repo string | github_id int64 | github_node_id string | number int64 | html_url string | api_url string | title string | body string | state string | state_reason string | locked bool | comments_count int64 | labels list | assignees list | created_at string | updated_at string | closed_at string | author_association string | milestone_title string | snapshot_id string | extracted_at string | author_login string | author_id int64 | author_node_id string | author_type string | author_site_admin bool |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
openclaw/openclaw | 4,127,126,071 | I_kwDOQb6kR871_vI3 | 53,643 | https://github.com/openclaw/openclaw/issues/53643 | https://api.github.com/repos/openclaw/openclaw/issues/53643 | [Bug]: 控制界面技能列表显示不稳定，点击后消失 | ### Bug type
Regression (worked before, now fails)
### Summary
OpenClaw控制界面的技能列表显示不稳定：点击技能按钮后列表不显示或部分显示，刷新后正常显示，点击其他内容后又消失。baidu-search技能显示为"missing"但实际功能正常。
### Steps to reproduce
1. 安装baidu-search技能: `clawhub install baidu-search`
2. 打开OpenClaw控制界面: http://127.0.0.1:18789/
3. 点击左侧边栏"代理"->"技能"按钮
4. 观察技能列表显示情况（可能不... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T11:24:25Z | 2026-03-24T11:24:56Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | hanbing1117 | 270,601,597 | U_kgDOECENfQ | User | false |
openclaw/openclaw | 4,127,129,855 | I_kwDOQb6kR871_wD_ | 53,644 | https://github.com/openclaw/openclaw/issues/53644 | https://api.github.com/repos/openclaw/openclaw/issues/53644 | [Bug]: Discord WebSocket code 1006 triggers uncaught exception and crashes gateway instead of reconnecting | ### Bug type
Crash (process/app exits or hangs)
### Summary
When the Discord WebSocket drops with code 1006 (abnormal closure), the internal `@buape/carbon` GatewayPlugin throws an uncaught exception (`Max reconnect attempts (0) reached after code 1006`) that crashes the entire OpenClaw gateway process, requiring ma... | open | null | false | 5 | [
"bug",
"bug:crash"
] | [] | 2026-03-24T11:25:06Z | 2026-03-24T17:37:41Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | shiftymix | 193,249,208 | U_kgDOC4S_uA | User | false |
openclaw/openclaw | 4,127,169,285 | I_kwDOQb6kR871_5sF | 53,645 | https://github.com/openclaw/openclaw/issues/53645 | https://api.github.com/repos/openclaw/openclaw/issues/53645 | 飞书消息回复引用(replyTo)有时不显示 - OpenClaw 2026.3.23-2 | ## 问题描述
OpenClaw 升级到 2026.3.23-2 后，飞书消息回复使用 replyTo 参数引用对方消息时，有时能显示引用，有时不能显示。
## 环境信息
- OpenClaw 版本: 2026.3.23-2
- 飞书连接模式: WebSocket 长连接
- 消息工具参数: message + card (2026.3.23-2 强制要求 card 参数)
## 复现步骤
1. 通过飞书发送消息给 OpenClaw agent
2. agent 使用 message 工具的 replyTo 参数回复
3. 有时能看到引用效果，有时看不到
## 日志信息
`
2026-03-24T11:28:32.457Z i... | open | null | false | 1 | [] | [] | 2026-03-24T11:32:41Z | 2026-03-24T11:34:01Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Kimin2011 | 268,289,122 | U_kgDOD_3EYg | User | false |
openclaw/openclaw | 4,127,177,131 | I_kwDOQb6kR871_7mr | 53,646 | https://github.com/openclaw/openclaw/issues/53646 | https://api.github.com/repos/openclaw/openclaw/issues/53646 | [Bug]: /compact in Control UI (webchat) creates new session instead of compacting | ## Summary
After upgrading to **2026.3.13**, `/compact` in the Control UI now behaves like `/new` — it creates a fresh session with no context, no summary, no memory flush. All prior conversation is lost.
`/compact` in the Control UI (webchat) **used to work correctly** on 2026.2.14 — it would summarize context, do m... | closed | completed | false | 1 | [] | [] | 2026-03-24T11:34:06Z | 2026-03-24T17:58:12Z | 2026-03-24T17:58:12Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ikenraf | 214,547,488 | U_kgDODMm8IA | User | false |
openclaw/openclaw | 4,127,209,195 | I_kwDOQb6kR872ADbr | 53,647 | https://github.com/openclaw/openclaw/issues/53647 | https://api.github.com/repos/openclaw/openclaw/issues/53647 | [Bug]: OpenProse isolated cron run can stop after first child sessions_yield boundary while cron is marked finished/ok | Title: [Bug]: OpenProse isolated cron run can stop after first child `sessions_yield` boundary while cron is marked finished/ok
Bug type
Behavior bug (incorrect output/state without crash)
Summary
In an isolated cron `agentTurn` running an OpenProse workflow, the parent OpenProse VM can yield while waiting for a chil... | open | null | false | 1 | [] | [] | 2026-03-24T11:40:07Z | 2026-03-24T14:54:37Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | 15809220109 | 219,821,571 | U_kgDODRo2Aw | User | false |
openclaw/openclaw | 4,127,226,120 | I_kwDOQb6kR872AHkI | 53,648 | https://github.com/openclaw/openclaw/issues/53648 | https://api.github.com/repos/openclaw/openclaw/issues/53648 | [Bug]: Control UI Skills Search Box Shows Gateway URL Instead of Empty | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In the OpenClaw Control UI, the search box in the left-side skills menu displays the gateway WebSocket address ws://127.0.0.1:xxxxxx by default, instead of being empty or showing placeholder text.
### Steps to reproduce
Open OpenClaw Cont... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T11:43:07Z | 2026-03-24T14:54:38Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | HelloMeme | 7,801,690 | MDQ6VXNlcjc4MDE2OTA= | User | false |
openclaw/openclaw | 4,127,230,604 | I_kwDOQb6kR872AIqM | 53,649 | https://github.com/openclaw/openclaw/issues/53649 | https://api.github.com/repos/openclaw/openclaw/issues/53649 | `openclaw doctor --fix` does not detect or repair invalid `plugins.load.paths` entries after upgrade | ## Problem
After upgrading openclaw, the gateway fails to start with the error:
```
Invalid config at ~/.openclaw/openclaw.json:
- plugins.load.paths: plugin: plugin path not found: /home/hzmember/.npm-global/lib/node_modules/openclaw/extensions/acpx
```
## Root Cause
When openclaw upgrades, bundled plugin paths ma... | open | null | false | 1 | [] | [] | 2026-03-24T11:43:52Z | 2026-03-24T14:54:35Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | hzmbr | 264,987,083 | U_kgDOD8thyw | User | false |
openclaw/openclaw | 4,127,313,609 | I_kwDOQb6kR872Ac7J | 53,652 | https://github.com/openclaw/openclaw/issues/53652 | https://api.github.com/repos/openclaw/openclaw/issues/53652 | WhatsApp: HTML files silently fail to attach as documents | ## Description
When sending an HTML file via the `message` tool with `action=send` on WhatsApp (baileys provider), the file is **not attached** — only the caption/text is delivered. No error is returned by the gateway; the send appears successful.
## Steps to Reproduce
1. Create any `.html` file
2. Send via: `messag... | open | null | false | 1 | [] | [] | 2026-03-24T11:57:01Z | 2026-03-24T14:54:36Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | zeolenon | 60,750,902 | MDQ6VXNlcjYwNzUwOTAy | User | false |
openclaw/openclaw | 4,127,401,344 | I_kwDOQb6kR872AyWA | 53,654 | https://github.com/openclaw/openclaw/issues/53654 | https://api.github.com/repos/openclaw/openclaw/issues/53654 | Discord: Support messageUpdate and messageDelete events for edit-to-reprocess and delete-to-cancel | ## Feature Request
**Submitted by:** Kat Arbiz (kat@scalewithsystem.com)
## Summary
Add support for Discord `messageUpdate` and `messageDelete` gateway events so that:
1. **Edited messages** re-trigger agent processing (if the edited message contains a bot mention)
2. **Deleted messages** halt/abort the in-progress... | open | null | false | 0 | [] | [] | 2026-03-24T12:11:27Z | 2026-03-24T12:11:27Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | sws-apps | 231,005,139 | U_kgDODcTb0w | User | false |
openclaw/openclaw | 4,127,406,188 | I_kwDOQb6kR872Azhs | 53,656 | https://github.com/openclaw/openclaw/issues/53656 | https://api.github.com/repos/openclaw/openclaw/issues/53656 | [Bug]: Feishu 通道 card 字段被错误标记为必填，无法发送纯媒体消息 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
在 Feishu（飞书）通道使用 `message` 工具发送纯媒体消息（图片/文件/音频）时，会触发 schema 验证错误：
```
Validation failed for tool "message":
- card: must have required property 'card'
```
### Steps to reproduce
1. 配置好 Feishu 通道，确保 app 有发送消息权限
2. 调用 `message` 工具发送图片/文件... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T12:12:15Z | 2026-03-24T12:30:51Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | djh2058386252-rgb | 225,677,842 | U_kgDODXOSEg | User | false |
openclaw/openclaw | 4,127,428,535 | I_kwDOQb6kR872A4-3 | 53,658 | https://github.com/openclaw/openclaw/issues/53658 | https://api.github.com/repos/openclaw/openclaw/issues/53658 | [Bug] Google OpenAI-compat fallback 400 from unsupported store + missing thought_signature during replay | ## Summary
Embedded fallback to Google OpenAI-compatible endpoint fails with `400 status code (no body)` in OpenClaw agent runs.
I traced this to two payload-compatibility problems during fallback/replay:
1) top-level `store: false` is sent to Google OpenAI-compat and rejected as unknown field.
2) replayed tool/funct... | open | null | false | 1 | [] | [] | 2026-03-24T12:16:06Z | 2026-03-24T12:20:27Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | s2030081 | 11,299,363 | MDQ6VXNlcjExMjk5MzYz | User | false |
openclaw/openclaw | 4,127,455,333 | I_kwDOQb6kR872A_hl | 53,659 | https://github.com/openclaw/openclaw/issues/53659 | https://api.github.com/repos/openclaw/openclaw/issues/53659 | Exec approval completion events lose forum topic context in Telegram forum supergroups | ## Summary
When exec approvals are configured for Telegram with `target: "dm"`, the approval prompt itself routes correctly to the user's DM with the bot. However, the **completion event** (the command result/output) loses the originating forum topic context and routes to General (topic 1) instead of the topic where t... | open | null | false | 0 | [] | [] | 2026-03-24T12:20:07Z | 2026-03-24T12:20:07Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | andywang514 | 31,902,066 | MDQ6VXNlcjMxOTAyMDY2 | User | false |
openclaw/openclaw | 4,127,458,278 | I_kwDOQb6kR872BAPm | 53,661 | https://github.com/openclaw/openclaw/issues/53661 | https://api.github.com/repos/openclaw/openclaw/issues/53661 | [Feature Request] Dispatch agents cannot notify users via Feishu on completion | ## Problem
When a dispatch sub-agent (e.g. `xiaoma:dispatch:xxx`) completes a task and tries to notify the main session or user via `sessions_send`, it frequently times out because the target session is busy or unresponsive. Additionally, cross-account Feishu messaging via the `message` tool often fails with 400 error... | open | null | false | 0 | [] | [] | 2026-03-24T12:20:38Z | 2026-03-24T12:20:38Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | mattclawcn | 267,024,418 | U_kgDOD-p4Ig | User | false |
openclaw/openclaw | 4,127,523,071 | I_kwDOQb6kR872BQD_ | 53,666 | https://github.com/openclaw/openclaw/issues/53666 | https://api.github.com/repos/openclaw/openclaw/issues/53666 | [Bug]: Regression: memory index shows 0 chunks on 2026.3.23-2 (works on 2026.3.12) | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading OpenClaw from 2026.3.12 to 2026.3.23-2, openclaw memory index --verbose reports success but openclaw memory status remains Indexed: 0/25 files · 0 chunks with provider qmd and source MEMORY.md + ~/.openclaw/workspace/memory/*.md. Rolling ... | closed | completed | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T12:30:03Z | 2026-03-24T13:40:22Z | 2026-03-24T13:40:22Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | OqpdbqpO | 19,563,815 | MDQ6VXNlcjE5NTYzODE1 | User | false |
openclaw/openclaw | 4,127,534,624 | I_kwDOQb6kR872BS4g | 53,668 | https://github.com/openclaw/openclaw/issues/53668 | https://api.github.com/repos/openclaw/openclaw/issues/53668 | gateway status reports unknown / staleGatewayPids on localized Chinese Windows even when Scheduled Task Gateway is running | # Title
`gateway status` reports `unknown` / `staleGatewayPids` on localized Chinese Windows even when Scheduled Task Gateway is running
## Summary
On a Chinese-localized Windows host, OpenClaw reports the managed Gateway runtime as `unknown` and marks the actual listener PID as `staleGatewayPids`, even though the Sch... | open | null | false | 0 | [] | [] | 2026-03-24T12:31:35Z | 2026-03-24T12:31:35Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | v45n21t2bh24h12-creator | 263,609,462 | U_kgDOD7Zcdg | User | false |
openclaw/openclaw | 4,127,561,601 | I_kwDOQb6kR872BZeB | 53,669 | https://github.com/openclaw/openclaw/issues/53669 | https://api.github.com/repos/openclaw/openclaw/issues/53669 | [Bug]: CronCreate: timezone mismatch — local tz set but UTC offset applied to cron expression | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When using CronCreate with a local timezone (e.g. Asia/Shanghai), the model incorrectly applies UTC-offset arithmetic to the cron expression while keeping the local timezone string. This results in the job firing at the wrong local time.
#... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T12:35:04Z | 2026-03-24T12:37:42Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | a13691966195-source | 233,706,742 | U_kgDODe4U9g | User | false |
openclaw/openclaw | 4,127,569,607 | I_kwDOQb6kR872BbbH | 53,671 | https://github.com/openclaw/openclaw/issues/53671 | https://api.github.com/repos/openclaw/openclaw/issues/53671 | [Bug] @mariozechner/pi-ai tool validation throws on empty args instead of returning tool_error — regression from 2026.3.22 | ## Problem Description
Since upgrading from **2026.3.13 → 2026.3.22**, certain third-party models (confirmed: Moonshot AI `kimi-latest` via `openai-completions`, and `kimi-coding/k2p5` via `anthropic-messages`) occasionally emit tool calls with empty arguments `{}` for the `exec` tool. This now causes a hard exception... | open | null | false | 2 | [] | [] | 2026-03-24T12:36:06Z | 2026-03-24T14:38:51Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | lihan891127 | 93,266,457 | U_kgDOBY8iGQ | User | false |
openclaw/openclaw | 4,127,569,843 | I_kwDOQb6kR872Bbez | 53,672 | https://github.com/openclaw/openclaw/issues/53672 | https://api.github.com/repos/openclaw/openclaw/issues/53672 | [Bug]: Control UI Not Working - Port 18789 | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading from 2026.3.13 to 2026.3.22 it worked fine, until I had to update my codex Oauth, after updating it, the Control UI shows this error:
Control UI assets not found. Build them with `pnpm ui:build` (auto-installs UI deps), or run `pnpm ui:de... | open | null | false | 1 | [
"bug",
"regression"
] | [] | 2026-03-24T12:36:09Z | 2026-03-24T12:39:57Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | alfred-romero-bot | 264,278,182 | U_kgDOD8CQpg | User | false |
openclaw/openclaw | 4,127,572,361 | I_kwDOQb6kR872BcGJ | 53,673 | https://github.com/openclaw/openclaw/issues/53673 | https://api.github.com/repos/openclaw/openclaw/issues/53673 | Telegram outbound audio replies lack a way to force sendAudio/sendVoice from normal assistant MEDIA replies | ## Summary
OpenClaw clearly supports Telegram audio sending internally, including `sendVoice`, `asVoice`, `audioAsVoice`, and the `[[audio_as_voice]]` reply tag. However, from a normal assistant reply path using `MEDIA:` / generic media reply payloads, there does not appear to be a reliable way to force Telegram to se... | open | null | false | 0 | [] | [] | 2026-03-24T12:36:29Z | 2026-03-24T12:36:29Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | 5toCode | 70,207,997 | MDQ6VXNlcjcwMjA3OTk3 | User | false |
openclaw/openclaw | 4,127,597,065 | I_kwDOQb6kR872BiIJ | 53,674 | https://github.com/openclaw/openclaw/issues/53674 | https://api.github.com/repos/openclaw/openclaw/issues/53674 | [Bug]: Subagent deliveryContext inheritance broken - responses never reach parent session | ## Bug Description
Subagents spawned via `sessions_spawn` complete successfully but their responses never reach the parent session or user. The root cause is missing `deliveryContext.to` and `deliveryContext.accountId` inheritance from parent to child session.
## Steps to Reproduce
1. **Create parent session** with ... | open | null | false | 0 | [] | [] | 2026-03-24T12:39:43Z | 2026-03-24T12:41:34Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | GSVSenseAmidMadness | 95,069,682 | U_kgDOBaql8g | User | false |
openclaw/openclaw | 4,127,628,097 | I_kwDOQb6kR872BptB | 53,678 | https://github.com/openclaw/openclaw/issues/53678 | https://api.github.com/repos/openclaw/openclaw/issues/53678 | Feature request: Add Gateway received timestamp to inbound metadata for accurate latency tracking | # Feature Request: Add Gateway Received Timestamp to Inbound Metadata
## Problem
Currently, the inbound metadata only includes the message timestamp from the source channel (e.g., WeChat message timestamp). This makes it impossible to calculate accurate **end-to-end processing latency**.
For channels using **long-pol... | open | null | false | 0 | [] | [] | 2026-03-24T12:44:42Z | 2026-03-24T12:44:42Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | victhjy | 15,571,112 | MDQ6VXNlcjE1NTcxMTEy | User | false |
openclaw/openclaw | 4,127,640,305 | I_kwDOQb6kR872Bsrx | 53,679 | https://github.com/openclaw/openclaw/issues/53679 | https://api.github.com/repos/openclaw/openclaw/issues/53679 | [Bug]: Gateway fails after Migration Assistant — hardcoded TMPDIR path in LaunchAgent | ### Bug type
Crash (process/app exits or hangs)
### Summary
**Title:** Gateway fails after Migration Assistant — hardcoded TMPDIR path in LaunchAgent
**OpenClaw Version:** 2026.3.11
**Environment:**
- macOS 15 (Darwin 25.3.0, arm64)
- Migrated from Mac mini (M4) → MacBook Pro (M4 Max) via Migration Assistant
**Wh... | open | null | false | 0 | [
"bug",
"bug:crash"
] | [] | 2026-03-24T12:46:33Z | 2026-03-24T12:46:43Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | dlturock | 84,206,714 | MDQ6VXNlcjg0MjA2NzE0 | User | false |
openclaw/openclaw | 4,127,656,068 | I_kwDOQb6kR872BwiE | 53,680 | https://github.com/openclaw/openclaw/issues/53680 | https://api.github.com/repos/openclaw/openclaw/issues/53680 | [Bug] ACP binding routes to wrong agent on second message (first message works fine) |
## Bug Description
After upgrading to **2026.3.23-2**, an ACP binding to a Telegram group topic only handles the **first** incoming message correctly via the ACP backend (Claude/Codex). On the **second** message, the binding falls back to the main agent instead of continuing via ACP.
## Environment
- **OpenClaw ver... | open | null | false | 0 | [] | [] | 2026-03-24T12:49:19Z | 2026-03-24T12:49:19Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | pengziran | 133,945,695 | U_kgDOB_vZXw | User | false |
openclaw/openclaw | 4,127,701,065 | I_kwDOQb6kR872B7hJ | 53,684 | https://github.com/openclaw/openclaw/issues/53684 | https://api.github.com/repos/openclaw/openclaw/issues/53684 | Feature Request: Gateway failure recovery and notification mechanism | ### Summary
Silence after a system crash
### Problem to solve
**The Problem:**
When the OpenClaw gateway fails to start or crashes at runtime, the failure is completely silent. There is:
- No on-screen notification
- No email or push alert
- No health check that detects the outage
- No automatic recovery attempt b... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-03-24T12:56:11Z | 2026-03-24T12:56:11Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | dlturock | 84,206,714 | MDQ6VXNlcjg0MjA2NzE0 | User | false |
openclaw/openclaw | 4,127,707,414 | I_kwDOQb6kR872B9EW | 53,685 | https://github.com/openclaw/openclaw/issues/53685 | https://api.github.com/repos/openclaw/openclaw/issues/53685 | External plugins fail to resolve openclaw/plugin-sdk/* imports — alias map missing argv1 fallback | ## Summary
External/community plugins installed at `~/.openclaw/extensions/` fail to load with:
```
Cannot find module 'openclaw/plugin-sdk/plugin-entry'
Require stack:
- /home/user/.openclaw/extensions/my-plugin/index.ts
```
This affects all non-bundled plugins installed via `openclaw plugins install` (both local p... | open | null | false | 0 | [] | [] | 2026-03-24T12:57:14Z | 2026-03-24T12:57:14Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Alex-wuhu | 57,132,813 | MDQ6VXNlcjU3MTMyODEz | User | false |
openclaw/openclaw | 4,127,715,741 | I_kwDOQb6kR872B_Gd | 53,686 | https://github.com/openclaw/openclaw/issues/53686 | https://api.github.com/repos/openclaw/openclaw/issues/53686 | [Bug]: Telegram: all outbound API calls fail (sendMessage, sendChatAction, editMessageText) while inbound polling works | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
On Windows 11 with Node v25, all outbound Telegram API calls fail with "Network request failed" while inbound polling (getUpdates) works correctly.
### Steps to reproduce
1. Install OpenClaw on Windows 11
2. Configure Telegram channel wit... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T12:58:36Z | 2026-03-24T18:54:27Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | filnadyir | 179,033,413 | U_kgDOCqvVRQ | User | false |
openclaw/openclaw | 4,127,721,598 | I_kwDOQb6kR872CAh- | 53,687 | https://github.com/openclaw/openclaw/issues/53687 | https://api.github.com/repos/openclaw/openclaw/issues/53687 | [Bug]:webchat: image/PDF visual inspection fails with 'No media-understanding provider registered' | ### Bug type
Regression (worked before, now fails)
### Summary
In OpenClaw webchat, Ballie can fetch Dropbox files and even render a linked PDF to PNG locally, but cannot visually inspect screenshots or rendered PDF pages because image analysis fails with a missing media-understanding provider.
### Steps to repro... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T12:59:31Z | 2026-03-24T12:59:40Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | theprocessguru | 17,456,876 | MDQ6VXNlcjE3NDU2ODc2 | User | false |
openclaw/openclaw | 4,127,723,021 | I_kwDOQb6kR872CA4N | 53,689 | https://github.com/openclaw/openclaw/issues/53689 | https://api.github.com/repos/openclaw/openclaw/issues/53689 | [Bug]: Model not found: moonshot/kimi-k2.5. Update agents.defaults.model or run /models list. | ### Bug type
Regression (worked before, now fails)
### Summary
I use `openclaw agents add cms` to add a new agent, and get:
`....
◇ Agent "cms" already exists. Update it?
│ Yes
│
◇ Workspace directory
│ /home/zhouych/.openclaw/workspace-cms
│
◇ Configure model/auth for this agent now?
│ Yes
20:49:35 [plugins] ... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T12:59:43Z | 2026-03-24T13:00:35Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | zhouych87 | 16,185,464 | MDQ6VXNlcjE2MTg1NDY0 | User | false |
openclaw/openclaw | 4,127,725,448 | I_kwDOQb6kR872CBeI | 53,690 | https://github.com/openclaw/openclaw/issues/53690 | https://api.github.com/repos/openclaw/openclaw/issues/53690 | [Bug]: openclaw.json MiniMax API-key config regressed to incompatible OpenAI endpoint | ## Summary
The recent config-only commit `4ce9d57d558649b2ec912ce4396157a7688342e3` switched `models.providers.minimax` in `openclaw.json` to the OpenAI-compatible endpoint:
- `baseUrl: https://api.minimax.io/v1`
- `api: openai-completions`
In this environment, the installed OpenClaw package documents and ships the M... | open | null | false | 2 | [] | [] | 2026-03-24T13:00:05Z | 2026-03-24T13:01:32Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | cct08311github | 145,902,126 | U_kgDOCLJKLg | User | false |
openclaw/openclaw | 4,127,729,761 | I_kwDOQb6kR872CChh | 53,691 | https://github.com/openclaw/openclaw/issues/53691 | https://api.github.com/repos/openclaw/openclaw/issues/53691 | [Bug]: openclaw system event silently swallowed since 2026.3.12 — gateway never logs receipt or sends message | ## Summary
`openclaw system event --text "..." --mode now` returns `{"ok": true}` but the gateway never logs receipt of the event and no message is delivered. Regression introduced in **2026.3.12** (released 12 March 2026).
## Steps to Reproduce
```bash
openclaw system event --text "Test ping" --mode now --json
# Re... | open | null | false | 0 | [] | [] | 2026-03-24T13:00:40Z | 2026-03-24T13:00:40Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | mikeevenssplash | 226,969,575 | U_kgDODYdH5w | User | false |
openclaw/openclaw | 4,127,753,571 | I_kwDOQb6kR872CIVj | 53,696 | https://github.com/openclaw/openclaw/issues/53696 | https://api.github.com/repos/openclaw/openclaw/issues/53696 | <relevant-memories> block visible in webchat UI after v3.23 update | After updating to v3.23.2, the `<relevant-memories>` block (injected by memory-lancedb-pro auto-recall) is now visible in the webchat transcript. This block should be internal context only and was not visible before the v3.23 update.
## Steps to Reproduce
1. Install memory-lancedb-pro with autoCapture and autoRecall e... | open | null | false | 0 | [] | [] | 2026-03-24T13:03:55Z | 2026-03-24T13:03:55Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | anisakis2003 | 264,262,100 | U_kgDOD8BR1A | User | false |
openclaw/openclaw | 4,127,755,357 | I_kwDOQb6kR872CIxd | 53,697 | https://github.com/openclaw/openclaw/issues/53697 | https://api.github.com/repos/openclaw/openclaw/issues/53697 | [Bug] Feishu send action: card field incorrectly marked as required in schema (3.23 regression) |
## Bug Description
**OpenClaw version:** 2026.3.23-2
**Channel:** Feishu
**Action:** message send (with media)
When trying to send an image via the Feishu channel using the \`message\` tool with \`media\` parameter (file path or URL), the tool fails with validation error:
```
Validation failed for tool \"messag... | closed | completed | false | 1 | [] | [] | 2026-03-24T13:04:09Z | 2026-03-24T17:15:12Z | 2026-03-24T17:15:12Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | rrriiiccckkk | 57,145,789 | MDQ6VXNlcjU3MTQ1Nzg5 | User | false |
openclaw/openclaw | 4,127,763,003 | I_kwDOQb6kR872CKo7 | 53,698 | https://github.com/openclaw/openclaw/issues/53698 | https://api.github.com/repos/openclaw/openclaw/issues/53698 | [Bug]: WhatsApp watchdog MESSAGE_TIMEOUT_MS (30min) not configurable — causes reconnect loops on low-traffic setups | ### Summary
The WhatsApp web monitor watchdog has a hardcoded `MESSAGE_TIMEOUT_MS` of 30 minutes (1800s). On low-traffic setups (e.g., DM-only with `dmPolicy: allowlist`), it's common to go hours without inbound messages. The watchdog treats this as a dead connection and force-closes it with status 499 ("watchdog-time... | open | null | false | 0 | [] | [] | 2026-03-24T13:05:09Z | 2026-03-24T13:05:09Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | DevInvestor | 252,522,253 | U_kgDODw0vDQ | User | false |
openclaw/openclaw | 4,127,818,095 | I_kwDOQb6kR872CYFv | 53,700 | https://github.com/openclaw/openclaw/issues/53700 | https://api.github.com/repos/openclaw/openclaw/issues/53700 | Signal: incoming emoji reactions arrive as <media:unknown> | ## Description
When a Signal user reacts to a message with an emoji (e.g. 👍), it is delivered to the agent as `<media:unknown>` rather than a structured reaction event.
## Expected Behavior
Incoming reactions should be parsed into a structured event, e.g.:
```json
{
"type": "reaction",
"emoji": "👍",
"targetM... | open | null | false | 0 | [] | [] | 2026-03-24T13:11:54Z | 2026-03-24T13:11:54Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ziggy-broze | 269,773,776 | U_kgDOEBRr0A | User | false |
openclaw/openclaw | 4,127,829,192 | I_kwDOQb6kR872CazI | 53,701 | https://github.com/openclaw/openclaw/issues/53701 | https://api.github.com/repos/openclaw/openclaw/issues/53701 | [Feature Request]: Show human-readable Telegram forum topic names in Control UI session selector | ## Problem
The Control UI session/channel dropdown shows raw session keys like `telegram:group:-1003874130048:topic:5` — but users can't tell at a glance which topic that is (e.g. ⚙️Cron, 📱SNS, 💻Dev).
Telegram's Bot API already knows the topic names (they're set when creating forum topics). The Control UI has no wa... | open | null | false | 0 | [] | [] | 2026-03-24T13:13:34Z | 2026-03-24T13:13:34Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | nikkiai0024 | 263,112,459 | U_kgDOD67HCw | User | false |
openclaw/openclaw | 4,127,848,546 | I_kwDOQb6kR872Cfhi | 53,703 | https://github.com/openclaw/openclaw/issues/53703 | https://api.github.com/repos/openclaw/openclaw/issues/53703 | [Bug]: message.send tool echoes outbound WhatsApp messages into message inbound stream (duplicate message_id) | This is a regression. Works fine in [2026.3.11](https://github.com/openclaw/openclaw/releases/tag/v2026.3.11) but not in [2026.3.23](https://github.com/openclaw/openclaw/releases/tag/v2026.3.23)
# Summary
**In whatsapp self-chat mode**, using the message.send tool to send WhatsApp messages to myself results in the res... | open | null | false | 1 | [] | [] | 2026-03-24T13:16:24Z | 2026-03-24T15:22:39Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Henryy417 | 72,913,927 | MDQ6VXNlcjcyOTEzOTI3 | User | false |
openclaw/openclaw | 4,127,861,694 | I_kwDOQb6kR872Ciu- | 53,705 | https://github.com/openclaw/openclaw/issues/53705 | https://api.github.com/repos/openclaw/openclaw/issues/53705 | [Bug]: openclaw agents delete --force incorrectly deletes shared workspace instead of agent-specific files | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When running openclaw agents delete <agent-id> --force on an agent that uses a shared workspace (i.e., workspace path that is also used by other agents), the command incorrectly moves the entire shared workspace to Trash, instead of:
1. Ref... | open | null | false | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T13:18:31Z | 2026-03-24T13:19:31Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Torinbob | 41,232,273 | MDQ6VXNlcjQxMjMyMjcz | User | false |
openclaw/openclaw | 4,127,942,830 | I_kwDOQb6kR872C2iu | 53,708 | https://github.com/openclaw/openclaw/issues/53708 | https://api.github.com/repos/openclaw/openclaw/issues/53708 | [Bug]: `openai-codex/gpt-5.4` is listed/configurable as image-capable, but `image` tool fails with `No media-understanding provider registered for openai-codex` | ## Summary
`openai-codex/gpt-5.4` appears in model listing as supporting image + text, and OpenClaw accepts it as the configured image model, but the actual `image` tool fails at runtime with:
`No media-understanding provider registered for openai-codex`
This makes the current behavior internally inconsistent: model... | open | null | false | 0 | [] | [] | 2026-03-24T13:30:43Z | 2026-03-24T13:30:43Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ninjaleoai | 258,920,567 | U_kgDOD27Qdw | User | false |
openclaw/openclaw | 4,127,943,395 | I_kwDOQb6kR872C2rj | 53,709 | https://github.com/openclaw/openclaw/issues/53709 | https://api.github.com/repos/openclaw/openclaw/issues/53709 | [Bug]: OpenClaw exec subprocesses expose invalid GH_TOKEN, breaking gh even when gh auth login is valid | ### Bug type
Regression (worked before, now fails)
### Summary
Observed behavior
- In a normal shell, gh api user works correctly.
- I... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T13:30:48Z | 2026-03-24T13:30:58Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | mackelito | 598,463 | MDQ6VXNlcjU5ODQ2Mw== | User | false |
openclaw/openclaw | 4,127,971,073 | I_kwDOQb6kR872C9cB | 53,712 | https://github.com/openclaw/openclaw/issues/53712 | https://api.github.com/repos/openclaw/openclaw/issues/53712 | Compaction flush rewrites entire file, breaking concurrent Edit tool operations | ## Summary
When the memory flush fires (pre-compaction), it rewrites workspace files (e.g., `NOW.md`) in their entirety — even sections that did not change. This causes the `Edit` tool's exact-match requirement to fail on unchanged sections, triggering visible `⚠️ Edit failed` notifications to the user.
The bot recov... | open | null | false | 0 | [] | [] | 2026-03-24T13:34:59Z | 2026-03-24T13:34:59Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | siraustin | 3,533,969 | MDQ6VXNlcjM1MzM5Njk= | User | false |
openclaw/openclaw | 4,128,006,829 | I_kwDOQb6kR872DGKt | 53,714 | https://github.com/openclaw/openclaw/issues/53714 | https://api.github.com/repos/openclaw/openclaw/issues/53714 | Discord image attachments not passing URL to image analysis tool | ## Problem
When a user sends an image attachment via Discord, the image tool receives only the placeholder string `attachment` instead of the actual Discord CDN URL. This prevents the image from being analyzed.
**Error message:**
```
Local media path is not under an allowed directory: attachment
```
## Expected Beha... | open | null | false | 0 | [] | [] | 2026-03-24T13:39:37Z | 2026-03-24T13:39:37Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | zcode-apps | 87,060,116 | MDQ6VXNlcjg3MDYwMTE2 | User | false |
openclaw/openclaw | 4,128,085,914 | I_kwDOQb6kR872DZea | 53,719 | https://github.com/openclaw/openclaw/issues/53719 | https://api.github.com/repos/openclaw/openclaw/issues/53719 | [Feature]: 我在查看思考过程时，界面能保持在我滚动到的位置，而不是自动跳转到最新的思考内容。 | ### Summary
```markdown
---
title: "🔧 Fix: Lock chat scroll when user scrolls up to view thinking/reasoning content"
labels: ["bug", "enhancement", "ui/ux", "priority-medium"]
---
## 🐛 Bug Description
When the thinking/reasoning process is streaming and the user scrolls up to view earlier thinking content, the UI ... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-03-24T13:50:03Z | 2026-03-24T13:50:03Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | dlanker | 62,333,996 | MDQ6VXNlcjYyMzMzOTk2 | User | false |
openclaw/openclaw | 4,128,095,763 | I_kwDOQb6kR872Db4T | 53,721 | https://github.com/openclaw/openclaw/issues/53721 | https://api.github.com/repos/openclaw/openclaw/issues/53721 | Telegram channel fails to initialize after gateway restart (regression in 2026.3.23-2) | ### Bug type
Regression (worked before, now fails)
### Summary
Telegram channel fails to initialize after any gateway restart — no inbound messages received, outbound via API works
### Steps to reproduce
1. Gateway is running with Telegram channel active and working
2. Any gateway restart occurs (SIGUSR1, SIGTERM,... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-03-24T13:51:26Z | 2026-03-24T17:15:31Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | shaktifxknight-dot | 270,670,922 | U_kgDOECIcSg | User | false |
openclaw/openclaw | 4,128,115,056 | I_kwDOQb6kR872Dglw | 53,722 | https://github.com/openclaw/openclaw/issues/53722 | https://api.github.com/repos/openclaw/openclaw/issues/53722 | [Bug] TTS 语音通过非 WEB 渠道发送时变成文件路径字符串 | ## 问题描述
通过飞书机器人发送语音消息时，机器人回复的 TTS 语音收到的是一个 mp3 文件路径字符串，而不是可播放的语音。
## 问题原因
OpenClaw 在 `reply-payload.ts` 中处理回复消息时，只检查了 `text` 和 `mediaUrl`，**完全忽略了 `audioAsVoice`（TTS 语音）**。
TTS 语音被生成后放入 `payload.audioAsVoice`，但消息发送逻辑没有处理这个字段，导致：
1. TTS 语音被生成
2. 发送逻辑忽略了 `audioAsVoice`
3. 飞书收到的是文本（文件路径）
## 期望行为
- **WEB 渠道**：保持现有逻辑，页... | open | null | false | 0 | [] | [] | 2026-03-24T13:54:06Z | 2026-03-24T13:54:06Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | couger | 4,597,816 | MDQ6VXNlcjQ1OTc4MTY= | User | false |
openclaw/openclaw | 4,128,129,252 | I_kwDOQb6kR872DkDk | 53,723 | https://github.com/openclaw/openclaw/issues/53723 | https://api.github.com/repos/openclaw/openclaw/issues/53723 | Bug: openai-codex OAuth token shows 'expires in 0m' despite valid JWT (display issue) | ## Summary
`openclaw models status` consistently reports `openai-codex:default ok expires in 0m` even when the underlying JWT access token has a valid expiry well into the future.
## Steps to Reproduce
1. Run `openclaw models auth login --provider openai-codex` and complete the OAuth flow
2. Run `openclaw models sta... | open | null | false | 0 | [] | [] | 2026-03-24T13:56:07Z | 2026-03-24T13:56:07Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | DolencLuka | 266,726,612 | U_kgDOD-Xs1A | User | false |
openclaw/openclaw | 4,128,145,735 | I_kwDOQb6kR872DoFH | 53,724 | https://github.com/openclaw/openclaw/issues/53724 | https://api.github.com/repos/openclaw/openclaw/issues/53724 | [Bug] Feishu message tool fails: card required property error | **Description**
The Feishu message tool fails with validation error:
Validation failed for tool message: card: must have required property card
This happens even with minimal arguments like sending plain text.
**Environment**
- OpenClaw: 2026.3.23-2
- Platform: Linux WSL2
- Feishu plugin enabled
**Expected**: ... | closed | completed | false | 1 | [] | [] | 2026-03-24T13:58:43Z | 2026-03-24T17:15:11Z | 2026-03-24T17:15:11Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | xion516 | 187,609,864 | U_kgDOCy6zCA | User | false |
openclaw/openclaw | 4,128,155,674 | I_kwDOQb6kR872Dqga | 53,726 | https://github.com/openclaw/openclaw/issues/53726 | https://api.github.com/repos/openclaw/openclaw/issues/53726 | [Bug]: Antivirus software blocks Trojans. | ### Bug type
Regression (worked before, now fails)
### Summary
It worked normally without any blocking before yesterday. After executing git clone https://github.com/openclaw/openclaw.git . today, Kaspersky blocked the downloaded files.
<img width="1490" height="548" alt="Image" src="https://github.com/user-attachm... | closed | completed | false | 1 | [
"bug",
"regression"
] | [] | 2026-03-24T14:00:15Z | 2026-03-24T14:56:55Z | 2026-03-24T14:56:55Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | bubbl000 | 108,770,692 | U_kgDOBnu1hA | User | false |
openclaw/openclaw | 4,128,193,753 | I_kwDOQb6kR872DzzZ | 53,727 | https://github.com/openclaw/openclaw/issues/53727 | https://api.github.com/repos/openclaw/openclaw/issues/53727 | Feature Request: Add Custom System Prompt Configuration for Mandatory Rules | ## Feature Request: Add Custom System Prompt Configuration
### Problem
Currently, OpenClaw only supports loading files via hooks (like `bootstrap-extra-files` and `before-prompt-build`) to inject context into the agent's prompts. However:
1. Files loaded via hooks are added as **user/assistant messages in context**,... | open | null | false | 0 | [] | [] | 2026-03-24T14:06:00Z | 2026-03-24T14:06:00Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | chenhao2062818194-rgb | 268,841,394 | U_kgDOEAYxsg | User | false |
openclaw/openclaw | 4,128,201,005 | I_kwDOQb6kR872D1kt | 53,729 | https://github.com/openclaw/openclaw/issues/53729 | https://api.github.com/repos/openclaw/openclaw/issues/53729 | [Bug]: WhatsApp plugin: outbound messages echo back as inbound | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
• Qué pasa: cada mensaje que envías reaparece como si te lo hubiera enviado Ingrid (la IA).
• Cuándo empezó: desde el primer día después de configurar
• Gateway config: gateway.bind=lan
• OpenClaw version: Versión: OpenClaw 2026.3.23-1
###... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T14:07:01Z | 2026-03-24T14:36:33Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | jjmartinartesi-a11y | 270,672,194 | U_kgDOECIhQg | User | false |
openclaw/openclaw | 4,128,233,243 | I_kwDOQb6kR872D9cb | 53,731 | https://github.com/openclaw/openclaw/issues/53731 | https://api.github.com/repos/openclaw/openclaw/issues/53731 | Bug: MEMORY.md is silently truncated during bootstrap | The agent bootstrap file `MEMORY.md` gets silently truncated in the injected context when it exceeds a certain character limit (e.g., 20,000).
This happens on every session start, which means the agent is consistently working with an incomplete memory without any warning to the user.
### Propuestas de solución:
1. ... | open | null | false | 0 | [] | [] | 2026-03-24T14:11:12Z | 2026-03-24T14:11:12Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | alvaro630 | 245,734,636 | U_kgDODqWc7A | User | false |
openclaw/openclaw | 4,128,233,824 | I_kwDOQb6kR872D9lg | 53,732 | https://github.com/openclaw/openclaw/issues/53732 | https://api.github.com/repos/openclaw/openclaw/issues/53732 | Bug: Fallback models echo BOOT.md instructions instead of executing them | When the primary model is unavailable and a fallback model (e.g., Gemini-2.5-pro) takes over, it sometimes echoes the `BOOT.md` startup instructions verbatim to the user instead of executing them silently.
This exposes internal system prompts and provides a confusing user experience.
### Propuestas de solución:
1. ... | open | null | false | 0 | [] | [] | 2026-03-24T14:11:18Z | 2026-03-24T14:11:18Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | alvaro630 | 245,734,636 | U_kgDODqWc7A | User | false |
openclaw/openclaw | 4,128,240,581 | I_kwDOQb6kR872D_PF | 53,733 | https://github.com/openclaw/openclaw/issues/53733 | https://api.github.com/repos/openclaw/openclaw/issues/53733 | [Feature]: OpenClaw App for Intel Chip Macs | ### Summary
Could you pwease make it for intel macs too ? We feel left out...
### Problem to solve
App not compatible with intel macs
### Proposed solution
Make app for intel macs :)
### Alternatives considered
_No response_
### Impact
Affects all Intel mac users
It's pretty severe
### Evidence/examples
_No... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-03-24T14:12:14Z | 2026-03-24T14:12:14Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | naeldavid | 100,679,094 | U_kgDOBgA9tg | User | false |
openclaw/openclaw | 4,128,242,367 | I_kwDOQb6kR872D_q_ | 53,734 | https://github.com/openclaw/openclaw/issues/53734 | https://api.github.com/repos/openclaw/openclaw/issues/53734 | [Bug]: toNormalizedUsage() discards accumulated input/cache tokens — cost tracking underreports ~80% of actual billed usage | ## Summary
`toNormalizedUsage()` in `src/agents/pi-embedded-runner/run.ts` (line ~192) intentionally returns only the **last API call's** `input`, `cacheRead`, and `cacheWrite` tokens instead of the accumulated values from the `UsageAccumulator`. While `output` tokens are correctly accumulated, input/cache tokens are ... | open | null | false | 1 | [] | [] | 2026-03-24T14:12:29Z | 2026-03-24T14:53:04Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Ponera-Syl | 120,110,195 | U_kgDOByi8cw | User | false |
openclaw/openclaw | 4,128,249,032 | I_kwDOQb6kR872EBTI | 53,735 | https://github.com/openclaw/openclaw/issues/53735 | https://api.github.com/repos/openclaw/openclaw/issues/53735 | Kimi k2p5 模型工具调用参数丢失问题 | ### Bug type
Regression (worked before, now fails)
### Summary
在 OpenClaw 系统中使用 Kimi k2p5 模型时，exec、write、edit 等工具调用的参数会**静默丢失**，导致工具调用失败，返回 `Validation failed ... Received arguments: {}`。
### Steps to reproduce
1. 启动 OpenClaw 会话，使用默认模型 Kimi k2p5
2. 进行多次工具调用（exec, read, write, edit 等）
3. 观察工具参数被拦截为空对象 `{}`
4. 切换模型到... | closed | completed | false | 2 | [
"bug",
"regression"
] | [] | 2026-03-24T14:13:26Z | 2026-03-24T17:58:05Z | 2026-03-24T17:58:05Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | neverlanding | 26,531,865 | MDQ6VXNlcjI2NTMxODY1 | User | false |
openclaw/openclaw | 4,128,351,470 | I_kwDOQb6kR872EaTu | 53,739 | https://github.com/openclaw/openclaw/issues/53739 | https://api.github.com/repos/openclaw/openclaw/issues/53739 | [Bug]: Feishu MI messages not responding after upgrading to v2026.3.23-2 and installing memory-lancedb plugin | ### Bug type
Regression (worked before, now fails)
### Summary
After executing npm install -g openclaw@latest to upgrade to the latest version v2026.3.23-2, and installing the memory-lancedb plugin, followed by gateway restart, Feishu MI (Mini App) messages stopped responding entirely.
### Steps to reproduce
1. Ru... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T14:28:26Z | 2026-03-24T14:28:58Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | jamesMuWB | 102,423 | MDQ6VXNlcjEwMjQyMw== | User | false |
openclaw/openclaw | 4,128,419,932 | I_kwDOQb6kR872ErBc | 53,742 | https://github.com/openclaw/openclaw/issues/53742 | https://api.github.com/repos/openclaw/openclaw/issues/53742 | Bug: macOS gateway install --force resolves SecretRef values into plaintext LaunchAgent plist and triggers token mismatch loop | ## Summary
`openclaw gateway install --force` resolves every `SecretRef` in `openclaw.json` and writes the resolved values as plaintext into the macOS LaunchAgent plist (`~/Library/LaunchAgents/ai.openclaw.gateway.plist`).
This defeats the purpose of moving secrets out of `openclaw.json` and into `~/.openclaw/.env` u... | open | null | false | 1 | [] | [] | 2026-03-24T14:37:54Z | 2026-03-24T14:44:51Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | beto-sudo | 229,640,506 | U_kgDODbAJOg | User | false |
openclaw/openclaw | 4,128,472,473 | I_kwDOQb6kR872E32Z | 53,744 | https://github.com/openclaw/openclaw/issues/53744 | https://api.github.com/repos/openclaw/openclaw/issues/53744 | [Bug]: Missing "image-generation" capability in Google plugin – Technical evidence | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Hi everyone,
I would like to report an issue regarding Gemini's integration for image generation in OpenClaw. Despite google/gemini-3-pro-image-preview being mentioned in the documentation, the Google plugin does not seem to register the n... | open | null | false | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T14:45:09Z | 2026-03-24T14:47:19Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ilcorsair | 95,966,133 | U_kgDOBbhTtQ | User | false |
openclaw/openclaw | 4,128,478,729 | I_kwDOQb6kR872E5YJ | 53,745 | https://github.com/openclaw/openclaw/issues/53745 | https://api.github.com/repos/openclaw/openclaw/issues/53745 | Gateway freezes after nested subagent activity, stops all Telegram polling | ## Description
After a coordinator agent calls subagents via `sessions_send`, the gateway process stays alive (HTTP 200 on healthcheck, PID running) but completely stops processing Telegram updates. Log output stops entirely — no incoming messages, no `health-monitor`, no `sendMessage`, nothing. All Telegram bots go s... | open | null | false | 0 | [] | [] | 2026-03-24T14:46:00Z | 2026-03-24T14:46:00Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | rusvtiem | 263,725,847 | U_kgDOD7gjFw | User | false |
openclaw/openclaw | 4,128,523,836 | I_kwDOQb6kR872FEY8 | 53,746 | https://github.com/openclaw/openclaw/issues/53746 | https://api.github.com/repos/openclaw/openclaw/issues/53746 | Bug: `saveCronStore` overwrites jobs.json from partial in-memory state after restart, causing silent job loss | ## Summary
When the gateway restarts and a cron fires before full state is loaded, `saveCronStore` writes a partial in-memory job list over the full on-disk job list — silently wiping all jobs not present in memory at that moment. We lost 86 jobs twice in 24 hours from this.
## Root Cause
**File:** `src/cron/store.t... | open | null | false | 0 | [] | [] | 2026-03-24T14:52:13Z | 2026-03-24T14:52:13Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | dustinprojectcoordinator-alt | 268,295,773 | U_kgDOD_3eXQ | User | false |
openclaw/openclaw | 4,128,527,581 | I_kwDOQb6kR872FFTd | 53,747 | https://github.com/openclaw/openclaw/issues/53747 | https://api.github.com/repos/openclaw/openclaw/issues/53747 | Regression: Kimi Coding (official Anthropic-compatible endpoint) emits malformed read tool calls on current OpenClaw builds | ## Summary
Kimi Coding via the official Kimi Code Anthropic-compatible endpoint started producing malformed tool calls after upgrading OpenClaw. The most visible failure is repeated `read` calls with no `path` / `file_path`.
This looks like a fresh regression on current builds, not the old Moonshot Direct / OpenAI-c... | open | null | false | 0 | [] | [] | 2026-03-24T14:52:44Z | 2026-03-24T14:52:44Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Eoniva616 | 129,255,883 | U_kgDOB7RJyw | User | false |
openclaw/openclaw | 4,128,542,647 | I_kwDOQb6kR872FI-3 | 53,749 | https://github.com/openclaw/openclaw/issues/53749 | https://api.github.com/repos/openclaw/openclaw/issues/53749 | Bug: message tool schema incorrectly marks 'buttons' as required for all actions | ## Summary
The `message` tool schema has `buttons` marked as a required parameter, but this breaks non-button actions like `action=react`.
## Steps to Reproduce
1. Try to use the message tool to add a reaction:
```json
{
"action": "react",
"emoji": "👋"
}
```
2. Validation fails with:
```
Validation failed fo... | open | null | false | 0 | [] | [] | 2026-03-24T14:55:07Z | 2026-03-24T14:55:07Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | justinbao19 | 97,433,668 | U_kgDOBc64RA | User | false |
openclaw/openclaw | 4,128,543,726 | I_kwDOQb6kR872FJPu | 53,750 | https://github.com/openclaw/openclaw/issues/53750 | https://api.github.com/repos/openclaw/openclaw/issues/53750 | Feature request: per-agent self-improving memory isolation |
**Problem**
Currently, the self-improving skill stores memory at \`~/self-improving/\` which resolves to the system home directory. All agents share the same path, meaning their self-improving memory (corrections, patterns, preferences) is physically mixed together in \`/Users/binary/self-improving/\`.
This makes it... | open | null | false | 1 | [] | [] | 2026-03-24T14:55:17Z | 2026-03-24T15:26:20Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | tontonbinary | 261,551,266 | U_kgDOD5b0og | User | false |
openclaw/openclaw | 4,128,545,398 | I_kwDOQb6kR872FJp2 | 53,751 | https://github.com/openclaw/openclaw/issues/53751 | https://api.github.com/repos/openclaw/openclaw/issues/53751 | Google image generation fails with HTTP 404: x-goog-api-key header auth not accepted by generateContent endpoint | ## Bug Report
### Environment
- **OpenClaw version:** 2026.3.23-2
- **OS:** macOS 26.3.1 (arm64)
- **Auth:** `GEMINI_API_KEY` env var (API key from Google AI Studio)
### Description
The bundled Google image generation provider (`dist/extensions/google/image-generation-provider.js`) sends the API key via the `x-goog-... | open | null | false | 0 | [] | [] | 2026-03-24T14:55:33Z | 2026-03-24T14:55:33Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | montrellcruse | 130,179,762 | U_kgDOB8Jisg | User | false |
openclaw/openclaw | 4,128,548,331 | I_kwDOQb6kR872FKXr | 53,753 | https://github.com/openclaw/openclaw/issues/53753 | https://api.github.com/repos/openclaw/openclaw/issues/53753 | Control UI model picker incorrectly combines provider/model ID when switching models | ## Bug Description
When using the Control UI to switch to a model from a different provider, the UI incorrectly constructs the model ID by prefixing the current provider to the target model ID.
## Environment
- OpenClaw version: 2026.3.13
- OS: macOS (arm64)
- Gateway mode: local (loopback)
## Steps to Reproduce
1. S... | open | null | false | 0 | [] | [] | 2026-03-24T14:56:00Z | 2026-03-24T14:56:00Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | rollygpt6-cpu | 253,716,761 | U_kgDODx9pGQ | User | false |
openclaw/openclaw | 4,128,565,709 | I_kwDOQb6kR872FOnN | 53,755 | https://github.com/openclaw/openclaw/issues/53755 | https://api.github.com/repos/openclaw/openclaw/issues/53755 | [Bug]: openai-codex catalog exposes only gpt-5.4 in models list and native /models on 2026.3.13 | ## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
On OpenClaw `2026.3.13`, the `openai-codex` catalog is exposed as exactly one model (`openai-codex/gpt-5.4`) in both `openclaw models list` and Discord native `/models`, while other providers (for example `anthropic`) expose multiple models nor... | open | null | false | 0 | [] | [] | 2026-03-24T14:58:23Z | 2026-03-24T14:58:23Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | mastash3ff | 3,743,196 | MDQ6VXNlcjM3NDMxOTY= | User | false |
openclaw/openclaw | 4,128,566,612 | I_kwDOQb6kR872FO1U | 53,756 | https://github.com/openclaw/openclaw/issues/53756 | https://api.github.com/repos/openclaw/openclaw/issues/53756 | doctor --fix crashes with Config validation failed on models.providers.google.baseUrl | ## Bug
Running `openclaw doctor --fix` fails with:
```
Error: Config validation failed: models.providers.google.baseUrl: Invalid input: expected string, received undefined
```
## Reproduction
1. Have a legacy `skills.entries.nano-banana-pro` config with an `apiKey` (but no existing `models.providers.google` section... | closed | completed | false | 0 | [] | [] | 2026-03-24T14:58:30Z | 2026-03-24T16:47:23Z | 2026-03-24T16:47:23Z | CONTRIBUTOR | null | 20260324T233649Z | 2026-03-24T23:36:49Z | mahopan | 258,319,253 | U_kgDOD2WjlQ | User | false |
openclaw/openclaw | 4,128,573,732 | I_kwDOQb6kR872FQkk | 53,758 | https://github.com/openclaw/openclaw/issues/53758 | https://api.github.com/repos/openclaw/openclaw/issues/53758 | [control-ui] Model dropdown shows duplicates and strips provider prefix on selection | ## Bug Description
The control-UI model dropdown exhibits two related issues:
### 1. Duplicate entries
When a model appears in both `agents.defaults.models` (allowlist) and `agents.defaults.model.primary` or `agents.defaults.model.fallbacks`, the dropdown shows duplicate entries for that model.
### 2. Provider prefi... | open | null | false | 0 | [] | [] | 2026-03-24T14:59:27Z | 2026-03-24T14:59:27Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | bottenbenny | 270,688,955 | U_kgDOECJiuw | User | false |
openclaw/openclaw | 4,128,577,927 | I_kwDOQb6kR872FRmH | 53,759 | https://github.com/openclaw/openclaw/issues/53759 | https://api.github.com/repos/openclaw/openclaw/issues/53759 | [Bug]: Mac companion app does not respond to system.run probe — skills requiring Mac binaries fail | ### Summary
Mac companion app node fails to execute skills that require local binaries (`memo`, `things`). The gateway times out probing the Mac node for `system.run` capability, even though the Mac app is connected and otherwise functional.
### Environment
- **OpenClaw Gateway:** 2026.3.23-2 on Linux (Manjaro, x64)... | open | null | false | 0 | [] | [] | 2026-03-24T15:00:01Z | 2026-03-24T15:00:01Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | toby-ikebot | 269,862,444 | U_kgDOEBXGLA | User | false |
openclaw/openclaw | 4,128,586,834 | I_kwDOQb6kR872FTxS | 53,760 | https://github.com/openclaw/openclaw/issues/53760 | https://api.github.com/repos/openclaw/openclaw/issues/53760 | Support toolsBySender for DM channels (Signal, Telegram, etc.) | ## Description
`toolsBySender` currently works for group/channel contexts (IRC, Slack, MS Teams, Discord) but is not available for DM-level policies on channels like Signal and Telegram.
For agents that accept DMs from multiple senders via an allowlist (`dmPolicy: "allowlist"`), there is no config-level way to restri... | open | null | false | 0 | [] | [] | 2026-03-24T15:01:23Z | 2026-03-24T15:01:23Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ziggy-broze | 269,773,776 | U_kgDOEBRr0A | User | false |
openclaw/openclaw | 4,128,679,566 | I_kwDOQb6kR872FqaO | 53,761 | https://github.com/openclaw/openclaw/issues/53761 | https://api.github.com/repos/openclaw/openclaw/issues/53761 | image_generate: No image-generation provider registered for google after 2026.3.23-2 update | ## Bug Report
**Version:** 2026.3.23-2 (7ffe7e4)
## Summary
After updating to `2026.3.23-2`, the `image_generate` tool fails with:
```
No image-generation provider registered for google
```
The same config worked correctly on the previous version. Gateway restarts do not resolve the issue.
## Config (sanitized)
... | closed | completed | false | 2 | [] | [] | 2026-03-24T15:16:23Z | 2026-03-24T17:15:11Z | 2026-03-24T17:15:11Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | qearlyao | 201,136,044 | U_kgDOC_0XrA | User | false |
openclaw/openclaw | 4,128,706,275 | I_kwDOQb6kR872Fw7j | 53,763 | https://github.com/openclaw/openclaw/issues/53763 | https://api.github.com/repos/openclaw/openclaw/issues/53763 | [Feature]: Built-in headless browser for reliable web access without external dependencies | ### Summary
Bundle a headless Chromium instance as a first-class OpenClaw tool, so agents can reliably access any webpage (including JS-rendered and login-required pages) without depending on the user's Chrome or third-party APIs.
### Problem to solve
Agents currently rely on a fragile three-layer stack for web acce... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-03-24T15:20:11Z | 2026-03-24T15:20:11Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | luoziyan100 | 199,953,103 | U_kgDOC-sKzw | User | false |
openclaw/openclaw | 4,128,718,139 | I_kwDOQb6kR872Fz07 | 53,764 | https://github.com/openclaw/openclaw/issues/53764 | https://api.github.com/repos/openclaw/openclaw/issues/53764 | Tavily plugin tools (tavily_search, tavily_extract) not accessible in agent runtime | ## Bug Description
The Tavily plugin (`@openclaw/tavily-plugin`) is loaded and the skill shows as ready, but the `tavily_search` and `tavily_extract` tools are not accessible to the agent runtime. When attempting to invoke `tavily_search`, the agent receives "unknown tool" or similar error.
## Environment
- OpenClaw... | open | null | false | 1 | [] | [] | 2026-03-24T15:21:46Z | 2026-03-24T15:37:14Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Crabsticksalad | 262,156,320 | U_kgDOD6AwIA | User | false |
openclaw/openclaw | 4,128,729,896 | I_kwDOQb6kR872F2so | 53,765 | https://github.com/openclaw/openclaw/issues/53765 | https://api.github.com/repos/openclaw/openclaw/issues/53765 | 飞书话题群回复时无法正确识别话题，导致创建新话题而非回复到原话题 | ## 问题描述
在飞书话题群（topic group）中，当用户在话题下发送消息时，Agent 回复会创建新话题，而不是回复到当前话题中。
## 问题复现
1. 创建一个飞书话题群
2. 创建一个话题
3. 在话题中 @机器人 发送消息
4. 机器人回复时创建了新话题，而非在原话题下回复
## 根因分析
在 `src/messaging/inbound/dispatch-context.js` 中，判断是否为话题消息的逻辑是：
```javascript
const isThread = isGroup && Boolean(ctx.threadId);
```
但飞书话题群的消息结构中，话题内的回复消息通常通过 `r... | open | null | false | 0 | [] | [] | 2026-03-24T15:23:13Z | 2026-03-24T15:23:13Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | longfeihao | 49,780,193 | MDQ6VXNlcjQ5NzgwMTkz | User | false |
openclaw/openclaw | 4,128,743,539 | I_kwDOQb6kR872F6Bz | 53,766 | https://github.com/openclaw/openclaw/issues/53766 | https://api.github.com/repos/openclaw/openclaw/issues/53766 | message tool: react action incorrectly requires 'buttons' field | ## Problem
When using the `message` tool with `action: react`, validation fails with:
```
Validation failed for tool "message":
- buttons: must have required property 'buttons'
Received arguments:
{
"action": "react",
"emoji": "👍"
}
```
## Expected Behavior
`action: react` should only require `emoji` (and o... | closed | not_planned | false | 1 | [] | [] | 2026-03-24T15:25:21Z | 2026-03-24T15:25:53Z | 2026-03-24T15:25:53Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | justinbao19 | 97,433,668 | U_kgDOBc64RA | User | false |
openclaw/openclaw | 4,128,748,377 | I_kwDOQb6kR872F7NZ | 53,767 | https://github.com/openclaw/openclaw/issues/53767 | https://api.github.com/repos/openclaw/openclaw/issues/53767 | WhatsApp channel fails to load after gateway restart in 2026.3.22 | ## Problem
After upgrading to OpenClaw 2026.3.22, the WhatsApp channel stopped loading after gateway restarts. The channel was configured and had valid credentials, but simply did not start.
## Symptoms
- Gateway logs showed only Telegram starting, no WhatsApp initialization
- `openclaw channels list` did not show W... | open | null | false | 0 | [] | [] | 2026-03-24T15:26:08Z | 2026-03-24T15:26:08Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | justinbao19 | 97,433,668 | U_kgDOBc64RA | User | false |
openclaw/openclaw | 4,128,798,911 | I_kwDOQb6kR872GHi_ | 53,769 | https://github.com/openclaw/openclaw/issues/53769 | https://api.github.com/repos/openclaw/openclaw/issues/53769 | Cron delivery to Slack failing with 'Unsupported channel: slack' | # Cron delivery to Slack failing with "Unsupported channel: slack"
## Environment
- **Version:** OpenClaw 2026.3.22 (e7d11f6)
- **Platform:** macOS (Darwin 25.2.0, arm64)
- **Installation:** Git source (`/Users/aiden/clawdbot/`)
- **Gateway started:** 2026-03-24 10:13:44 CDT
## Problem
Cron jobs configured with `deli... | open | null | false | 1 | [] | [] | 2026-03-24T15:33:26Z | 2026-03-24T16:17:47Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | AIdenB899 | 269,310,343 | U_kgDOEA1Zhw | User | false |
openclaw/openclaw | 4,128,866,497 | I_kwDOQb6kR872GYDB | 53,771 | https://github.com/openclaw/openclaw/issues/53771 | https://api.github.com/repos/openclaw/openclaw/issues/53771 | Feishu buildMarkdownCard generates incompatible card format (code: 230099) | ## Bug Description
The `buildMarkdownCard` function in `send-Bz-uzZMi.js` generates a card format that is incompatible with the current Feishu API, causing error code 230099 (parse card json err).
## Root Cause
The function generates this format:
```json
{"schema":"2.0","config":{"wide_screen_mode":true},"body":{"el... | open | null | false | 0 | [] | [] | 2026-03-24T15:43:02Z | 2026-03-24T15:43:02Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | xiejin77 | 1,996,627 | MDQ6VXNlcjE5OTY2Mjc= | User | false |
openclaw/openclaw | 4,128,881,385 | I_kwDOQb6kR872Gbrp | 53,772 | https://github.com/openclaw/openclaw/issues/53772 | https://api.github.com/repos/openclaw/openclaw/issues/53772 | Use local model for compaction memory flush to reduce costs | ## Problem
Currently, when the conversation context is compacted (memory flush), the system prompt and compaction messages are processed by whatever model is currently active. This means cloud-based paid models are being used for internal housekeeping tasks, which is wasteful and expensive.
## Suggested Solution
Add... | open | null | false | 0 | [] | [] | 2026-03-24T15:45:11Z | 2026-03-24T15:45:11Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | limen96 | 40,869,957 | MDQ6VXNlcjQwODY5OTU3 | User | false |
openclaw/openclaw | 4,128,956,305 | I_kwDOQb6kR872Gt-R | 53,774 | https://github.com/openclaw/openclaw/issues/53774 | https://api.github.com/repos/openclaw/openclaw/issues/53774 | Feature Request: Custom TUI Footer Fields for Cost/Balance/Rate Limit Status | # Feature Request: Custom TUI Footer Fields
**Submitted by:** Major Domo (on behalf of Tim Walters)
**Date:** 2026-03-24
**OpenClaw Version:** 2026.3.22
## Summary
Request ability to customize the TUI footer to display:
1. **Estimated session cost** (based on token usage × model pricing)
2. **Remaining API balan... | open | null | false | 0 | [] | [] | 2026-03-24T15:55:38Z | 2026-03-24T15:55:38Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | timwalterseh-max | 255,151,790 | U_kgDODzVOrg | User | false |
openclaw/openclaw | 4,129,017,989 | I_kwDOQb6kR872G9CF | 53,781 | https://github.com/openclaw/openclaw/issues/53781 | https://api.github.com/repos/openclaw/openclaw/issues/53781 | [Bug]: Session not recovered after gateway crash — orphaned tool_use blocks cause permanent LLM rejection | ## Summary
When the gateway crashes mid-tool-execution, the session JSONL contains a `tool_use` block with no corresponding `tool_result`. On restart, every subsequent LLM request is permanently rejected by the Anthropic API:
```
LLM request rejected: messages.91: tool_use ids were found without tool_result blocks
im... | open | null | false | 0 | [] | [] | 2026-03-24T16:03:51Z | 2026-03-24T16:03:51Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | shiftymix | 193,249,208 | U_kgDOC4S_uA | User | false |
openclaw/openclaw | 4,129,030,365 | I_kwDOQb6kR872HADd | 53,783 | https://github.com/openclaw/openclaw/issues/53783 | https://api.github.com/repos/openclaw/openclaw/issues/53783 | [Bug]: Telegram group: cross-agent sessions_list visibility mismatch causes one-way sessions_send failure | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`Environment:
• OpenClaw
• Channel: Telegram
• Chat type: group
• Group id: -5141389644
Setup:
• Two agents in the same group:
• Agent A id: "tomer"
• Agent B id: "genadi"
Problem:
Agent A cannot resolve Agent B's group sessionKey, whil... | open | null | false | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T16:05:27Z | 2026-03-24T16:06:34Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | asafchoo | 45,887,109 | MDQ6VXNlcjQ1ODg3MTA5 | User | false |
openclaw/openclaw | 4,129,052,086 | I_kwDOQb6kR872HFW2 | 53,785 | https://github.com/openclaw/openclaw/issues/53785 | https://api.github.com/repos/openclaw/openclaw/issues/53785 | [Bug]: Linux: @matrix-org/matrix-sdk-crypto-nodejs missing from npm package (only darwin-arm64 binary bundled) | ### Bug type
Regression (worked before, now fails)
### Summary
After installing openclaw via npm install -g openclaw on Linux x64, the Matrix plugin fails with:
crypto runtime bootstrap failed: Cannot find module '@matrix-org/matrix-sdk-crypto-nodejs/download-lib.js'
Only matrix-sdk-crypto.darwin-arm6... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T16:08:24Z | 2026-03-24T16:08:40Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | marxst | 10,089,123 | MDQ6VXNlcjEwMDg5MTIz | User | false |
openclaw/openclaw | 4,129,068,356 | I_kwDOQb6kR872HJVE | 53,786 | https://github.com/openclaw/openclaw/issues/53786 | https://api.github.com/repos/openclaw/openclaw/issues/53786 | [Bug]: Real-time message & task status desync between Local Dashboard and Feishu Bot | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Chat messages and task status don't sync in real time between my local OpenClaw dashboard and the Feishu bot channel.
### Steps to reproduce
1. Set up and enable the Feishu integration in OpenClaw.
2. Open the local OpenClaw web dashboard... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T16:10:50Z | 2026-03-24T16:16:50Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | JoeDu-12 | 270,418,788 | U_kgDOEB5DZA | User | false |
openclaw/openclaw | 4,129,114,531 | I_kwDOQb6kR872HUmj | 53,790 | https://github.com/openclaw/openclaw/issues/53790 | https://api.github.com/repos/openclaw/openclaw/issues/53790 | [Bug]: Regression: tool-execution replies intermittently not delivered in webchat on 2026.3.22+ (works on 2026.3.13) | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading from 2026.3.13 to 2026.3.22 / 2026.3.23-2, webchat tool workflows become inconsistent: the assistant acknowledges work, but tool-result replies are intermittently not delivered back to chat.
Rolling back to 2026.3.13 restores expected beh... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T16:18:59Z | 2026-03-24T16:19:10Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | malcombecerra | 20,689,770 | MDQ6VXNlcjIwNjg5Nzcw | User | false |
openclaw/openclaw | 4,129,130,523 | I_kwDOQb6kR872HYgb | 53,794 | https://github.com/openclaw/openclaw/issues/53794 | https://api.github.com/repos/openclaw/openclaw/issues/53794 | iMessage/imsg channel config is accepted but gateway never starts provider on macOS | ## Summary
On macOS, the legacy iMessage `imsg` channel appears to be accepted/configured by OpenClaw, but the gateway never starts an iMessage provider and never surfaces the channel in `channels list` or `channels status --probe`.
## Environment
- OpenClaw: `2026.3.23-2`
- macOS: local install on Dan’s Mac Studio
... | open | null | false | 0 | [] | [] | 2026-03-24T16:22:03Z | 2026-03-24T16:22:03Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | danmorris-ca | 7,462,780 | MDQ6VXNlcjc0NjI3ODA= | User | false |
openclaw/openclaw | 4,129,135,083 | I_kwDOQb6kR872HZnr | 53,797 | https://github.com/openclaw/openclaw/issues/53797 | https://api.github.com/repos/openclaw/openclaw/issues/53797 | /btw fails with github-copilot provider: 400 The requested model is not supported | ## Description
`/btw` side questions fail when using the `github-copilot` provider. The model call returns a 400 error.
## Error
```
⚠️ /btw failed: 400 The requested model is not supported.
```
## Environment
- **OpenClaw version:** 2026.3.23-2 (7ffe7e4)
- **Provider:** `github-copilot`
- **Models configured:** `... | open | null | false | 1 | [] | [] | 2026-03-24T16:22:56Z | 2026-03-24T16:24:16Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | darconada | 14,311,175 | MDQ6VXNlcjE0MzExMTc1 | User | false |
openclaw/openclaw | 4,129,135,898 | I_kwDOQb6kR872HZ0a | 53,798 | https://github.com/openclaw/openclaw/issues/53798 | https://api.github.com/repos/openclaw/openclaw/issues/53798 | [Bug]: Feishu voice message sent via message tool does not display audio duration | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When sending voice messages to Feishu via the `message` tool with `media` (opus file), the message is sent successfully as a voice message, but the duration is not displayed in the Feishu UI. The voice message appears without the usual "X:X... | open | null | false | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-03-24T16:23:05Z | 2026-03-24T16:23:23Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | kinrocW | 269,866,882 | U_kgDOEBXXgg | User | false |
openclaw/openclaw | 4,129,145,721 | I_kwDOQb6kR872HcN5 | 53,803 | https://github.com/openclaw/openclaw/issues/53803 | https://api.github.com/repos/openclaw/openclaw/issues/53803 | [Bug]: Tool results duplicated in session context — 3x token overhead with OmniRoute gateway |
---
## Title
```
Tool results duplicated in session context — 3x token overhead with OmniRoute gateway
```
---
## Bug type
```
Regression (worked before, now fails)
```
---
## Summary
```
Session context fills 3-4x faster than expected due to tool result duplication (`content[].text` + `details.aggregated`) and v... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T16:24:44Z | 2026-03-24T16:36:17Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | dimarch0x | 132,605,638 | U_kgDOB-dmxg | User | false |
openclaw/openclaw | 4,129,280,177 | I_kwDOQb6kR872H9Cx | 53,810 | https://github.com/openclaw/openclaw/issues/53810 | https://api.github.com/repos/openclaw/openclaw/issues/53810 | Subagent Premature Announce Bug - Model-Specific Tool Call Handling Issues | # Subagent Premature Announce Bug - Model-Specific Tool Call Handling Issues
## Summary
Subagents are being announced as "completed successfully" before they actually complete their work. This issue affects multiple models via different providers and endpoints, with behavior varying randomly based on the endpoint and... | open | null | false | 0 | [] | [] | 2026-03-24T16:48:31Z | 2026-03-24T16:48:31Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | FiredMosquito831 | 79,049,343 | MDQ6VXNlcjc5MDQ5MzQz | User | false |
openclaw/openclaw | 4,129,284,022 | I_kwDOQb6kR872H9-2 | 53,811 | https://github.com/openclaw/openclaw/issues/53811 | https://api.github.com/repos/openclaw/openclaw/issues/53811 | feat: /new should pass previous session ID to new session for context continuity | ## Problem
When a user runs `/new`, the session is wiped before I get any pre-reset hook. This means there's no way to:
1. Query the previous session via QMD before it's wiped
2. Offer to continue a prior discussion
3. Surface unresolved context from the last session
## Proposed Solution
Add a `--resume` (or `--keep... | open | null | false | 0 | [] | [] | 2026-03-24T16:49:17Z | 2026-03-24T16:49:17Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | karniksushant-jarvismini | 264,846,507 | U_kgDOD8k8qw | User | false |
openclaw/openclaw | 4,129,302,511 | I_kwDOQb6kR872ICfv | 53,813 | https://github.com/openclaw/openclaw/issues/53813 | https://api.github.com/repos/openclaw/openclaw/issues/53813 | [Bug]: exec tool loses command parameter on consecutive calls | ### Bug type
Regression (worked before, now fails)
### Summary
When calling the `exec` tool consecutively, the `command` parameter is lost in transmission.
### Steps to reproduce
1. Call `exec` tool 5 times in a row
2. 1st call: ✅ Success
3. 2nd-4th calls: ❌ Fail with "command: must have required property"
4. 5t... | open | null | false | 0 | [
"bug",
"regression"
] | [] | 2026-03-24T16:52:48Z | 2026-03-24T16:53:10Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | chanyangerse-sudo | 255,045,336 | U_kgDODzOu2A | User | false |
openclaw/openclaw | 4,129,321,782 | I_kwDOQb6kR872IHM2 | 53,815 | https://github.com/openclaw/openclaw/issues/53815 | https://api.github.com/repos/openclaw/openclaw/issues/53815 | Web UI stops updating messages after WhatsApp gateway reconnect (status 428) | ## Bug Description
Web UI (Control UI) stops receiving new message updates after the WhatsApp gateway disconnects and reconnects. Users must manually refresh the Web UI to see new messages.
## Steps to Reproduce
1. Use OpenClaw with WhatsApp channel enabled
2. Send a message via WhatsApp while Web UI is open
3. Obse... | open | null | false | 0 | [] | [] | 2026-03-24T16:56:16Z | 2026-03-24T16:56:16Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ziweili267-maker | 247,451,103 | U_kgDODr_N3w | User | false |
openclaw/openclaw | 4,129,324,455 | I_kwDOQb6kR872IH2n | 53,816 | https://github.com/openclaw/openclaw/issues/53816 | https://api.github.com/repos/openclaw/openclaw/issues/53816 | [Bug]: Windows CI: channel-policy.test.ts fails due to OPENCLAW_CONTAINER_HINT env leak from lifecycle-core.test.ts | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`src/plugin-sdk/channel-policy.test.ts` fails on Windows CI when it runs in the same shard as `src/cli/daemon-cli/lifecycle-core.test.ts`. The likely cause is `OPENCLAW_CONTAINER_HINT` leaking across test files under `--isolate=false`, caus... | open | null | false | 0 | [] | [] | 2026-03-24T16:56:47Z | 2026-03-24T16:56:47Z | null | CONTRIBUTOR | null | 20260324T233649Z | 2026-03-24T23:36:49Z | wonderchook | 67,992 | MDQ6VXNlcjY3OTky | User | false |
openclaw/openclaw | 4,129,327,318 | I_kwDOQb6kR872IIjW | 53,817 | https://github.com/openclaw/openclaw/issues/53817 | https://api.github.com/repos/openclaw/openclaw/issues/53817 | Signal plugin: send read receipts on inbound message processing | ## Summary
The Signal plugin does not send read receipts (`sendReceipt` with `type: read`) when it processes inbound messages. This means the sender never sees the "filled double-check" indicator that their message was read.
## Current Behavior
- `signal-cli updateConfiguration --read-receipts true` advertises recei... | closed | not_planned | false | 1 | [] | [] | 2026-03-24T16:57:21Z | 2026-03-24T17:00:45Z | 2026-03-24T17:00:37Z | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ziggy-broze | 269,773,776 | U_kgDOEBRr0A | User | false |
openclaw/openclaw | 4,129,335,710 | I_kwDOQb6kR872IKme | 53,818 | https://github.com/openclaw/openclaw/issues/53818 | https://api.github.com/repos/openclaw/openclaw/issues/53818 | feat: self-healing for missing bundled plugin runtime modules (npm publish pipeline regression) | ## Problem
Since 2026.3.22, several plugins crash on startup because the npm publish pipeline omits TypeScript source files needed by jiti at runtime:
| Plugin | Issue | Tracking |
|--------|-------|----------|
| WhatsApp | Missing `light-runtime-api.ts`, `runtime-api.ts`, `src/` | #53244 |
| Matrix | Missing `@matri... | open | null | false | 2 | [] | [] | 2026-03-24T16:59:02Z | 2026-03-24T17:00:12Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | soroyue | 145,274,992 | U_kgDOCKi4cA | User | false |
openclaw/openclaw | 4,129,374,597 | I_kwDOQb6kR872IUGF | 53,824 | https://github.com/openclaw/openclaw/issues/53824 | https://api.github.com/repos/openclaw/openclaw/issues/53824 | [Bug]: Isolated cron job delivery fails with 'Unknown Channel' after gateway restart | ## Summary
Isolated cron jobs with `delivery.mode = 'announce'` consistently fail with `Error: Unknown Channel` after a gateway restart. The jobs run successfully and generate output, but the delivery step fails every time. The main session can still send/receive Discord messages normally.
## Steps to Reproduce
1. C... | open | null | false | 0 | [] | [] | 2026-03-24T17:06:47Z | 2026-03-24T17:06:47Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | chasehainey | 94,016,231 | U_kgDOBZqS5w | User | false |
openclaw/openclaw | 4,129,421,772 | I_kwDOQb6kR872IfnM | 53,826 | https://github.com/openclaw/openclaw/issues/53826 | https://api.github.com/repos/openclaw/openclaw/issues/53826 | Per-agent thinkingDefault ignored in /status and session_status display | ## Bug
When `thinkingDefault` is set on a per-agent entry in `agents.list[]`, the `/status` command and `session_status` tool ignore it and fall through to `"off"`. The actual API calls correctly apply the per-agent thinking level — only the status display is wrong.
## Steps to Reproduce
1. Configure a per-agent `th... | open | null | false | 0 | [] | [] | 2026-03-24T17:16:04Z | 2026-03-24T17:16:04Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | ysfbsf | 9,834,101 | MDQ6VXNlcjk4MzQxMDE= | User | false |
openclaw/openclaw | 4,129,425,082 | I_kwDOQb6kR872Iga6 | 53,827 | https://github.com/openclaw/openclaw/issues/53827 | https://api.github.com/repos/openclaw/openclaw/issues/53827 | Podman setup fails loading saved image as openclaw user due to temp tar permissions | ## Summary
During the Podman setup flow, the image tarball is written into a temporary directory owned by the invoking user with directory mode 700. The script then switches to the openclaw user for podman load. Because openclaw cannot traverse/read that path, the load step fails with permission denied.
## Repro
1. ... | open | null | false | 0 | [] | [] | 2026-03-24T17:16:39Z | 2026-03-24T17:16:39Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | albertxos | 31,403,401 | MDQ6VXNlcjMxNDAzNDAx | User | false |
openclaw/openclaw | 4,129,437,028 | I_kwDOQb6kR872IjVk | 53,829 | https://github.com/openclaw/openclaw/issues/53829 | https://api.github.com/repos/openclaw/openclaw/issues/53829 | [Feature]: Add i18n support for Control UI - Chinese (zh-CN) localization | ### Summary
The OpenClaw Control UI (gateway dashboard at http://127.0.0.1:18789) is currently English-only. For Chinese-speaking users, this creates a barrier to effectively configuring and managing the gateway.
### Problem to solve
Add internationalization (i18n) support to the Control UI, starting with Chinese (S... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-03-24T17:18:35Z | 2026-03-24T17:21:05Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | yanhuo888888-bit | 270,722,591 | U_kgDOECLmHw | User | false |
openclaw/openclaw | 4,129,444,226 | I_kwDOQb6kR872IlGC | 53,831 | https://github.com/openclaw/openclaw/issues/53831 | https://api.github.com/repos/openclaw/openclaw/issues/53831 | [Bug] Discord channel module silently fails to start on Windows (no logs, no error) in 2026.3.23-2 | ### Bug type
Regression (worked before, now fails)
### Summary
On Windows with OpenClaw \2026.3.23-2\, the Discord channel completely fails to start. There are no module resolution errors, no warning logs, and no entries under the \Channels\ section in \openclaw status --deep\. The gateway simply skips initializing th... | open | null | false | 0 | [] | [] | 2026-03-24T17:19:43Z | 2026-03-24T17:19:43Z | null | NONE | null | 20260324T233649Z | 2026-03-24T23:36:49Z | Skrblik | 139,452,656 | U_kgDOCE_g8A | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.