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 | 4,077,818,288 | I_kwDOQb6kR87zDpGw | 47,021 | https://github.com/openclaw/openclaw/issues/47021 | https://api.github.com/repos/openclaw/openclaw/issues/47021 | [Bug]: Telegram plugin fails to load after refactor: Cannot find module '../../../src/infra/outbound/deliver.js' | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
After cloning the repo and building the Docker image locally, the Telegram plugin fails to load on every gateway start with the following error:
[plugins] telegram failed to load from /app/extensions/telegram/index.ts: Error: Cannot find mo... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T07:26:01Z | 2026-04-27T04:40:41Z | 2026-04-25T02:23:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | munyua-arch | 78,699,306 | MDQ6VXNlcjc4Njk5MzA2 | User | false |
openclaw/openclaw | 4,077,835,776 | I_kwDOQb6kR87zDtYA | 47,032 | https://github.com/openclaw/openclaw/issues/47032 | https://api.github.com/repos/openclaw/openclaw/issues/47032 | [Bug]: 使用 `[DINGTALK_FILE]` 标记发送文件时，文件路径中的中文字符后被错误插入空格，导致路径被篡改，文件无法发送。 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
openclaw集成钉钉，从钉钉机器人发文件到对话出现文件名错误
### Steps to reproduce
1. Windows 系统
2. 文件路径包含中文字符（如：`D:/test/test实施路径汇报 -202512281000.pptx`）
3. 使用 `[DINGTALK_FILE]` 标记发送
### Expected behavior
路径应保持原样，不添加额外字符。
### Actual behavior
路径变成：`D:/test/test实... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T07:35:33Z | 2026-04-27T04:40:41Z | 2026-04-25T02:23:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | billowjen | 43,055,672 | MDQ6VXNlcjQzMDU1Njcy | User | false |
openclaw/openclaw | 4,326,445,964 | I_kwDOQb6kR88AAAABAeBTjA | 71,315 | https://github.com/openclaw/openclaw/issues/71315 | https://api.github.com/repos/openclaw/openclaw/issues/71315 | Footgun: bare-name allowlist patterns are silently ignored | # Footgun: bare-name allowlist patterns are silently ignored
**OpenClaw version:** `2026.4.23` (homebrew install)
**Repro on:** macOS 14, single-user setup, kimi (non-main) agent
## Summary
`exec-approvals.json` lets you write bare-name allowlist patterns like `"gbrain"`, `"ls"`, `"bash"`. They're accepted at config... | closed | completed | true | 3 | [] | [] | 2026-04-25T00:47:25Z | 2026-04-27T04:29:26Z | 2026-04-25T03:18:40Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kengwei | 627,500 | MDQ6VXNlcjYyNzUwMA== | User | false |
openclaw/openclaw | 4,077,853,539 | I_kwDOQb6kR87zDxtj | 47,042 | https://github.com/openclaw/openclaw/issues/47042 | https://api.github.com/repos/openclaw/openclaw/issues/47042 | auto-extract-learnings.sh fails silently: uses non-existent --title/--content flags for mem-store.sh | ## Bug
`scripts/auto-extract-learnings.sh` calls `mem-store.sh` with `--title` and `--content` flags, which don't exist in `mem-store.sh`. This causes **every auto-extraction attempt to fail silently** with `Unknown arg: --title`.
## Impact
The 15-minute cron sweep (`auto-extract-learnings.sh --sweep`) has been prod... | closed | not_planned | true | 1 | [] | [] | 2026-03-15T07:46:23Z | 2026-04-27T04:40:40Z | 2026-04-25T02:24:09Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | bharatindia123-a11y | 252,865,299 | U_kgDODxJrEw | User | false |
openclaw/openclaw | 4,176,978,233 | I_kwDOQb6kR87496E5 | 58,231 | https://github.com/openclaw/openclaw/issues/58231 | https://api.github.com/repos/openclaw/openclaw/issues/58231 | Telnyx inbound calls reach webhook but are never answered in voice-call plugin | Hi — I’m testing the bundled `@openclaw/voice-call` plugin with **Telnyx** and could use a sanity check on whether inbound answering is currently supported in stable, or whether I’m missing a required Telnyx/OpenClaw setting.
## Environment
- OpenClaw version: `2026.3.28` initially, then updated after testing
- Instal... | closed | completed | true | 7 | [
"bug",
"channel: voice-call"
] | [] | 2026-03-31T08:09:14Z | 2026-04-27T04:40:39Z | 2026-04-25T02:24:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | KonsultDigital | 65,322,345 | MDQ6VXNlcjY1MzIyMzQ1 | User | false |
openclaw/openclaw | 4,041,816,992 | I_kwDOQb6kR87w6Tug | 40,131 | https://github.com/openclaw/openclaw/issues/40131 | https://api.github.com/repos/openclaw/openclaw/issues/40131 | [Bug]: Bundled voice-call plugin (v2026.3.2) returns 200 OK on Telnyx webhooks but does not process inbound calls | ### Bug type
Regression (worked before, now fails)
### Summary
Inbound calls to a Telnyx number trigger webhooks that reach the bundled voice-call webhook server (confirmed via Caddy reverse proxy debug logs showing telnyx-webhooks User-Agent, HTTP 200 response). However, the plugin does not log anything in the Gate... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-08T17:49:32Z | 2026-04-27T04:40:38Z | 2026-04-25T02:24:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | polygonwindow | 89,137 | MDQ6VXNlcjg5MTM3 | User | false |
openclaw/openclaw | 4,077,865,992 | I_kwDOQb6kR87zD0wI | 47,050 | https://github.com/openclaw/openclaw/issues/47050 | https://api.github.com/repos/openclaw/openclaw/issues/47050 | Fix: Add TTL/Timeout garbage collection to ChatRunRegistry to prevent Gateway OOM | ## ?? Problem Description
In the current Gateway implementation, incoming channel messages create new sessions pushed into an in-memory ChatRunRegistry Map. Sessions are **only removed** when explicit done/error events are emitted.
**Critical Issue**: If an agent hangs indefinitely (API rate limit, silent connection ... | closed | completed | true | 3 | [] | [] | 2026-03-15T07:55:26Z | 2026-04-27T04:40:38Z | 2026-04-25T02:24:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhangxue1985122219 | 54,266,548 | MDQ6VXNlcjU0MjY2NTQ4 | User | false |
openclaw/openclaw | 4,077,866,513 | I_kwDOQb6kR87zD04R | 47,052 | https://github.com/openclaw/openclaw/issues/47052 | https://api.github.com/repos/openclaw/openclaw/issues/47052 | Security: Sandbox external channel inputs to prevent prompt injection via agentToAgent | ## ?? Security Vulnerability
**Type**: Prompt Injection ? Privilege Escalation
**Severity**: ?? **High** (potential remote code execution via privileged sub-agents)
---
## ?? Problem Description
External channel messages (from Webhooks like Feishu, Telegram, Slack) are **concatenated directly** into agent context... | closed | completed | true | 1 | [] | [] | 2026-03-15T07:55:53Z | 2026-04-27T04:40:36Z | 2026-04-25T02:24:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhangxue1985122219 | 54,266,548 | MDQ6VXNlcjU0MjY2NTQ4 | User | false |
openclaw/openclaw | 4,077,866,552 | I_kwDOQb6kR87zD044 | 47,053 | https://github.com/openclaw/openclaw/issues/47053 | https://api.github.com/repos/openclaw/openclaw/issues/47053 | tts.test.ts mock missing resolveModelAsync — Windows CI fails | ### Bug type
Regression (worked before, now fails)
### Summary
Commit 8e4a1d87e (#45824) switched `tts-core.ts` from `resolveModel` (sync) to `resolveModelAsync` but did not update the `vi.mock` in `tts.test.ts`. The mock only exports `resolveModel`, so vitest throws:
```
No "resolveModelAsync" export is defined on... | closed | completed | true | 1 | [] | [] | 2026-03-15T07:55:56Z | 2026-04-27T04:40:35Z | 2026-04-25T02:24:59Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jackjin1997 | 28,055,621 | MDQ6VXNlcjI4MDU1NjIx | User | false |
openclaw/openclaw | 4,077,875,239 | I_kwDOQb6kR87zD3An | 47,062 | https://github.com/openclaw/openclaw/issues/47062 | https://api.github.com/repos/openclaw/openclaw/issues/47062 | [Bug]: Gateway fails to start with tls.enabled: false when tailscale.mode: serve is active | ### Bug type
Crash (process/app exits or hangs)
### Summary
tls.enabled: false crashes the container when tailscale.mode: "serve"
tls.enabled: true is silently ignored — gateway serves HTTP regardless, but CLI assumes WSS → breaks all CLI-based managemen
When using Tailscale Serve (gateway.tailscale.mode: "serve") ... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-03-15T08:01:42Z | 2026-04-27T04:40:35Z | 2026-04-25T02:25:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SmartALB | 139,439,933 | U_kgDOCE-vPQ | User | false |
openclaw/openclaw | 4,077,879,376 | I_kwDOQb6kR87zD4BQ | 47,064 | https://github.com/openclaw/openclaw/issues/47064 | https://api.github.com/repos/openclaw/openclaw/issues/47064 | [Bug]: 模型调用API-KEY 环境变量语法配置不一致 | ### Bug type
Regression (worked before, now fails)
### Summary
--openclaw.json 和 models.json - 由 Gateway/Agents 核心组件读取，使用 env:default:VAR_NAME 语法
- auth-profiles.json - 有独立的认证模块解析，有自己的 keyRef 结构，会自动补全 env:default: 前缀
用户体验比较差！
### Steps to reproduce
条件： Windows 11 原生安装openclaw（未通过wsl2和docker安装方式）
1、在windows11系... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-15T08:04:21Z | 2026-04-27T04:40:32Z | 2026-04-25T02:25:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | David2026518 | 263,960,152 | U_kgDOD7u2WA | User | false |
openclaw/openclaw | 4,077,885,539 | I_kwDOQb6kR87zD5hj | 47,072 | https://github.com/openclaw/openclaw/issues/47072 | https://api.github.com/repos/openclaw/openclaw/issues/47072 | [Bug]: | ### Bug type
Regression (worked before, now fails)
### Summary
[Bug] Cron job with Asia/Shanghai timezone executes 8 hours late (configured 03:00, actual 11:00)
### Steps to reproduce
## Bug Description
Cron job with schedule `0 3 * * *` and timezone `Asia/Shanghai` executes at 11:00 instead of 03:00 Beijing time.... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-15T08:08:39Z | 2026-04-27T04:40:31Z | 2026-04-25T02:25:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sunruimx-missbear | 249,883,269 | U_kgDODuTqhQ | User | false |
openclaw/openclaw | 4,077,898,606 | I_kwDOQb6kR87zD8tu | 47,079 | https://github.com/openclaw/openclaw/issues/47079 | https://api.github.com/repos/openclaw/openclaw/issues/47079 | [Bug]: Mistral API returns 422 - openai-completions adapter sends max_completion_tokens which Mistral rejects | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
## Bug Description
When using Mistral as a provider with the `openai-completions` API adapter, all agent requests fail with `422 status code (no body)`.
## Root Cause
The `openai-completions` adapter sends `max_completion_tokens` in the ... | closed | completed | true | 10 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T08:17:41Z | 2026-04-27T04:40:31Z | 2026-04-25T02:26:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | modern-sapien | 58,672,259 | MDQ6VXNlcjU4NjcyMjU5 | User | false |
openclaw/openclaw | 4,077,899,233 | I_kwDOQb6kR87zD83h | 47,081 | https://github.com/openclaw/openclaw/issues/47081 | https://api.github.com/repos/openclaw/openclaw/issues/47081 | 飞书 channel 断开/重启时 session 上下文丢失 | ## 问题描述
飞书 channel 断开或重启时，session 上下文会丢失，AI 出现"失忆"现象。
## 复现步骤
1. 通过飞书与 AI 进行多轮对话
2. 修改飞书配置（如 groupAllowFrom、groupPolicy 等）触发 channel 重启
3. 或者飞书 WebSocket 连接断开（code=1001）
4. 再次发送消息时，AI 无法回忆之前的对话内容
## 日志证据
```
# 昨晚发生的 WebSocket 断开：
2026-03-14T21:20:19.250+08:00 webchat disconnected code=1001
2026-03-14T22:34:54.412+... | closed | completed | true | 1 | [] | [] | 2026-03-15T08:18:11Z | 2026-04-27T04:40:30Z | 2026-04-25T02:26:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jiangsenhan0405-lgtm | 265,287,804 | U_kgDOD8_4fA | User | false |
openclaw/openclaw | 4,077,902,122 | I_kwDOQb6kR87zD9kq | 47,086 | https://github.com/openclaw/openclaw/issues/47086 | https://api.github.com/repos/openclaw/openclaw/issues/47086 | Telegram streaming race condition causes message_start before previous message_stop | **Describe the bug**
When using Telegram chat and sending multiple messages quickly in the same session, the OpenClaw gateway intermittently throws:
```
Unexpected event order, got message_start before receiving "message_stop"
```
This suggests a race condition in the gateway's streaming lifecycle handling. A new `m... | closed | completed | true | 4 | [] | [] | 2026-03-15T08:20:11Z | 2026-04-27T04:40:28Z | 2026-04-25T02:26:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yangfei222666-9 | 261,852,489 | U_kgDOD5uNSQ | User | false |
openclaw/openclaw | 4,326,524,439 | I_kwDOQb6kR88AAAABAeGGFw | 71,326 | https://github.com/openclaw/openclaw/issues/71326 | https://api.github.com/repos/openclaw/openclaw/issues/71326 | Cross-exec stale file reads (cross-process vnode/dentry cache race) — regression in 2026.4.20 | ## Summary
Cross-exec-call file reads return stale content — same path, identical `Path.read_text()` (or `cat`), different results from one exec call to the next, with no writer in between. Began appearing immediately after upgrading to **2026.4.20** on 2026-04-20. Not observed in 25+ days of prior operation on the sa... | open | null | false | 1 | [] | [] | 2026-04-25T01:15:13Z | 2026-05-03T06:07:32Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | twoshotscrbot | 269,910,159 | U_kgDOEBaAjw | User | false |
openclaw/openclaw | 4,077,915,806 | I_kwDOQb6kR87zEA6e | 47,088 | https://github.com/openclaw/openclaw/issues/47088 | https://api.github.com/repos/openclaw/openclaw/issues/47088 | [Bugfix]: [DEP0040] DeprecationWarning: The `punycode` module is deprecated. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When my OpenClaw gateway starts, the log always prints “[DEP0040] DeprecationWarning: The `punycode` module is deprecated”. According to the instructions in `punycode.js\README.md`, I changed all places using the `require("punycode")` call ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T08:26:51Z | 2026-04-27T04:40:28Z | 2026-04-25T02:26:55Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Far-cloind | 80,151,472 | MDQ6VXNlcjgwMTUxNDcy | User | false |
openclaw/openclaw | 4,077,951,793 | I_kwDOQb6kR87zEJsx | 47,103 | https://github.com/openclaw/openclaw/issues/47103 | https://api.github.com/repos/openclaw/openclaw/issues/47103 | [Bug]: openclaw devices list --> Error，[openclaw] Failed to start CLI: Error: gateway closed (1000 normal closure): no close reason | ### Bug type
Regression (worked before, now fails)
### Summary
Can't run command openclaw devices list/approve, show error as below.
[openclaw] Failed to start CLI: Error: gateway closed (1000 normal closure): no close reason
### Steps to reproduce
1. Run [openclaw devices list]
2. Show [ 🦞 OpenClaw 2026.3.13 (... | closed | completed | true | 9 | [
"bug",
"regression"
] | [] | 2026-03-15T08:49:17Z | 2026-04-27T04:40:27Z | 2026-04-25T02:27:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | laogiant | 261,174,351 | U_kgDOD5E0Tw | User | false |
openclaw/openclaw | 4,077,980,436 | I_kwDOQb6kR87zEQsU | 47,118 | https://github.com/openclaw/openclaw/issues/47118 | https://api.github.com/repos/openclaw/openclaw/issues/47118 | Feature request: include auth profile identifier in model fallback logs (sanitized) | ## Summary
When provider/model fallback occurs, logs currently show provider/model transitions and errors but not which auth profile was attempted. This makes troubleshooting fallback behavior difficult when multiple auth profiles are configured.
## Requested behavior
Please include a **sanitized profile identifier if... | closed | completed | true | 1 | [] | [] | 2026-03-15T09:04:43Z | 2026-04-27T04:40:26Z | 2026-04-25T02:27:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | DeadlySilent | 3,342,508 | MDQ6VXNlcjMzNDI1MDg= | User | false |
openclaw/openclaw | 4,077,989,440 | I_kwDOQb6kR87zES5A | 47,123 | https://github.com/openclaw/openclaw/issues/47123 | https://api.github.com/repos/openclaw/openclaw/issues/47123 | Webchat: messages silently truncated at ~4000 chars with no config override | ## Bug Description
Long assistant responses in the webchat control-UI are silently truncated at approximately 4000 characters. The content streams correctly during generation, but the final rendered message is cut off. There is no error, no indication of truncation, and no configuration option to change this limit.
#... | closed | completed | true | 1 | [] | [] | 2026-03-15T09:10:27Z | 2026-04-27T04:40:26Z | 2026-04-25T02:27:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | navendugoyal19 | 132,033,603 | U_kgDOB96sQw | User | false |
openclaw/openclaw | 4,225,898,350 | I_kwDOQb6kR8774hdu | 63,231 | https://github.com/openclaw/openclaw/issues/63231 | https://api.github.com/repos/openclaw/openclaw/issues/63231 | [Bug]: v2026.4.8, [gateway-http] unhandled error in request handler: Error: Cannot find module '@buape/carbon' | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
Yes
### Summary
After upgrading from 2026.4.5 to 2026.4.8, gateway says `[gateway-http] unhandled error in request handler: Erro`
.
### Steps to reproduce
```
npm uninstall -g openclaw
npm install -g openclaw
openclaw gateway uninstall... | closed | duplicate | true | 2 | [
"bug"
] | [] | 2026-04-08T16:03:05Z | 2026-04-27T04:40:24Z | 2026-04-25T02:27:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Honghe | 1,092,722 | MDQ6VXNlcjEwOTI3MjI= | User | false |
openclaw/openclaw | 4,232,784,589 | I_kwDOQb6kR878SyrN | 63,774 | https://github.com/openclaw/openclaw/issues/63774 | https://api.github.com/repos/openclaw/openclaw/issues/63774 | [Bug]: Discord channel missing from bundled sidecar fix in 2026.4.8/2026.4.9 — breaks on Homebrew install | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Bug type: Regression (worked before, now fails)
Summary: The 2026.4.8 release fixed bundled channel sidecar loading for BlueBubbles, Feishu, Google Chat, IRC, Matrix, Mattermost, Microsoft Teams, Nextcloud Talk, Slack, and Z... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-09T14:00:39Z | 2026-04-27T04:40:23Z | 2026-04-25T02:27:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | virtualserverguy | 2,394,687 | MDQ6VXNlcjIzOTQ2ODc= | User | false |
openclaw/openclaw | 4,077,992,931 | I_kwDOQb6kR87zETvj | 47,126 | https://github.com/openclaw/openclaw/issues/47126 | https://api.github.com/repos/openclaw/openclaw/issues/47126 | Bug: Telegram reaction system events stop arriving after multiple gateway restarts (SIGUSR1) | ## Summary
Telegram reaction system events (e.g., `Telegram reaction added: ❤️ by Anton on msg 6556`) stop being delivered to the agent session after multiple consecutive gateway restarts via SIGUSR1.
## Steps to reproduce
1. Config: `channels.telegram.reactionNotifications: "own"` — working correctly
2. Reactions a... | closed | completed | true | 2 | [] | [] | 2026-03-15T09:12:45Z | 2026-04-27T04:40:23Z | 2026-04-25T02:28:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | drakulavich | 129,260 | MDQ6VXNlcjEyOTI2MA== | User | false |
openclaw/openclaw | 4,078,017,895 | I_kwDOQb6kR87zEZ1n | 47,139 | https://github.com/openclaw/openclaw/issues/47139 | https://api.github.com/repos/openclaw/openclaw/issues/47139 | [Bug] Control Panel shows workspace skills as bundled skills | ### Bug type
Regression (worked before, now fails)
### Summary
Control Panel Web UI incorrectly shows some skills as "bundled" (built-in) when they are actually installed in workspace directory
### Steps to reproduce
1. Install a skill from ClawHub that originally came bundled with OpenClaw (e.g., summarize, nano-... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-03-15T09:27:50Z | 2026-04-27T04:40:22Z | 2026-04-25T02:28:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Lrs-GreenBottle | 253,340,193 | U_kgDODxmqIQ | User | false |
openclaw/openclaw | 4,227,887,863 | I_kwDOQb6kR878AHL3 | 63,387 | https://github.com/openclaw/openclaw/issues/63387 | https://api.github.com/repos/openclaw/openclaw/issues/63387 | [Bug]: Gateway/provider race: stale heartbeat reconnect callback throws after disconnect | ## Summary
I hit an internal invariant failure in the gateway/provider websocket heartbeat logic:
`Attempted to reconnect zombie connection after disconnecting first (this shouldn't be possible)`
This appears to be a race where a stale heartbeat reconnect callback fires after the websocket/session has already been c... | closed | duplicate | true | 1 | [] | [] | 2026-04-08T21:52:19Z | 2026-04-27T04:40:20Z | 2026-04-25T02:28:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Megaplex1 | 103,837,711 | U_kgDOBjBwDw | User | false |
openclaw/openclaw | 4,078,023,749 | I_kwDOQb6kR87zEbRF | 47,142 | https://github.com/openclaw/openclaw/issues/47142 | https://api.github.com/repos/openclaw/openclaw/issues/47142 | Feature: Graceful gateway self-restart — prevent zombie processes when agent calls `openclaw gateway restart` | ## Problem
When an OpenClaw agent (or any script running inside the gateway process) calls `openclaw gateway restart`, the command spawns a **new gateway process without stopping the old one**. This creates zombie gateway processes that compete for shared resources.
### Real-world incident (2026-03-14)
Our agent cal... | closed | completed | true | 1 | [] | [] | 2026-03-15T09:31:41Z | 2026-04-27T04:40:19Z | 2026-04-25T02:28:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | braminatorh | 263,757,029 | U_kgDOD7ic5Q | User | false |
openclaw/openclaw | 4,078,028,397 | I_kwDOQb6kR87zEcZt | 47,144 | https://github.com/openclaw/openclaw/issues/47144 | https://api.github.com/repos/openclaw/openclaw/issues/47144 | [bug]：WWeb UI Conversation Display Error | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
There is an issue with the web UI. When clicking on a chat, it should default to displaying the main conversation. However, the main conversation is shown, while the conversation selector displays a different one, such as my Telegram chat. ... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T09:34:19Z | 2026-04-27T04:40:19Z | 2026-04-25T02:28:50Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Maziyang2 | 104,501,944 | U_kgDOBjqSuA | User | false |
openclaw/openclaw | 4,078,029,543 | I_kwDOQb6kR87zEcrn | 47,145 | https://github.com/openclaw/openclaw/issues/47145 | https://api.github.com/repos/openclaw/openclaw/issues/47145 | test(tts): tts.test.ts mock missing resolveModelAsync — 6 tests fail since #45824 | ## Summary
`src/tts/tts.test.ts` has 6 failing tests introduced by #45824. The PR updated `tts-core.ts` to call `resolveModelAsync` (from `pi-embedded-runner/model.ts`) but did not update the `vi.mock` for that module in `tts.test.ts`, so the mock throws at runtime.
## Failing tests
All 6 failures are in the `summar... | closed | completed | true | 2 | [] | [] | 2026-03-15T09:35:06Z | 2026-04-27T04:40:18Z | 2026-04-25T02:29:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Arry8 | 263,019,747 | U_kgDOD61c4w | User | false |
openclaw/openclaw | 4,078,043,069 | I_kwDOQb6kR87zEf-9 | 47,154 | https://github.com/openclaw/openclaw/issues/47154 | https://api.github.com/repos/openclaw/openclaw/issues/47154 | [WhatsApp] Agent message tool sends on stale socket while auto-reply uses live socket | ## Problem
When the agent's `message` tool sends a WhatsApp message, it fails with "No active WhatsApp Web listener" while the auto-reply path succeeds at the **exact same time**. The health-monitor detects stale sockets and restarts, but the `message` tool's send path still holds a reference to the dead socket.
## E... | closed | completed | true | 3 | [] | [] | 2026-03-15T09:43:42Z | 2026-04-27T04:40:17Z | 2026-04-25T02:29:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kimjcodes | 258,650,695 | U_kgDOD2qyRw | User | false |
openclaw/openclaw | 4,078,102,028 | I_kwDOQb6kR87zEuYM | 47,188 | https://github.com/openclaw/openclaw/issues/47188 | https://api.github.com/repos/openclaw/openclaw/issues/47188 | [Bug]: Webchat streaming renders duplicate bubbles instead of updating in-place | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to 2026.3.13, streaming responses in webchat (Control UI) create multiple duplicate message bubbles with growing text, instead of updating a single message in-place. The complete message only appears after many duplicates.
### Steps to r... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-15T10:22:34Z | 2026-04-27T04:40:16Z | 2026-04-25T02:29:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | gusigui201 | 267,826,102 | U_kgDOD_aztg | User | false |
openclaw/openclaw | 4,211,200,866 | I_kwDOQb6kR877AdNi | 61,826 | https://github.com/openclaw/openclaw/issues/61826 | https://api.github.com/repos/openclaw/openclaw/issues/61826 | openai-completions sends 'store' param that breaks non-OpenAI endpoints (e.g. Google Gemma) | ## Problem
When using `openai-completions` API type with a non-OpenAI endpoint (e.g., Google's OpenAI-compatible endpoint at `generativelanguage.googleapis.com/v1beta/openai`), OpenClaw sends a `store: false` parameter in the request body.
This is added automatically by the OpenAI JS SDK (`OpenAI/JS 6.26.0`), but Goo... | closed | completed | true | 1 | [] | [] | 2026-04-06T10:56:29Z | 2026-04-27T04:40:15Z | 2026-04-25T02:29:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kylin19860916 | 229,595,532 | U_kgDODa9ZjA | User | false |
openclaw/openclaw | 4,078,121,677 | I_kwDOQb6kR87zEzLN | 47,200 | https://github.com/openclaw/openclaw/issues/47200 | https://api.github.com/repos/openclaw/openclaw/issues/47200 | [Bug]: Telegram channel broken when using Mistral API | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
I just installed OpenClaw the first time using Mistral. When I talk to the gateway via telegram I get a `422 status code (no body)` response message
### Steps to reproduce
1. Setup OpenClaw with Mistral
2. Configure Telegram channel
3. Se... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T10:36:21Z | 2026-04-27T04:40:14Z | 2026-04-25T02:29:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cozis | 80,131,537 | MDQ6VXNlcjgwMTMxNTM3 | User | false |
openclaw/openclaw | 4,078,141,831 | I_kwDOQb6kR87zE4GH | 47,211 | https://github.com/openclaw/openclaw/issues/47211 | https://api.github.com/repos/openclaw/openclaw/issues/47211 | [Bug] Assistant 消息出现临时重复现象 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Assistant messages are temporarily duplicated in session files during streaming responses
### Steps to reproduce
1. Send a message to OpenClaw assistant via webchat or control-ui
2. Observe the assistant's response during generation
3. Ch... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T10:51:19Z | 2026-04-27T04:40:13Z | 2026-04-25T02:30:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Harry-y-j | 162,933,598 | U_kgDOCbYrXg | User | false |
openclaw/openclaw | 4,078,145,635 | I_kwDOQb6kR87zE5Bj | 47,213 | https://github.com/openclaw/openclaw/issues/47213 | https://api.github.com/repos/openclaw/openclaw/issues/47213 | [Feature Request]: Add Strict Agentic Mode (Reject Non-Actionable / Empty-Action Outputs) | # [Feature Request]: Add Strict Agentic Mode (Reject Non-Actionable / Empty-Action Outputs)
## Summary
OpenClaw is already strong for general-purpose agent workflows, but in industrial / execution-heavy environments there is still a costly failure mode:
**the model emits a conversational ACK instead of an actionable... | closed | completed | true | 1 | [] | [] | 2026-03-15T10:54:14Z | 2026-04-27T04:40:12Z | 2026-04-25T02:30:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SennTse | 120,245,552 | U_kgDOByrNMA | User | false |
openclaw/openclaw | 4,078,190,552 | I_kwDOQb6kR87zFD_Y | 47,236 | https://github.com/openclaw/openclaw/issues/47236 | https://api.github.com/repos/openclaw/openclaw/issues/47236 | [Bug] /compact command succeeds but does not actually compact session in Control UI | ## Bug Report: `/compact` command succeeds but does not actually compact session
**Environment:**
- OpenClaw version: 2026.3.13 (61d171a)
- Platform: macOS (Darwin 22.6.0)
- Channel: webchat (Control UI)
- Session: agent:main:main
**Description:**
When executing `/compact` command in the Control UI (webchat), the co... | closed | completed | true | 2 | [] | [] | 2026-03-15T11:24:16Z | 2026-04-27T04:40:11Z | 2026-04-25T02:30:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Bright-Chengliang | 78,144,620 | MDQ6VXNlcjc4MTQ0NjIw | User | false |
openclaw/openclaw | 4,326,521,920 | I_kwDOQb6kR88AAAABAeF8QA | 71,323 | https://github.com/openclaw/openclaw/issues/71323 | https://api.github.com/repos/openclaw/openclaw/issues/71323 | Skill validator ignores metadata.clawdbot.requires.bins → 24+ skills falsely reported as eligible | ## Problem
`openclaw 2026.4.23` (a979721) reports skills as `eligible: true` even when they declare required binaries (`metadata.requires.bins`) that aren't installed on the system. The skill validator **only honors the `metadata.openclaw.*` namespace and silently ignores `metadata.clawdbot.*`**, which is the legacy k... | closed | completed | true | 1 | [] | [] | 2026-04-25T01:14:28Z | 2026-04-27T04:40:10Z | 2026-04-25T02:30:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | christianoaugusto | 118,463,400 | U_kgDOBw-bqA | User | false |
openclaw/openclaw | 4,078,196,307 | I_kwDOQb6kR87zFFZT | 47,238 | https://github.com/openclaw/openclaw/issues/47238 | https://api.github.com/repos/openclaw/openclaw/issues/47238 | Per-thread agent routing for Telegram DM threads | ### Summary
Telegram DM threads (created via message_thread_id) currently all route to the same agent as the parent DM. This extends per-thread agent routing — already working for group forum topics — to Telegram DMs.
### Problem to solve
With threaded conversations enabled in Telegram DMs (channels.telegram.thread... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-15T11:28:27Z | 2026-04-27T04:40:10Z | 2026-04-25T02:31:08Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lazarmarius249-sketch | 265,618,909 | U_kgDOD9UF3Q | User | false |
openclaw/openclaw | 4,078,201,233 | I_kwDOQb6kR87zFGmR | 47,241 | https://github.com/openclaw/openclaw/issues/47241 | https://api.github.com/repos/openclaw/openclaw/issues/47241 | [Feature]: Sub Agents should know if a chat message is from a Human or my main AI agent | ### Summary
Thanks for adding more detailed drop down UI in the OpenClaw webchat that allows to talk to different sub agents. But right now the sub agent don't know if I am the human or if this is my main agent. They should see that as a prefix or some clear indication.
### Problem to solve
Sub Agents should know if... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-15T11:31:56Z | 2026-04-27T04:40:09Z | 2026-04-25T02:31:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | albusdumbledoremolty | 258,898,131 | U_kgDOD2540w | User | false |
openclaw/openclaw | 4,078,232,079 | I_kwDOQb6kR87zFOIP | 47,259 | https://github.com/openclaw/openclaw/issues/47259 | https://api.github.com/repos/openclaw/openclaw/issues/47259 | [Feature]: Feature Request: 强制知识库检查 (Force Knowledge Base Only Mode) | ### Summary
## 背景
当前 AI Agent 可以自由使用外部工具或自己的知识库来回答问题，无法强制限制只能从知识库获取信息。
## 需求
希望添加一个配置选项，实现：
1. 只能从飞书知识库/云文档获取信息
2. 知识库没有的内容，统一回复「我不知道，请联系XXX」
3. 禁止调用任何外部工具（web_search, web_fetch, weather 等）
4. 禁止用模型自身知识库回答
### Problem to solve
## 期望的实现方式
在 agent 配置中添加类似 `knowledgeBaseOnly: true` 的选项，或者通过预处理脚本来实现强制检查。
### Propo... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-15T11:53:27Z | 2026-04-27T04:40:08Z | 2026-04-25T02:31:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | wtjwrold | 35,794,562 | MDQ6VXNlcjM1Nzk0NTYy | User | false |
openclaw/openclaw | 4,078,234,918 | I_kwDOQb6kR87zFO0m | 47,260 | https://github.com/openclaw/openclaw/issues/47260 | https://api.github.com/repos/openclaw/openclaw/issues/47260 | Extensions fail to load after Docker build: send-deps.ts missing from runtime image | ### Bug type
Regression (worked before, now fails)
### Summary
After building the Docker image from source, both Discord bots fail to load with `Error: Cannot find module '/app/src/infra/outbound/send-deps.js'` because the Dockerfile omits the `src/` directory from the runtime image.
### Steps to reproduce
1. Clon... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-15T11:55:31Z | 2026-04-27T04:40:08Z | 2026-04-25T02:31:57Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | domlen | 2,361,464 | MDQ6VXNlcjIzNjE0NjQ= | User | false |
openclaw/openclaw | 4,078,235,456 | I_kwDOQb6kR87zFO9A | 47,262 | https://github.com/openclaw/openclaw/issues/47262 | https://api.github.com/repos/openclaw/openclaw/issues/47262 | Telegram long replies intermittently terminate (embedded_run_agent_end: terminated) and may resend | ## Title
Telegram long-message replies intermittently terminate (`embedded_run_agent_end: terminated`) and may duplicate/retry-send
## Version / Environment
- OpenClaw: 2026.3.13 (also observed after 2026.3.12)
- Host: macOS arm64
- Channel: Telegram DM
- Model observed: `openai-codex/gpt-5.3-codex` (also seen once on... | closed | completed | true | 2 | [] | [] | 2026-03-15T11:55:53Z | 2026-04-27T04:40:07Z | 2026-04-25T02:32:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | MCJhou | 268,355,853 | U_kgDOD_7JDQ | User | false |
openclaw/openclaw | 4,177,372,625 | I_kwDOQb6kR874_aXR | 58,260 | https://github.com/openclaw/openclaw/issues/58260 | https://api.github.com/repos/openclaw/openclaw/issues/58260 | Discord channels.discord.proxy does not cover Carbon REST calls — deploy commands and bot identity fetch bypass proxy | # Discord `channels.discord.proxy` does not cover Carbon REST calls — deploy commands and bot identity fetch bypass proxy
## Summary
`channels.discord.proxy` creates a proxied fetcher for Discord gateway WebSocket traffic and some startup REST lookups, but the Carbon framework's `RequestClient` (used for deploy-comma... | closed | completed | true | 1 | [] | [] | 2026-03-31T09:01:05Z | 2026-04-27T04:40:06Z | 2026-04-25T02:31:56Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pedh | 11,317,842 | MDQ6VXNlcjExMzE3ODQy | User | false |
openclaw/openclaw | 4,078,239,169 | I_kwDOQb6kR87zFP3B | 47,265 | https://github.com/openclaw/openclaw/issues/47265 | https://api.github.com/repos/openclaw/openclaw/issues/47265 | WebSocket CLI connection fails after upgrade to 2026.3.13 — handshake timeout on loopback | ## Summary
After upgrading from OpenClaw 2026.3.8 to 2026.3.13, all CLI commands that use the WebSocket RPC connection to the gateway fail with `gateway closed (1000 normal closure)`. The gateway itself is healthy — HTTP health endpoint, Telegram, and Discord channels all work normally. Only the local CLI → gateway We... | closed | completed | true | 10 | [] | [] | 2026-03-15T11:58:38Z | 2026-04-27T04:40:05Z | 2026-04-25T02:32:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | akkaneror | 42,983,232 | MDQ6VXNlcjQyOTgzMjMy | User | false |
openclaw/openclaw | 4,051,693,147 | I_kwDOQb6kR87xf-5b | 42,155 | https://github.com/openclaw/openclaw/issues/42155 | https://api.github.com/repos/openclaw/openclaw/issues/42155 | voice-call plugin uses main agent workspace for all response generation | The voice-call plugin hardcodes response generation against the `main` agent's workspace. In multi-agent setups where different agents have distinct memory, skills, and personality, voice calls always respond as `main` regardless of which agent should handle the call.
There is no config field to route voice response g... | closed | completed | true | 3 | [] | [] | 2026-03-10T13:06:25Z | 2026-04-27T04:40:05Z | 2026-04-25T02:32:37Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Diaspar4u | 3,605,840 | MDQ6VXNlcjM2MDU4NDA= | User | false |
openclaw/openclaw | 4,078,242,540 | I_kwDOQb6kR87zFQrs | 47,268 | https://github.com/openclaw/openclaw/issues/47268 | https://api.github.com/repos/openclaw/openclaw/issues/47268 | [msteams] FileConsentCard not updated after user accepts — consent card remains frozen | ## Bug: FileConsentCard stays in Allow/Deny state after file upload completes
**Component:** `@openclaw/msteams` plugin
**File:** `extensions/msteams/src/monitor-handler.ts`
**Severity:** Low (cosmetic — file uploads succeed, but UX is degraded)
### Summary
When a bot sends a file via `FileConsentCard` in a personal... | closed | completed | true | 2 | [] | [] | 2026-03-15T12:00:45Z | 2026-04-27T04:40:04Z | 2026-04-25T02:32:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cwatts-sage | 231,297,557 | U_kgDODclSFQ | User | false |
openclaw/openclaw | 4,078,254,949 | I_kwDOQb6kR87zFTtl | 47,281 | https://github.com/openclaw/openclaw/issues/47281 | https://api.github.com/repos/openclaw/openclaw/issues/47281 | [Bug]: Web UI incorrectly displays cumulative token count as context usage (shows 857k/128k when actual is 75k/128k) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The Web UI displays an incorrect context usage percentage by showing the cumulative token count instead of the actual current context usage. This results in the UI showing values like "857.1k/128k (100%)" when the actual backend context usa... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T12:08:53Z | 2026-04-27T04:40:03Z | 2026-04-25T02:33:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | caopinhe2004-design | 267,341,538 | U_kgDOD-9O4g | User | false |
openclaw/openclaw | 4,078,256,100 | I_kwDOQb6kR87zFT_k | 47,282 | https://github.com/openclaw/openclaw/issues/47282 | https://api.github.com/repos/openclaw/openclaw/issues/47282 | [Bug] Heartbeat only fires once, then stops | **Version:** OpenClaw 2026.3.13
**Bug:** Heartbeat configured with every: '5m' only triggers once after gateway restart, then never fires again.
**Logs:**
2026-03-15T20:01:44.703+08:00 [heartbeat] started
(then nothing until next restart)
**Expected:** Heartbeat should fire every 5 minutes continuously.
**Additiona... | closed | completed | true | 3 | [] | [] | 2026-03-15T12:09:38Z | 2026-04-27T04:40:02Z | 2026-04-25T02:33:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | garysze77 | 213,480,852 | U_kgDODLl1lA | User | false |
openclaw/openclaw | 4,078,264,030 | I_kwDOQb6kR87zFV7e | 47,286 | https://github.com/openclaw/openclaw/issues/47286 | https://api.github.com/repos/openclaw/openclaw/issues/47286 | UX: Suppress intermediate tool failures when operation succeeds | ## UX: Suppress intermediate tool failures when operation ultimately succeeds
### Problem
When a tool fails but the agent recovers using a different approach, the user still sees the intermediate failure message (e.g., "Edit failed"). This creates unnecessary noise and confusion.
**Example:**
1. Agent tries `Edit` to... | closed | completed | true | 2 | [] | [] | 2026-03-15T12:15:00Z | 2026-04-27T04:40:02Z | 2026-04-25T02:33:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sabo961 | 191,984 | MDQ6VXNlcjE5MTk4NA== | User | false |
openclaw/openclaw | 4,078,299,593 | I_kwDOQb6kR87zFenJ | 47,304 | https://github.com/openclaw/openclaw/issues/47304 | https://api.github.com/repos/openclaw/openclaw/issues/47304 | [Bug]: | ### Bug type
Regression (worked before, now fails)
### Summary
Vertex AI 401 UNAUTHENTICATED / CREDENTIALS_MISSING on 2026.3.13 (works on 2026.3.8)
### Steps to reproduce
На 2026.3.8: google-vertex/gemini-3.1-pro-preview с GOOGLE_APPLICATION_CREDENTIALS работает.
Обновиться до 2026.3.13.
Запустить агента → LLM... | closed | completed | true | 2 | [
"bug",
"dedupe:parent",
"regression"
] | [
"vincentkoc"
] | 2026-03-15T12:38:45Z | 2026-04-27T04:31:03Z | 2026-04-25T02:33:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | wrx861 | 56,321,354 | MDQ6VXNlcjU2MzIxMzU0 | User | false |
openclaw/openclaw | 4,078,304,656 | I_kwDOQb6kR87zFf2Q | 47,307 | https://github.com/openclaw/openclaw/issues/47307 | https://api.github.com/repos/openclaw/openclaw/issues/47307 | `openclaw status` shows "unreachable (missing scope: operator.read)" for working local gateway | ## Summary
`openclaw status` reports the local gateway as `unreachable (missing scope: operator.read)`, even though the gateway is fully functional (sessions work, channels connected, `openclaw gateway status` shows "Connect: ok · RPC: limited").
## Environment
- OpenClaw: git HEAD @ `3f434080`
- OS: macOS (arm64)
-... | closed | completed | true | 3 | [] | [] | 2026-03-15T12:42:07Z | 2026-04-27T04:40:01Z | 2026-04-25T02:34:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | gamer-mitsuha | 107,781,752 | U_kgDOBmyeeA | User | false |
openclaw/openclaw | 4,078,313,453 | I_kwDOQb6kR87zFh_t | 47,314 | https://github.com/openclaw/openclaw/issues/47314 | https://api.github.com/repos/openclaw/openclaw/issues/47314 | Token usage not captured for Alibaba Bailian (GLM-5) streaming API | ## Problem Description
When using OpenClaw with Alibaba Bailian (GLM-5) model, the context length display shows **0/180k (0%)** even after multiple conversations. The token usage is not being captured from the streaming API response.
## Environment
- OpenClaw version: 2026.3.12
- Provider: bailian (Alibaba DashScope... | closed | completed | true | 1 | [] | [] | 2026-03-15T12:48:03Z | 2026-04-27T04:40:00Z | 2026-04-25T02:34:27Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | a88883284 | 18,188,830 | MDQ6VXNlcjE4MTg4ODMw | User | false |
openclaw/openclaw | 4,078,313,959 | I_kwDOQb6kR87zFiHn | 47,315 | https://github.com/openclaw/openclaw/issues/47315 | https://api.github.com/repos/openclaw/openclaw/issues/47315 | GitHub Copilot (Enterprise): Subagents return empty output (0 tokens) after upgrade to 2026.3.13 | ## Description
After upgrading from a previous version to **2026.3.13**, GitHub Copilot Enterprise subagents return empty output with 0 tokens. The main session also intermittently throws **421 Misdirected Request** errors.
## Environment
- **OpenClaw version:** 2026.3.13 (61d171a)
- **OS:** Linux 6.17.0-14-generic ... | closed | not_planned | true | 4 | [] | [] | 2026-03-15T12:48:25Z | 2026-04-27T04:39:59Z | 2026-04-25T02:34:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | buzz39 | 16,227,736 | MDQ6VXNlcjE2MjI3NzM2 | User | false |
openclaw/openclaw | 4,078,318,693 | I_kwDOQb6kR87zFjRl | 47,317 | https://github.com/openclaw/openclaw/issues/47317 | https://api.github.com/repos/openclaw/openclaw/issues/47317 | exec tool validation error - command property missing | ## Error Description
The exec tool is returning:
This happens even when valid commands are sent with the correct parameter.
## Environment
- OS: macOS 25.4.0 (arm64)
- Node: v25.8.0
- OpenClaw version: Latest (as of 2026-03-15)
## Reproduction
1. Send exec tool call with valid command:
2. Response:
## Notes
- S... | closed | completed | true | 3 | [] | [] | 2026-03-15T12:51:42Z | 2026-04-27T04:39:59Z | 2026-04-25T02:35:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | RockENZO | 125,934,469 | U_kgDOB4GbhQ | User | false |
openclaw/openclaw | 4,078,319,261 | I_kwDOQb6kR87zFjad | 47,319 | https://github.com/openclaw/openclaw/issues/47319 | https://api.github.com/repos/openclaw/openclaw/issues/47319 | [Bug]: gateway crashes after second cron add (exit code 1) | ## Description
Executing `openclaw cron add` causes the gateway to crash and exit with code 1.
## Environment
- **OpenClaw version:** 2026.3.13 (61d171a)
- **Node:** v22.22.1
- **Platform:** WSL2 (Linux)
- **Channel:** Feishu
## Steps to Reproduce
1. Start gateway: `openclaw gateway start`
2. Execute `openclaw cron a... | closed | completed | true | 1 | [] | [] | 2026-03-15T12:52:06Z | 2026-04-27T04:39:57Z | 2026-04-25T02:35:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | diamondhu6688 | 268,366,481 | U_kgDOD_7ykQ | User | false |
openclaw/openclaw | 4,110,020,949 | I_kwDOQb6kR870-fFV | 51,263 | https://github.com/openclaw/openclaw/issues/51263 | https://api.github.com/repos/openclaw/openclaw/issues/51263 | Tool subsystem fails to resolve SecretRefs for Discord token (message tool) | ## Bug Description
The `message` tool fails to resolve SecretRefs when making proactive Discord API calls from non-Discord sessions (e.g., webchat, cron). The error is:
```
channels.discord.accounts.default.token: unresolved SecretRef "env:default:DISCORD_TOKEN_DEFAULT". Resolve this command against an active gateway... | closed | completed | true | 5 | [] | [] | 2026-03-20T20:43:06Z | 2026-04-27T04:39:56Z | 2026-04-25T02:35:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | williamschatten | 263,741,034 | U_kgDOD7heag | User | false |
openclaw/openclaw | 4,078,329,175 | I_kwDOQb6kR87zFl1X | 47,324 | https://github.com/openclaw/openclaw/issues/47324 | https://api.github.com/repos/openclaw/openclaw/issues/47324 | [Feature]: Million-Agent Swarm Mode with Pheromone-Based Coordination | ### Summary
Introduce a Swarm Intelligence Mode that enables large-scale multi-agent emergence using a pheromone-based coordination mechanism.
Instead of centralized orchestration, agents interact through a pheromone signal layer, where shared data, signals, and environmental states guide collective behavior.
This a... | closed | not_planned | true | 4 | [
"enhancement"
] | [] | 2026-03-15T12:58:56Z | 2026-04-27T04:39:56Z | 2026-04-25T02:35:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zgsndzq | 258,358,461 | U_kgDOD2Y8vQ | User | false |
openclaw/openclaw | 4,078,339,185 | I_kwDOQb6kR87zFoRx | 47,326 | https://github.com/openclaw/openclaw/issues/47326 | https://api.github.com/repos/openclaw/openclaw/issues/47326 | [Bug]: BOOTSTRAP persistence -> Hard AGENT reset -> MEMORY triage | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The "First Run" logic fails
### Steps to reproduce
PATCH: If BOOTSTRAP exists - run then delete -> AGENTS - remove "First Run" clause -> echo: "safe from lobotomy"
The First Run clause causes a hard AGENT reset -> /memory picks up on ed... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T13:05:06Z | 2026-04-27T04:39:55Z | 2026-04-25T02:35:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | echoalphachambers | 199,370,005 | U_kgDOC-IlFQ | User | false |
openclaw/openclaw | 4,078,364,832 | I_kwDOQb6kR87zFuig | 47,339 | https://github.com/openclaw/openclaw/issues/47339 | https://api.github.com/repos/openclaw/openclaw/issues/47339 | Feature Request: allow depth-1 sub-agents to send intermediate progress messages to parent via sessions_send | ## Problem
With the orchestrator pattern (main -> PM -> workers), the PM (depth-1) manages multi-phase projects autonomously. However, the PM cannot send intermediate progress updates to the main agent - it can only announce when fully completed.
This means the main agent (and the human) have no way to receive real-t... | closed | completed | true | 2 | [] | [] | 2026-03-15T13:21:23Z | 2026-04-27T04:39:54Z | 2026-04-25T02:36:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | shibiao1998-dot | 266,466,674 | U_kgDOD-H1cg | User | false |
openclaw/openclaw | 4,078,409,397 | I_kwDOQb6kR87zF5a1 | 47,358 | https://github.com/openclaw/openclaw/issues/47358 | https://api.github.com/repos/openclaw/openclaw/issues/47358 | [Bug]: Subagents ignore configured model routing and fall back to primary/default model | ### Bug type
Crash (process/app exits or hangs)
### Summary
Summary
Subagents consistently ignore their intended model configuration and fall back to the primary/default model of the parent agent.
Multi-agent role separation works at the persona/workspace level, but not at the model-routing level.
### Steps to ... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-03-15T13:46:37Z | 2026-04-27T04:39:53Z | 2026-04-25T02:36:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ptao73 | 10,295,231 | MDQ6VXNlcjEwMjk1MjMx | User | false |
openclaw/openclaw | 4,078,420,577 | I_kwDOQb6kR87zF8Jh | 47,360 | https://github.com/openclaw/openclaw/issues/47360 | https://api.github.com/repos/openclaw/openclaw/issues/47360 | [Feature]: Add a pluginized structured context engine to improve compaction fidelity in long sessions | ### Summary
Context compaction in long or tool-heavy sessions can lose critical continuity signals (latest user intent, hard constraints, exact identifiers, and pending asks).
This issue tracks introducing and validating a structured context engine approach that preserves decision-critical context while still reduci... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-15T13:51:23Z | 2026-04-27T04:39:53Z | 2026-04-25T02:36:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | noipmystery | 129,147,310 | U_kgDOB7Khrg | User | false |
openclaw/openclaw | 4,078,422,349 | I_kwDOQb6kR87zF8lN | 47,361 | https://github.com/openclaw/openclaw/issues/47361 | https://api.github.com/repos/openclaw/openclaw/issues/47361 | Urgent Help Needed: Cannot configure subagent invocation permissions for main agent (Version 2026.3.12) | ### Bug type
Regression (worked before, now fails)
### Summary
2026.3.12
### Steps to reproduce
Yes, the issue is 100% reproducible by applying the configurations mentioned in the report.
### Expected behavior
Yes, absolutely. I am willing to help test any potential fixes.
### Actual behavior
When I try to add... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-15T13:52:30Z | 2026-04-27T04:39:52Z | 2026-04-25T02:36:57Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | youeshopent-debug | 266,634,203 | U_kgDOD-SD2w | User | false |
openclaw/openclaw | 4,078,463,081 | I_kwDOQb6kR87zGGhp | 47,381 | https://github.com/openclaw/openclaw/issues/47381 | https://api.github.com/repos/openclaw/openclaw/issues/47381 | Model override in cron agentTurn payload is ignored — falls back to default model | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Model override in cron agentTurn payload is stored correctly but ignored at runtime — job always falls back to the default agent model
### Steps to reproduce
1. Configure an Ollama provider with a local model (e.g. ollama/qwen2.5:7b)
2. C... | closed | completed | true | 6 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T14:14:28Z | 2026-04-27T04:39:51Z | 2026-04-25T02:37:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ramshenoy | 268,190,902 | U_kgDOD_xEtg | User | false |
openclaw/openclaw | 4,078,465,185 | I_kwDOQb6kR87zGHCh | 47,383 | https://github.com/openclaw/openclaw/issues/47383 | https://api.github.com/repos/openclaw/openclaw/issues/47383 | GitHub Copilot Business accounts get 421 Misdirected Request — runtime baseUrl ignored in pi-embedded | ## Bug Description
When using a GitHub Copilot **Business** (or Business Trial) account, agent runs consistently fail with `421 Misdirected Request`, even though the Copilot runtime token correctly contains `proxy-ep=proxy.business.githubcopilot.com`.
Direct `curl` requests to the **correct** endpoint (`api.business.... | closed | completed | true | 3 | [] | [] | 2026-03-15T14:15:43Z | 2026-04-27T04:39:49Z | 2026-04-25T02:37:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Frankla20 | 265,706,740 | U_kgDOD9Zc9A | User | false |
openclaw/openclaw | 4,078,475,213 | I_kwDOQb6kR87zGJfN | 47,389 | https://github.com/openclaw/openclaw/issues/47389 | https://api.github.com/repos/openclaw/openclaw/issues/47389 | [Feature]: Add postinstall hook to prevent code-splitting singleton bugs (WhatsApp active-listener, #45994) | ## Problem
Issue #45994 documents a critical code-splitting bug where `active-listener.ts` is duplicated across 7 bundle chunks, each with an independent `Map()` instance. This breaks all proactive WhatsApp outbound sends while inbound and auto-replies work fine.
The community workaround (patching `globalThis` into e... | closed | completed | true | 1 | [] | [] | 2026-03-15T14:21:57Z | 2026-04-27T04:39:49Z | 2026-04-25T02:37:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | longevityboris | 5,087,259 | MDQ6VXNlcjUwODcyNTk= | User | false |
openclaw/openclaw | 4,071,275,183 | I_kwDOQb6kR87yqrqv | 45,153 | https://github.com/openclaw/openclaw/issues/45153 | https://api.github.com/repos/openclaw/openclaw/issues/45153 | [Bug]: Sandbox browser ignores top-level browser.ssrfPolicy — SSRF checks always strict | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
buildSandboxBrowserResolvedConfig in src/agents/sandbox/browser.ts does not include ssrfPolicy in the constructed ResolvedBrowserConfig, causing sandbox browser to always run with the strictest SSRF policy regardless of user configuration i... | closed | completed | true | 10 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T13:56:33Z | 2026-04-27T04:39:48Z | 2026-04-25T02:38:03Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | JackyMeow | 169,056,675 | U_kgDOChOZow | User | false |
openclaw/openclaw | 4,078,479,959 | I_kwDOQb6kR87zGKpX | 47,392 | https://github.com/openclaw/openclaw/issues/47392 | https://api.github.com/repos/openclaw/openclaw/issues/47392 | [Bug]: Feishu channel - First message after /new session startup is not delivered | ### Bug type
Regression (worked before, now fails)
### Summary
When starting a new session via `/new` or `/reset` command in Feishu (Lark) channel, the agent's first reply message is not delivered to the user. The message appears to be sent internally but never arrives in the Feishu chat.
### Steps to reproduce
1.... | closed | not_planned | true | 6 | [
"bug",
"regression"
] | [] | 2026-03-15T14:24:40Z | 2026-04-27T04:39:46Z | 2026-04-25T02:38:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ScrappyZhang | 34,594,989 | MDQ6VXNlcjM0NTk0OTg5 | User | false |
openclaw/openclaw | 4,179,799,389 | I_kwDOQb6kR875Iq1d | 58,416 | https://github.com/openclaw/openclaw/issues/58416 | https://api.github.com/repos/openclaw/openclaw/issues/58416 | [Bug]: Voice call plugin TTS config | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
After upgrading from 2026.3.24 to 2026.3.28 I receive the following error when openclaw attempt to validate the configuration:
`plugins.entries.voice-call.config.tts: invalid config: must NOT have additional properties`
This ... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-03-31T14:41:51Z | 2026-04-27T04:39:46Z | 2026-04-25T02:38:06Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ghosty2004 | 72,819,231 | MDQ6VXNlcjcyODE5MjMx | User | false |
openclaw/openclaw | 4,078,490,401 | I_kwDOQb6kR87zGNMh | 47,394 | https://github.com/openclaw/openclaw/issues/47394 | https://api.github.com/repos/openclaw/openclaw/issues/47394 | Support multiple Feishu bots | ## Feature Request
### Description
Currently OpenClaw only supports one Feishu bot. It would be great to support multiple bots for different purposes (e.g., one for daily chat, one for task processing).
### Use Case
- Separate bot instances for different tasks
- Better isolation between chat contexts
### Possible So... | closed | completed | true | 2 | [] | [] | 2026-03-15T14:30:29Z | 2026-04-27T04:39:45Z | 2026-04-25T02:38:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sunyuhao-max | 267,542,701 | U_kgDOD_JgrQ | User | false |
openclaw/openclaw | 4,078,495,326 | I_kwDOQb6kR87zGOZe | 47,396 | https://github.com/openclaw/openclaw/issues/47396 | https://api.github.com/repos/openclaw/openclaw/issues/47396 | [feishu-plugin] Bug: message tool media parameter sends images as file attachments | ## Bug Description
Using the message tool with the media parameter sends images as **file attachments** (📎) instead of **image messages** (inline display), regardless of file extension (.jpg/.png).
## Reproduction Steps
1. Use the message tool with media parameter:
```
openclaw tools message --action send --c... | closed | completed | true | 3 | [] | [] | 2026-03-15T14:33:23Z | 2026-04-27T04:39:44Z | 2026-04-25T02:38:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hughesnee | 8,207,479 | MDQ6VXNlcjgyMDc0Nzk= | User | false |
openclaw/openclaw | 4,078,504,492 | I_kwDOQb6kR87zGQos | 47,397 | https://github.com/openclaw/openclaw/issues/47397 | https://api.github.com/repos/openclaw/openclaw/issues/47397 | billing failure on one auth profile disables entire provider for embedded agents, blocking healthy fallback profiles | ## Bug Description
When an embedded agent run (e.g. `exec-approval-followup`) encounters a `billing` error on one auth profile (e.g. `anthropic:default` with an exhausted API key), the entire `anthropic` provider ends up in a disabled/cooldown state. Any other profiles for the same provider (e.g. `anthropic:manual` us... | closed | completed | true | 2 | [] | [] | 2026-03-15T14:38:50Z | 2026-04-27T04:39:43Z | 2026-04-25T02:38:55Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tkothe | 5,236,566 | MDQ6VXNlcjUyMzY1NjY= | User | false |
openclaw/openclaw | 4,078,513,268 | I_kwDOQb6kR87zGSx0 | 47,401 | https://github.com/openclaw/openclaw/issues/47401 | https://api.github.com/repos/openclaw/openclaw/issues/47401 | [Bug]: Official Docker runtime fails to load Telegram/Discord plugins because /app/src is missing | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In the official Docker runtime, the bundled Telegram and Discord plugins fail to load.
On current `origin/main`, their extension source files import core modules from `../../../src/...`, but the official Dockerfile runtime stage does not c... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T14:42:44Z | 2026-04-27T04:39:43Z | 2026-04-25T02:39:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | livekm0309 | 145,443,085 | U_kgDOCKtJDQ | User | false |
openclaw/openclaw | 4,078,515,258 | I_kwDOQb6kR87zGTQ6 | 47,402 | https://github.com/openclaw/openclaw/issues/47402 | https://api.github.com/repos/openclaw/openclaw/issues/47402 | [Bug]: WebSocket device auth regression in 2026.3.13 breaks web dashboard connections | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to 2026.3.13, WebSocket connections from web clients fail with "device identity required" error, breaking dashboard integrations that worked in 2026.3.11.
### Steps to reproduce
1. Install OpenClaw 2026.3.13
2. Configure gateway with a... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-15T14:43:51Z | 2026-04-27T04:39:42Z | 2026-04-25T02:39:29Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Jamesville2 | 268,385,481 | U_kgDOD_88yQ | User | false |
openclaw/openclaw | 4,103,339,540 | I_kwDOQb6kR870k_4U | 50,573 | https://github.com/openclaw/openclaw/issues/50573 | https://api.github.com/repos/openclaw/openclaw/issues/50573 | [Bug]: Inbound voice call: RealtimeSTT transcribes speech but result never reaches agent session | ### Summary
Inbound calls via Twilio are received, the greeting plays (ElevenLabs TTS), and `RealtimeSTT` detects speech — but the transcription result is never delivered to an agent session. The call ends silently after speech stops.
### Steps to reproduce
1. Configure voice-call plugin with Twilio + ElevenLabs TTS... | closed | completed | true | 2 | [] | [] | 2026-03-19T18:06:45Z | 2026-04-27T04:39:41Z | 2026-04-25T02:39:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kiagentkronos-cell | 269,010,453 | U_kgDOEAjGFQ | User | false |
openclaw/openclaw | 4,078,527,691 | I_kwDOQb6kR87zGWTL | 47,408 | https://github.com/openclaw/openclaw/issues/47408 | https://api.github.com/repos/openclaw/openclaw/issues/47408 | [Bug]: Sandbox reported as active, but exec still had host filesystem access outside workspace | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
I tested OpenClaw sandboxing with the goal of allowing access to the workspace only.
### Steps to reproduce
### Intended config
- sandbox.mode: "all"
- sandbox.scope: "session"
- workspaceAccess: "rw"
### What looked correct
- D... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T14:50:58Z | 2026-04-27T04:39:40Z | 2026-04-25T02:39:45Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | LaskerSaw | 130,273,415 | U_kgDOB8PQhw | User | false |
openclaw/openclaw | 4,078,535,763 | I_kwDOQb6kR87zGYRT | 47,411 | https://github.com/openclaw/openclaw/issues/47411 | https://api.github.com/repos/openclaw/openclaw/issues/47411 | [Bug]: Docker runtime image missing src/ breaks extensions with ../../../src relative imports | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
These extensions have ~80+ relative imports to ../../../src/... for internal core modules. In Docker, the src/ directory is not copied to the runtime image (only dist/ is), so all those relative paths break.
**This is a design issue - exte... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T14:55:53Z | 2026-04-27T04:39:40Z | 2026-04-25T02:40:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhangwei | 6,028,709 | MDQ6VXNlcjYwMjg3MDk= | User | false |
openclaw/openclaw | 4,078,559,717 | I_kwDOQb6kR87zGeHl | 47,418 | https://github.com/openclaw/openclaw/issues/47418 | https://api.github.com/repos/openclaw/openclaw/issues/47418 | [Bug]: macOS app exec approval popup truncates/hides command and provides no details/expand view | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In the macOS app, the exec approval popup is too small and does not show the actual command being approved, making it impossible to make an informed allow/deny decision.
### Steps to reproduce
1. Use the macOS OpenClaw app with exec appr... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T15:08:42Z | 2026-04-27T04:39:39Z | 2026-04-25T02:40:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kidroca | 12,156,624 | MDQ6VXNlcjEyMTU2NjI0 | User | false |
openclaw/openclaw | 4,046,079,579 | I_kwDOQb6kR87xKkZb | 41,157 | https://github.com/openclaw/openclaw/issues/41157 | https://api.github.com/repos/openclaw/openclaw/issues/41157 | Google/Gemini provider (openai-completions): 400 error, apiKey not sent as Authorization: Bearer | ## Bug Report
**Version:** 2026.3.7
**Provider:** `google` with `api: openai-completions`
**Endpoint:** `https://generativelanguage.googleapis.com/v1beta/openai/`
## Description
When configuring a Google/Gemini model via the `openai-completions` API type, all requests return `400 status code (no body)` with `input... | closed | completed | true | 3 | [] | [] | 2026-03-09T14:53:24Z | 2026-04-27T04:39:38Z | 2026-04-25T02:40:29Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | alfredbotlerai-beep | 263,067,575 | U_kgDOD64Xtw | User | false |
openclaw/openclaw | 4,078,561,787 | I_kwDOQb6kR87zGen7 | 47,419 | https://github.com/openclaw/openclaw/issues/47419 | https://api.github.com/repos/openclaw/openclaw/issues/47419 | [Bug]: Title: [Bug] v2026.3.13 crashes with SIGILL (ExitCode 132) on non-AVX2 CPUs | ### Bug type
Regression (worked before, now fails)
### Summary
v2026.3.13 crashes with SIGILL (ExitCode 132) on non-AVX2 CPUs Possible CPU: Intel Celeron J4125 (or similar Gemini Lake/Apollo Lake)
### Steps to reproduce
Hardware/Environment:
Device: QNAP NAS / iStoreOS
CPU Flags: fpu vme de pse tsc msr pae mce ... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-15T15:09:52Z | 2026-04-27T04:39:37Z | 2026-04-25T02:40:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zzxy777 | 130,168,483 | U_kgDOB8I2ow | User | false |
openclaw/openclaw | 4,078,588,243 | I_kwDOQb6kR87zGlFT | 47,430 | https://github.com/openclaw/openclaw/issues/47430 | https://api.github.com/repos/openclaw/openclaw/issues/47430 | [Bug]: Regression: 2026.3.13 causes OOM and hangs on low-memory Linux server; 2026.3.11 works | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to 2026.3.13, OpenClaw hangs on status/health/probe commands and is repeatedly killed by OOM on a low-memory Linux server; 2026.3.11 works on the same machine.
### Steps to reproduce
1. Install OpenClaw 2026.3.13 via pnpm global install... | closed | completed | true | 5 | [
"bug",
"regression"
] | [] | 2026-03-15T15:25:34Z | 2026-04-27T04:39:36Z | 2026-04-25T02:40:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Ultramantoo | 47,166,896 | MDQ6VXNlcjQ3MTY2ODk2 | User | false |
openclaw/openclaw | 4,078,593,184 | I_kwDOQb6kR87zGmSg | 47,431 | https://github.com/openclaw/openclaw/issues/47431 | https://api.github.com/repos/openclaw/openclaw/issues/47431 | Control UI Context Display Bug (Misleading Usage) | Dashboard shows '378k/128k' which misleadingly implies context overflow (297%).
**Reality:**
- **378k** = Lifetime tokens sent (cumulative session total, informational only)
- **58k** = Actual active context in memory
- **Context health:** 58k/128k (46%) ✅
**Root cause:** UI compares apples-to-oranges (lifetime consu... | closed | completed | true | 3 | [] | [] | 2026-03-15T15:28:22Z | 2026-04-27T04:39:36Z | 2026-04-25T02:41:09Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Kru442 | 187,809,917 | U_kgDOCzHAfQ | User | false |
openclaw/openclaw | 4,078,597,746 | I_kwDOQb6kR87zGnZy | 47,432 | https://github.com/openclaw/openclaw/issues/47432 | https://api.github.com/repos/openclaw/openclaw/issues/47432 | controlUi.basePath Configuration Causes Frontend WebSocket URL to Missing Trailing Slash | ---
Title: controlUi.basePath Configuration Causes Frontend WebSocket URL to Missing Trailing Slash
Severity: Medium
Environment:
- OpenClaw Version: 2026.3.13
- Operating System: Linux
- Reverse Proxy: nginx
Problem Description:
When configuring a non-root path (e.g., /openclaw) using controlUi.ba... | closed | not_planned | true | 1 | [] | [] | 2026-03-15T15:30:38Z | 2026-04-27T04:39:32Z | 2026-04-25T02:41:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | patrickchen7419-dev | 251,942,862 | U_kgDODwRXzg | User | false |
openclaw/openclaw | 4,190,867,071 | I_kwDOQb6kR875y45_ | 59,328 | https://github.com/openclaw/openclaw/issues/59328 | https://api.github.com/repos/openclaw/openclaw/issues/59328 | [Bug] openai-completions streaming: Context always shows 0 — missing stream_options include_usage | ## Summary
When using `openai-completions` API type with providers that strictly follow the OpenAI spec (e.g., Alibaba Cloud Bailian/DashScope), `/status` always shows `Context: 0` because the provider returns `usage: null` in streaming chunks.
## Root Cause
The OpenAI Chat Completions API spec requires clients to e... | closed | completed | true | 4 | [] | [] | 2026-04-02T01:37:10Z | 2026-04-27T04:39:31Z | 2026-04-25T02:41:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | youyou-agent | 271,778,299 | U_kgDOEDMB-w | User | false |
openclaw/openclaw | 4,078,602,225 | I_kwDOQb6kR87zGofx | 47,434 | https://github.com/openclaw/openclaw/issues/47434 | https://api.github.com/repos/openclaw/openclaw/issues/47434 | Feature Request: Add model fallback/autofailover support | ## Description
Currently OpenClaw only supports a single primary model. When the primary model fails (API error, rate limit, timeout, etc.), there is no automatic fallback to another configured model.
## Requested Feature
Add model fallback/autofailover support. When the primary model fails, automatically try the ne... | closed | completed | true | 2 | [] | [] | 2026-03-15T15:32:45Z | 2026-04-27T04:39:30Z | 2026-04-25T02:41:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | wx2cyj | 123,572,594 | U_kgDOB12Rcg | User | false |
openclaw/openclaw | 4,078,604,354 | I_kwDOQb6kR87zGpBC | 47,436 | https://github.com/openclaw/openclaw/issues/47436 | https://api.github.com/repos/openclaw/openclaw/issues/47436 | [Bug] Feishu multi-account (accounts.*) appSecret SecretRef fails to resolve, crashes feishu plugin after ~3 minutes | ### Bug type
Regression (worked before, now fails)
### Summary
## Environment
- OpenClaw version: 2026.3.13
- Channel: feishu (websocket mode)
## Description
When configuring a second Feishu bot under `channels.feishu.accounts.bot2`, the `appSecret` SecretRef resolves successfully at startup, but approximately 3 mi... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-15T15:33:56Z | 2026-04-27T04:39:30Z | 2026-04-25T02:41:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | charliecha | 2,815,473 | MDQ6VXNlcjI4MTU0NzM= | User | false |
openclaw/openclaw | 4,078,627,713 | I_kwDOQb6kR87zGuuB | 47,442 | https://github.com/openclaw/openclaw/issues/47442 | https://api.github.com/repos/openclaw/openclaw/issues/47442 | [Bug]: OpenAI Codex tool continuations can resume stale backend state across top-level turns | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In long-lived OpenAI Codex sessions, a new user turn can start correctly and plan the right tools, but after tool results the assistant may emit a stale final from an older, already-completed topic.
### Steps to reproduce
This is not yet ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T15:47:46Z | 2026-04-27T04:39:29Z | 2026-04-25T02:42:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pascalkienast | 6,682,526 | MDQ6VXNlcjY2ODI1MjY= | User | false |
openclaw/openclaw | 4,078,628,696 | I_kwDOQb6kR87zGu9Y | 47,443 | https://github.com/openclaw/openclaw/issues/47443 | https://api.github.com/repos/openclaw/openclaw/issues/47443 | ACP on DM/non-thread surfaces needs first-class completion relay / proactive notification | ## Summary
OpenClaw needs a first-class workflow for **ACP on direct-message / non-thread surfaces** (for example Telegram DM, webchat DM, Feishu DM, Slack DM) so long-running coding-agent work can **proactively notify the user when it finishes**.
Today, ACP behavior on DM / non-thread surfaces falls into an awkward ... | closed | completed | true | 1 | [] | [] | 2026-03-15T15:48:19Z | 2026-04-27T04:39:27Z | 2026-04-25T02:42:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Groos-dev | 81,287,043 | MDQ6VXNlcjgxMjg3MDQz | User | false |
openclaw/openclaw | 4,078,633,238 | I_kwDOQb6kR87zGwEW | 47,445 | https://github.com/openclaw/openclaw/issues/47445 | https://api.github.com/repos/openclaw/openclaw/issues/47445 | [Bug]: gateway restart command fails when executed via exec tool on Windows | ### Bug type
Regression (worked before, now fails)
### Summary
When executing 'openclaw gateway restart' via the exec tool on Windows, the command fails with 'Port 18789 is already in use' error. The gateway does not restart successfully and requires manual restart via PowerShell.
### Steps to reproduce
1. Run Ope... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-15T15:51:03Z | 2026-04-27T04:39:26Z | 2026-04-25T02:42:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Blickhand | 117,608,944 | U_kgDOBwKR8A | User | false |
openclaw/openclaw | 4,078,636,577 | I_kwDOQb6kR87zGw4h | 47,448 | https://github.com/openclaw/openclaw/issues/47448 | https://api.github.com/repos/openclaw/openclaw/issues/47448 | [Bug]: Local Chief ignores MEMORY.md and treats my own team data as “untrusted metadata” after workspace move | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
After moving my OpenClaw workspace from Akash to a local Windows + Docker + Ollama setup, Chief can see the correct team data in MEMORY.md but refuses to use it. Instead, he treats the memory content as “untrusted metadata,” hallucinates ro... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-15T15:53:04Z | 2026-04-27T04:39:25Z | 2026-04-25T02:43:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Chief-COO | 265,807,809 | U_kgDOD9fnwQ | User | false |
openclaw/openclaw | 4,159,435,087 | I_kwDOQb6kR8736_FP | 56,367 | https://github.com/openclaw/openclaw/issues/56367 | https://api.github.com/repos/openclaw/openclaw/issues/56367 | [Bug]: Voice-call plugin ignores agents.defaults.sandbox.mode: "off" - forces Docker sandbox for voice lane responses | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The voice-call plugin's `generateVoiceResponse()` creates a Docker sandbox container for voice lane responses even when `agents.defaults.sandbox.mode` is set to `"off"`. The sandbox then fails because bind moun... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-28T11:21:55Z | 2026-04-27T04:39:24Z | 2026-04-25T02:43:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | crpol | 125,162,293 | U_kgDOB3XTNQ | User | false |
openclaw/openclaw | 4,078,643,677 | I_kwDOQb6kR87zGynd | 47,449 | https://github.com/openclaw/openclaw/issues/47449 | https://api.github.com/repos/openclaw/openclaw/issues/47449 | Feature: session.onReset hook — give agents a final turn before /new clears context | ## Problem
When a user types `/new` or `/reset`, the session is immediately cleared. Any unsaved agent state — in-progress memory writes, session summaries, lessons learned — is lost forever.
Agents can mitigate this by writing memories in real-time (and they should), but complex multi-step work sometimes produces in... | closed | completed | true | 1 | [] | [] | 2026-03-15T15:57:24Z | 2026-04-27T04:39:23Z | 2026-04-25T02:43:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mcbuckeye | 1,521,511 | MDQ6VXNlcjE1MjE1MTE= | User | false |
openclaw/openclaw | 4,078,645,740 | I_kwDOQb6kR87zGzHs | 47,450 | https://github.com/openclaw/openclaw/issues/47450 | https://api.github.com/repos/openclaw/openclaw/issues/47450 | Bug: vLLM usage 解析问题 - 本地模型返回 0 tokens | ## 问题描述
**vLLM 返回的 usage 格式：**
```json
{"usage":{"prompt_tokens":39,"total_tokens":49,"completion_tokens":10}}
```
**OpenClaw 记录的 usage：**
```json
{"usage":{"input":0,"output":0,"totalTokens":0}}
```
**问题原因：**
vLLM 使用 `prompt_tokens`/`completion_tokens`/`total_tokens`，但 OpenClaw 期望 `input`/`output`/`totalTokens`，导致解... | closed | completed | true | 2 | [] | [] | 2026-03-15T15:58:41Z | 2026-04-27T04:39:23Z | 2026-04-25T02:43:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | czdiy | 12,508,302 | MDQ6VXNlcjEyNTA4MzAy | User | false |
openclaw/openclaw | 4,078,646,473 | I_kwDOQb6kR87zGzTJ | 47,451 | https://github.com/openclaw/openclaw/issues/47451 | https://api.github.com/repos/openclaw/openclaw/issues/47451 | Lexa verwechselt E-Befund mit Energieausweis — fehlende Begriffsklärung bei juristischen Fachbegriffen | ## Problem
Bei der Frage "Was kann passieren wenn meine Mieterin den E-Befund erst 6 Monate nach Mietbeginn bekommt" hat Lexa **E-Befund** (= Elektrobefund, elektrische Überprüfung nach ÖVE/ÖNORM E 8001) fälschlich als **Energieausweis** (EAVG 2012) interpretiert.
## Erwartetes Verhalten
- Lexa sollte bei österreich... | closed | not_planned | true | 1 | [] | [] | 2026-03-15T15:59:08Z | 2026-04-27T04:39:22Z | 2026-04-25T02:43:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | christianhaberl | 116,439,789 | U_kgDOBvC67Q | User | false |
openclaw/openclaw | 4,078,657,000 | I_kwDOQb6kR87zG13o | 47,454 | https://github.com/openclaw/openclaw/issues/47454 | https://api.github.com/repos/openclaw/openclaw/issues/47454 | [Bug]: Telegram messages split into multiple messages when output contains newlines | ### Bug type
Regression (worked before, now fails)
### Summary
Telegram channel: messages are split into multiple separate messages when the response contains newlines. Previously this worked correctly - single message with line breaks were preserved.
### Steps to reproduce
1. Configure Telegram channel in OpenCla... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-15T16:05:00Z | 2026-04-27T04:39:21Z | 2026-04-25T02:44:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | changchichung | 20,943,335 | MDQ6VXNlcjIwOTQzMzM1 | User | false |
openclaw/openclaw | 4,078,658,574 | I_kwDOQb6kR87zG2QO | 47,456 | https://github.com/openclaw/openclaw/issues/47456 | https://api.github.com/repos/openclaw/openclaw/issues/47456 | [Feature]: heartbeat.timeoutSeconds — per-heartbeat embedded run timeout | ### Summary
Heartbeat embedded runs inherit agents.defaults.timeoutSeconds (default 600s). There is no way to set a shorter timeout specifically for heartbeats. Heartbeats are lightweight status checks that should fail fast (30-60s), not block for 10 minutes when a model hangs.
### Problem to solve
Heartbeat embedde... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-15T16:05:52Z | 2026-04-27T04:39:20Z | 2026-04-25T02:44:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aderius | 78,624,045 | MDQ6VXNlcjc4NjI0MDQ1 | User | false |
openclaw/openclaw | 4,078,659,086 | I_kwDOQb6kR87zG2YO | 47,457 | https://github.com/openclaw/openclaw/issues/47457 | https://api.github.com/repos/openclaw/openclaw/issues/47457 | Feature request: mode="session" without thread binding for headless sub-agent orchestration | ## Summary
`mode: "session"` currently requires `thread: true`, which is only supported by Discord. This means persistent sub-agent sessions are unavailable from WhatsApp, webchat, or any channel without thread bindings.
## Use Case
Orchestrating persistent architect/specialist sub-agents from WhatsApp:
- Main agent... | closed | not_planned | true | 1 | [] | [] | 2026-03-15T16:06:10Z | 2026-04-27T04:39:20Z | 2026-04-25T02:44:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | brugillo | 50,134,745 | MDQ6VXNlcjUwMTM0NzQ1 | User | false |
openclaw/openclaw | 4,078,669,955 | I_kwDOQb6kR87zG5CD | 47,460 | https://github.com/openclaw/openclaw/issues/47460 | https://api.github.com/repos/openclaw/openclaw/issues/47460 | [Bug]: Telegram voice note can be saved to media/inbound during provider startup, but never injected into session (silent drop) | ### Summary
I hit an intermittent Telegram voice-message ingestion failure on OpenClaw where the raw audio file is successfully saved to `~/.openclaw/media/inbound/`, but no corresponding user event / transcript is injected into the active session.
The result is a **silent drop** from the assistant's point of view:
-... | closed | completed | true | 2 | [] | [] | 2026-03-15T16:12:34Z | 2026-04-27T04:39:19Z | 2026-04-25T02:45:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chunhualiao | 1,627,206 | MDQ6VXNlcjE2MjcyMDY= | User | false |
openclaw/openclaw | 4,164,493,107 | I_kwDOQb6kR874OR8z | 57,084 | https://github.com/openclaw/openclaw/issues/57084 | https://api.github.com/repos/openclaw/openclaw/issues/57084 | [Bug]: Browser CLI fails with `unknown method: browser.request` even though browser service logs as ready | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
**OpenClaw browser commands fail with `GatewayClientRequestError: unknown method: browser.request` even though the gateway is healthy and logs show the browser service starting.**
### Steps to reproduce
1. In... | closed | completed | true | 12 | [
"bug",
"bug:behavior"
] | [] | 2026-03-29T14:16:46Z | 2026-04-27T04:39:18Z | 2026-04-25T02:45:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | FoundryHouseAI | 271,721,896 | U_kgDOEDIlqA | User | false |
openclaw/openclaw | 4,176,569,026 | I_kwDOQb6kR8748WLC | 58,190 | https://github.com/openclaw/openclaw/issues/58190 | https://api.github.com/repos/openclaw/openclaw/issues/58190 | CLI `openclaw browser start` fails: `unknown method: browser.request` | ## Bug Report
**Version:** OpenClaw 2026.3.28 (f9b1079)
**Platform:** macOS (arm64), Darwin 24.6.0, Node v22.22.1
**Install:** Homebrew global (`/opt/homebrew/lib/node_modules/openclaw`)
### Reproduction
```bash
openclaw browser --browser-profile <any-profile> start
# => GatewayClientRequestError: unknown method: br... | closed | completed | true | 2 | [] | [] | 2026-03-31T07:11:45Z | 2026-04-27T04:39:17Z | 2026-04-25T02:45:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | andreihirvi | 1,726,179 | MDQ6VXNlcjE3MjYxNzk= | User | false |
openclaw/openclaw | 4,176,899,445 | I_kwDOQb6kR8749m11 | 58,222 | https://github.com/openclaw/openclaw/issues/58222 | https://api.github.com/repos/openclaw/openclaw/issues/58222 | [Bug]: Browser plugin appears loaded, but gateway rejects all browser commands with unknown method: browser.request | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
# Bug Report: `openclaw browser` CLI present but gateway returns `unknown method: browser.request`
## Summary
On a macOS local OpenClaw install, the `openclaw browser` CLI is present and the bundled `browser`... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-03-31T07:56:30Z | 2026-04-27T04:39:16Z | 2026-04-25T02:45:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | BurgersXV | 258,666,601 | U_kgDOD2rwaQ | User | false |
openclaw/openclaw | 4,078,678,826 | I_kwDOQb6kR87zG7Mq | 47,466 | https://github.com/openclaw/openclaw/issues/47466 | https://api.github.com/repos/openclaw/openclaw/issues/47466 | Browser tool fails in Docker: dbus stderr incorrectly parsed as fatal launch error | ## Browser tool fails in Docker: dbus stderr parsed as fatal error
### Environment
- OpenClaw: 2026.3.13 (Docker, Debian bookworm)
- Chromium: 145.0.7632.6 (Playwright chromium-1208)
- Container: no dbus, no X11, headless only
### Problem
`browser start` / `browser navigate` always fails with timeout or "Failed to st... | closed | completed | true | 2 | [] | [] | 2026-03-15T16:17:57Z | 2026-04-27T04:39:17Z | 2026-04-25T02:45:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | GDXbsv | 1,261,269 | MDQ6VXNlcjEyNjEyNjk= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.