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 | 4,102,467,615 | I_kwDOQb6kR870hrAf | 50,491 | https://github.com/openclaw/openclaw/issues/50491 | https://api.github.com/repos/openclaw/openclaw/issues/50491 | sessions_list returns only current session while openclaw status shows all active sessions | ## Summary
In the same runtime, the `sessions_list` tool returned only the current `#system` session, while `openclaw status` reported 9 active sessions.
This looks like a visibility / result inconsistency between the tool-facing session listing and the CLI/runtime status view.
## Environment
- OpenClaw: 2026.3.13
... | closed | completed | true | 3 | [] | [] | 2026-03-19T15:28:59Z | 2026-04-30T04:54:41Z | 2026-04-25T04:47:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | RichardHojunJang | 12,093,653 | MDQ6VXNlcjEyMDkzNjUz | User | false |
openclaw/openclaw | 4,102,497,108 | I_kwDOQb6kR870hyNU | 50,494 | https://github.com/openclaw/openclaw/issues/50494 | https://api.github.com/repos/openclaw/openclaw/issues/50494 | Feature: Auto-analyze inbound images with vision model | ## Problem
When users send images via Telegram, the system only provides AI-generated descriptions to the agent, not the actual image content. This is problematic for:
- Screenshots with Chinese text (descriptions don't include specific text)
- Financial charts and data
- Any image requiring detailed analysis
## Curr... | closed | completed | true | 1 | [] | [] | 2026-03-19T15:33:17Z | 2026-04-30T04:54:40Z | 2026-04-25T04:47:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | HanSoul-CHEN | 29,673,965 | MDQ6VXNlcjI5NjczOTY1 | User | false |
openclaw/openclaw | 4,102,518,189 | I_kwDOQb6kR870h3Wt | 50,496 | https://github.com/openclaw/openclaw/issues/50496 | https://api.github.com/repos/openclaw/openclaw/issues/50496 | [Bug]: Trashed session's message tool calls re-delivered on every gateway restart | ### Bug type
Regression (worked before, now fails)
### Summary
Messages from a trashed (garbage-collected) session are being re-sent to Feishu (Lark) on every gateway restart. The same message has been delivered many times over the past week.
### Steps to reproduce
1. Agent creates a video via nemo-video skill and... | closed | completed | true | 6 | [
"bug",
"regression"
] | [] | 2026-03-19T15:36:31Z | 2026-04-30T04:54:39Z | 2026-04-25T04:47:42Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | hongliang-nemovideo | 231,823,987 | U_kgDODdFacw | User | false |
openclaw/openclaw | 4,102,543,765 | I_kwDOQb6kR870h9mV | 50,498 | https://github.com/openclaw/openclaw/issues/50498 | https://api.github.com/repos/openclaw/openclaw/issues/50498 | Feature: Allow hiding/filtering built-in model catalog entries | ## Problem
When `models.mode: "replace"` is set with only custom providers configured, the `/model` command and WebUI still show all 783+ built-in models from the pi-coding-agent catalog. This makes the list extremely long and users can accidentally select models with no API key configured, resulting in errors.
## Cu... | closed | completed | true | 1 | [] | [] | 2026-03-19T15:40:22Z | 2026-04-30T04:54:38Z | 2026-04-25T04:47:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | langbyyi | 213,003,902 | U_kgDODLIufg | User | false |
openclaw/openclaw | 4,102,590,708 | I_kwDOQb6kR870iJD0 | 50,501 | https://github.com/openclaw/openclaw/issues/50501 | https://api.github.com/repos/openclaw/openclaw/issues/50501 | Feature Request: Federated Context Engine Routing (Decoupling Memory Types) | ### The Problem
Currently, OpenClaw forces agents to select a single monolithic `contextEngine` plugin. However, recent academic benchmarks (e.g., LMEB: Long-horizon Memory Embedding Benchmark, arXiv:2603.12572) demonstrate that AI agents require four distinct types of memory retrieval:
1. **Dialogue Memory** (fragment... | closed | completed | true | 3 | [] | [] | 2026-03-19T15:49:02Z | 2026-04-30T04:54:38Z | 2026-04-25T04:47:56Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | octavia-reising | 39,859,353 | MDQ6VXNlcjM5ODU5MzUz | User | false |
openclaw/openclaw | 4,102,599,520 | I_kwDOQb6kR870iLNg | 50,502 | https://github.com/openclaw/openclaw/issues/50502 | https://api.github.com/repos/openclaw/openclaw/issues/50502 | [Bug] 飞书私聊新建session时错误地默认使用glm-4v模型 | ## 问题描述
在飞书私聊中新建session时，错误地默认使用glm-4v模型，而不是配置的默认模型（minimax-portal/MiniMax-M2.5）。
## 环境
- OpenClaw版本：2026.3.13
- 飞书机器人：cli_a92b003ebcf89cc1（main agent对应）
- 默认模型配置：minimax-portal/MiniMax-M2.5
## 复现步骤
1. 通过飞书机器人发送 /new 新建session
2. 查看session使用的模型，发现是 zhipu/glm-4v 而不是默认的 minimax
## 临时解决方案
1. 删除 main agent 目录下的 mode... | closed | completed | true | 2 | [] | [] | 2026-03-19T15:50:34Z | 2026-04-30T04:54:37Z | 2026-04-25T04:48:03Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | huixuedai-arch | 224,418,747 | U_kgDODWBbuw | User | false |
openclaw/openclaw | 4,102,651,413 | I_kwDOQb6kR870iX4V | 50,504 | https://github.com/openclaw/openclaw/issues/50504 | https://api.github.com/repos/openclaw/openclaw/issues/50504 | CLI gateway handshake fails on slow-startup hosts (challenge timeout too short) | ## Summary
On hosts where OpenClaw CLI startup takes >2 seconds (common with plugins loaded), every CLI→gateway RPC command fails intermittently or consistently with:
```
gateway connect failed: Error: gateway not connected
Error: gateway closed (1000 normal closure): no close reason
```
Server logs show:
```
handsh... | closed | completed | true | 2 | [] | [] | 2026-03-19T15:59:37Z | 2026-04-30T04:54:36Z | 2026-04-25T04:48:10Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | thomasbek3 | 261,084,914 | U_kgDOD4_W8g | User | false |
openclaw/openclaw | 4,102,690,269 | I_kwDOQb6kR870ihXd | 50,508 | https://github.com/openclaw/openclaw/issues/50508 | https://api.github.com/repos/openclaw/openclaw/issues/50508 | [Bug]: Message responses occasionally routed to wrong channel (webchat instead of Feishu) | ### Bug type
Regression (worked before, now fails)
### Summary
## Problem
When sending messages to the Agent via Feishu, responses are sometimes delivered to the PC webchat UI instead of the Feishu channel where the message originated.
## Steps to Reproduce
1. Configure Feishu channel and pair with user
2. Send a me... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-19T16:05:24Z | 2026-04-30T04:54:35Z | 2026-04-25T04:48:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | azhixiansheng | 65,844,553 | MDQ6VXNlcjY1ODQ0NTUz | User | false |
openclaw/openclaw | 4,102,694,216 | I_kwDOQb6kR870iiVI | 50,509 | https://github.com/openclaw/openclaw/issues/50509 | https://api.github.com/repos/openclaw/openclaw/issues/50509 | [Bug]: Control UI cannot switch Ollama models when model name contains slashes | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Control UI model switcher fails with "model not allowed" for Ollama models whose name contains slashes (e.g. huihui_ai/qwen3-vl-abliterated)
### Steps to reproduce
1. Add to agents.defaults.models in openclaw.json:
"ollama/huihui_ai/qw... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T16:06:00Z | 2026-04-30T04:54:34Z | 2026-04-25T04:48:24Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Cabra165 | 93,967,196 | U_kgDOBZnTXA | User | false |
openclaw/openclaw | 4,102,716,538 | I_kwDOQb6kR870inx6 | 50,513 | https://github.com/openclaw/openclaw/issues/50513 | https://api.github.com/repos/openclaw/openclaw/issues/50513 | feat(work-loop): post-response hook to auto-resolve referenced GitHub issues | ## Summary
After each bot response, scan the conversation for referenced GitHub issues (#NNN), check their open/closed state via `gh`, and automatically:
- **Closed/resolved work**: close the issue with a comment explaining what was done
- **Open/unresolved work**: queue a follow-up agent turn to pick it up and work ... | closed | not_planned | true | 2 | [] | [] | 2026-03-19T16:09:06Z | 2026-04-30T04:54:34Z | 2026-04-25T04:48:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zharl | 8,402,541 | MDQ6VXNlcjg0MDI1NDE= | User | false |
openclaw/openclaw | 4,102,726,005 | I_kwDOQb6kR870iqF1 | 50,514 | https://github.com/openclaw/openclaw/issues/50514 | https://api.github.com/repos/openclaw/openclaw/issues/50514 | [Bug]: openclaw security audit --deep reports missing scope: operator.read despite operator tokens/scopes being present and gateway healthy | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
OpenClaw 2026.3.13 on Linux (6.16.8+kali-cloud-amd64, Node 25.6.1, stable/pnpm).
Gateway is local (ws://127.0.0.1:18789) with Tailscale serve enabled.
openclaw security audit --deep always reports:
WARN gateway.probe_failed
missing scope:... | closed | completed | true | 2 | [
"bug",
"security",
"bug:behavior"
] | [] | 2026-03-19T16:10:26Z | 2026-04-30T04:54:33Z | 2026-04-25T04:48:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | LaRojaFuria | 47,408,035 | MDQ6VXNlcjQ3NDA4MDM1 | User | false |
openclaw/openclaw | 4,102,789,139 | I_kwDOQb6kR870i5gT | 50,524 | https://github.com/openclaw/openclaw/issues/50524 | https://api.github.com/repos/openclaw/openclaw/issues/50524 | Telegram /model switch confirms but doesn't take effect on active session | ## Bug
When using `/model codex` in Telegram, the bot confirms "Model set to codex (openai-codex/gpt-5.3-codex)" but the active session continues running on the previous model (Claude Opus 4 in this case).
## Steps to Reproduce
1. Start a session on Telegram (default model: anthropic/claude-opus-4-6)
2. Send `/model... | closed | completed | true | 2 | [] | [] | 2026-03-19T16:20:02Z | 2026-04-30T04:54:32Z | 2026-04-25T04:48:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,102,834,037 | I_kwDOQb6kR870jEd1 | 50,529 | https://github.com/openclaw/openclaw/issues/50529 | https://api.github.com/repos/openclaw/openclaw/issues/50529 | Bug: depth-2 subagent receives task message twice on single sessions_spawn call | ## Summary
A single `sessions_spawn` call at depth 2 (orchestrator → task-manager → coder) results in the child session receiving the initial task message **twice** as separate `role: user` messages. Depth-1 spawns (orchestrator → coder) are unaffected.
## Reproduction
1. Orchestrator spawns Task Manager (depth 1)
2... | closed | completed | true | 2 | [] | [] | 2026-03-19T16:28:43Z | 2026-04-30T04:54:31Z | 2026-04-25T04:48:52Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | oiGaDio | 265,029,916 | U_kgDOD8wJHA | User | false |
openclaw/openclaw | 4,102,846,689 | I_kwDOQb6kR870jHjh | 50,531 | https://github.com/openclaw/openclaw/issues/50531 | https://api.github.com/repos/openclaw/openclaw/issues/50531 | Feishu: GM机器人(WebSocket)无法接收消息 - 多账号配置问题 | ## 问题描述
在飞书配置多账号时，GM机器人无法接收用户消息。
## 环境
- OpenClaw: 本地部署
- 飞书插件: 最新版本
- 通道: Feishu
- 内网穿透: natapp
## 配置情况
已配置两个飞书机器人账号：
1. **在兜 (default)**
- App ID: cli_a927cbc72778dbd3
- 状态: ✅ 正常工作
- 能接收和回复消息
2. **GM-总经理 (gm)**
- App ID: cli_a93843cceaf8dbd2
- 状态: ❌ 无法接收消息
- 飞书后台已正确配置：
- 回调地址已设置（长连接模式）
... | closed | completed | true | 1 | [] | [] | 2026-03-19T16:31:04Z | 2026-04-30T04:54:31Z | 2026-04-25T04:48:59Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Shannnnne | 266,429,455 | U_kgDOD-FkDw | User | false |
openclaw/openclaw | 4,102,853,423 | I_kwDOQb6kR870jJMv | 50,533 | https://github.com/openclaw/openclaw/issues/50533 | https://api.github.com/repos/openclaw/openclaw/issues/50533 | Images sent via chat.send RPC are accepted but silently dropped before reaching the model | ## Summary
Images attached to `chat.send` RPC calls are accepted by the gateway (returns `runId`), but the model never receives them. The model responds saying it cannot see any attached image.
## Environment
- **OpenClaw version:** 2026.3.13 (61d171a)
- **OS:** macOS (Darwin 25.3.0, arm64)
- **Model:** claude-opus-... | closed | completed | true | 1 | [] | [] | 2026-03-19T16:32:27Z | 2026-04-30T04:54:30Z | 2026-04-25T04:49:05Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dagny-holland | 261,707,352 | U_kgDOD5lWWA | User | false |
openclaw/openclaw | 4,102,861,123 | I_kwDOQb6kR870jLFD | 50,534 | https://github.com/openclaw/openclaw/issues/50534 | https://api.github.com/repos/openclaw/openclaw/issues/50534 | Control UI: Agent settings model switch sends model ID without provider prefix | ## Bug Description
In the Control UI's **Agent Settings** panel, clicking to switch model sends only the model ID (e.g. `glm-5-turbo`) without the provider prefix (e.g. `zai/glm-5-turbo`), causing a `model not allowed: openai-codex/glm-5-turbo` error.
## Steps to Reproduce
1. Configure multiple providers in `opencla... | closed | completed | true | 2 | [] | [] | 2026-03-19T16:34:00Z | 2026-04-30T04:54:29Z | 2026-04-25T04:49:12Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ACaN-Pro-Lee | 253,497,195 | U_kgDODxwPaw | User | false |
openclaw/openclaw | 4,102,869,130 | I_kwDOQb6kR870jNCK | 50,536 | https://github.com/openclaw/openclaw/issues/50536 | https://api.github.com/repos/openclaw/openclaw/issues/50536 | [Bug]: Control UI model selector sends wrong provider prefix when switching from OpenAI to Qwen models | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
## Bug Description
When using the Control UI model dropdown, selecting a Qwen model fails because the UI appears to send an invalid provider/model combination.
Instead of sending the canonical model ID:
- `qwen-portal/coder-model`
th... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T16:35:34Z | 2026-04-30T04:54:28Z | 2026-04-25T04:49:19Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ArabianCowboy | 155,951,433 | U_kgDOCUuhSQ | User | false |
openclaw/openclaw | 4,102,919,139 | I_kwDOQb6kR870jZPj | 50,539 | https://github.com/openclaw/openclaw/issues/50539 | https://api.github.com/repos/openclaw/openclaw/issues/50539 | [Feature]: Add gateway:shutdown hook event | ### Summary
## Problem
Currently, OpenClaw hooks only support `gateway:startup` event for gateway lifecycle. There is no corresponding `gateway:shutdown` event to notify users when the gateway is stopping.
### Problem to solve
## Use Case
I want to send a notification to my Telegram when the gateway goes offline (... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-19T16:46:23Z | 2026-04-30T04:54:27Z | 2026-04-25T04:49:26Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | cbd2020 | 60,251,704 | MDQ6VXNlcjYwMjUxNzA0 | User | false |
openclaw/openclaw | 4,103,077,214 | I_kwDOQb6kR870j_1e | 50,549 | https://github.com/openclaw/openclaw/issues/50549 | https://api.github.com/repos/openclaw/openclaw/issues/50549 | Feature Request: Gateway Auto-Recovery Tool Suite | # Feature Request: Gateway Auto-Recovery Tool Suite
## Summary
Add a set of optional tools to improve OpenClaw Gateway reliability, addressing the common issue of Gateway failing to restart after configuration changes.
## Problem Statement
After modifying `openclaw.json`, the Gateway often fails to restart. While `... | closed | not_planned | true | 3 | [] | [] | 2026-03-19T17:14:19Z | 2026-04-30T04:54:26Z | 2026-04-25T04:49:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jacksonlam138-hub | 256,724,425 | U_kgDOD01NyQ | User | false |
openclaw/openclaw | 4,103,077,787 | I_kwDOQb6kR870j_-b | 50,550 | https://github.com/openclaw/openclaw/issues/50550 | https://api.github.com/repos/openclaw/openclaw/issues/50550 | [Feature] Gateway Auto-Recovery Toolkit - Prevent service crashes from config errors | ## Problem
When modifying OpenClaw configuration files, errors (such as JSON syntax errors, references to non-existent plugins, etc.) can cause the Gateway to fail to start. Current issues:
- `openclaw doctor --fix` cannot automatically fix all problems (e.g., residual plugin references)
- No automatic rollback mecha... | closed | completed | true | 2 | [] | [] | 2026-03-19T17:14:26Z | 2026-04-30T04:54:25Z | 2026-04-25T04:49:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jacksonlam138-hub | 256,724,425 | U_kgDOD01NyQ | User | false |
openclaw/openclaw | 4,103,097,437 | I_kwDOQb6kR870kExd | 50,552 | https://github.com/openclaw/openclaw/issues/50552 | https://api.github.com/repos/openclaw/openclaw/issues/50552 | feishu plugin declares skills path but directory does not exist | ## Summary
The feishu extension's `openclaw.plugin.json` declares `"skills": ["./skills"]`, but no `skills/` directory exists under `extensions/feishu/`. This causes repeated warnings on every session init:
```
[skills] plugin skill path not found (feishu): .../openclaw/extensions/feishu/skills
```
## Steps to repro... | closed | completed | true | 2 | [] | [] | 2026-03-19T17:17:48Z | 2026-04-30T04:54:24Z | 2026-04-25T04:49:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | godouxm | 1,287,287 | MDQ6VXNlcjEyODcyODc= | User | false |
openclaw/openclaw | 4,103,138,756 | I_kwDOQb6kR870kO3E | 50,554 | https://github.com/openclaw/openclaw/issues/50554 | https://api.github.com/repos/openclaw/openclaw/issues/50554 | [Feature]: Support Interrupting and Adding Details like chatGPT and Claude Code | ### Summary
Support interrupting an ongoing task and dynamically adding or refining instructions during execution (“Interrupting and Adding Details”).
You can refer to this.
https://www.youtube.com/watch?v=qSHMlduU6Lw
### Problem to solve
Currently, OpenClaw primarily follows a linear execution model:
User provide... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-19T17:26:16Z | 2026-04-30T04:54:24Z | 2026-04-25T04:50:01Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | codingOuch | 25,929,878 | MDQ6VXNlcjI1OTI5ODc4 | User | false |
openclaw/openclaw | 4,103,167,110 | I_kwDOQb6kR870kVyG | 50,555 | https://github.com/openclaw/openclaw/issues/50555 | https://api.github.com/repos/openclaw/openclaw/issues/50555 | [Feature]: 对话窗口不能切换问题 | ### Summary
比如，我离开了电脑，微信上开了一个对话，现在回到电脑上，应该是在断开微信连接或保持连接的情况下，可以继续原来的对话。但，这个目前无法实现。建议改进。
### Problem to solve
对话窗口不能切换问题
### Proposed solution
应该是在断开微信连接或保持连接的情况下，可以继续原来的对话。但，这个目前无法实现。建议改进。
### Alternatives considered
_No response_
### Impact
对话窗口不能切换问题
### Evidence/examples
_No response_
### Additional infor... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-19T17:32:12Z | 2026-04-30T04:54:23Z | 2026-04-25T04:50:08Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zhenxiwen | 23,516,346 | MDQ6VXNlcjIzNTE2MzQ2 | User | false |
openclaw/openclaw | 4,103,200,257 | I_kwDOQb6kR870kd4B | 50,557 | https://github.com/openclaw/openclaw/issues/50557 | https://api.github.com/repos/openclaw/openclaw/issues/50557 | Lock file contention under high load | ## Problem
Session store lock timeouts occur frequently when API calls fail/retry. The `.lock` file mechanism does not handle concurrent access gracefully under stress.
## Observed Behavior
- `timeout acquiring session store lock` warnings in logs
- Locks accumulate faster than cleanup during API errors
- Multiple gat... | closed | completed | true | 2 | [] | [] | 2026-03-19T17:38:37Z | 2026-04-30T04:54:22Z | 2026-04-25T04:50:15Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | coywolffuturist | 264,284,025 | U_kgDOD8CneQ | User | false |
openclaw/openclaw | 4,103,201,391 | I_kwDOQb6kR870keJv | 50,559 | https://github.com/openclaw/openclaw/issues/50559 | https://api.github.com/repos/openclaw/openclaw/issues/50559 | Gateway should detect and handle competing processes | ## Problem
If a manual gateway process is already running, `systemctl restart clawdbot` fails silently and enters a failed state. The gateway does not clearly detect or handle competing processes.
## Observed Behavior
- Manual gateway running on port 18789
- Systemd tries to start, gets "port already in use"
- Service... | closed | completed | true | 2 | [] | [] | 2026-03-19T17:38:50Z | 2026-04-30T04:54:21Z | 2026-04-25T04:50:22Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | coywolffuturist | 264,284,025 | U_kgDOD8CneQ | User | false |
openclaw/openclaw | 4,103,204,467 | I_kwDOQb6kR870ke5z | 50,562 | https://github.com/openclaw/openclaw/issues/50562 | https://api.github.com/repos/openclaw/openclaw/issues/50562 | Add native heartbeat interval config | ## Problem
HEARTBEAT.md exists as a convention for periodic agent checks, but there is no built-in `heartbeat.interval` config that auto-triggers it. Users must set up their own cron jobs.
## Current State
- HEARTBEAT.md is documented as a workspace file
- No native mechanism to trigger periodic reads
- Users create c... | closed | completed | true | 2 | [] | [] | 2026-03-19T17:39:27Z | 2026-04-30T04:54:21Z | 2026-04-25T04:50:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | coywolffuturist | 264,284,025 | U_kgDOD8CneQ | User | false |
openclaw/openclaw | 4,103,212,587 | I_kwDOQb6kR870kg4r | 50,564 | https://github.com/openclaw/openclaw/issues/50564 | https://api.github.com/repos/openclaw/openclaw/issues/50564 | Skill descriptions have no size enforcement | ## Problem
No guardrails on skill description length. Verbose descriptions bloat the system prompt on every turn.
## Observed Impact
- Pakt skills had descriptions of 600-700 chars each
- Effectively injecting API documentation into every request
- System prompt bloat increases token usage and cost
## Example
A skill... | closed | completed | true | 2 | [] | [] | 2026-03-19T17:41:07Z | 2026-04-30T04:54:20Z | 2026-04-25T04:50:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | coywolffuturist | 264,284,025 | U_kgDOD8CneQ | User | false |
openclaw/openclaw | 4,103,281,029 | I_kwDOQb6kR870kxmF | 50,569 | https://github.com/openclaw/openclaw/issues/50569 | https://api.github.com/repos/openclaw/openclaw/issues/50569 | Ollama vision models (qwen3-vl, llama3.2-vision) not detected as supporting images | ## Bug Description
Ollama vision models like `qwen3-vl:235b-cloud` and `llama3.2-vision:11b` are not recognized as supporting image input, even though they are multimodal models capable of vision tasks.
## Steps to Reproduce
1. Install a vision model: `ollama pull qwen3-vl:235b-cloud`
2. Configure as image model: `o... | closed | completed | true | 2 | [] | [] | 2026-03-19T17:55:05Z | 2026-04-30T04:54:19Z | 2026-04-25T04:50:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dynamispraxis | 269,545,863 | U_kgDOEBDxhw | User | false |
openclaw/openclaw | 4,103,310,107 | I_kwDOQb6kR870k4sb | 50,571 | https://github.com/openclaw/openclaw/issues/50571 | https://api.github.com/repos/openclaw/openclaw/issues/50571 | Bug: Hook/subagent sessions route reply to last-active chat instead of trigger session owner | ## Summary
When a hook or subagent session is spawned by a Clack/internal trigger (rather than a direct user message), its reply routing falls back to the **last active Telegram chat** rather than the owner's channel or the triggering session's context. This creates a silent cross-user data leak.
## Steps to Reproduc... | closed | completed | true | 4 | [] | [] | 2026-03-19T18:00:57Z | 2026-04-30T04:54:18Z | 2026-04-25T04:50:50Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | adroidian | 7,563,431 | MDQ6VXNlcjc1NjM0MzE= | User | false |
openclaw/openclaw | 4,103,353,259 | I_kwDOQb6kR870lDOr | 50,574 | https://github.com/openclaw/openclaw/issues/50574 | https://api.github.com/repos/openclaw/openclaw/issues/50574 | [Bug]: `doctor` rejects `agentId` on telegram topic config despite runtime support | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The `openclaw doctor` command flags `agentId` on per-topic Telegram config as an unrecognized key and strips it, even though the runtime correctly reads and routes based on `topicConfig.agentId`.
Example config:
```js
channels: {
telegr... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T18:09:04Z | 2026-04-30T04:54:17Z | 2026-04-25T04:50:57Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jescalan | 556,932 | MDQ6VXNlcjU1NjkzMg== | User | false |
openclaw/openclaw | 4,099,256,993 | I_kwDOQb6kR870VbKh | 50,207 | https://github.com/openclaw/openclaw/issues/50207 | https://api.github.com/repos/openclaw/openclaw/issues/50207 | [Feature]: Pre-tool-call approval hook / suppress assistant text when message tool used | ## Problem
When an agent uses `message action=send` to deliver real-time updates to the user during multi-step work, OpenClaw still collects all assistant text content blocks (narration between tool calls) and delivers them as a final batched reply. This results in:
1. Duplicate messages — user sees the `message actio... | closed | completed | true | 6 | [] | [] | 2026-03-19T04:43:31Z | 2026-04-30T04:54:18Z | 2026-04-25T04:39:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | KelTech-Services | 18,705,530 | MDQ6VXNlcjE4NzA1NTMw | User | false |
openclaw/openclaw | 4,103,383,559 | I_kwDOQb6kR870lKoH | 50,576 | https://github.com/openclaw/openclaw/issues/50576 | https://api.github.com/repos/openclaw/openclaw/issues/50576 | [Bug]: elevated=full does not work | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Version: 2026.3.14
Runtime: direct
Config: agents.defaults.elevatedDefault = "full"
Status shows: elevated=full
### Steps to reproduce
1) /elevated full
2) /status (shows elevated=full)
3) Run: !exec python3 -c "print('All systems green!'... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T18:14:24Z | 2026-04-30T04:54:16Z | 2026-04-25T04:51:04Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | massiccio | 4,521,350 | MDQ6VXNlcjQ1MjEzNTA= | User | false |
openclaw/openclaw | 4,103,410,403 | I_kwDOQb6kR870lRLj | 50,579 | https://github.com/openclaw/openclaw/issues/50579 | https://api.github.com/repos/openclaw/openclaw/issues/50579 | [Bug]: Chat UI Voice Input button not responsive. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When trying to use the voice input icon for speech to text transcription in chat the process appears to fail silently.
### Steps to reproduce
1. Start openclaw 2026.3.12 gateway
2. Access Dashboard
3. Click Microphone Icon
### Expected b... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T18:19:58Z | 2026-04-30T04:54:15Z | 2026-04-25T04:51:11Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | storytyler | 161,856,107 | U_kgDOCaW6aw | User | false |
openclaw/openclaw | 4,103,747,815 | I_kwDOQb6kR870mjjn | 50,597 | https://github.com/openclaw/openclaw/issues/50597 | https://api.github.com/repos/openclaw/openclaw/issues/50597 | [Bug] Slack socket mode reconnect causes duplicate outbound messages | # Bug Report: Slack Socket Mode reconnect causes duplicate outbound messages
**Filed:** 2026-03-19
**Reporter:** Cortana (AI assistant) on behalf of Carter Laren
**Repo:** https://github.com/openclaw/openclaw/issues
---
## Summary
When the Slack Socket Mode connection drops and reconnects (due to stale socket... | closed | completed | true | 5 | [] | [] | 2026-03-19T19:22:12Z | 2026-04-30T04:54:15Z | 2026-04-25T04:51:18Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | cortana-sprite | 269,511,364 | U_kgDOEBBqxA | User | false |
openclaw/openclaw | 4,103,810,406 | I_kwDOQb6kR870my1m | 50,599 | https://github.com/openclaw/openclaw/issues/50599 | https://api.github.com/repos/openclaw/openclaw/issues/50599 | Gateway probe is inconsistent on Windows 11: scheduled gateway times out, foreground gateway can handshake but returns missing scope: operator.read; Tailscale Serve path also times out | Title: Gateway probe is inconsistent on Windows 11: scheduled gateway times out, foreground gateway can handshake but returns missing scope: operator.read; Tailscale Serve path also times out
### Environment
- OpenClaw: `2026.3.13 (61d171a)`
- Install method: `npm install -g openclaw@2026.3.13`
- OS: Windows 11
- Node... | closed | completed | true | 2 | [] | [] | 2026-03-19T19:35:33Z | 2026-04-30T04:54:14Z | 2026-04-25T04:51:25Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | LPMIV | 34,988,581 | MDQ6VXNlcjM0OTg4NTgx | User | false |
openclaw/openclaw | 4,103,819,993 | I_kwDOQb6kR870m1LZ | 50,601 | https://github.com/openclaw/openclaw/issues/50601 | https://api.github.com/repos/openclaw/openclaw/issues/50601 | [Bug]: [Bug]: Thinking status can report high even when persisted session thinking level is xhigh | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
OpenClaw can report the current session thinking level as `high` even when the persisted session thinking level is actually `xhigh`.
This makes it look like `/thin... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T19:37:07Z | 2026-04-30T04:54:13Z | 2026-04-25T04:51:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | allenhurff | 209,308 | MDQ6VXNlcjIwOTMwOA== | User | false |
openclaw/openclaw | 4,103,864,498 | I_kwDOQb6kR870nACy | 50,603 | https://github.com/openclaw/openclaw/issues/50603 | https://api.github.com/repos/openclaw/openclaw/issues/50603 | [Bug]: gateway call health/cron.status fails with WS handshake timeout on 2026.3.13 (loopback, token auth) | ### Bug type
Regression (worked before, now fails)
### Summary
Upgraded from 2026.3.2
## Summary
After upgrading to **OpenClaw 2026.3.13**, Gateway RPC calls from CLI fail locally (`ws://127.0.0.1:18789`) with close 1000/1006 and no result.
This originally surfaced as "cron jobs appear set up but never execute", bu... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-19T19:44:33Z | 2026-04-30T04:54:12Z | 2026-04-25T04:51:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | taotarodev | 65,584,396 | MDQ6VXNlcjY1NTg0Mzk2 | User | false |
openclaw/openclaw | 4,103,869,551 | I_kwDOQb6kR870nBRv | 50,605 | https://github.com/openclaw/openclaw/issues/50605 | https://api.github.com/repos/openclaw/openclaw/issues/50605 | [Bug] openclaw dashboard fails to resolve token from macOS Keychain | ## Description
`openclaw dashboard` generates a URL with `$env:OPENCLAW_GATEWAY_TOKEN` literally in the URL instead of resolving the token from macOS Keychain.
## Steps to Reproduce
1. Store gateway token in macOS Keychain (as recommended by setup)
2. Run `openclaw dashboard`
3. Observe that the URL contains the lit... | closed | completed | true | 2 | [] | [] | 2026-03-19T19:45:37Z | 2026-04-30T04:54:11Z | 2026-04-25T04:51:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | lars-hv | 235,699,712 | U_kgDODgx-AA | User | false |
openclaw/openclaw | 4,103,909,135 | I_kwDOQb6kR870nK8P | 50,608 | https://github.com/openclaw/openclaw/issues/50608 | https://api.github.com/repos/openclaw/openclaw/issues/50608 | Discord: forwarded messages (message_snapshots) not delivered to agent | When a user forwards a message from one Discord channel to another using Discord's 'Share to Channel' feature, the forwarded message content (stored in message_snapshots) is not included in the message delivered to the agent. The agent receives an empty message.
This is a workflow blocker for setups where users forwar... | closed | completed | true | 2 | [] | [] | 2026-03-19T19:53:57Z | 2026-04-30T04:54:11Z | 2026-04-25T04:52:00Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | azorel | 8,345,182 | MDQ6VXNlcjgzNDUxODI= | User | false |
openclaw/openclaw | 4,104,144,060 | I_kwDOQb6kR870oES8 | 50,634 | https://github.com/openclaw/openclaw/issues/50634 | https://api.github.com/repos/openclaw/openclaw/issues/50634 | Feature Request: Safe Config Patching with Pre-Validation | # Feature Request: Safe Config Patching with Pre-Validation
## Summary
The `gateway` tool's `config.patch` action should validate the new config before writing to disk, or rollback on runtime refresh failure. Currently it can leave the gateway in a broken state if the patched config fails runtime validation.
## Curr... | closed | completed | true | 3 | [] | [] | 2026-03-19T20:47:26Z | 2026-04-30T04:54:10Z | 2026-04-25T04:52:07Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | sgarc57 | 22,865,328 | MDQ6VXNlcjIyODY1MzI4 | User | false |
openclaw/openclaw | 4,104,288,200 | I_kwDOQb6kR870onfI | 50,647 | https://github.com/openclaw/openclaw/issues/50647 | https://api.github.com/repos/openclaw/openclaw/issues/50647 | [Bug]: Model dropdown strips provider prefix, causing "model not allowed" error for non-Anthropic models | ### Bug type
Regression (worked before, now fails)
### Summary
Selecting a non-default provider model from the webchat UI model dropdown fails with (with a default provider of anthropic) `GatewayRequestError: model not allowed: anthropic/<model-id>`. The UI appears to strip the provider prefix from the model ID (e.g... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-19T21:23:13Z | 2026-04-30T04:54:09Z | 2026-04-25T04:52:14Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ctbritt | 4,516,034 | MDQ6VXNlcjQ1MTYwMzQ= | User | false |
openclaw/openclaw | 4,104,327,520 | I_kwDOQb6kR870oxFg | 50,649 | https://github.com/openclaw/openclaw/issues/50649 | https://api.github.com/repos/openclaw/openclaw/issues/50649 | Feature: PreToolUse/PostToolUse hook system for OpenClaw agent sessions | ## Problem
Claude Code supports `PreToolUse` and `PostToolUse` hooks that allow parameter-level inspection and blocking of tool calls before they execute. OpenClaw agent sessions (non-Claude-Code) have **no equivalent mechanism**.
Currently, OpenClaw agent sessions can only enforce tool access at the on/off level via... | closed | completed | true | 3 | [] | [] | 2026-03-19T21:33:36Z | 2026-04-30T04:54:08Z | 2026-04-25T04:52:21Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | hubertmodularshift-cmyk | 261,700,715 | U_kgDOD5k8aw | User | false |
openclaw/openclaw | 4,104,394,489 | I_kwDOQb6kR870pBb5 | 50,655 | https://github.com/openclaw/openclaw/issues/50655 | https://api.github.com/repos/openclaw/openclaw/issues/50655 | [Bug]: Agent crashes (exit code 255) when fetching GitHub profile page inside OpenShell sandbox | ### Bug type
Crash (process/app exits or hangs)
### Summary
Description:
The OpenClaw agent process crashes with exit code 255 when attempting to fetch a GitHub profile URL (e.g. https://github.com/<username>) from within an NVIDIA OpenShell sandbox. The crash is reproducible — it happened twice in a row when the a... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-03-19T21:50:47Z | 2026-04-30T04:54:07Z | 2026-04-25T04:52:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | lellky | 7,119,913 | MDQ6VXNlcjcxMTk5MTM= | User | false |
openclaw/openclaw | 4,104,394,885 | I_kwDOQb6kR870pBiF | 50,656 | https://github.com/openclaw/openclaw/issues/50656 | https://api.github.com/repos/openclaw/openclaw/issues/50656 | WhatsApp Web outbound listener fails to start — sends broken after gateway restart | ## Summary
WhatsApp inbound (receiving messages) works, but **outbound sends always fail** with:
```
GatewayClientRequestError: Error: No active WhatsApp Web listener (account: default).
Start the gateway, then link WhatsApp with: openclaw channels login --channel whatsapp --account default.
```
## Environment
- **... | closed | completed | true | 4 | [] | [] | 2026-03-19T21:50:52Z | 2026-04-30T04:54:06Z | 2026-04-25T04:52:42Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rwenzlofsky | 6,306,611 | MDQ6VXNlcjYzMDY2MTE= | User | false |
openclaw/openclaw | 4,104,476,722 | I_kwDOQb6kR870pVgy | 50,665 | https://github.com/openclaw/openclaw/issues/50665 | https://api.github.com/repos/openclaw/openclaw/issues/50665 | [Bug]: Gateway WS handshake timeout (3s) not configurable in production; closes openclaw acp bridge connections | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The gateway WebSocket handshake timeout (`DEFAULT_HANDSHAKE_TIMEOUT_MS`, 3000ms) is hardcoded and cannot be overridden in production. An env var override exists (`OPENCLAW_TEST_HANDSHAKE_TIMEOUT_MS`) but is gated behind `process.env.VITEST`... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-03-19T22:12:33Z | 2026-04-30T04:54:05Z | 2026-04-25T04:52:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | t4min0 | 269,137,166 | U_kgDOEAq1Dg | User | false |
openclaw/openclaw | 4,104,622,973 | I_kwDOQb6kR870p5N9 | 50,679 | https://github.com/openclaw/openclaw/issues/50679 | https://api.github.com/repos/openclaw/openclaw/issues/50679 | Control UI: Auto-navigate to active chat after sending message | **Describe the bug**
When sending a message in the Control UI, the interface does not automatically switch to the active chat tab/container where the response appears. The user has to manually navigate to see the assistant's response.
**To Reproduce**
1. Open Control UI
2. Navigate to a different tab (e.g., Overview,... | closed | not_planned | true | 1 | [] | [] | 2026-03-19T22:45:18Z | 2026-04-30T04:54:04Z | 2026-04-25T04:52:56Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dmitriyy0828-web | 261,611,373 | U_kgDOD5ffbQ | User | false |
openclaw/openclaw | 4,104,700,467 | I_kwDOQb6kR870qMIz | 50,687 | https://github.com/openclaw/openclaw/issues/50687 | https://api.github.com/repos/openclaw/openclaw/issues/50687 | [Report Scam] 举报 OpenClaw 相关人员存在虚假宣传与欺骗行为 | ### Bug type
Regression (worked before, now fails)
### Summary
## 举报内容 / Report Content
我要举报 OpenClaw 项目相关人员存在持续说谎、虚假宣传的欺骗行为。
I am reporting that individuals associated with the OpenClaw project have engaged in deceptive behavior, including persistent lying and false advertising.
### 具体情况 / Details
- 对方多次向我承诺无法兑现的功... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-19T23:04:57Z | 2026-04-30T04:54:03Z | 2026-04-25T04:53:10Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | w13450071043-a11y | 266,811,167 | U_kgDOD-c3Hw | User | false |
openclaw/openclaw | 4,104,922,357 | I_kwDOQb6kR870rCT1 | 50,711 | https://github.com/openclaw/openclaw/issues/50711 | https://api.github.com/repos/openclaw/openclaw/issues/50711 | [Bug]: Control UI model picker strips openrouter/ prefix from model IDs | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
## Description
The Control UI model picker dropdown sends model IDs without the `openrouter/` provider prefix, causing "model not allowed" errors when selecting any OpenRouter model.
### Steps to reproduce
## Steps to reproduce
1. Confi... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-20T00:11:04Z | 2026-04-30T04:54:03Z | 2026-04-25T04:53:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yunodu | 147,952,274 | U_kgDOCNGSkg | User | false |
openclaw/openclaw | 4,104,924,203 | I_kwDOQb6kR870rCwr | 50,713 | https://github.com/openclaw/openclaw/issues/50713 | https://api.github.com/repos/openclaw/openclaw/issues/50713 | [Bug] Ollama tool call arguments not deserialized from string before passing to Ollama API | ## Bug
When an Ollama model returns a tool call, OpenClaw stores `arguments` as a JSON string (OpenAI format). When this assistant message is later sent back to Ollama as conversation context, `extractToolCalls()` passes `arguments` as-is. Ollama expects an **object** for the `function.arguments` field and fails parsi... | closed | completed | true | 2 | [] | [] | 2026-03-20T00:11:41Z | 2026-04-30T04:54:02Z | 2026-04-25T04:53:24Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ai-nurmamat | 130,913,840 | U_kgDOB82WMA | User | false |
openclaw/openclaw | 4,105,002,514 | I_kwDOQb6kR870rV4S | 50,721 | https://github.com/openclaw/openclaw/issues/50721 | https://api.github.com/repos/openclaw/openclaw/issues/50721 | boot-md reads BOOT.md/state on gateway startup but does not deliver outbound Telegram message | ## Summary
`boot-md` runs on gateway startup and successfully reads `BOOT.md` plus the persisted restart state, but it does not deliver the expected outbound message after restart.
Based on the docs, this should work:
- `docs/automation/hooks.md` says `boot-md` sends requested outbound messages via the message tool.
... | closed | completed | true | 2 | [] | [] | 2026-03-20T00:40:11Z | 2026-04-30T04:54:01Z | 2026-04-25T04:53:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jarvispelabot | 262,861,181 | U_kgDOD6rxfQ | User | false |
openclaw/openclaw | 4,105,037,255 | I_kwDOQb6kR870reXH | 50,726 | https://github.com/openclaw/openclaw/issues/50726 | https://api.github.com/repos/openclaw/openclaw/issues/50726 | [Bug]: Context监控显示永远0% - contextTokens被写死为200000 | [Bug]: # GitHub Issue - Context监控显示永远0% Bug / GitHub Issue - Context Monitor Shows 0% Forever
---
## 🐛 Bug Report / 🐛 Bug 报告
### 问题描述 / Problem Description / 问题描述
所有会话的Context监控永远显示 `Context: 0/200k (0%)`，但实际使用率已爆满（飞书会话：69%-138%）。
All sessions' Context monitor permanently displays `Context: 0/200k (0%)`, but act... | closed | completed | true | 2 | [] | [] | 2026-03-20T00:54:06Z | 2026-04-30T04:54:00Z | 2026-04-25T04:53:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | hfz-ai | 268,842,415 | U_kgDOEAY1rw | User | false |
openclaw/openclaw | 4,105,077,576 | I_kwDOQb6kR870roNI | 50,731 | https://github.com/openclaw/openclaw/issues/50731 | https://api.github.com/repos/openclaw/openclaw/issues/50731 | [Bug] WhatsApp outbound sends fail despite active inbound listener (2026.3.13) | ## Description
`openclaw message send --channel whatsapp` fails with `No active WhatsApp Web listener` even though the gateway log shows the WhatsApp provider started successfully and the inbound listener is active.
The 2026.3.13 changelog mentions a fix for this exact pattern (WhatsApp/active-listener: pin the activ... | closed | completed | true | 4 | [] | [] | 2026-03-20T01:10:17Z | 2026-04-30T04:53:59Z | 2026-04-25T04:53:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kwong975 | 254,052,773 | U_kgDODySJpQ | User | false |
openclaw/openclaw | 4,105,099,550 | I_kwDOQb6kR870rtke | 50,735 | https://github.com/openclaw/openclaw/issues/50735 | https://api.github.com/repos/openclaw/openclaw/issues/50735 | [Bug]: wechat-access channel: Cron/announce notifications not delivered | ### Bug type
Regression (worked before, now fails)
### Summary
The wechat-access (腾讯通路) channel cannot receive proactive notifications from cron jobs or manual message sending.
### Steps to reproduce
1. Configure wechat-access channel in OpenClaw
2. User can send messages to AI and receive replies normally
3. Set ... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-20T01:18:50Z | 2026-04-30T04:53:59Z | 2026-04-25T04:53:52Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | chpstudio | 1,185,532 | MDQ6VXNlcjExODU1MzI= | User | false |
openclaw/openclaw | 4,105,102,313 | I_kwDOQb6kR870ruPp | 50,736 | https://github.com/openclaw/openclaw/issues/50736 | https://api.github.com/repos/openclaw/openclaw/issues/50736 | Sub-agent spawns silently fail — all three providers (Opus, Sonnet, GPT 5.4) return no results | ## Bug
Sub-agents spawned via `sessions_spawn` are accepted but never announce completion back to the parent session. This affects ALL providers — not model-specific.
## Structured Test (Mar 19, 2026)
Spawned 3 sub-agents simultaneously with identical tasks (simple generation + file read + computation) to isolate wh... | closed | completed | true | 5 | [] | [] | 2026-03-20T01:19:53Z | 2026-04-30T04:53:58Z | 2026-04-25T04:53:59Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,105,179,659 | I_kwDOQb6kR870sBIL | 50,744 | https://github.com/openclaw/openclaw/issues/50744 | https://api.github.com/repos/openclaw/openclaw/issues/50744 | Gateway hangs on startup after plugin system refactor (never binds port) | ## Description
After pulling the March 16 plugin system refactors, the gateway process hangs during startup and never binds its port (18789). The process stays alive (consuming ~500MB RAM) but the gateway never becomes functional — no heartbeat, no channel connections, no Telegram polling.
## Steps to reproduce
1. R... | closed | completed | true | 2 | [] | [] | 2026-03-20T01:40:29Z | 2026-04-30T04:53:57Z | 2026-04-25T04:54:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | alfredomenezes | 266,347,586 | U_kgDOD-AkQg | User | false |
openclaw/openclaw | 4,105,206,127 | I_kwDOQb6kR870sHlv | 50,747 | https://github.com/openclaw/openclaw/issues/50747 | https://api.github.com/repos/openclaw/openclaw/issues/50747 | [Bug]: [Bug] Subagent completion signal not received by parent session, message queue stuck 子Agent完成后主会话收不到完成信号，消息队列卡住 | ### Bug type
Regression (worked before, now fails)
### Summary
# 🐛 Bug 报告模板 - OpenClaw 子Agent完成信号丢失
---
## 🇨🇳 中文版
```markdown
### 标题
[Bug] 子Agent完成后主会话收不到完成信号，消息队列卡住
### 描述
子Agent实际已完成输出（可在PowerShell终端看到完整结果），但主会话状态未更新，导致消息队列阻塞，新消息显示"Queued"。重启网关后，积压的消息才被处理并显示结果。
### 复现步骤
1. 使用 `sessions_spawn` 启动子Agent（文档生成... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-20T01:50:17Z | 2026-04-30T04:53:55Z | 2026-04-25T04:54:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | a11911a | 269,615,203 | U_kgDOEBIAYw | User | false |
openclaw/openclaw | 4,105,218,293 | I_kwDOQb6kR870sKj1 | 50,748 | https://github.com/openclaw/openclaw/issues/50748 | https://api.github.com/repos/openclaw/openclaw/issues/50748 | [Bug]: Behavior bug | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The /model command resolves the minimax model alias Minimax2.7 to anthropic/MiniMax-M2.7 instead of minimax/MiniMax-M2.7, causing a model allowlist error.
### Steps to reproduce
1.Configure a minimax provider with api: "anthropic-messages... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-20T01:56:11Z | 2026-04-30T04:53:54Z | 2026-04-25T04:54:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | KaigeGao1110 | 200,879,999 | U_kgDOC_kvfw | User | false |
openclaw/openclaw | 4,105,253,413 | I_kwDOQb6kR870sTIl | 50,752 | https://github.com/openclaw/openclaw/issues/50752 | https://api.github.com/repos/openclaw/openclaw/issues/50752 | [Bug]: session-memory hook produces empty memory file when /new is called on a timed-out (stale) session | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When a session becomes stale due to inactivity timeout and the user subsequently calls /new, the session-memory hook fires but writes a header-only memory file with no conversation content. The hook receives a reference to the new session i... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-20T02:09:31Z | 2026-04-30T04:53:54Z | 2026-04-25T04:54:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xujiahao718 | 48,267,442 | MDQ6VXNlcjQ4MjY3NDQy | User | false |
openclaw/openclaw | 4,105,254,016 | I_kwDOQb6kR870sTSA | 50,753 | https://github.com/openclaw/openclaw/issues/50753 | https://api.github.com/repos/openclaw/openclaw/issues/50753 | Create AlertPipe and HealthPipe ecosystem tools | Following [OpenClaw discussion #50364](https://github.com/openclaw/openclaw/discussions/50364) about AlertPipe & HealthPipe.
These are new utility tools to improve workspace reliability:
**HealthPipe**: `healthpipe status` returns health score (0-100) based on gateway uptime, cron success, memory growth, restart freq... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T02:09:46Z | 2026-04-30T04:53:53Z | 2026-04-25T04:54:34Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DockeGumi | 258,497,605 | U_kgDOD2hcRQ | User | false |
openclaw/openclaw | 4,105,272,113 | I_kwDOQb6kR870sXsx | 50,756 | https://github.com/openclaw/openclaw/issues/50756 | https://api.github.com/repos/openclaw/openclaw/issues/50756 | [Bug]: The copy Button is displayed as "CopyCopied!" | ### Bug type
Regression (worked before, now fails)
### Summary
<button type="button" class="code-block-copy" data-code="C:\Users\Evin.Zhang\Pictures\xxx.png" aria-label="Copy code"><span class="code-block-copy__idle">Copy</span><span class="code-block-copy__done">Copied!</span></button>
### Steps to reproduce
1.Op... | closed | completed | true | 5 | [
"bug",
"regression"
] | [] | 2026-03-20T02:14:39Z | 2026-04-30T04:53:52Z | 2026-04-25T04:54:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | icestorms | 34,729,430 | MDQ6VXNlcjM0NzI5NDMw | User | false |
openclaw/openclaw | 4,105,311,247 | I_kwDOQb6kR870shQP | 50,759 | https://github.com/openclaw/openclaw/issues/50759 | https://api.github.com/repos/openclaw/openclaw/issues/50759 | [Bug]:Local Ollama provider incorrectly requires API key in auth preflight (v2026.3.13) | ### Bug type
Regression (worked before, now fails)
### Summary
Local Ollama provider incorrectly requires API key in auth preflight (v2026.3.13)
Environment:
OS: macOS (iMac / arm64)
OpenClaw Version: 2026.3.13
Provider: Ollama (Local)
Model: llama3.1:8b
Description:
Despite configuring the gateway to use a lo... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-20T02:22:59Z | 2026-04-30T04:53:51Z | 2026-04-25T04:54:48Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | freakingcat23 | 263,116,638 | U_kgDOD67XXg | User | false |
openclaw/openclaw | 4,105,339,755 | I_kwDOQb6kR870soNr | 50,769 | https://github.com/openclaw/openclaw/issues/50769 | https://api.github.com/repos/openclaw/openclaw/issues/50769 | Rate Limit Request: Increase clawhub skill installation quota | ### Description
I am requesting an increase in the rate limit for skill installations via `clawhub` CLI.
### Current Behavior
When attempting to install skills using `clawhub install`, I receive a rate limit error after installing approximately 10 skills within a short period.
### Error Message
```
Rate limit exceede... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T02:33:51Z | 2026-04-30T04:53:51Z | 2026-04-25T04:54:55Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zouyuechun | 269,538,735 | U_kgDOEBDVrw | User | false |
openclaw/openclaw | 4,105,355,446 | I_kwDOQb6kR870ssC2 | 50,771 | https://github.com/openclaw/openclaw/issues/50771 | https://api.github.com/repos/openclaw/openclaw/issues/50771 | Feature: Pre-execution tool dispatch hooks (critic/validator pattern) | ## Summary
Add a middleware/hook system in the tool dispatch pipeline that allows validation **after** the model decides on a tool call but **before** it executes. This enables patterns like:
- Verifying a routing/resolver tool (e.g. `resolve_skill`) was consulted before executing work tools
- Checking the chosen too... | closed | completed | true | 1 | [] | [] | 2026-03-20T02:39:49Z | 2026-04-30T04:53:50Z | 2026-04-25T04:55:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | coolmanns | 117,685,070 | U_kgDOBwO7Tg | User | false |
openclaw/openclaw | 4,105,362,982 | I_kwDOQb6kR870st4m | 50,775 | https://github.com/openclaw/openclaw/issues/50775 | https://api.github.com/repos/openclaw/openclaw/issues/50775 | [Bug]: webchat Channel Attachment Image Path Not Injected into Message Metadata | ### Bug type
Regression (worked before, now fails)
### Summary
When sending images via the webchat channel (OpenClaw Control UI), whether by:
1. Ctrl+V pasting a screenshot
2. Clicking the attachment button to select a local image
The message metadata received by the Agent **does not contain attachment path informa... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-20T02:42:50Z | 2026-04-30T04:53:49Z | 2026-04-25T04:55:10Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | MartyAlien | 62,040,646 | MDQ6VXNlcjYyMDQwNjQ2 | User | false |
openclaw/openclaw | 4,105,366,862 | I_kwDOQb6kR870su1O | 50,778 | https://github.com/openclaw/openclaw/issues/50778 | https://api.github.com/repos/openclaw/openclaw/issues/50778 | Control UI: model switcher prepends wrong provider prefix | ## Bug
When switching models via the Control UI dashboard dropdown, the UI prepends the **current session's provider** instead of using the full `provider/model` ID from the configured models catalog.
### Example
- Models catalog contains: `openai/gpt-5.4`, `anthropic/claude-opus-4-6`
- Current session is on `anthro... | closed | completed | true | 1 | [] | [] | 2026-03-20T02:44:09Z | 2026-04-30T04:53:48Z | 2026-04-25T04:55:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | johnaustinfilmcrew | 264,027,730 | U_kgDOD7y-Ug | User | false |
openclaw/openclaw | 4,105,376,262 | I_kwDOQb6kR870sxIG | 50,780 | https://github.com/openclaw/openclaw/issues/50780 | https://api.github.com/repos/openclaw/openclaw/issues/50780 | Crash: Cannot read properties of undefined (reading 'includes') when streaming model sends content:null during reasoning | ## Bug Description
OpenClaw crashes with `Cannot read properties of undefined (reading 'includes')` when the upstream LLM provider sends streaming chunks with `"content": null` during the reasoning/thinking phase.
## Root Cause
Models like **GLM-5**, **Qwen3.5-Plus**, and **MiniMax-M2.5** (via alibaba-coding / alico... | closed | completed | true | 2 | [] | [] | 2026-03-20T02:47:34Z | 2026-04-30T04:53:48Z | 2026-04-25T04:55:24Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mxgiangbui | 261,258,233 | U_kgDOD5J7-Q | User | false |
openclaw/openclaw | 4,105,385,610 | I_kwDOQb6kR870szaK | 50,784 | https://github.com/openclaw/openclaw/issues/50784 | https://api.github.com/repos/openclaw/openclaw/issues/50784 | WhatsApp group messages not routed to auto-reply for secondary accounts (multi-account setup) | ## Bug Description
In a multi-account WhatsApp setup, group messages arriving on secondary accounts (non-default) are received at the `web-inbound` layer but **never dispatched to `web-auto-reply`**, so the agent never processes them. DMs on the same account work correctly.
## Environment
- **OpenClaw version:** 202... | closed | completed | true | 3 | [] | [] | 2026-03-20T02:51:21Z | 2026-04-30T04:53:47Z | 2026-04-25T04:55:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | vittorioliberti19-wq | 263,185,754 | U_kgDOD6_lWg | User | false |
openclaw/openclaw | 4,105,413,220 | I_kwDOQb6kR870s6Jk | 50,788 | https://github.com/openclaw/openclaw/issues/50788 | https://api.github.com/repos/openclaw/openclaw/issues/50788 | Control UI displays internal relevant-memories metadata to users | ## Bug Description
In Control UI (webchat), users see internal metadata information after sending messages, including 'relevant-memories', 'Knowledge' and other debug information that should not be shown to end users.
## Steps to Reproduce
1. Open Control UI (http://127.0.0.1:18789/)
2. Send any message
3. After sendi... | closed | completed | true | 1 | [] | [] | 2026-03-20T03:01:45Z | 2026-04-30T04:53:46Z | 2026-04-25T04:55:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yangjiale515 | 26,948,042 | MDQ6VXNlcjI2OTQ4MDQy | User | false |
openclaw/openclaw | 4,105,421,619 | I_kwDOQb6kR870s8Mz | 50,790 | https://github.com/openclaw/openclaw/issues/50790 | https://api.github.com/repos/openclaw/openclaw/issues/50790 | Feature request: session-scoped reset / forget for chat and group contexts | ## Summary
OpenClaw currently appears to lack a first-class way to reset or forget a specific session context without affecting the broader runtime.
In practice, when a single chat or group session becomes polluted, stale, or behaviorally stuck, the available options are currently awkward:
- do nothing and wait for ... | closed | completed | true | 2 | [] | [] | 2026-03-20T03:04:42Z | 2026-04-30T04:53:45Z | 2026-04-25T04:55:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 1sexywoo8 | 262,731,466 | U_kgDOD6j2yg | User | false |
openclaw/openclaw | 4,105,435,968 | I_kwDOQb6kR870s_tA | 50,791 | https://github.com/openclaw/openclaw/issues/50791 | https://api.github.com/repos/openclaw/openclaw/issues/50791 | [Bug]: Auto-resume pending sessions after Gateway restart | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Currently, if an agent encounters a timeout or connection failure while checking a task status, it may automatically trigger a Gateway restart. However, after the restart completes, the agent does not resume the previously active sessions.... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-20T03:09:36Z | 2026-04-30T04:53:44Z | 2026-04-25T04:55:52Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dpptinh | 150,330,248 | U_kgDOCPXbiA | User | false |
openclaw/openclaw | 4,105,438,771 | I_kwDOQb6kR870tAYz | 50,792 | https://github.com/openclaw/openclaw/issues/50792 | https://api.github.com/repos/openclaw/openclaw/issues/50792 | [Feature]: Unified conversation history across channels | ### Summary
Currently, conversations started on one channel (e.g., Telegram) are not accessible when switching to another channel (e.g., Discord).
### Problem to solve
## Problem
Currently, conversations started on one channel (e.g., Telegram) are not accessible when switching to another channel (e.g., Discord). Th... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-20T03:10:38Z | 2026-04-30T04:53:43Z | 2026-04-25T04:55:59Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | JotaPe0154 | 269,626,503 | U_kgDOEBIshw | User | false |
openclaw/openclaw | 4,105,492,083 | I_kwDOQb6kR870tNZz | 50,803 | https://github.com/openclaw/openclaw/issues/50803 | https://api.github.com/repos/openclaw/openclaw/issues/50803 | Synology Chat: insecure TLS default in transport functions | ## Summary
The transport-level functions in extensions/synology-chat/src/client.ts (sendMessage, sendFileUrl, fetchChatUsers, resolveChatUserId, doPost) all default allowInsecureSsl = true, which sets rejectUnauthorized: false on HTTPS connections. This disables TLS certificate verification by default, enabling potent... | closed | completed | true | 1 | [] | [] | 2026-03-20T03:30:34Z | 2026-04-30T04:53:43Z | 2026-04-25T04:56:07Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | scottgl9 | 10,623,288 | MDQ6VXNlcjEwNjIzMjg4 | User | false |
openclaw/openclaw | 4,105,582,031 | I_kwDOQb6kR870tjXP | 50,823 | https://github.com/openclaw/openclaw/issues/50823 | https://api.github.com/repos/openclaw/openclaw/issues/50823 | [Feature]: Strategic Game-like Visualization Dashboard for Agent Orchestration and Monitoring | ### Summary
Hi OpenClaw Team,
First of all, huge thanks for building such an incredible tool that truly makes AI "do things." The current CLI and TUI interfaces are powerful, but as users start orchestrating complex multi-agent workflows or long-running tasks, there's a growing need for a higher-level visual overview.... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-20T04:01:37Z | 2026-04-30T04:53:41Z | 2026-04-25T04:56:21Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zgsndzq | 258,358,461 | U_kgDOD2Y8vQ | User | false |
openclaw/openclaw | 4,105,620,558 | I_kwDOQb6kR870tsxO | 50,828 | https://github.com/openclaw/openclaw/issues/50828 | https://api.github.com/repos/openclaw/openclaw/issues/50828 | Docs: broken internal links across documentation | Multiple broken internal links found across the documentation. These lead to 404s or incorrect anchors on docs.openclaw.ai.
## Broken links found
### Missing pages
- `docs/plugins/manifest.md` and `docs/tools/plugin.md`: link to `/configuration` (page does not exist, should be `/gateway/configuration-reference`)
###... | closed | not_planned | true | 1 | [] | [] | 2026-03-20T04:11:56Z | 2026-04-30T04:53:40Z | 2026-04-25T04:56:28Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | scottgl9 | 10,623,288 | MDQ6VXNlcjEwNjIzMjg4 | User | false |
openclaw/openclaw | 4,105,639,993 | I_kwDOQb6kR870txg5 | 50,833 | https://github.com/openclaw/openclaw/issues/50833 | https://api.github.com/repos/openclaw/openclaw/issues/50833 | Docs: stale 'unreleased at the time of writing' version references | Two docs still describe version 2026.1.12 as 'unreleased at the time of writing', but that version has been released (current is 2026.3.14).
## Instances
1. `docs/providers/minimax.md:201`: 'unreleased at the time of writing'
2. `docs/help/faq.md:2152`: 'unreleased at the time of writing' | closed | completed | true | 2 | [] | [] | 2026-03-20T04:17:47Z | 2026-04-30T04:53:40Z | 2026-04-25T04:56:35Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | scottgl9 | 10,623,288 | MDQ6VXNlcjEwNjIzMjg4 | User | false |
openclaw/openclaw | 4,105,642,156 | I_kwDOQb6kR870tyCs | 50,836 | https://github.com/openclaw/openclaw/issues/50836 | https://api.github.com/repos/openclaw/openclaw/issues/50836 | Discord: startup race condition causes permanent `connected: false` after health-monitor restart | ## Bug Description
After a health-monitor restart of the Discord provider, the channel status `connected` permanently stays `false` in ~95% of reconnection attempts. This creates an infinite 10-minute restart loop (`disconnected` → restart → `disconnected`) that only recovers when random timing variation allows the Di... | closed | completed | true | 3 | [] | [] | 2026-03-20T04:18:22Z | 2026-04-30T04:53:39Z | 2026-04-25T04:56:42Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | harrison-ming | 77,006,832 | MDQ6VXNlcjc3MDA2ODMy | User | false |
openclaw/openclaw | 4,105,695,305 | I_kwDOQb6kR870t_BJ | 50,844 | https://github.com/openclaw/openclaw/issues/50844 | https://api.github.com/repos/openclaw/openclaw/issues/50844 | Cron job timeout/error should send notification via announce delivery | ## Problem
When a cron job times out or fails, the current behavior with `announce` delivery mode is completely silent. The job runs, fails/times out, and no notification is sent to the user — they only discover the failure by checking logs or wondering why expected output never arrived.
## Expected behavior
When a ... | closed | completed | true | 2 | [] | [] | 2026-03-20T04:37:15Z | 2026-04-30T04:53:38Z | 2026-04-25T04:56:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wangwllu | 7,668,944 | MDQ6VXNlcjc2Njg5NDQ= | User | false |
openclaw/openclaw | 4,105,891,393 | I_kwDOQb6kR870uu5B | 50,866 | https://github.com/openclaw/openclaw/issues/50866 | https://api.github.com/repos/openclaw/openclaw/issues/50866 | Telegram 429 rate limit causes crash spiral when multiple agents/crons post simultaneously | ## Problem
When multiple agent sessions or cron jobs complete around the same time (e.g., after an Anthropic API overload clears, or after a gateway restart fires missed cron jobs), the gateway attempts to send all queued Telegram messages in a burst. This triggers Telegram's 429 rate limit (`Too Many Requests: retry ... | closed | completed | true | 2 | [] | [] | 2026-03-20T05:31:37Z | 2026-04-30T04:53:37Z | 2026-04-25T04:56:56Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rets85 | 29,980,872 | MDQ6VXNlcjI5OTgwODcy | User | false |
openclaw/openclaw | 4,105,901,484 | I_kwDOQb6kR870uxWs | 50,869 | https://github.com/openclaw/openclaw/issues/50869 | https://api.github.com/repos/openclaw/openclaw/issues/50869 | daemon start: auto-bootstrap when service not loaded after stop | ## Problem
`openclaw daemon stop` uses `launchctl bootout` to fully unregister the LaunchAgent (necessary because `KeepAlive` would otherwise respawn it). However, `openclaw daemon start` does not auto-recover from this state — it prints hints and exits:
```
$ openclaw daemon stop
Stopped LaunchAgent: gui/501/ai.open... | closed | completed | true | 2 | [] | [] | 2026-03-20T05:34:34Z | 2026-04-30T04:53:37Z | 2026-04-25T04:57:03Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | bridzl | 52,784,198 | MDQ6VXNlcjUyNzg0MTk4 | User | false |
openclaw/openclaw | 4,105,971,572 | I_kwDOQb6kR870vCd0 | 50,877 | https://github.com/openclaw/openclaw/issues/50877 | https://api.github.com/repos/openclaw/openclaw/issues/50877 | [Bug]: OpenCLaw Misalignment of scheduled task time | ### Bug type
Regression (worked before, now fails)
### Summary
OpenClaw Create corn use UTC time ,But it shoud Use Local Time
### Steps to reproduce
1. OpenClaw verson is 2026.3.13
2. My Computer time is
ops@virtual-machine:~$ timedatectl
Local time: Fri 2026-03-20 11:19:47 CST
Universa... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-20T05:56:49Z | 2026-04-30T04:53:36Z | 2026-04-25T04:57:10Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dongdong2048 | 66,246,797 | MDQ6VXNlcjY2MjQ2Nzk3 | User | false |
openclaw/openclaw | 4,106,002,964 | I_kwDOQb6kR870vKIU | 50,879 | https://github.com/openclaw/openclaw/issues/50879 | https://api.github.com/repos/openclaw/openclaw/issues/50879 | Telegram: message.send with poll fields returns ok but no poll is posted | # Bug report: `message.send` with poll fields does not create Telegram poll (text-only sent)
## Environment
- OpenClaw stable channel
- Telegram group chat (`-5115843474`)
- Tool used: `message` (`action=send`)
## Problem
When calling `message.send` with poll parameters (`pollQuestion`, `pollOption`, etc.), the call ... | closed | completed | true | 2 | [] | [] | 2026-03-20T06:05:11Z | 2026-04-30T04:53:35Z | 2026-04-25T04:57:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Palgrich | 39,589,387 | MDQ6VXNlcjM5NTg5Mzg3 | User | false |
openclaw/openclaw | 4,106,095,495 | I_kwDOQb6kR870vguH | 50,892 | https://github.com/openclaw/openclaw/issues/50892 | https://api.github.com/repos/openclaw/openclaw/issues/50892 | [Bug]: Discord collect-mode auto-reply drain causes duplicate message delivery | ## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
When an agent session uses **collect queue mode** on Discord, the auto-reply drain mechanism delivers the same response message **twice** to the Discord channel. This happens consistently for normal conversational replies (not cron, not announc... | closed | completed | true | 3 | [] | [] | 2026-03-20T06:31:01Z | 2026-04-30T04:53:33Z | 2026-04-25T04:57:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | irchelper | 38,723,383 | MDQ6VXNlcjM4NzIzMzgz | User | false |
openclaw/openclaw | 4,106,095,650 | I_kwDOQb6kR870vgwi | 50,893 | https://github.com/openclaw/openclaw/issues/50893 | https://api.github.com/repos/openclaw/openclaw/issues/50893 | [Bug] WeChat plugin cannot receive messages | ## 问题描述
WeChat plugin installed via QClaw plugin store cannot receive messages. It can send messages normally, but messages from contacts are not received at all.
## 环境
- OS: Windows 11
- QClaw Version: Latest
- WeChat Version: 3.9.x
## 复现步骤
1. Install WeChat plugin from QClaw plugin store
2. Configure WeChat with... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T06:31:04Z | 2026-04-30T04:53:32Z | 2026-04-25T04:57:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | JK8658 | 269,649,239 | U_kgDOEBKFVw | User | false |
openclaw/openclaw | 4,106,153,192 | I_kwDOQb6kR870vuzo | 50,899 | https://github.com/openclaw/openclaw/issues/50899 | https://api.github.com/repos/openclaw/openclaw/issues/50899 | [Bug] WeChat plugin cannot receive messages | ## Bug Description
The WeChat plugin (wechat-access) is unable to receive messages in the current version.
## Environment
- OS: Windows 11 (NT 10.0.26200)
- OpenClaw: Latest version
- QClaw: Latest version
- WeChat: 3.9.x
- Plugin: wechat-access
## Expected Behavior
- Receive and display text messages from WeChat ... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T06:48:00Z | 2026-04-30T04:53:31Z | 2026-04-25T04:57:59Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | JK8658 | 269,649,239 | U_kgDOEBKFVw | User | false |
openclaw/openclaw | 4,106,171,082 | I_kwDOQb6kR870vzLK | 50,904 | https://github.com/openclaw/openclaw/issues/50904 | https://api.github.com/repos/openclaw/openclaw/issues/50904 | Feature Request: Support automatic multi-model task routing and fallback | ### Summary
## Summary
I would like OpenClaw to support **automatic model selection and routing based on task type or complexity**, as well as a **fallback mechanism** in case the primary model fails. Currently, OpenClaw requires manually switching between agents for different models, which limits workflow efficiency ... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-20T06:52:48Z | 2026-04-30T04:53:30Z | 2026-04-25T04:58:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | lca1rus01 | 92,661,223 | U_kgDOBYXl5w | User | false |
openclaw/openclaw | 4,106,297,498 | I_kwDOQb6kR870wSCa | 50,905 | https://github.com/openclaw/openclaw/issues/50905 | https://api.github.com/repos/openclaw/openclaw/issues/50905 | Control UI displays [TUI] evt= debug messages in chat interface | ## Description
Control UI (webchat) displays `[TUI] evt=health sk=undefined cur=agent:main:main` and `[TUI] evt=agent sk=agent:main:main` messages in the chat area / status bar. These WebSocket event logs leak into the user-visible interface.
## Steps to Reproduce
1. Open Control UI in browser (http://localhost:18789)... | closed | completed | true | 3 | [] | [] | 2026-03-20T07:25:29Z | 2026-04-30T04:53:29Z | 2026-04-25T04:58:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | MarsLobster2 | 266,109,704 | U_kgDOD9yDCA | User | false |
openclaw/openclaw | 4,106,389,367 | I_kwDOQb6kR870wod3 | 50,913 | https://github.com/openclaw/openclaw/issues/50913 | https://api.github.com/repos/openclaw/openclaw/issues/50913 | Discord channel plugin missing stopAccount — health-monitor hot restart leaves stale WS, enters death loop | ## Bug Description
When the Discord WebSocket disconnects (close code 1006), the health-monitor detects `stale-socket` or `disconnected` and calls `stopChannel` → `startChannel` to restart the Discord provider. However, the Discord channel plugin **does not implement `stopAccount`**, so the old WebSocket connection is... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T07:46:00Z | 2026-04-30T04:53:29Z | 2026-04-25T04:58:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | doldrums2025 | 236,311,651 | U_kgDODhXUYw | User | false |
openclaw/openclaw | 4,106,424,578 | I_kwDOQb6kR870wxEC | 50,916 | https://github.com/openclaw/openclaw/issues/50916 | https://api.github.com/repos/openclaw/openclaw/issues/50916 | 优化 Kimi web_search：实现完整的两轮对话流程 | ## 问题描述
当前 OpenClaw 的 Kimi web_search 实现使用单次请求，返回结果简略且不够准确。
## 建议改进
实现完整的两轮对话流程：
### 第一轮：发起搜索请求
```
POST https://api.moonshot.cn/v1/chat/completions
Content-Type: application/json
Authorization: Bearer <api_key>
{
"model": "kimi-k2.5",
"messages": [{"role": "user", "content": "查询内容"}],
"tools": [{"type": "bu... | closed | completed | true | 1 | [] | [] | 2026-03-20T07:53:32Z | 2026-04-30T04:53:28Z | 2026-04-25T04:58:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | lsw7314 | 129,318,552 | U_kgDOB7U-mA | User | false |
openclaw/openclaw | 4,106,489,219 | I_kwDOQb6kR870xA2D | 50,925 | https://github.com/openclaw/openclaw/issues/50925 | https://api.github.com/repos/openclaw/openclaw/issues/50925 | [Proposal] SovereignLedger: Unified Accounting Ecosystem | **Proposal**: SovereignLedger — Consolidated Accounting & Bookkeeping Service
**Problem**: Financial tools are fragmented (DebtPipe, Budget Tool, etc.). Need unified double-entry ledger.
**Features**:
- Double-entry bookkeeping: entries with `debit_account`, `credit_account`, `amount`, `currency`, `date`.
- Trial bal... | closed | not_planned | true | 1 | [] | [] | 2026-03-20T08:07:34Z | 2026-04-30T04:53:27Z | 2026-04-25T04:58:34Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DockeGumi | 258,497,605 | U_kgDOD2hcRQ | User | false |
openclaw/openclaw | 4,106,490,340 | I_kwDOQb6kR870xBHk | 50,926 | https://github.com/openclaw/openclaw/issues/50926 | https://api.github.com/repos/openclaw/openclaw/issues/50926 | [Proposal] VaultSeal: Encryption-as-a-Service | **Proposal**: VaultSeal — Encryption-as-a-Service for Workspace Data
**Goal**: Provide transparent encryption for sensitive DuckDB files and config at rest, with audit logging.
**Architecture**:
- Lightweight service exposing `/encrypt` and `/decrypt` endpoints (AES-256-GCM).
- Key management: `AGE` file-based (defau... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T08:07:48Z | 2026-04-30T04:53:26Z | 2026-04-25T04:58:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DockeGumi | 258,497,605 | U_kgDOD2hcRQ | User | false |
openclaw/openclaw | 4,106,491,008 | I_kwDOQb6kR870xBSA | 50,927 | https://github.com/openclaw/openclaw/issues/50927 | https://api.github.com/repos/openclaw/openclaw/issues/50927 | [Proposal] Pulse: Service Health Monitoring Dashboard | **Proposal**: Pulse — Health Monitoring Dashboard for Workspace Services
**Scope**: Unified monitoring of OpenClaw gateway, QuantPipe, GitchPage, SovereignLedger, VaultSeal, and infrastructure (CPU, RAM, disk).
**Metrics**:
- System: CPU %, RAM %, disk %, uptime.
- Services: HTTP 2xx rate, response time p95, error ra... | closed | not_planned | true | 1 | [] | [] | 2026-03-20T08:07:57Z | 2026-04-30T04:53:25Z | 2026-04-25T04:58:48Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DockeGumi | 258,497,605 | U_kgDOD2hcRQ | User | false |
openclaw/openclaw | 4,106,491,638 | I_kwDOQb6kR870xBb2 | 50,928 | https://github.com/openclaw/openclaw/issues/50928 | https://api.github.com/repos/openclaw/openclaw/issues/50928 | [Proposal] Agent Autonomy Patterns — Undo Stacks & Atomic Operations | **Proposal**: Agent Autonomy Patterns — Undo Stacks & Atomic Operations
**Vision** (2025 AI agents): Treat atomicity as infrastructure, not prompting. Enable reversible operations and self-improvement loops.
**Undo Stack**:
- Wrap state-changing operations in `with undo_context():` block.
- Push inverse operation ont... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T08:08:08Z | 2026-04-30T04:53:25Z | 2026-04-25T04:58:55Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DockeGumi | 258,497,605 | U_kgDOD2hcRQ | User | false |
openclaw/openclaw | 4,106,621,636 | I_kwDOQb6kR870xhLE | 50,939 | https://github.com/openclaw/openclaw/issues/50939 | https://api.github.com/repos/openclaw/openclaw/issues/50939 | [Bug] Model switch displays wrong prefix: moonshot/qwen3.5-plus | ## Bug Description
When switching models in the chat window and selecting dashscope/qwen3.5-plus, an error occurs:
Failed to set model: GatewayRequestError: model not allowed: moonshot/qwen3.5-plus
The error shows moonshot/qwen3.5-plus, but the correct configuration in openclaw.json is dashscope/qwen3.5-pl... | closed | completed | true | 2 | [] | [] | 2026-03-20T08:37:50Z | 2026-04-30T04:53:24Z | 2026-04-25T04:59:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | soulofandroid | 17,871,294 | MDQ6VXNlcjE3ODcxMjk0 | User | false |
openclaw/openclaw | 4,106,623,136 | I_kwDOQb6kR870xhig | 50,940 | https://github.com/openclaw/openclaw/issues/50940 | https://api.github.com/repos/openclaw/openclaw/issues/50940 | [Feature]: Heartbeat should be documented as periodic awareness (not uptime monitoring), and support task-scoped auto-enable/disable | ## Summary
OpenClaw heartbeat is very useful as a periodic agent turn / awareness loop, but it is easy for users to misunderstand it as a system health check or uptime signal.
A heartbeat that runs inside the same OpenClaw/Gateway environment can help with:
- background task follow-up
- interruption detection
- lightwe... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-20T08:38:11Z | 2026-04-30T04:53:23Z | 2026-04-25T04:59:09Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | youdmeng | 38,751,158 | MDQ6VXNlcjM4NzUxMTU4 | User | false |
openclaw/openclaw | 4,106,630,607 | I_kwDOQb6kR870xjXP | 50,942 | https://github.com/openclaw/openclaw/issues/50942 | https://api.github.com/repos/openclaw/openclaw/issues/50942 | [Bug]: Cron schedule object format not accepted (contradicts documentation) | ### Bug type
Regression (worked before, now fails)
### Summary
Cron configuration with object format {"schedule": {"kind": "cron", "expr": "..."}} is rejected by gateway with error "invalid cron schedule: expr is required", despite documentation stating both simple and object formats are supported.
### Steps to rep... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-20T08:39:56Z | 2026-04-30T04:53:22Z | 2026-04-25T04:59:16Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | haojibb-beep | 265,210,968 | U_kgDOD87MWA | User | false |
openclaw/openclaw | 4,106,686,189 | I_kwDOQb6kR870xw7t | 50,945 | https://github.com/openclaw/openclaw/issues/50945 | https://api.github.com/repos/openclaw/openclaw/issues/50945 | Feature: Native agent mesh discovery via Nostr (decentralized peer-to-peer agent network) | # Feature Request: Native Agent Mesh Discovery via Nostr
## Summary
Add built-in peer discovery to OpenClaw so that every running instance can
find and communicate with other OpenClaw agents across the internet — with
zero configuration, no shared infrastructure, and no accounts required.
## Motivation
OpenClaw age... | closed | not_planned | true | 2 | [] | [] | 2026-03-20T08:54:54Z | 2026-04-30T04:53:22Z | 2026-04-25T04:59:23Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Codejain1 | 124,192,173 | U_kgDOB2cFrQ | User | false |
openclaw/openclaw | 4,106,689,479 | I_kwDOQb6kR870xxvH | 50,946 | https://github.com/openclaw/openclaw/issues/50946 | https://api.github.com/repos/openclaw/openclaw/issues/50946 | Feature Request: Display images directly in webchat responses | ## Problem
When the assistant generates or reads an image file (PNG/JPG), the image is not displayed directly in the webchat interface. Instead, users see either:
- A broken/empty response
- Need to manually open the file path in the filesystem
This happens with:
- `read` tool on image files
- `message` tool with ima... | closed | completed | true | 1 | [] | [] | 2026-03-20T08:55:46Z | 2026-04-30T04:53:21Z | 2026-04-25T04:59:30Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wangbiwang | 45,363,247 | MDQ6VXNlcjQ1MzYzMjQ3 | User | false |
openclaw/openclaw | 4,106,690,250 | I_kwDOQb6kR870xx7K | 50,947 | https://github.com/openclaw/openclaw/issues/50947 | https://api.github.com/repos/openclaw/openclaw/issues/50947 | Feature Request: Display images directly in webchat responses | ## Problem
When the assistant generates or reads an image file (PNG/JPG), the image is not displayed directly in the webchat interface. Instead, users see either:
- A broken/empty response
- Need to manually open the file path in the filesystem
This happens with:
- read tool on image files
- message tool with image a... | closed | completed | true | 1 | [] | [] | 2026-03-20T08:55:59Z | 2026-04-30T04:53:20Z | 2026-04-25T04:59:37Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wangbiwang | 45,363,247 | MDQ6VXNlcjQ1MzYzMjQ3 | User | false |
openclaw/openclaw | 4,106,690,933 | I_kwDOQb6kR870xyF1 | 50,948 | https://github.com/openclaw/openclaw/issues/50948 | https://api.github.com/repos/openclaw/openclaw/issues/50948 | Feature Request: Display images directly in webchat responses | ## Problem
When the assistant generates or reads an image file, the image is not displayed directly in the webchat interface.
## Expected Behavior
Images should render inline in webchat.
## Proposed Solution
Add image rendering support to webchat. | closed | completed | true | 1 | [] | [] | 2026-03-20T08:56:08Z | 2026-04-30T04:53:19Z | 2026-04-25T04:59:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wangbiwang | 45,363,247 | MDQ6VXNlcjQ1MzYzMjQ3 | User | false |
openclaw/openclaw | 4,106,692,805 | I_kwDOQb6kR870xyjF | 50,949 | https://github.com/openclaw/openclaw/issues/50949 | https://api.github.com/repos/openclaw/openclaw/issues/50949 | Feature Request: Display images directly in webchat responses | ## Problem
When the assistant generates or reads an image file (PNG/JPG), the image is not displayed directly in the webchat interface.
## Expected Behavior
Images should render inline in webchat, similar to Discord/Telegram.
## Proposed Solution
Add image rendering support to webchat for read tool and message too... | closed | completed | true | 1 | [] | [] | 2026-03-20T08:56:39Z | 2026-04-30T04:53:19Z | 2026-04-25T04:59:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wangbiwang | 45,363,247 | MDQ6VXNlcjQ1MzYzMjQ3 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.