openclaw/openclaw | 4,345,363,372 | I_kwDOQb6kR88AAAABAwD7rA | 73,731 | https://github.com/openclaw/openclaw/issues/73731 | https://api.github.com/repos/openclaw/openclaw/issues/73731 | resolveBuiltInModelSuppression runs the same 48-plugin load twice back-to-back via independent caches | ## Environment
- openclaw `2026.4.26`
## Summary
During a single message dispatch, the 48-bundled-provider load for "built-in model suppression" runs **twice**, back-to-back (~125s + ~125s = ~250s wasted) when both catalog-hook and compat-hook paths are triggered. The two paths maintain independent caches and neither ... | closed | completed | true | 1 | [] | [] | 2026-04-28T18:28:05Z | 2026-05-01T04:42:17Z | 2026-04-28T18:40:17Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | sg1416-zg | 259,938,361 | U_kgDOD35YOQ | User | false |
openclaw/openclaw | 4,044,566,878 | I_kwDOQb6kR87xEzFe | 40,838 | https://github.com/openclaw/openclaw/issues/40838 | https://api.github.com/repos/openclaw/openclaw/issues/40838 | [Feature]: requirement: Expand Telegram message field information for the agent. | ### Summary
Key fields of Telegram messages (is_bot, msg.from, etc.) are exported to the Agent's context, enabling the Agent to automatically identify message sources and types, and implement message classification and differentiated processing. By enhancing contextual information, hard-coded conditional judgments are... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-09T09:53:04Z | 2026-05-04T09:50:32Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ohyeah521 | 1,449,374 | MDQ6VXNlcjE0NDkzNzQ= | User | false |
openclaw/openclaw | 4,046,847,741 | I_kwDOQb6kR87xNf79 | 41,294 | https://github.com/openclaw/openclaw/issues/41294 | https://api.github.com/repos/openclaw/openclaw/issues/41294 | [Feature]: Interactive fallback confirmation mod | ### Summary
## Feature Request
### Problem
When the primary model is unavailable and OpenClaw falls back to another model,
this happens automatically without user confirmation. In some cases, users may
want to be asked before switching to a fallback model – for example to control
costs or to decide case-by-case w... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-09T17:15:35Z | 2026-05-03T04:36:07Z | 2026-04-30T16:16:16Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | alfuboo | 109,141,047 | U_kgDOBoFcNw | User | false |
openclaw/openclaw | 4,047,081,757 | I_kwDOQb6kR87xOZEd | 41,330 | https://github.com/openclaw/openclaw/issues/41330 | https://api.github.com/repos/openclaw/openclaw/issues/41330 | iMessage channel duplicate message loop - messages sent by agent are re-ingested as inbound | **Issue Description:**
When using the iMessage channel with `imsg` provider, there is a critical bug where messages sent by the agent (replies) are being read back from the Messages database and treated as new inbound messages. This creates an infinite loop where each user message triggers two database entries: one `is... | open | null | false | 10 | [] | [
"vincentkoc"
] | 2026-03-09T18:02:12Z | 2026-05-04T09:14:32Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ajkwok322 | 188,639,818 | U_kgDOCz5qSg | User | false |
openclaw/openclaw | 4,048,063,455 | I_kwDOQb6kR87xSIvf | 41,459 | https://github.com/openclaw/openclaw/issues/41459 | https://api.github.com/repos/openclaw/openclaw/issues/41459 | [Feature]: Add per-channel identity override for Discord bindings | ### Summary
Add optional nickname field to binding config for Discord so agents can display different names per channel when using one bot token.
### Problem to solve
When using a single Discord bot token with multiple agents bound to different channels (via bindings), users can't display different bot nicknames pe... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-09T21:44:08Z | 2026-05-04T10:35:44Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | MaxBumbaclot | 264,505,132 | U_kgDOD8QHLA | User | false |
openclaw/openclaw | 4,048,750,049 | I_kwDOQb6kR87xUwXh | 41,579 | https://github.com/openclaw/openclaw/issues/41579 | https://api.github.com/repos/openclaw/openclaw/issues/41579 | [Feature]: Add File Gallery button in Mission Control left nav | ### Summary
Add a permanent “File Gallery” nav item in Mission Control so users can view and open workspace documents/images from one place.
### Problem to solve
Current behaviour requires users to generate and open ALL_FILES_GALLERY.html separately in the browser/terminal. This breaks flow, adds friction, and is ha... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-03-10T01:08:47Z | 2026-05-04T10:36:09Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | HMpsychEd | 266,874,550 | U_kgDOD-gutg | User | false |
openclaw/openclaw | 4,048,992,594 | I_kwDOQb6kR87xVrlS | 41,645 | https://github.com/openclaw/openclaw/issues/41645 | https://api.github.com/repos/openclaw/openclaw/issues/41645 | [Feature]: [Feature Request] 按用户维度统计 Token 使用量 | ### Summary
## 背景
目前 OpenClaw 没有按用户统计 Token 使用量的功能，无法满足多用户场景下的计费/配额管理需求。
## 期望功能
1. 按用户统计 input/output/total tokens 使用量
2. 支持按天/周/月统计
3. 可通过 /usage 或 Dashboard 查看
## 建议实现
在每次模型调用后，日志中增加 token 使用量字段：
```json
{
"user_id": "ou_xxx",
"session_id": "xxx",
"model": "MiniMax-M2.5",
"input_tokens": 1234,
"output_t... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-10T02:34:19Z | 2026-05-04T11:26:23Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | rogeryangyc | 39,161,553 | MDQ6VXNlcjM5MTYxNTUz | User | false |
openclaw/openclaw | 4,049,568,563 | I_kwDOQb6kR87xX4Mz | 41,805 | https://github.com/openclaw/openclaw/issues/41805 | https://api.github.com/repos/openclaw/openclaw/issues/41805 | [Feature]: Include interaction metadata in Discord button callbacks | ### Summary
When a Discord button is clicked, the agent callback receives no metadata about the interaction — no message ID, no button identifier, no user ID. This forces developers to embed identifiers directly in button labels as a workaround.
### Problem to solve
Button callback arrives as a plain message with ju... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-10T05:48:54Z | 2026-05-04T10:36:58Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | irman | 6,107,943 | MDQ6VXNlcjYxMDc5NDM= | User | false |
openclaw/openclaw | 4,052,320,011 | I_kwDOQb6kR87xiX8L | 42,246 | https://github.com/openclaw/openclaw/issues/42246 | https://api.github.com/repos/openclaw/openclaw/issues/42246 | [Feature]: Configurable batching/aggregation of outbound Telegram notifications | ### Summary
Add a configurable batching layer for outbound OpenClaw notifications sent to Telegram, so that many short updates are aggregated into fewer, more readable messages instead of flooding the channel.
### Problem to solve
When OpenClaw performs background tasks — running routines, executing skills, processi... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-10T14:48:54Z | 2026-05-04T11:26:50Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | fireniko | 211,585,889 | U_kgDODJyLYQ | User | false |
openclaw/openclaw | 4,052,509,116 | I_kwDOQb6kR87xjGG8 | 42,276 | https://github.com/openclaw/openclaw/issues/42276 | https://api.github.com/repos/openclaw/openclaw/issues/42276 | Reasoning stream | ### Summary
Is it possible to have openclaw write and overwrite lines like openai and grok do to indicate the thinking process? Normally, messages appear like "searching google, scanning code, installing program ..."
There is /reason stream but openclaw is telling me that it cannot overwrite lines (i am using opencl... | open | null | false | 6 | [
"enhancement"
] | [] | 2026-03-10T15:19:16Z | 2026-05-04T11:26:56Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | cryptopsy0 | 32,229,149 | MDQ6VXNlcjMyMjI5MTQ5 | User | false |
openclaw/openclaw | 4,052,820,510 | I_kwDOQb6kR87xkSIe | 42,332 | https://github.com/openclaw/openclaw/issues/42332 | https://api.github.com/repos/openclaw/openclaw/issues/42332 | [Feature]: Support sessionKey-based concurrency guard for webhook deliveries | ### Summary
Add sessionKey-based concurrency guard for webhook deliveries
### Problem to solve
When a webhook source delivers messages on a recurring schedule (e.g., polling every 2 min), there's no way to know if a previous webhook turn for the same sessionKey is still running. This causes duplicate LLM turns pro... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-10T16:17:12Z | 2026-05-04T11:27:08Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ethanbeard | 60,940,730 | MDQ6VXNlcjYwOTQwNzMw | User | false |
openclaw/openclaw | 4,053,185,922 | I_kwDOQb6kR87xlrWC | 42,397 | https://github.com/openclaw/openclaw/issues/42397 | https://api.github.com/repos/openclaw/openclaw/issues/42397 | [Feature]: Per-task model override in HEARTBEAT.md | ### Summary
Currently, the heartbeat.model setting in openclaw.json applies a single model to the entire heartbeat run. There is no way to assign a different (e.g. more capable) model to specific tasks defined in HEARTBEAT.md.
### Problem to solve
Use case:
I run a cost-optimized setup where routine heartbeat checks... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-10T17:27:20Z | 2026-05-04T11:27:20Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Cologran | 72,316,053 | MDQ6VXNlcjcyMzE2MDUz | User | false |
openclaw/openclaw | 4,053,492,068 | I_kwDOQb6kR87xm2Fk | 42,438 | https://github.com/openclaw/openclaw/issues/42438 | https://api.github.com/repos/openclaw/openclaw/issues/42438 | [Feature]: Silent/Notify DM Policy Option for allowlist mode | ### Summary
## Use Case
Personal assistant users who want to know if strangers try to reach them, without the stranger knowing the bot is active. This is a common privacy/security request for personal AI assistants.
### Problem to solve
## Problem
Currently, `dmPolicy: "pairing"` sends an "access not configured" m... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-10T18:29:36Z | 2026-05-03T04:35:44Z | 2026-04-30T17:59:12Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | djktopdj | 35,653,223 | MDQ6VXNlcjM1NjUzMjIz | User | false |
openclaw/openclaw | 4,054,028,694 | I_kwDOQb6kR87xo5GW | 42,488 | https://github.com/openclaw/openclaw/issues/42488 | https://api.github.com/repos/openclaw/openclaw/issues/42488 | [Feature]: Add old_text/new_text parameter aliases for edit tool | ### Summary
```
Add old_text and new_text as accepted parameter aliases for the edit tool, alongside existing oldText/old_string and newText/new_string.
```
### Problem to solve
```
The edit tool accepts oldText/old_string and newText/new_string. Some LLMs (e.g. qwen3.5:35b, other models that output snake_case) send... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-10T20:21:07Z | 2026-05-04T11:27:35Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ipitchkhadze | 5,347,534 | MDQ6VXNlcjUzNDc1MzQ= | User | false |
openclaw/openclaw | 4,345,417,890 | I_kwDOQb6kR88AAAABAwHQog | 73,745 | https://github.com/openclaw/openclaw/issues/73745 | https://api.github.com/repos/openclaw/openclaw/issues/73745 | [Bug]: Discord /reset sends empty prompt on 2026.4.26 | Discord native /reset appears to send an empty model prompt on OpenClaw 2026.4.26 (be8c246), causing OpenAI Responses API to reject the request with: One of input or previous_response_id or prompt or conversation_id must be provided. Sometimes Discord also shows The application did not respond.
Repro: run 2026.4.26, u... | closed | completed | true | 1 | [] | [] | 2026-04-28T18:38:34Z | 2026-05-01T04:42:17Z | 2026-04-28T18:43:15Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | iannwu | 5,016,913 | MDQ6VXNlcjUwMTY5MTM= | User | false |
openclaw/openclaw | 4,055,210,023 | I_kwDOQb6kR87xtZgn | 42,665 | https://github.com/openclaw/openclaw/issues/42665 | https://api.github.com/repos/openclaw/openclaw/issues/42665 | [Feature]: memory-lancedb: No way to obtain full UUID for memory deletion - all ID fields truncated to 8 chars | ### Summary
The memory_recall tool provides no way to obtain the full memory UUID required by memory_forget. All ID fields are truncated to 8 characters, making it impossible for agents to delete specific memories using the memoryId parameter.
### Problem to solve
The Problem:
- memory_recall returns only 8-charac... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-11T01:59:45Z | 2026-05-02T04:26:26Z | 2026-04-30T03:04:20Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | gnokit | 155,364 | MDQ6VXNlcjE1NTM2NA== | User | false |
openclaw/openclaw | 4,055,486,566 | I_kwDOQb6kR87xudBm | 42,747 | https://github.com/openclaw/openclaw/issues/42747 | https://api.github.com/repos/openclaw/openclaw/issues/42747 | Native Slack token rotation support (oauth.v2.access refresh flow) | ### Summary
Add native support for Slack's token rotation feature, enabling OpenClaw to automatically refresh expiring bot tokens using the oauth.v2.access refresh_token grant without manual intervention or external tooling.
### Problem to solve
OpenClaw stores the Slack bot token as a static string in openclaw.json... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-11T03:28:20Z | 2026-05-04T00:42:47Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | ewood70 | 26,074,458 | MDQ6VXNlcjI2MDc0NDU4 | User | false |
openclaw/openclaw | 4,055,521,222 | I_kwDOQb6kR87xulfG | 42,754 | https://github.com/openclaw/openclaw/issues/42754 | https://api.github.com/repos/openclaw/openclaw/issues/42754 | [Feature]: Feishu card action handler discards action.option, action.options, and action.form_value | ### Summary
Extract `action.option`, `action.options`, and `action.form_value` from Feishu card action callbacks in `handleFeishuCardAction`, so plugins can receive dropdown selections, multi-select choices, and form submissions.
### Problem to solve
`handleFeishuCardAction` in `extensions/feishu/src/card-action.ts`... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-11T03:40:03Z | 2026-05-04T11:32:24Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | littleWalikrence | 41,196,524 | MDQ6VXNlcjQxMTk2NTI0 | User | false |
openclaw/openclaw | 4,055,843,806 | I_kwDOQb6kR87xv0Pe | 42,816 | https://github.com/openclaw/openclaw/issues/42816 | https://api.github.com/repos/openclaw/openclaw/issues/42816 | [Feature]: Implement task state persistence and proactive result reporting across restarts | ### Summary
Feature Request: State persistence and proactive task reporting after system/service restarts
### Problem to solve
When instructing OpenClaw to perform configuration tasks that require a system reboot or restarting the OpenClaw service itself (e.g., updating core components or modifying kernel parameters... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-11T05:03:35Z | 2026-05-04T13:18:28Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | gmvp3 | 27,907,782 | MDQ6VXNlcjI3OTA3Nzgy | User | false |
openclaw/openclaw | 4,211,590,542 | I_kwDOQb6kR877B8WO | 61,859 | https://github.com/openclaw/openclaw/issues/61859 | https://api.github.com/repos/openclaw/openclaw/issues/61859 | [Bug] 飞书图片与消息错配 - 图片路由错误 | ## Bug 描述
飞书消息中的图片附件与文字消息出现错配，导致用户收到的图片与发送内容不匹配。
## 复现环境
- OpenClaw 版本：最新
- 操作系统：macOS
- 消息渠道：飞书
## 复现时间
2026-04-06 20:03 (GMT+8)
## 具体现象
1. 用户通过飞书发送消息："这个格力面板如何关闭长亮" 并附上格力空调控制面板的照片
2. OpenClaw 接收到的消息：
- 文字消息：正确显示 "格力面板如何关闭长亮"
- 附件图片：显示的却是一只边境牧羊犬（白毛棕斑）躺在地毯上的照片
3. 用户确认从未发送过狗狗照片，该图片来源不明
## 影响评估
- **隐私风险**：... | closed | completed | true | 1 | [] | [] | 2026-04-06T12:22:01Z | 2026-05-01T04:42:15Z | 2026-04-28T18:45:30Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | WwNeXst | 268,033,705 | U_kgDOD_neqQ | User | false |
openclaw/openclaw | 4,345,384,685 | I_kwDOQb6kR88AAAABAwFO7Q | 73,736 | https://github.com/openclaw/openclaw/issues/73736 | https://api.github.com/repos/openclaw/openclaw/issues/73736 | [Bug]: 2026.4.26 ships incomplete typebox bundles in extensions/xai and extensions/browser, breaking plugin load | ## Summary
The `ghcr.io/openclaw/openclaw:2026.4.26` image ships **incomplete copies of `typebox` 1.1.28** in `extensions/xai/node_modules/typebox/` and `extensions/browser/node_modules/typebox/`. Multiple subdirectories are missing files that the bundled `index.mjs` barrel-re-exports try to require:
- `system/memory... | closed | completed | true | 1 | [] | [] | 2026-04-28T18:32:12Z | 2026-05-03T04:36:12Z | 2026-04-30T15:55:57Z | CONTRIBUTOR | null | 20260503T060030Z | 2026-05-03T06:00:30Z | perlowja | 162,375,972 | U_kgDOCa2pJA | User | false |
openclaw/openclaw | 4,345,321,184 | I_kwDOQb6kR88AAAABAwBW4A | 73,723 | https://github.com/openclaw/openclaw/issues/73723 | https://api.github.com/repos/openclaw/openclaw/issues/73723 | Native hook relay remains unavailable after gateway restart and hooks report 5/5 ready | ### Summary
On OpenClaw `2026.4.26`, Codex-native tool execution can get stuck behind the native hook relay even when:
- `commands.native = true`
- `commands.nativeSkills = true`
- all internal hooks are enabled
- the gateway has been restarted
- the hook loader reports `loaded 5 internal hook handlers`
- `openclaw h... | open | null | false | 4 | [] | [] | 2026-04-28T18:19:34Z | 2026-05-03T12:31:54Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | kaimo455 | 23,258,141 | MDQ6VXNlcjIzMjU4MTQx | User | false |
openclaw/openclaw | 4,345,446,087 | I_kwDOQb6kR88AAAABAwI-xw | 73,748 | https://github.com/openclaw/openclaw/issues/73748 | https://api.github.com/repos/openclaw/openclaw/issues/73748 | MemOS stopped capturing after 4.26 upgrade - fix: hooks.allowConversationAccess | ## Symptom · 症状
After upgrading to OpenClaw 4.26, `memos-local-openclaw-plugin` silently stopped indexing new conversations. No errors, no warnings.
升级 OpenClaw 4.26 后，memos-local 插件静默停止收录新对话。无报错、无警告。
## Root Cause · 根因
OpenClaw 4.26 added a security policy (see `docs/plugins/hooks.md` line ~200):
4.26 新增了插件安全策略：
... | closed | completed | true | 2 | [] | [] | 2026-04-28T18:44:02Z | 2026-05-01T04:39:07Z | 2026-04-28T18:48:14Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | 17329971 | 83,849,762 | MDQ6VXNlcjgzODQ5NzYy | User | false |
openclaw/openclaw | 4,345,475,468 | I_kwDOQb6kR88AAAABAwKxjA | 73,752 | https://github.com/openclaw/openclaw/issues/73752 | https://api.github.com/repos/openclaw/openclaw/issues/73752 | [Bug]: Windows 2026.4.26 gateway/WebChat/TUI handshake timeouts + stuck agent:main:main queue | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
OpenClaw 2026.4.26 on Windows 11 / Node 24.14.1.
Symptoms:
- Gateway starts and reaches `[gateway] ready`.
- Control UI can connect and load sessions/models/config.
- Sending a simple WebChat message queues forever.
- `open... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-28T18:49:59Z | 2026-05-01T04:42:13Z | 2026-04-28T18:54:39Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | hcdc070165 | 280,114,626 | U_kgDOELI1wg | User | false |
openclaw/openclaw | 4,345,556,370 | I_kwDOQb6kR88AAAABAwPtkg | 73,756 | https://github.com/openclaw/openclaw/issues/73756 | https://api.github.com/repos/openclaw/openclaw/issues/73756 | Telegram dmPolicy=open bypasses visible allowFrom allowlist and can expose bots publicly | ## Summary
Telegram channel permissions can be configured in a way that looks allowlisted but is effectively public. In a real local deployment, a non-allowlisted Telegram user was able to start an OpenClaw chat/session and issue agent commands because the effective Telegram account policy was `dmPolicy: "open"` with ... | closed | completed | true | 2 | [] | [] | 2026-04-28T19:06:05Z | 2026-05-03T04:37:30Z | 2026-04-30T13:49:58Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | Hilo-Hilo | 57,138,830 | MDQ6VXNlcjU3MTM4ODMw | User | false |
openclaw/openclaw | 4,337,523,980 | I_kwDOQb6kR88AAAABAoldDA | 72,938 | https://github.com/openclaw/openclaw/issues/72938 | https://api.github.com/repos/openclaw/openclaw/issues/72938 | openclaw plugins install <path> rejects valid plugin paths with cryptic 'must stay within extensions directory' error | > **In plain English:** when you try to install a plugin from a path on disk via `openclaw plugins install <path>`, the CLI rejects it with a confusing error message that mentions "extensions directory" but doesn't say *where* that directory is. The actual constraint — that the source path has to already be located und... | closed | not_planned | true | 1 | [] | [] | 2026-04-27T17:07:56Z | 2026-04-30T04:30:24Z | 2026-04-27T17:29:18Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mayank6136 | 16,468,779 | MDQ6VXNlcjE2NDY4Nzc5 | User | false |
openclaw/openclaw | 4,337,530,743 | I_kwDOQb6kR88AAAABAol3dw | 72,941 | https://github.com/openclaw/openclaw/issues/72941 | https://api.github.com/repos/openclaw/openclaw/issues/72941 | Plugin loader silently truncates async register() — 'returned a promise; async registration is ignored' | > **In plain English:** the gateway's plugin loader calls a plugin's `register()` synchronously and ignores its returned promise — so any setup the plugin does after the first `await` runs *after* registration is reported complete. Hooks that fire during the gap see a half-initialized plugin. The gateway log even says ... | closed | not_planned | true | 1 | [] | [] | 2026-04-27T17:09:08Z | 2026-04-30T04:30:23Z | 2026-04-27T17:29:22Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mayank6136 | 16,468,779 | MDQ6VXNlcjE2NDY4Nzc5 | User | false |
openclaw/openclaw | 4,337,538,521 | I_kwDOQb6kR88AAAABAomV2Q | 72,943 | https://github.com/openclaw/openclaw/issues/72943 | https://api.github.com/repos/openclaw/openclaw/issues/72943 | Plugin hooks register twice per gateway run (gateway loader + embedded acpx loader) — first ~7s window has telemetry-null hooks | > **In plain English:** the same plugin's hooks register twice during gateway startup — once via the gateway's plugin loader at ~+2 seconds, and a second time via a separate "plugins" loader at ~+9 seconds (after the embedded acpx runtime starts). The plugin's `service.start()` callback only runs on the second pass. So... | closed | not_planned | true | 1 | [] | [] | 2026-04-27T17:10:26Z | 2026-04-30T04:30:22Z | 2026-04-27T17:29:26Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mayank6136 | 16,468,779 | MDQ6VXNlcjE2NDY4Nzc5 | User | false |
openclaw/openclaw | 4,337,538,616 | I_kwDOQb6kR88AAAABAomWOA | 72,944 | https://github.com/openclaw/openclaw/issues/72944 | https://api.github.com/repos/openclaw/openclaw/issues/72944 | Plugin config in openclaw.json is RO inside NemoClaw sandbox — needs writable overlay path or env-var fallback for runtime reconfig | > **In plain English:** plugin configuration today lives in `openclaw.json:plugins.entries.<id>.config`, which is editable on a normal host install but **read-only at runtime in NemoClaw / OpenShell sandboxes** (the `/sandbox/.openclaw/` directory is Landlock-RO). That means every config change requires rebuilding the ... | closed | not_planned | true | 1 | [] | [] | 2026-04-27T17:10:27Z | 2026-04-30T04:30:21Z | 2026-04-27T17:29:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mayank6136 | 16,468,779 | MDQ6VXNlcjE2NDY4Nzc5 | User | false |
openclaw/openclaw | 4,337,538,823 | I_kwDOQb6kR88AAAABAomXBw | 72,946 | https://github.com/openclaw/openclaw/issues/72946 | https://api.github.com/repos/openclaw/openclaw/issues/72946 | openclaw plugins install <path> rejects valid plugin paths with cryptic error; correct install path nowhere documented | ## Summary
> **In plain English:** `openclaw plugins install <path>` rejects valid plugin paths with the error "Invalid path: must stay within extensions directory" — but it never tells you *where* that directory is. Users hit this trying to install a plugin from a build-output dir; the error gives no actionable guida... | closed | completed | true | 1 | [] | [] | 2026-04-27T17:10:30Z | 2026-04-30T04:30:21Z | 2026-04-27T19:52:52Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mayank6136 | 16,468,779 | MDQ6VXNlcjE2NDY4Nzc5 | User | false |
openclaw/openclaw | 4,345,657,911 | I_kwDOQb6kR88AAAABAwV6Nw | 73,765 | https://github.com/openclaw/openclaw/issues/73765 | https://api.github.com/repos/openclaw/openclaw/issues/73765 | skills.entries.*.enabled: false silently ignored on existing sessions due to stale skillsSnapshot (cache-invalidation gap) | ### Summary
`skills.entries[name].enabled: false` is silently ignored for **existing
long-lived sessions** if the user adds the override after the session's
`skillsSnapshot` was first persisted. The disabled skills keep appearing
in the model's `<available_skills>` block until the session expires past
the freshness wi... | open | null | false | 1 | [] | [] | 2026-04-28T19:24:08Z | 2026-05-03T10:19:45Z | null | CONTRIBUTOR | null | 20260503T120038Z | 2026-05-03T12:00:38Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,338,707,036 | I_kwDOQb6kR88AAAABAptqXA | 73,024 | https://github.com/openclaw/openclaw/issues/73024 | https://api.github.com/repos/openclaw/openclaw/issues/73024 | infer model run fails before model call when tools.alsoAllow is set and modelRun disables tools | ## Summary
`openclaw infer model run --local --model <id> --prompt <p>` can abort before attempting any model call when `~/.openclaw/openclaw.json` has a non-empty `tools.alsoAllow`.
The error is:
```
No callable tools remain after resolving explicit tool allowlist (tools.allow: *, lobster, llm-task)
```
This happe... | closed | completed | true | 1 | [] | [] | 2026-04-27T20:33:15Z | 2026-05-01T04:42:09Z | 2026-04-28T19:33:55Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | aa-on-ai | 94,061,050 | U_kgDOBZtB-g | User | false |
openclaw/openclaw | 4,345,621,018 | I_kwDOQb6kR88AAAABAwTqGg | 73,761 | https://github.com/openclaw/openclaw/issues/73761 | https://api.github.com/repos/openclaw/openclaw/issues/73761 | [Bug]: can't get "start talk" to work with a local whisper server . | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
AI generated , The "openai" realtime voice provider is reported as not configured even when its STT baseUrl and apiKey are explicitly defined within the plugins configuration block.
### Steps to reproduce
AI generated ,... | closed | not_planned | true | 1 | [
"bug"
] | [] | 2026-04-28T19:17:37Z | 2026-05-04T04:36:23Z | 2026-05-01T13:14:08Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | peter247 | 3,247,383 | MDQ6VXNlcjMyNDczODM= | User | false |
openclaw/openclaw | 4,199,060,554 | I_kwDOQb6kR876SJRK | 60,143 | https://github.com/openclaw/openclaw/issues/60143 | https://api.github.com/repos/openclaw/openclaw/issues/60143 | Plugin config error causes hard gateway crash instead of graceful disable | ## Description
When a plugin config key is invalid (e.g. after a schema migration), the gateway throws a hard config validation error and exits with code 1 instead of disabling only the affected plugin and continuing to start.
## Real-World Incident
After upgrading from 2026.3.28 → 2026.4.1, the `tools.web.search` c... | closed | completed | true | 2 | [] | [] | 2026-04-03T07:52:08Z | 2026-05-01T04:42:05Z | 2026-04-28T19:43:16Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | yhyatt | 10,474,956 | MDQ6VXNlcjEwNDc0OTU2 | User | false |
openclaw/openclaw | 4,345,594,550 | I_kwDOQb6kR88AAAABAwSCtg | 73,758 | https://github.com/openclaw/openclaw/issues/73758 | https://api.github.com/repos/openclaw/openclaw/issues/73758 | [Bug]: TTS directives ( + [[tts:text]]) silently dropped on claude-cli backend; no voice delivere | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On openclaw 2026.4.26 with the claude-cli backend, model output containing
and [[tts:text]]…[[/tts:text]] is parsed but never synthesized,
so Telegram receives no voice message and the reply is dropped.
### Steps to reprod... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-28T19:13:26Z | 2026-05-05T04:28:05Z | 2026-05-02T08:25:02Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | asidko | 22,843,881 | MDQ6VXNlcjIyODQzODgx | User | false |
openclaw/openclaw | 4,345,777,276 | I_kwDOQb6kR88AAAABAwdMfA | 73,777 | https://github.com/openclaw/openclaw/issues/73777 | https://api.github.com/repos/openclaw/openclaw/issues/73777 | [Bug] cron run + systemEvent + sessionTarget=main still results in deliveryStatus: "not-requested" on v2026.4.26 (regression from #60262 / #62296) | ## Bug Description
On OpenClaw **v2026.4.26**, cron jobs with `payload.kind: "systemEvent"` and `sessionTarget: "main"` still silently fail when manually triggered via `cron run` (both CLI and Control UI "Run" button). The run completes in ~12-31ms with `deliveryStatus: "not-requested"` and no actual execution.
This ... | closed | completed | true | 1 | [] | [] | 2026-04-28T19:45:45Z | 2026-05-03T04:34:41Z | 2026-04-30T23:29:18Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | vokasug | 280,127,440 | U_kgDOELJn0A | User | false |
openclaw/openclaw | 4,345,785,045 | I_kwDOQb6kR88AAAABAwdq1Q | 73,778 | https://github.com/openclaw/openclaw/issues/73778 | https://api.github.com/repos/openclaw/openclaw/issues/73778 | [Bug]: Telegram reply context captured in session JSONL but stripped from LLM context — agent cannot see replied message | ## Bug Description
When a user replies to a message in Telegram (using the native reply/quote feature), the reply context data is **correctly captured and stored** in the session JSONL as a `custom_message` (`openclaw.runtime-context`), but **it is stripped before reaching the LLM context**.
The LLM only sees the run... | closed | completed | true | 1 | [] | [] | 2026-04-28T19:47:19Z | 2026-05-01T04:42:04Z | 2026-04-28T19:51:50Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | superandylin | 56,207,675 | MDQ6VXNlcjU2MjA3Njc1 | User | false |
openclaw/openclaw | 4,345,598,480 | I_kwDOQb6kR88AAAABAwSSEA | 73,759 | https://github.com/openclaw/openclaw/issues/73759 | https://api.github.com/repos/openclaw/openclaw/issues/73759 | openclaw infer model run hangs at secrets.resolve in 2026.4.26 (both --local and --gateway paths) | ## Summary
After upgrading to **2026.4.26 (be8c246)** and registering an Anthropic auth profile via `openclaw models auth login --provider anthropic --method cli --set-default`, `openclaw infer model run` hangs indefinitely for any `anthropic/*` model. **Both `--local` and `--gateway` execution paths fail** at the sec... | closed | completed | true | 2 | [] | [] | 2026-04-28T19:14:10Z | 2026-05-01T04:42:03Z | 2026-04-28T19:56:21Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | chrislro | 7,277,502 | MDQ6VXNlcjcyNzc1MDI= | User | false |
openclaw/openclaw | 4,345,882,308 | I_kwDOQb6kR88AAAABAwjmxA | 73,784 | https://github.com/openclaw/openclaw/issues/73784 | https://api.github.com/repos/openclaw/openclaw/issues/73784 | [Bug]: /new Discord session can receive stale lossless-claw summaries as active context | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
After `/new` in a Discord channel session using `lossless-claw` as the context engine, compacted/summary context from the prior logical channel conversation can be injected into the fresh session and treated as... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T20:05:25Z | 2026-05-03T04:36:33Z | 2026-04-30T14:59:00Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | BooRy95 | 257,679,890 | U_kgDOD1viEg | User | false |
openclaw/openclaw | 4,345,361,123 | I_kwDOQb6kR88AAAABAwDy4w | 73,730 | https://github.com/openclaw/openclaw/issues/73730 | https://api.github.com/repos/openclaw/openclaw/issues/73730 | resolvePluginWebProviders snapshot cache never hits because callers pass fresh config objects | ## Environment
- openclaw `2026.4.26`
## Summary
The web-fetch / web-search provider layer has a `memoizeSnapshot` WeakMap cache keyed by `params.config`. In practice the cache never hits because upstream callers construct a fresh `config` object on every call. Every message dispatch pays the full ~30 second web-searc... | closed | completed | true | 5 | [] | [] | 2026-04-28T18:27:35Z | 2026-05-03T04:37:21Z | 2026-04-30T13:55:49Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | sg1416-zg | 259,938,361 | U_kgDOD35YOQ | User | false |
openclaw/openclaw | 4,345,794,470 | I_kwDOQb6kR88AAAABAwePpg | 73,782 | https://github.com/openclaw/openclaw/issues/73782 | https://api.github.com/repos/openclaw/openclaw/issues/73782 | [Bug]: Discord channel missing from bundled sidecar in 2026.4.26 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Discord plugin fails to load on core 2026.4.26 with @openclaw/discord 2026.3.13. Same root cause as #63774 (Cannot find module .../plugin-sdk/root-alias.cjs/discord), but on npm install rather than Homebrew.
### Steps to r... | closed | completed | true | 0 | [
"bug"
] | [] | 2026-04-28T19:49:18Z | 2026-05-01T04:42:00Z | 2026-04-28T20:09:44Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | lilaclabs | 4,230,696 | MDQ6VXNlcjQyMzA2OTY= | User | false |
openclaw/openclaw | 4,345,921,102 | I_kwDOQb6kR88AAAABAwl-Tg | 73,786 | https://github.com/openclaw/openclaw/issues/73786 | https://api.github.com/repos/openclaw/openclaw/issues/73786 | [Feature]: Token optimization flags: compact tool descriptions and system prompt sections for effective use of local hosted models | ### Summary
Add experimental token-saving flags (compact tool descriptions, compact system prompt sections, and optional memory injection skip) to reduce context consumption for resource-constrained deployments.
### Problem to solve
Users running OpenClaw with smaller language models (GPT-4, Sonnet, Haiku), local m... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-28T20:12:09Z | 2026-05-03T04:36:32Z | 2026-04-30T14:59:08Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | varunsaravagi | 725,323 | MDQ6VXNlcjcyNTMyMw== | User | false |
openclaw/openclaw | 4,340,892,977 | I_kwDOQb6kR88AAAABArzFMQ | 73,314 | https://github.com/openclaw/openclaw/issues/73314 | https://api.github.com/repos/openclaw/openclaw/issues/73314 | Regression in 2026.4.26: /new command fails with 'messages: at least one message is required' | ## Bug Report
**Version:** 2026.4.26 (be8c246)
**Previous working version:** 2026.4.25
**Platform:** macOS arm64 (Mac mini M4 Pro)
**Channel:** Telegram (direct chat)
## Summary
After updating to 2026.4.26, using the bare `/new` command (with no text) consistently fails. This was not an issue in 2026.4.25.
## Steps... | closed | not_planned | true | 2 | [] | [] | 2026-04-28T05:47:32Z | 2026-05-01T04:41:58Z | 2026-04-28T05:57:12Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | moorerussel | 267,125,914 | U_kgDOD-wEmg | User | false |
openclaw/openclaw | 4,345,635,277 | I_kwDOQb6kR88AAAABAwUhzQ | 73,763 | https://github.com/openclaw/openclaw/issues/73763 | https://api.github.com/repos/openclaw/openclaw/issues/73763 | [Bug]: session_status tool-call loop not hard-aborted — agent repeats 30+ identical calls despite loop detector warnings | ## Environment
- OpenClaw: `2026.4.26 (be8c246)`
- OS: Linux 6.8.0-107-generic (x64)
- Node: v22.22.1
- Model: `claude-opus-4.6` (via custom provider)
- Session type: human-facing `agent:main:main` (Telegram direct)
## Bug summary
When the agent calls `session_status` as part of assembling a status dashboard, it ent... | closed | not_planned | true | 2 | [] | [] | 2026-04-28T19:20:04Z | 2026-05-03T04:36:01Z | 2026-04-30T16:18:17Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | MagnaAI | 265,843,828 | U_kgDOD9h0dA | User | false |
openclaw/openclaw | 4,345,967,213 | I_kwDOQb6kR88AAAABAwoybQ | 73,789 | https://github.com/openclaw/openclaw/issues/73789 | https://api.github.com/repos/openclaw/openclaw/issues/73789 | Bug: bare /new can submit empty prompt after session reset | ## Summary
Bare `/new` (and likely bare `/reset`) can create a fresh session and then submit an empty prompt to the model. With `openai-codex-responses` this fails immediately with:
```text
One of "input" or "previous_response_id" or 'prompt' or 'conversation_id' must be provided.
```
## Environment
- OpenClaw: `20... | closed | completed | true | 6 | [] | [] | 2026-04-28T20:19:50Z | 2026-05-03T04:37:08Z | 2026-04-30T14:08:16Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | chac4l | 9,748,597 | MDQ6VXNlcjk3NDg1OTc= | User | false |
openclaw/openclaw | 4,346,045,153 | I_kwDOQb6kR88AAAABAwti4Q | 73,795 | https://github.com/openclaw/openclaw/issues/73795 | https://api.github.com/repos/openclaw/openclaw/issues/73795 | toolProgress config stripped from channels without streaming.mode | ## Bug
Setting `streaming.preview.toolProgress: false` on a channel in `openclaw.json` is silently stripped by the config normalizer if the channel does not already have a `streaming.mode` key. This causes internal tool execution details (exec commands, curl invocations, tool names) to leak to end users on affected ch... | open | null | false | 1 | [] | [] | 2026-04-28T20:33:40Z | 2026-05-03T09:13:23Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | victoryoung1 | 11,039,178 | MDQ6VXNlcjExMDM5MTc4 | User | false |
openclaw/openclaw | 4,346,027,613 | I_kwDOQb6kR88AAAABAwseXQ | 73,793 | https://github.com/openclaw/openclaw/issues/73793 | https://api.github.com/repos/openclaw/openclaw/issues/73793 | Capability-provider lookups bypass cache when plugins.entries is non-empty (~25-30s latency per turn) | ### Bug type
Behavior bug (incorrect output/state without crash) — performance regression-class issue (latency stacking).
### Beta release blocker
No
### Summary
`resolvePluginCapabilityProviders` bypasses its active-registry cache on every call when `cfg.plugins.entries` is non-empty, causing 4–5 full `loadOpenClawP... | closed | completed | true | 1 | [] | [] | 2026-04-28T20:30:29Z | 2026-05-02T04:29:02Z | 2026-04-29T06:52:35Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | poolside-ventures | 215,486,291 | U_kgDODNgPUw | User | false |
openclaw/openclaw | 4,074,339,708 | I_kwDOQb6kR87y2X18 | 45,637 | https://github.com/openclaw/openclaw/issues/45637 | https://api.github.com/repos/openclaw/openclaw/issues/45637 | [Feature]: Add timeoutMs config support for LLM Providers | ### Summary
OpenClaw's LLM request timeout is currently hardcoded to 30 seconds (`DEFAULT_TIMEOUT_MS$2 = 3e4`), which is too short for certain reasoning models (such as Zhipu GLM-5), causing frequent timeout failures.
Add `timeoutMs` parameter to `models.providers.*` configuration, allowing users to customize the time... | closed | completed | true | 4 | [
"enhancement"
] | [] | 2026-03-14T02:13:12Z | 2026-05-01T04:41:56Z | 2026-04-28T20:42:24Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | lhtpluto | 132,045,861 | U_kgDOB97cJQ | User | false |
openclaw/openclaw | 4,074,623,267 | I_kwDOQb6kR87y3dEj | 45,731 | https://github.com/openclaw/openclaw/issues/45731 | https://api.github.com/repos/openclaw/openclaw/issues/45731 | [Bug]: Web UI extremely slow or unresponsive after upgrading to v2026.3.12 | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to OpenClaw v2026.3.12, the Web UI (Control UI at http://127.0.0.1:18789) becomes extremely slow or completely unresponsive when opened in browser. Both Chrome and Safari are affected.
### Steps to reproduce
1. Upgrade to v2026.3.12
2. ... | closed | not_planned | true | 4 | [
"bug",
"regression"
] | [] | 2026-03-14T04:19:54Z | 2026-05-01T04:41:55Z | 2026-04-28T20:42:32Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | 6300browndust2-ship-it | 267,835,940 | U_kgDOD_baJA | User | false |
openclaw/openclaw | 4,075,320,253 | I_kwDOQb6kR87y6HO9 | 45,943 | https://github.com/openclaw/openclaw/issues/45943 | https://api.github.com/repos/openclaw/openclaw/issues/45943 | Feature Request: Auto-restart Gateway after npm upgrade | ---
name: Feature Request: Auto-restart Gateway after npm upgrade
about: Suggest an improvement to automatically restart Gateway after npm upgrade
title: 'Feature Request: Auto-restart Gateway after npm upgrade'
labels: enhancement, gateway, npm
assignees: ''
---
## 🎯 Problem Statement
When upgrading OpenClaw via np... | closed | completed | true | 1 | [] | [] | 2026-03-14T09:40:21Z | 2026-05-01T04:41:55Z | 2026-04-28T20:42:39Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | luffys1984 | 127,186,644 | U_kgDOB5S21A | User | false |
openclaw/openclaw | 4,075,653,205 | I_kwDOQb6kR87y7YhV | 46,070 | https://github.com/openclaw/openclaw/issues/46070 | https://api.github.com/repos/openclaw/openclaw/issues/46070 | [Regression] clearUnboundScopes() in 2026.3.12 breaks Paperclip adapter connections | ## Description
OpenClaw 2026.3.12 introduced `clearUnboundScopes()` which clears operator scopes to `[]` when a WebSocket client connects without device identity and `sharedAuthOk` is false.
This breaks the Paperclip `openclaw_gateway` adapter, which connects with `disableDeviceAuth: true` and relies on scopes declar... | closed | not_planned | true | 3 | [] | [] | 2026-03-14T11:56:17Z | 2026-05-01T04:41:54Z | 2026-04-28T20:42:44Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | jean-francois-bot[bot] | 260,262,996 | BOT_kgDOD4NMVA | Bot | false |
openclaw/openclaw | 4,076,123,516 | I_kwDOQb6kR87y9LV8 | 46,315 | https://github.com/openclaw/openclaw/issues/46315 | https://api.github.com/repos/openclaw/openclaw/issues/46315 | [Feature]: 跨通道消息同步 | ### Summary
希望实现飞书（Feishu）和 WebChat（Control UI）之间的消息双向同步。
### Problem to solve
当前行为：
1. session.dmScope: "main" 配置下，飞书和 WebChat 共享主会话历史
2. 消息从哪个通道来，就回复回哪个通道
3. 无法实现：网页发消息 → 飞书收到回复（目前飞书发消息网页可以收到并同步）
期望行为：在任一通道发送消息，另一通道也能收到回复（消息互通）
使用场景：
1.用户在网页端提问，收到飞书回复（反之亦然）
2.多设备、多平台用户希望统一对话入口
### Proposed solution
1.新增配置项如 se... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-14T16:07:40Z | 2026-05-01T04:41:53Z | 2026-04-28T20:42:50Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | rollegc | 208,344,968 | U_kgDODGsXiA | User | false |
openclaw/openclaw | 4,076,834,668 | I_kwDOQb6kR87y_49s | 46,604 | https://github.com/openclaw/openclaw/issues/46604 | https://api.github.com/repos/openclaw/openclaw/issues/46604 | [Feature]: Split Dashboard into Communication Panel + Tool Activity Panel | ### Summary
Split the webchat dashboard main container into two independent panels: a stable Communication Panel for agent↔user messages, and a separate Tool Activity Panel for tool execution output.
### Problem to solve
During multi-step agent operations, tool execution output (file reads, shell commands, API calls... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-14T21:55:37Z | 2026-05-01T04:41:52Z | 2026-04-28T20:42:56Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | gjoham | 181,919,056 | U_kgDOCtfdUA | User | false |
openclaw/openclaw | 4,077,284,827 | I_kwDOQb6kR87zBm3b | 46,703 | https://github.com/openclaw/openclaw/issues/46703 | https://api.github.com/repos/openclaw/openclaw/issues/46703 | Bug: session-memory hook uses UTC date for filenames instead of local timezone | ## Problem
The session-memory hook saves session summaries using UTC date for filenames, ignoring the system timezone.
### Example
- Session started at **18:17 PDT on March 14, 2026** (local time)
- Filename created: `2026-03-15-skill-cleanup.md` (UTC date)
- Expected filename: `2026-03-14-skill-cleanup.md` (local da... | closed | completed | true | 1 | [] | [] | 2026-03-15T01:24:41Z | 2026-05-01T04:41:52Z | 2026-04-28T20:43:06Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | razhangwei | 4,484,566 | MDQ6VXNlcjQ0ODQ1NjY= | User | false |
openclaw/openclaw | 4,077,508,246 | I_kwDOQb6kR87zCdaW | 46,833 | https://github.com/openclaw/openclaw/issues/46833 | https://api.github.com/repos/openclaw/openclaw/issues/46833 | Display multiple channel sessions in Control UI sidebar | Currently the OpenClaw Control UI only shows one conversation window at a time. Users cannot see multiple Feishu conversations (private chats and group chats) simultaneously in the sidebar.
Expected behavior:
- Show all active sessions in the left sidebar
- Allow switching between different channel conversations
Curr... | closed | not_planned | true | 1 | [] | [] | 2026-03-15T03:53:13Z | 2026-05-01T04:41:51Z | 2026-04-28T20:43:13Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | fasignal-svg | 268,292,782 | U_kgDOD_3Srg | User | false |
openclaw/openclaw | 4,345,792,398 | I_kwDOQb6kR88AAAABAweHjg | 73,781 | https://github.com/openclaw/openclaw/issues/73781 | https://api.github.com/repos/openclaw/openclaw/issues/73781 | [Bug]: Tool call replay loop: runtime repeats agent's previous tool call after failure, causing infinite loops | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Bug Description
After upgrading to OpenClaw 2026.4.25, the agent enters an infinite tool loop where the **runtime replays the agent's previous tool call** after a tool failure, rather than allowing the agent to reason an... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-04-28T19:48:52Z | 2026-05-03T07:45:30Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | jackycao2000 | 255,738,170 | U_kgDODz5BOg | User | false |
openclaw/openclaw | 4,077,730,585 | I_kwDOQb6kR87zDTsZ | 46,965 | https://github.com/openclaw/openclaw/issues/46965 | https://api.github.com/repos/openclaw/openclaw/issues/46965 | Feature: global suppressToolErrorWarnings (not just heartbeat) | ## Problem
When an agent's tool call fails (e.g. `edit` with non-matching text, sandbox path escape, or invalid config schema path), the Gateway emits a fallback tool error reply to the user. This is correct behavior for cases where the user needs to know — but creates noise for:
1. **Internal agent operation errors*... | closed | not_planned | true | 1 | [] | [] | 2026-03-15T06:28:09Z | 2026-05-01T04:41:50Z | 2026-04-28T20:43:18Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | agentbuilder1999 | 266,951,717 | U_kgDOD-lcJQ | User | false |
openclaw/openclaw | 4,081,119,207 | I_kwDOQb6kR87zQO_n | 48,066 | https://github.com/openclaw/openclaw/issues/48066 | https://api.github.com/repos/openclaw/openclaw/issues/48066 | [Bug] Feishu plugin causes Gateway crash | ### Bug type
Regression (worked before, now fails)
### Summary
Feishu plugin causes Gateway to crash with WebSocket handshake timeout error. This happens not only when using feishu_doc tool, but also when the Feishu plugin is simply loaded.
### Steps to reproduce
标题: [Bug] Feishu plugin causes Gateway crash (WebS... | closed | completed | true | 6 | [
"bug",
"regression"
] | [] | 2026-03-16T08:56:20Z | 2026-05-01T04:39:56Z | 2026-04-28T20:43:27Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | zzzgd | 35,446,203 | MDQ6VXNlcjM1NDQ2MjAz | User | false |
openclaw/openclaw | 4,086,498,457 | I_kwDOQb6kR87zkwSZ | 48,782 | https://github.com/openclaw/openclaw/issues/48782 | https://api.github.com/repos/openclaw/openclaw/issues/48782 | Feature Request: Project-level memory for multi-project workflows | ## Problem
When working on multiple projects in the same session, the agent may inadvertently mix up tech stacks and features between projects. There is no project-level context awareness - the agent only knows about the current workspace files but does not explicitly track which project is currently being worked on.
... | closed | not_planned | true | 1 | [] | [] | 2026-03-17T06:26:14Z | 2026-05-01T04:41:49Z | 2026-04-28T20:43:33Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | wujinyuan | 635,435 | MDQ6VXNlcjYzNTQzNQ== | User | false |
openclaw/openclaw | 4,217,641,429 | I_kwDOQb6kR877ZBnV | 62,486 | https://github.com/openclaw/openclaw/issues/62486 | https://api.github.com/repos/openclaw/openclaw/issues/62486 | [Bug]: 2026.4.2 版本三个影响体验的问题 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## 问题描述
作为 OpenClaw 用户，在使用 2026.4.2 版本时遇到以下三个影响体验的问题：
### 问题1：QQ 和 Web 聊天记录无法同步
**现象：**
- QQ 端（NapCat/OneBot）和 OpenClaw Web/电脑端聊天记录完全独立
- 在 Web 端聊的内容，QQ 上看不到历史
- 在 QQ 上聊的内容，Web 端看不到历史
**影响：**
- 需要重复说明情况
- 用户体验割裂
**期望：**... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-07T12:28:33Z | 2026-05-01T04:39:58Z | 2026-04-28T20:43:38Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | yangbo1986 | 38,829,320 | MDQ6VXNlcjM4ODI5MzIw | User | false |
openclaw/openclaw | 4,220,523,387 | I_kwDOQb6kR877kBN7 | 62,702 | https://github.com/openclaw/openclaw/issues/62702 | https://api.github.com/repos/openclaw/openclaw/issues/62702 | feat(plugins): Non-blocking async plugin hooks with configurable timeout + fallback | ## Problem
Plugin hooks that perform I/O (cloud API calls, database queries, file reads) currently block the turn pipeline. If any hook is slow or the external service is degraded, the entire agent turn waits — the user sees no typing indicator and no response until the hook completes or fails.
**Real-world example:*... | closed | completed | true | 1 | [] | [] | 2026-04-07T20:50:54Z | 2026-05-01T04:41:48Z | 2026-04-28T20:43:43Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | issaba1 | 9,679,850 | MDQ6VXNlcjk2Nzk4NTA= | User | false |
openclaw/openclaw | 4,221,125,350 | I_kwDOQb6kR877mULm | 62,754 | https://github.com/openclaw/openclaw/issues/62754 | https://api.github.com/repos/openclaw/openclaw/issues/62754 | Gateway: unhandled promise rejection from pi-agent-core Agent.processEvents after run abort/timeout | # Upstream Bug 1 — pi-agent-core lifecycle race: "Agent listener invoked outside active run"
**Repo:** github.com/openclaw/openclaw
**Suggested labels:** `bug`, `gateway`, `pi-agent-core`, `stability`
**OpenClaw version:** 2026.4.5
**Node:** 22.22.1
**OS:** macOS (Apple Silicon)
---
## Title
Gateway: unhandled promi... | closed | completed | true | 1 | [] | [] | 2026-04-07T23:23:32Z | 2026-05-01T04:41:48Z | 2026-04-28T20:43:50Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | jdroblee-afk | 266,868,297 | U_kgDOD-gWSQ | User | false |
openclaw/openclaw | 4,221,214,142 | I_kwDOQb6kR877mp2- | 62,764 | https://github.com/openclaw/openclaw/issues/62764 | https://api.github.com/repos/openclaw/openclaw/issues/62764 | openclaw doctor --fix regenerates user-systemd unit even when system unit owns the gateway, causing dual-instance deadlock | ## Bug: `openclaw doctor --fix` regenerates a user-systemd unit even when a system-systemd unit owns the gateway, causing dual-instance deadlock
### Environment
- OpenClaw `2026.4.5 (3e72c03)`
- Ubuntu 24.04.4 LTS, systemd 255
- Node v24.14.1 via nvm
- openclaw-gateway running as a **system-systemd** unit at `/etc/sy... | closed | completed | true | 2 | [] | [] | 2026-04-07T23:49:21Z | 2026-05-01T04:41:47Z | 2026-04-28T20:43:58Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | tmote | 21,976,980 | MDQ6VXNlcjIxOTc2OTgw | User | false |
openclaw/openclaw | 4,346,092,528 | I_kwDOQb6kR88AAAABAwwb8A | 73,797 | https://github.com/openclaw/openclaw/issues/73797 | https://api.github.com/repos/openclaw/openclaw/issues/73797 | WhatsApp pairing: verification message sent without inbound text message | ## Problem
When Master sends an outbound WhatsApp message to a new number (not in allowFrom), the recipient sometimes receives an unsolicited pairing/verification message — even though they never sent a message to Master.
## Expected behavior
The pairing verification message should **only** be sent when someone acti... | closed | completed | true | 1 | [] | [] | 2026-04-28T20:43:00Z | 2026-05-01T04:40:58Z | 2026-04-29T01:09:24Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | hbmasters | 264,673,599 | U_kgDOD8aZPw | User | false |
openclaw/openclaw | 4,345,788,625 | I_kwDOQb6kR88AAAABAwd40Q | 73,780 | https://github.com/openclaw/openclaw/issues/73780 | https://api.github.com/repos/openclaw/openclaw/issues/73780 | 🔒 Repeated API rate limit errors on Telegram | $(cat /tmp/github_issue_updated.txt) | open | null | false | 2 | [] | [] | 2026-04-28T19:48:02Z | 2026-05-03T11:50:05Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | xace1825 | 199,889,897 | U_kgDOC-oT6Q | User | false |
openclaw/openclaw | 4,209,104,832 | I_kwDOQb6kR8764dfA | 61,610 | https://github.com/openclaw/openclaw/issues/61610 | https://api.github.com/repos/openclaw/openclaw/issues/61610 | [Bug]: tasks cancel not supported + tasks maintenance --apply prunes 0 stuck running tasks | ## Summary
`openclaw tasks cancel` returns "Task runtime does not support cancellation yet." and `openclaw tasks maintenance --apply` prunes **0** stuck tasks. Full gateway `stop` + `start` also fails to clear these persisted task records.
I have 6 background tasks stuck in `running` state that cannot be recovered.
... | open | null | false | 3 | [] | [] | 2026-04-06T02:06:48Z | 2026-05-02T18:31:08Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | nickcn2n-dot | 268,765,385 | U_kgDOEAUIyQ | User | false |
openclaw/openclaw | 4,345,408,137 | I_kwDOQb6kR88AAAABAwGqiQ | 73,740 | https://github.com/openclaw/openclaw/issues/73740 | https://api.github.com/repos/openclaw/openclaw/issues/73740 | [Feature]: Support outbound-only messaging for specific WhatsApp peers (no inbound processing) | ### Summary
Currently, OpenClaw allows sending WhatsApp messages to specific contacts via configured bindings or allowlists. However I could not find a way to separate outbound messaging permissions from inbound message processing on a per-contact basis.
This means that if a number is allowed for outbound communicati... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-28T18:36:44Z | 2026-05-03T04:33:39Z | 2026-04-30T14:58:42Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | 3sken | 99,713,976 | U_kgDOBfGDuA | User | false |
openclaw/openclaw | 4,345,410,709 | I_kwDOQb6kR88AAAABAwG0lQ | 73,743 | https://github.com/openclaw/openclaw/issues/73743 | https://api.github.com/repos/openclaw/openclaw/issues/73743 | [Bug]: openclaw CLI subprocess startup ~25s on idle macOS; worker-tick fan-out under launchd compounds to 5+ min ticks | ## Summary
`openclaw <subcommand>` CLI subprocesses take ~25 s of wall time even on an idle macOS host with no concurrent activity. When the same machine fans out 21 launchd-driven workflow ticks every 60 s (`recipes workflows worker-tick` × 19 + `runner-tick` × 2), the per-tick runtime stretches from ~25 s in isolati... | open | null | false | 1 | [] | [] | 2026-04-28T18:37:16Z | 2026-05-03T06:07:45Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | rjdjohnston | 293,686 | MDQ6VXNlcjI5MzY4Ng== | User | false |
openclaw/openclaw | 4,345,430,260 | I_kwDOQb6kR88AAAABAwIA9A | 73,747 | https://github.com/openclaw/openclaw/issues/73747 | https://api.github.com/repos/openclaw/openclaw/issues/73747 | [Bug]: Workspace plugin HTTP routes register successfully but dispatcher returns 404 on all paths | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
A workspace plugin's api.registerHttpRoute(...) call succeeds, the route appears in the plugin registry per openclaw plugins inspect, and the registration is logged at startup. However the HTTP dispatcher retur... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-04-28T18:41:05Z | 2026-05-05T04:28:34Z | 2026-05-02T06:16:59Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | fmsxuser | 253,172,214 | U_kgDODxcZ9g | User | false |
openclaw/openclaw | 4,345,485,387 | I_kwDOQb6kR88AAAABAwLYSw | 73,753 | https://github.com/openclaw/openclaw/issues/73753 | https://api.github.com/repos/openclaw/openclaw/issues/73753 | Discord voice enabled causes sustained gateway CPU spin after embedded Discord connects | ## Bug Report
### Summary
When `channels.discord.voice.enabled=true` (the default), the embedded Discord provider causes `openclaw-gateway` to sustain ~100% CPU on Linux after bot accounts connect. This is a separate issue from #73709 (which covers the intent-selection bug when voice is *disabled*) — this is the unde... | closed | completed | true | 2 | [] | [] | 2026-04-28T18:52:14Z | 2026-05-04T04:36:31Z | 2026-05-01T11:49:36Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | sanchezm86 | 66,266,304 | MDQ6VXNlcjY2MjY2MzA0 | User | false |
openclaw/openclaw | 4,345,490,207 | I_kwDOQb6kR88AAAABAwLrHw | 73,754 | https://github.com/openclaw/openclaw/issues/73754 | https://api.github.com/repos/openclaw/openclaw/issues/73754 | msteams: v2026.4.26 ships broken JWT validator (jwt.verify is not a function); fix b3bc60ae missed the cut | ## Summary
The msteams plugin bundled with `openclaw@2026.4.26` (which pins `@openclaw/msteams@2026.4.25`) silently drops every inbound Teams message because `jwt.verify is not a function`. Outbound messages from the bot continue to work, masking the failure — bots appear "online" but never receive user messages.
A fi... | closed | completed | true | 2 | [] | [] | 2026-04-28T18:53:14Z | 2026-05-03T04:36:33Z | 2026-04-30T14:58:50Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | mroboff | 81,203,001 | MDQ6VXNlcjgxMjAzMDAx | User | false |
openclaw/openclaw | 4,346,056,562 | I_kwDOQb6kR88AAAABAwuPcg | 73,796 | https://github.com/openclaw/openclaw/issues/73796 | https://api.github.com/repos/openclaw/openclaw/issues/73796 | Browser screenshot delivery to chat fails — MEDIA:~/... paths rejected by media-source validator | ### Summary
When the browser plugin saves a screenshot and the agent emits a `MEDIA:` line referencing it with a `~/`-prefixed path, the channel outbound silently drops the attachment. The user receives `⚠️ Media failed.` instead of the photo. This makes "send me a screenshot" requests over Telegram unable to deliver ... | closed | completed | true | 1 | [] | [] | 2026-04-28T20:35:46Z | 2026-05-05T04:26:09Z | 2026-05-02T22:21:59Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | fabkury | 9,516,939 | MDQ6VXNlcjk1MTY5Mzk= | User | false |
openclaw/openclaw | 4,078,766,329 | I_kwDOQb6kR87zHQj5 | 47,504 | https://github.com/openclaw/openclaw/issues/47504 | https://api.github.com/repos/openclaw/openclaw/issues/47504 | [Channel] Session freezes after /new command - multiple channels affected | ## 问题描述
在 channel 中发送 /new 指令创建新会话后，该 channel 下的所有 agent 都无法接收新消息（假死状态），但其他 channel 不受影响。
此外，即使不发送 /new，agent 在 channel 中聊天时也偶尔会出现这种假死状态。
## 环境信息
- OpenClaw 版本：2026.3.13
- 操作系统：Linux
- Channel：Feishu, Telegram, DingTalk
- Agent：main, claudebot, im-operator
## 复现步骤
1. 在某个 channel（如飞书）中与 agent 对话
2. 发送 /new 指令创建新会话... | closed | completed | true | 3 | [] | [] | 2026-03-15T17:05:06Z | 2026-05-01T04:41:44Z | 2026-04-28T21:13:55Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | smallmj | 23,742,160 | MDQ6VXNlcjIzNzQyMTYw | User | false |
openclaw/openclaw | 4,083,727,716 | I_kwDOQb6kR87zaL1k | 48,390 | https://github.com/openclaw/openclaw/issues/48390 | https://api.github.com/repos/openclaw/openclaw/issues/48390 | WhatsApp Web: chronic 440 session conflict causes stale-socket loop and delivery failures | ## Summary
WhatsApp Web connection enters a self-reinforcing loop: the health monitor restarts the provider after 30 min of no events, but each new connection gets a **440 session conflict** from WhatsApp and immediately stops. This repeats indefinitely (51+ times in a single day), causing all outbound delivery to fai... | closed | completed | true | 6 | [] | [] | 2026-03-16T17:10:42Z | 2026-05-01T04:41:43Z | 2026-04-28T21:14:01Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | mgonto | 723,723 | MDQ6VXNlcjcyMzcyMw== | User | false |
openclaw/openclaw | 4,084,039,113 | I_kwDOQb6kR87zbX3J | 48,420 | https://github.com/openclaw/openclaw/issues/48420 | https://api.github.com/repos/openclaw/openclaw/issues/48420 | [Bug]: OAuth authentication broken for Anthropic and OpenAI providers in v2026.2.17+ | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to OpenClaw 2026.2.17+, OAuth authentication for Anthropic and OpenAI providers fails with "invalid x-api-key" error. API key authentication continues to work correctly.
### Steps to reproduce
1. Configure OAuth authentication for Anthr... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-16T18:16:20Z | 2026-05-01T04:41:42Z | 2026-04-28T21:14:10Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | AnthonyFrancis | 3,010,099 | MDQ6VXNlcjMwMTAwOTk= | User | false |
openclaw/openclaw | 4,087,265,578 | I_kwDOQb6kR87znrkq | 48,894 | https://github.com/openclaw/openclaw/issues/48894 | https://api.github.com/repos/openclaw/openclaw/issues/48894 | [Feature]: Support MESSAGE_UPDATE for new @mention detection | ### Summary
When a Discord message is posted without an @mention and then edited to add one, the mentioned agent is never triggered. OpenClaw only processes MESSAGE_CREATE events — MESSAGE_UPDATE with newly added @mentions is silently ignored.
Expected behaviour: If a user edits a message to add a bot @mention that w... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-17T09:26:00Z | 2026-05-01T04:41:41Z | 2026-04-28T21:14:16Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | nexus-666 | 265,620,448 | U_kgDOD9UL4A | User | false |
openclaw/openclaw | 4,346,294,906 | I_kwDOQb6kR88AAAABAw8yeg | 73,805 | https://github.com/openclaw/openclaw/issues/73805 | https://api.github.com/repos/openclaw/openclaw/issues/73805 | prewarmConfiguredPrimaryModel blocks 30s on startup (openai-codex provider name mismatch) | ## Bug Description
`prewarmConfiguredPrimaryModel` blocks gateway startup for exactly 30 seconds when the configured primary model uses a provider name that doesn't match the models registry.
## Root Cause
The sync path in `resolveModel` (called with `skipProviderRuntimeHooks: true`) looks up the provider by the nam... | closed | completed | true | 1 | [] | [] | 2026-04-28T21:19:41Z | 2026-05-03T04:36:00Z | 2026-04-30T16:18:28Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | dertbv | 3,845,627 | MDQ6VXNlcjM4NDU2Mjc= | User | false |
openclaw/openclaw | 4,346,312,835 | I_kwDOQb6kR88AAAABAw94gw | 73,808 | https://github.com/openclaw/openclaw/issues/73808 | https://api.github.com/repos/openclaw/openclaw/issues/73808 | First message after restart stalls ~42s (bundled provider tool factories timeout sequentially) | ## Bug Description
After a clean gateway startup (~6s), the first user message stalls for ~42-55 seconds before the assistant begins responding. Subsequent messages are fast (~3s API round-trip). The stall occurs in the tool initialization phase where bundled provider plugins run tool factories sequentially, and uncon... | closed | completed | true | 1 | [] | [] | 2026-04-28T21:22:50Z | 2026-05-03T04:36:31Z | 2026-04-30T14:59:28Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | dertbv | 3,845,627 | MDQ6VXNlcjM4NDU2Mjc= | User | false |
openclaw/openclaw | 4,346,308,598 | I_kwDOQb6kR88AAAABAw9n9g | 73,806 | https://github.com/openclaw/openclaw/issues/73806 | https://api.github.com/repos/openclaw/openclaw/issues/73806 | OC 2026.4.23 schema validator rejects hooks.allowConversationAccess (loader expects it) | ## Summary
Schema/loader drift in OC 2026.4.23: the zod config validator only accepts `plugins.entries.<id>.hooks.allowPromptInjection`, but the plugin loader reads and **enforces** `allowConversationAccess` for non-bundled plugins that register conversation-class typed hooks (`llm_input`, `llm_output`, `agent_end`).
... | closed | completed | true | 3 | [] | [
"vincentkoc"
] | 2026-04-28T21:22:04Z | 2026-05-02T04:28:15Z | 2026-04-29T13:09:23Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | GodotH | 804,151 | MDQ6VXNlcjgwNDE1MQ== | User | false |
openclaw/openclaw | 4,093,522,845 | I_kwDOQb6kR87z_jOd | 49,613 | https://github.com/openclaw/openclaw/issues/49613 | https://api.github.com/repos/openclaw/openclaw/issues/49613 | [Bug]: Non-default agents do not inherit agents.defaults.heartbeat unless they define agents.list[].heartbeat | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`agents.defaults.heartbeat` merges heartbeat values for non-default agents, but it does not enable heartbeat for them unless they also define `agents.list[].heartbeat`.
### Steps to reproduce
1. Configure a default agent and a non-default... | closed | not_planned | true | 3 | [] | [] | 2026-03-18T08:19:17Z | 2026-05-01T04:41:40Z | 2026-04-28T21:40:13Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | zanmato1984 | 1,061,004 | MDQ6VXNlcjEwNjEwMDQ= | User | false |
openclaw/openclaw | 4,095,934,003 | I_kwDOQb6kR870Iv4z | 49,878 | https://github.com/openclaw/openclaw/issues/49878 | https://api.github.com/repos/openclaw/openclaw/issues/49878 | Bug: Heartbeat session caches old HEARTBEAT.md content and doesn't re-read after update | ## Bug: Heartbeat session caches old HEARTBEAT.md content and doesn't re-read after update
### Description
When `HEARTBEAT.md` is updated, the heartbeat session doesn't re-read the file. It continues using the cached content from when the session started, causing old tasks to continue running even after `HEARTBEAT.md`... | closed | completed | true | 4 | [] | [] | 2026-03-18T15:24:52Z | 2026-05-01T04:41:39Z | 2026-04-28T21:40:20Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | smallmj | 23,742,160 | MDQ6VXNlcjIzNzQyMTYw | User | false |
openclaw/openclaw | 4,346,386,380 | I_kwDOQb6kR88AAAABAxCXzA | 73,813 | https://github.com/openclaw/openclaw/issues/73813 | https://api.github.com/repos/openclaw/openclaw/issues/73813 | [Bug] Cron job reports status 'ok' immediately after sessions_yield, before subagent completes | ## Summary
When a cron job with `sessionTarget: isolated` calls `sessions_yield` after spawning a subagent, the cron infrastructure immediately marks the job as `status: "ok"` and records a short `durationMs` (~25 seconds), even though the subagent continues running for several minutes and the parent session later res... | open | null | false | 2 | [] | [] | 2026-04-28T21:36:26Z | 2026-05-03T13:22:04Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | lunan-stellaraether | 266,393,738 | U_kgDOD-DYig | User | false |
openclaw/openclaw | 4,098,902,071 | I_kwDOQb6kR870UEg3 | 50,154 | https://github.com/openclaw/openclaw/issues/50154 | https://api.github.com/repos/openclaw/openclaw/issues/50154 | [Bug]: macOS LaunchAgent gateway fails to push to LAN GitLab over SSH, foreground gateway works | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
On macOS, the LaunchAgent-managed OpenClaw gateway cannot reliably access/push to a LAN-hosted GitLab over SSH, while a foreground gateway started manually with `openclaw gateway --port 18789 --verbose` works on the same machine.
### Steps... | closed | not_planned | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T02:45:23Z | 2026-05-01T04:41:38Z | 2026-04-28T21:40:26Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Jarvis-Zongao-Bian | 61,961,213 | MDQ6VXNlcjYxOTYxMjEz | User | false |
openclaw/openclaw | 4,099,261,191 | I_kwDOQb6kR870VcMH | 50,211 | https://github.com/openclaw/openclaw/issues/50211 | https://api.github.com/repos/openclaw/openclaw/issues/50211 | [Bug]: Shared CI regressions in loader, Feishu, and Google OAuth tests are failing unrelated PRs | ### Bug type
Regression (worked before, now fails)
### Summary
Recent upstream CI changes appear to have introduced shared test instability on unrelated PRs, with failures concentrated in src/plugins/loader.test.ts, extensions/feishu/src/{client,bot}.test.ts, extensions/google/oauth.test.ts, and Windows test shards.... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-19T04:45:22Z | 2026-05-01T04:41:37Z | 2026-04-28T21:40:34Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | hhhhao28 | 112,874,572 | U_kgDOBrpUTA | User | false |
openclaw/openclaw | 4,099,264,778 | I_kwDOQb6kR870VdEK | 50,213 | https://github.com/openclaw/openclaw/issues/50213 | https://api.github.com/repos/openclaw/openclaw/issues/50213 | [Feature]: add web search provider baseurl for gemini, so we could use Third-party GEMINI API provider | ### Summary
add web search provider baseurl for gemini, so we could use Third-party GEMINI API provider， for some reason, i can't use google original api
### Problem to solve
most of chinese user need proxy for google, so Third-party GEMINI API provider is necessary
### Proposed solution
"web": {
"search":... | closed | not_planned | true | 3 | [
"enhancement"
] | [] | 2026-03-19T04:46:55Z | 2026-05-01T04:41:37Z | 2026-04-28T21:40:41Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | TinyQAQ | 7,219,724 | MDQ6VXNlcjcyMTk3MjQ= | User | false |
openclaw/openclaw | 4,346,250,194 | I_kwDOQb6kR88AAAABAw6D0g | 73,801 | https://github.com/openclaw/openclaw/issues/73801 | https://api.github.com/repos/openclaw/openclaw/issues/73801 | Active Memory with Cerebras gpt-oss-120b times out and can pin gateway CPU | ## Summary
Active Memory reliably times out and can leave the local gateway CPU-bound when configured to use `cerebras/gpt-oss-120b`, even though the same Cerebras key/model works through direct `/v1/chat/completions` calls, including streaming and tool calls.
This looks related to reasoning-model handling inside the... | open | null | false | 1 | [] | [] | 2026-04-28T21:11:31Z | 2026-05-03T06:07:47Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | iannwu | 5,016,913 | MDQ6VXNlcjUwMTY5MTM= | User | false |
openclaw/openclaw | 4,102,080,168 | I_kwDOQb6kR870gMao | 50,468 | https://github.com/openclaw/openclaw/issues/50468 | https://api.github.com/repos/openclaw/openclaw/issues/50468 | [Bug]: Skills file watcher broken due to chokidar v5 glob incompatibility — installed skills not detected in existing sessions |
### Summary
`src/agents/skills/refresh.ts` → `resolveWatchTargets()` passes glob patterns (`*/SKILL.md`) to chokidar v5, which removed built-in glob support. The watcher silently registers nothing, `snapshotVersion` is never bumped, and existing sessions never refresh their skill snapshot after new skills are install... | closed | completed | true | 2 | [] | [] | 2026-03-19T14:24:03Z | 2026-05-01T04:41:36Z | 2026-04-28T21:40:51Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | ningzio | 45,959,659 | MDQ6VXNlcjQ1OTU5NjU5 | User | false |
openclaw/openclaw | 4,107,103,436 | I_kwDOQb6kR870zWzM | 50,989 | https://github.com/openclaw/openclaw/issues/50989 | https://api.github.com/repos/openclaw/openclaw/issues/50989 | Suppress intermediate text output before tool calls in group channels | ## Problem
In multi-agent group channels, LLMs frequently emit "thinking out loud" text before tool calls — messages like "Let me check...", "Looking into this...", "I should set up...". This text is the model's deliberate output (not a streaming artifact), so `blockStreamingDefault` cannot suppress it.
Each intermed... | closed | completed | true | 2 | [] | [] | 2026-03-20T10:33:03Z | 2026-05-01T04:41:35Z | 2026-04-28T21:40:58Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Asclepeion | 169,803,573 | U_kgDOCh7_NQ | User | false |
openclaw/openclaw | 4,346,352,514 | I_kwDOQb6kR88AAAABAxATgg | 73,811 | https://github.com/openclaw/openclaw/issues/73811 | https://api.github.com/repos/openclaw/openclaw/issues/73811 | [Bug]: sessions_spawn model parameter ignored due to write/read field mismatch | ## Description
When sessions_spawn is called with an explicit model parameter (e.g., "ollama/kimi-k2.6:cloud"), the subagent ignores the parameter and uses the agent's default model instead. The spawn-supplied model never reaches the inference path.
This is caused by a field-name mismatch: the spawn writer populates ... | closed | completed | true | 1 | [] | [] | 2026-04-28T21:29:49Z | 2026-05-01T04:41:32Z | 2026-04-28T21:41:43Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Hexy888 | 258,729,234 | U_kgDOD2vlEg | User | false |
openclaw/openclaw | 4,346,389,727 | I_kwDOQb6kR88AAAABAxCk3w | 73,815 | https://github.com/openclaw/openclaw/issues/73815 | https://api.github.com/repos/openclaw/openclaw/issues/73815 | OPENCLAW_PLUGIN_STAGE_DIR with two paths: memory status fails on missing chokidar in second path | OpenClaw version: 2026.4.26
Environment:
```text
OPENCLAW_PLUGIN_STAGE_DIR=/opt/openclaw/plugin-runtime-deps:/data/plugin-runtime-deps
```
Observed:
`openclaw status` shows:
```text
Memory: enabled (plugin memory-core) · unavailable
```
Running:
```bash
openclaw memory status
```
fails with:
```text
Cannot fin... | closed | completed | true | 1 | [] | [] | 2026-04-28T21:37:04Z | 2026-05-01T04:41:31Z | 2026-04-28T21:41:52Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | liorb-mountapps | 205,664,440 | U_kgDODEIwuA | User | false |
openclaw/openclaw | 4,346,416,907 | I_kwDOQb6kR88AAAABAxEPCw | 73,818 | https://github.com/openclaw/openclaw/issues/73818 | https://api.github.com/repos/openclaw/openclaw/issues/73818 | Add drag-and-drop and file-picker uploads to Glyph/WebChat | ## Summary
Add first-class file/image upload support to the Glyph/WebChat message composer, comparable to Slack or Telegram.
## User need
Users should be able to attach images, PDFs, and other files directly from the chat UI instead of needing an external channel or manual filesystem workaround.
Requested interacti... | closed | completed | true | 0 | [] | [] | 2026-04-28T21:42:32Z | 2026-05-01T04:41:30Z | 2026-04-28T21:43:13Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | ryosuzuki | 1,428,255 | MDQ6VXNlcjE0MjgyNTU= | User | false |
openclaw/openclaw | 4,346,425,036 | I_kwDOQb6kR88AAAABAxEuzA | 73,820 | https://github.com/openclaw/openclaw/issues/73820 | https://api.github.com/repos/openclaw/openclaw/issues/73820 | openai-codex provider sends empty input[] when context has only systemPrompt → ChatGPT 400 | ## Summary
When the `openai-codex` provider (api `openai-codex-responses`, baseUrl
`https://chatgpt.com/backend-api/codex`) is invoked with a context whose
`messages` array is empty but whose `systemPrompt` is set, the resulting
request body contains `input: []`. ChatGPT's codex backend rejects this with
a 400:
> One... | closed | completed | true | 1 | [] | [] | 2026-04-28T21:44:12Z | 2026-05-02T04:27:25Z | 2026-04-29T19:08:51Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | Gusty3055 | 262,864,832 | U_kgDOD6r_wA | User | false |
openclaw/openclaw | 4,346,423,316 | I_kwDOQb6kR88AAAABAxEoFA | 73,819 | https://github.com/openclaw/openclaw/issues/73819 | https://api.github.com/repos/openclaw/openclaw/issues/73819 | [Bug]: ESM import of chokidar fails in plugin runtime deps, causing embeddings "not ready" | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
The `chokidar` package files exist in `node_modules/chokidar/`, but the bare ESM specifier `import { watch } from 'chokidar'` fails to resolve. The module loader suggests `chokidar/index.js` instead.
This causes the entire ... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-28T21:43:52Z | 2026-05-01T04:41:30Z | 2026-04-28T21:48:40Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | taigoil | 93,786,041 | U_kgDOBZcPuQ | User | false |
openclaw/openclaw | 4,346,453,124 | I_kwDOQb6kR88AAAABAxGchA | 73,824 | https://github.com/openclaw/openclaw/issues/73824 | https://api.github.com/repos/openclaw/openclaw/issues/73824 | openclaw status does not show active Telegram channel because status command does not load plugin registry | OpenClaw version: 2026.4.26
Observed:
The agent responds on Telegram, but `openclaw status` shows an empty `Channels` table.
Live config has Telegram enabled:
```json
channels: {
telegram: {
enabled: true,
accounts: {
agent1: {
enabled: true,
dmPolicy: allowlist
}
}
}
}
`... | closed | completed | true | 1 | [] | [] | 2026-04-28T21:49:52Z | 2026-05-01T04:41:29Z | 2026-04-28T21:55:19Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | liorb-mountapps | 205,664,440 | U_kgDODEIwuA | User | false |
openclaw/openclaw | 4,346,503,795 | I_kwDOQb6kR88AAAABAxJicw | 73,827 | https://github.com/openclaw/openclaw/issues/73827 | https://api.github.com/repos/openclaw/openclaw/issues/73827 | slack: agents do not wake on <!subteam^...> user-group mentions | ## Summary
Mentioning a Slack user-group (subteam) does **not** wake any of the OpenClaw agents that belong to it. The Slack adapter's mention preflight only recognizes direct `<@UXXX>` tokens against each bot's own user id; it has no awareness of `<!subteam^SXXX>` tokens at all. So a message like:
```
<!subteam^S0B0... | closed | completed | true | 3 | [] | [] | 2026-04-28T21:59:25Z | 2026-05-04T04:34:09Z | 2026-05-02T03:46:39Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | CG-Intelligence-Agent-Jack | 277,066,800 | U_kgDOEIO0MA | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.