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 | 4,019,556,677 | I_kwDOQb6kR87vlZFF | 33,738 | https://github.com/openclaw/openclaw/issues/33738 | https://api.github.com/repos/openclaw/openclaw/issues/33738 | feat: 支持飞书语音消息发送 (Feishu audio message support) | ## 功能描述
希望在飞书 channel 中支持发送语音消息，而不是作为文件附件。
## 使用场景
- TTS 生成的语音可以直接作为语音气泡发送，用户无需下载即可播放
- 情报播报、语音提醒等场景更加友好
## 技术方案
飞书语音消息需要：
1. 音频格式转换为 OPUS：`ffmpeg -i input.mp3 -acodec libopus -ac 1 -ar 16000 output.opus`
2. 上传文件获取 file_key：`POST /open-apis/im/v1/files` (file_type=opus)
3. 发送消息：`msg_type: audio`, content 包含 file_k... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T01:36:22Z | 2026-04-22T04:31:46Z | 2026-04-19T04:24:23Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | hadiliu | 1,926,042 | MDQ6VXNlcjE5MjYwNDI= | User | false |
openclaw/openclaw | 4,019,562,453 | I_kwDOQb6kR87vlafV | 33,743 | https://github.com/openclaw/openclaw/issues/33743 | https://api.github.com/repos/openclaw/openclaw/issues/33743 | Harden sandbox buffered-send materialization against path escapes and filename collisions | ## Summary
The new shared `message action=send` buffer hydration restores buffered sends by materializing payloads into media before dispatch, but the sandbox-side materialization path still needs hardening.
Two follow-up issues remain in the sandbox branch:
1. writes under `sandboxRoot/outbound/...` should be harde... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T01:38:57Z | 2026-04-22T04:31:45Z | 2026-04-19T04:24:24Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | raw34 | 1,996,768 | MDQ6VXNlcjE5OTY3Njg= | User | false |
openclaw/openclaw | 4,019,594,830 | I_kwDOQb6kR87vliZO | 33,752 | https://github.com/openclaw/openclaw/issues/33752 | https://api.github.com/repos/openclaw/openclaw/issues/33752 | Feature Request: postAnnounce hook for sub-agent completion | ## Problem
When a sub-agent completes via `sessions_spawn(mode="run")`, the result is auto-announced back to the main session as a message. However, the main agent (LLM) must decide what to do next — and it frequently fails to act on the result (just forwards/echoes it instead of running verification and follow-up ste... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T01:52:24Z | 2026-04-22T04:31:45Z | 2026-04-19T04:24:26Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | nguathen | 213,979,139 | U_kgDODMEQAw | User | false |
openclaw/openclaw | 4,019,607,502 | I_kwDOQb6kR87vllfO | 33,758 | https://github.com/openclaw/openclaw/issues/33758 | https://api.github.com/repos/openclaw/openclaw/issues/33758 | [Bug]: TUI does not display real-time streaming responses with MiniMax-CN (anthropic-messages API) | ### Bug type
Regression (worked before, now fails)
### Summary
TUI does not render real-time streaming responses when using minimax-cn provider (anthropic-messages API); Dashboard works fine.
### Steps to reproduce
1. Fresh install OpenClaw 2026.3.2 via npm on WSL2
2. Run onboarding: `openclaw onboard`, select Min... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T01:57:57Z | 2026-04-22T04:31:44Z | 2026-04-19T04:24:28Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | alexzhu0 | 178,769,291 | U_kgDOCqfNiw | User | false |
openclaw/openclaw | 4,019,648,444 | I_kwDOQb6kR87vlve8 | 33,773 | https://github.com/openclaw/openclaw/issues/33773 | https://api.github.com/repos/openclaw/openclaw/issues/33773 | [Bug]: file operation failed somtime & no memory directory in 3.2 version | ### Bug type
Regression (worked before, now fails)
### Summary
version: I just installed the latest version of 3.2.
BUG1 :
I change the config profile.tools to "full"
But I tried to create some files in workspace dir, it worked only once, then failed.
I tried to change the IDENTITY via feishu, but it doesnt work.
... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T02:15:27Z | 2026-04-22T04:31:43Z | 2026-04-19T04:24:29Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | alexjigit | 10,066,850 | MDQ6VXNlcjEwMDY2ODUw | User | false |
openclaw/openclaw | 4,019,660,188 | I_kwDOQb6kR87vlyWc | 33,781 | https://github.com/openclaw/openclaw/issues/33781 | https://api.github.com/repos/openclaw/openclaw/issues/33781 | Feature Request: Auto-trigger AI response on Slack member_joined_channel event | ## Summary
When a bot is added to a new Slack channel, OpenClaw currently receives the `member_joined_channel` event and converts it to a system event (`enqueueSystemEvent`), but does **not** trigger an AI response. This means the bot silently joins without any acknowledgment or initial setup interaction.
## Use Case... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T02:20:17Z | 2026-04-22T04:31:42Z | 2026-04-19T04:24:31Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | peteradams2026 | 259,958,501 | U_kgDOD36m5Q | User | false |
openclaw/openclaw | 4,019,665,404 | I_kwDOQb6kR87vlzn8 | 33,782 | https://github.com/openclaw/openclaw/issues/33782 | https://api.github.com/repos/openclaw/openclaw/issues/33782 | Feature: Coalesce back-to-back user messages by aborting in-flight model request | Problem
When users send two short messages back-to-back, OpenClaw may already be processing the first one.
This can split what is logically one intent into two turns.
Current pain
- Context gets fragmented across separate requests.
- The first request may continue reasoning on incomplete information.
- Final responses... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T02:22:35Z | 2026-04-22T04:31:41Z | 2026-04-19T04:24:33Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | goxofy | 1,695,416 | MDQ6VXNlcjE2OTU0MTY= | User | false |
openclaw/openclaw | 4,019,673,568 | I_kwDOQb6kR87vl1ng | 33,784 | https://github.com/openclaw/openclaw/issues/33784 | https://api.github.com/repos/openclaw/openclaw/issues/33784 | Voice messages fallback to Gemini instead of local Whisper | ## Bug Report: Voice messages fallback to Gemini instead of local Whisper
### Summary
Telegram voice messages are being transcribed by Gemini API instead of local Whisper, despite the skill `tg-voice-transcribe` being configured to use local Whisper.
### Environment
- OpenClaw version: 2026.3.2
- OS: macOS 26.3 (Buil... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T02:25:27Z | 2026-04-22T04:31:41Z | 2026-04-19T04:24:34Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | hkurbexman | 253,662,397 | U_kgDODx6UvQ | User | false |
openclaw/openclaw | 4,019,706,099 | I_kwDOQb6kR87vl9jz | 33,798 | https://github.com/openclaw/openclaw/issues/33798 | https://api.github.com/repos/openclaw/openclaw/issues/33798 | [Feishu] Download media from quoted/replied messages | ## Problem
When a user **replies to an image message** in a Feishu group chat and @mentions the bot, the bot cannot see the image content.
### Current behavior
- Downloads media from the **current message** ✅
- Fetches quoted message content via `getMessageFeishu` → **only extracts text, not media** ❌
### Example sc... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T02:39:26Z | 2026-04-22T04:31:40Z | 2026-04-19T04:24:36Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | talywy | 1,817,470 | MDQ6VXNlcjE4MTc0NzA= | User | false |
openclaw/openclaw | 4,019,711,776 | I_kwDOQb6kR87vl-8g | 33,803 | https://github.com/openclaw/openclaw/issues/33803 | https://api.github.com/repos/openclaw/openclaw/issues/33803 | [Feature]: heartbeat: add jitter/startDelay to prevent simultaneous triggers across agents | ## Problem
When running multiple agents with the same `heartbeat.every` interval, all agents fire simultaneously after gateway restart. This causes request queue buildup on the main lane since all heartbeat turns compete for the same serialized lane.
**Observed impact:** 7 agents with `every: "120m"` caused lane wait... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T02:42:01Z | 2026-04-22T04:31:39Z | 2026-04-19T04:24:38Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | zhangsensen | 26,587,484 | MDQ6VXNlcjI2NTg3NDg0 | User | false |
openclaw/openclaw | 4,019,718,829 | I_kwDOQb6kR87vmAqt | 33,807 | https://github.com/openclaw/openclaw/issues/33807 | https://api.github.com/repos/openclaw/openclaw/issues/33807 | Chrome extension relay auth doesn't support SSH-tunneled/remote relays | ## Summary
The browser extension relay authentication architecture assumes the relay always runs **locally on the same machine** as the gateway. When accessing a relay through an SSH reverse tunnel (different port), three places fail:
## Environment
- OpenClaw v2026.3.2
- Gateway on GCP VM (Linux), Chrome extension ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T02:45:00Z | 2026-04-22T04:31:38Z | 2026-04-19T04:24:39Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | puneet1409 | 15,050,377 | MDQ6VXNlcjE1MDUwMzc3 | User | false |
openclaw/openclaw | 4,019,719,578 | I_kwDOQb6kR87vmA2a | 33,809 | https://github.com/openclaw/openclaw/issues/33809 | https://api.github.com/repos/openclaw/openclaw/issues/33809 | RFC: Self-Healing Tool Router — Deterministic Fallback Routing for Tool Failures | # RFC: Self-Healing Tool Router
## Summary
Add a deterministic fallback routing layer to OpenClaw's tool execution pipeline. When a tool fails mid-execution, the router checks a fallback table before returning the error to the LLM. Predictable failures (timeouts, network errors, permission issues) get deterministic r... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T02:45:22Z | 2026-04-22T04:31:38Z | 2026-04-19T04:24:41Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | nxrobins | 11,342,198 | MDQ6VXNlcjExMzQyMTk4 | User | false |
openclaw/openclaw | 4,019,720,271 | I_kwDOQb6kR87vmBBP | 33,811 | https://github.com/openclaw/openclaw/issues/33811 | https://api.github.com/repos/openclaw/openclaw/issues/33811 | Bundler emits duplicate chrome module pair — patches/debugging target wrong file | ## Summary
The dist output contains **two separate copies** of the chrome relay module with different content hashes. Only one is loaded at runtime, but both have valid import chains referencing them. This makes runtime patching and debugging extremely confusing — code changes to the wrong copy silently have no effect... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T02:45:41Z | 2026-04-22T04:31:37Z | 2026-04-19T04:24:43Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | puneet1409 | 15,050,377 | MDQ6VXNlcjE1MDUwMzc3 | User | false |
openclaw/openclaw | 4,019,726,645 | I_kwDOQb6kR87vmCk1 | 33,815 | https://github.com/openclaw/openclaw/issues/33815 | https://api.github.com/repos/openclaw/openclaw/issues/33815 | [Feature]: Sub-agent completion push notification to originating channel | ## Problem
When a user triggers a sub-agent task via Discord/Telegram (e.g., "analyze this codebase"), the assistant says "I'll notify you when done." But the main session is **passive** — it only activates on incoming messages. So the user always has to ask "is it done yet?" instead of being notified.
## Current Beh... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T02:48:41Z | 2026-04-22T04:31:36Z | 2026-04-19T04:24:45Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | imwyvern | 1,765,672 | MDQ6VXNlcjE3NjU2NzI= | User | false |
openclaw/openclaw | 4,019,744,707 | I_kwDOQb6kR87vmG_D | 33,827 | https://github.com/openclaw/openclaw/issues/33827 | https://api.github.com/repos/openclaw/openclaw/issues/33827 | [Bug]: Sub-agent timeout returns minimal error — no partial results or diagnostic info | ## Bug Description
When a sub-agent spawned via `sessions_spawn` with `mode=run` exceeds its `runTimeoutSeconds`, the system returns only a minimal "timed out" message. No partial results, no diagnostic information, and no actionable guidance are provided.
## Steps to Reproduce
1. Spawn a sub-agent with a complex an... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T02:57:15Z | 2026-04-22T04:31:35Z | 2026-04-19T04:24:46Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | imwyvern | 1,765,672 | MDQ6VXNlcjE3NjU2NzI= | User | false |
openclaw/openclaw | 4,019,744,854 | I_kwDOQb6kR87vmHBW | 33,828 | https://github.com/openclaw/openclaw/issues/33828 | https://api.github.com/repos/openclaw/openclaw/issues/33828 | [Feature]: Allow heartbeat results to push notifications to configured channels | ## Feature Description
Currently, heartbeat check results can only be returned as `HEARTBEAT_OK` or a text reply within the session. There is no mechanism to route heartbeat results to external notification channels (Discord, Telegram, Slack, etc.) based on the check outcome.
## Proposal
Allow `HEARTBEAT.md` to spec... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T02:57:18Z | 2026-04-22T04:31:35Z | 2026-04-19T04:24:48Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | imwyvern | 1,765,672 | MDQ6VXNlcjE3NjU2NzI= | User | false |
openclaw/openclaw | 4,019,781,350 | I_kwDOQb6kR87vmP7m | 33,837 | https://github.com/openclaw/openclaw/issues/33837 | https://api.github.com/repos/openclaw/openclaw/issues/33837 | [Feature]: A2UI v0.9 message support | ## Summary
Add support for A2UI v0.9 messages alongside existing v0.8 support.
## Problem to solve
OpenClaw's A2UI integration is pinned to v0.8 and explicitly rejects v0.9 messages — the Android handler returns an error on `createSurface`, and the JSONL validator whitelists only v0.8 action keys.
The A2UI v0.9 spec... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T03:12:44Z | 2026-04-22T04:31:34Z | 2026-04-19T04:24:50Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | zeroasterisk | 23,422 | MDQ6VXNlcjIzNDIy | User | false |
openclaw/openclaw | 4,019,793,153 | I_kwDOQb6kR87vmS0B | 33,841 | https://github.com/openclaw/openclaw/issues/33841 | https://api.github.com/repos/openclaw/openclaw/issues/33841 | Signal probe false-negative: status reports fetch failed while JSON-RPC daemon is healthy | ## Summary
`openclaw channels status --probe` can report Signal as failed (`fetch failed`) even when signal-cli daemon is alive and sending/receiving messages successfully.
## Environment
- OpenClaw: 2026.3.1
- signal-cli: 0.13.24 (daemon mode, `--http 127.0.0.1:8080`)
- Channel transport in use: JSON-RPC (`/api/v1/rp... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T03:18:02Z | 2026-04-22T04:31:33Z | 2026-04-19T04:24:51Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | bobsummerwill-openclaw-bot[bot] | 262,731,058 | BOT_kgDOD6j1Mg | Bot | false |
openclaw/openclaw | 4,019,802,794 | I_kwDOQb6kR87vmVKq | 33,843 | https://github.com/openclaw/openclaw/issues/33843 | https://api.github.com/repos/openclaw/openclaw/issues/33843 | Bug: sessions_spawn with runtime="acp" uses invalid acpx CLI syntax | ## Bug Description
When using `sessions_spawn` with `runtime="acp"` and `agentId="openai-codex"`, OpenClaw constructs an invalid acpx command line:
**Current (broken):**
```
acpx --agent openai-codex --task "prompt..."
```
**Expected (correct per acpx 0.1.15):**
```
acpx codex "prompt..."
# or for one-shot:
acpx cod... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T03:22:19Z | 2026-04-22T04:31:32Z | 2026-04-19T04:24:53Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | ambiguouStr | 59,325,440 | MDQ6VXNlcjU5MzI1NDQw | User | false |
openclaw/openclaw | 4,019,828,427 | I_kwDOQb6kR87vmbbL | 33,854 | https://github.com/openclaw/openclaw/issues/33854 | https://api.github.com/repos/openclaw/openclaw/issues/33854 | [Bug]: [REGRESSION] Intermittent Message Delivery Failure in Telegram after v2026.3.3 | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to v2026.3.3, agent replies in Telegram group topics are intermittently not delivered to the Telegram client, even though the agent turn completes successfully and the response is visible in the OpenClaw Web UI.
### Steps to reproduce
... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T03:33:29Z | 2026-04-22T04:31:32Z | 2026-04-19T04:24:55Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | namrop | 2,350,989 | MDQ6VXNlcjIzNTA5ODk= | User | false |
openclaw/openclaw | 4,019,835,028 | I_kwDOQb6kR87vmdCU | 33,856 | https://github.com/openclaw/openclaw/issues/33856 | https://api.github.com/repos/openclaw/openclaw/issues/33856 | [Bug]: Bug Report: Streaming Message Disappearance Due to Pinned Main DM Recipient Conflict | ### Bug type
Regression (worked before, now fails)
### Summary
Telegram streaming messages disappear when pinned main DM recipient mechanism is active
### Steps to reproduce
1. Configure Telegram with streaming enabled:
```json
"telegram": {
"streaming": "progress",
"dmPolicy": "open",
"allowFrom": ["*"]
}
... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T03:36:09Z | 2026-04-22T04:31:31Z | 2026-04-19T04:24:57Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | wanqianwin-jpg | 254,811,378 | U_kgDODzAc8g | User | false |
openclaw/openclaw | 4,019,848,653 | I_kwDOQb6kR87vmgXN | 33,862 | https://github.com/openclaw/openclaw/issues/33862 | https://api.github.com/repos/openclaw/openclaw/issues/33862 | [Bug]: Feishu groupPolicy allowlist not working on Windows - works on Linux | ## Bug Description
Feishu group messages are rejected with `group not in groupAllowFrom` even when `groupPolicy: allow` is configured.
## Environment
- **OS**: Windows 11
- **OpenClaw version**: 2026.2.26
- **Node version**: v22.22.0
- **Same config works on**: Ubuntu (Alibaba Cloud)
## Configuration
```json
{
"... | closed | not_planned | true | 8 | [
"stale"
] | [] | 2026-03-04T03:41:22Z | 2026-04-22T04:31:30Z | 2026-04-19T04:24:58Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | zng8418 | 162,404,177 | U_kgDOCa4XUQ | User | false |
openclaw/openclaw | 4,019,854,613 | I_kwDOQb6kR87vmh0V | 33,864 | https://github.com/openclaw/openclaw/issues/33864 | https://api.github.com/repos/openclaw/openclaw/issues/33864 | Bug: User messages silently deleted when session fails due to provider error | ## Bug Description
When an agent session fails due to a provider initialization error (e.g., missing API key for a model in the failover chain), subsequent user messages in the same channel can be silently deleted from the session context without any notification to the user.
## Steps to Reproduce
1. A session is co... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T03:43:30Z | 2026-04-22T04:31:29Z | 2026-04-19T04:25:00Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | p3nchan | 5,032,148 | MDQ6VXNlcjUwMzIxNDg= | User | false |
openclaw/openclaw | 4,019,874,580 | I_kwDOQb6kR87vmmsU | 33,875 | https://github.com/openclaw/openclaw/issues/33875 | https://api.github.com/repos/openclaw/openclaw/issues/33875 | [Bug]: Feishu all accounts replies=0 in v2026.3.2 — send layer fails silently (regression from v2026.3.1) | ## Summary
After upgrading from v2026.3.1 to v2026.3.2, all Feishu bot accounts fail to send replies. Messages are received and processed by agents, but outbound delivery shows `replies=0` with no error logs. Rolling back to v2026.3.1 immediately resolves the issue.
## Environment
- OpenClaw: v2026.3.2 (regression f... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T03:52:13Z | 2026-04-22T04:31:29Z | 2026-04-19T04:25:02Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | alansong63 | 129,918,616 | U_kgDOB75mmA | User | false |
openclaw/openclaw | 4,019,879,554 | I_kwDOQb6kR87vmn6C | 33,877 | https://github.com/openclaw/openclaw/issues/33877 | https://api.github.com/repos/openclaw/openclaw/issues/33877 | 🚨 Critical: Default sandbox configuration causes agent startup failure | # 🚨 Critical: Default sandbox configuration causes agent startup failure
## Problem
After a fresh OpenClaw installation or update, **all agents fail to start** with the following error:
```
⚠️ Agent failed before reply: Failed to inspect sandbox image:
Emulate Docker CLI using podman. Create /etc/containers/nodock... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T03:54:36Z | 2026-04-22T04:31:28Z | 2026-04-19T04:25:03Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | 2233admin | 57,929,895 | MDQ6VXNlcjU3OTI5ODk1 | User | false |
openclaw/openclaw | 4,019,883,022 | I_kwDOQb6kR87vmowO | 33,878 | https://github.com/openclaw/openclaw/issues/33878 | https://api.github.com/repos/openclaw/openclaw/issues/33878 | [Bug] Browser tool start times out on ARM64 Ubuntu despite browser control service ready | ## Summary
On ARM64 Ubuntu (Oracle Cloud), OpenClaw browser tool repeatedly fails with:
`Can't reach the OpenClaw browser control service (timed out after 15000ms)`
Even though the browser control service is listening and reports ready.
## Environment
- OpenClaw: `2026.3.2` (stable)
- OS: Ubuntu Linux `6.14.0-1018-o... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T03:56:02Z | 2026-04-22T04:31:27Z | 2026-04-19T04:25:05Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | lalalasyun | 90,920,222 | MDQ6VXNlcjkwOTIwMjIy | User | false |
openclaw/openclaw | 4,019,887,327 | I_kwDOQb6kR87vmpzf | 33,879 | https://github.com/openclaw/openclaw/issues/33879 | https://api.github.com/repos/openclaw/openclaw/issues/33879 | [Bug]: Docker installation, a TypeScript error is reported during building | ### Bug type
Regression (worked before, now fails)
### Summary
When I was executing ./docker-setup.sh from the root directory, a type-checking error occurred during the final build step. The error content:
> openclaw@2026.3.2 build:plugin-sdk:dts /app
> tsc -p tsconfig.plugin-sdk.dts.json
src/agents/pi-tools.read.t... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T03:57:54Z | 2026-04-22T04:31:26Z | 2026-04-19T04:25:07Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | ooweigan | 129,486,757 | U_kgDOB7fPpQ | User | false |
openclaw/openclaw | 4,019,888,238 | I_kwDOQb6kR87vmqBu | 33,881 | https://github.com/openclaw/openclaw/issues/33881 | https://api.github.com/repos/openclaw/openclaw/issues/33881 | [Proposal] Incremental reliability loop: failure → fix → verification trace | Hi OpenClaw team — I’m using OpenClaw daily (cron + Telegram + Feishu + custom skills), and I’ve seen recurring operational issues (timeouts, post-update pairing changes, config drift).
I’d like to propose a small, incremental reliability improvement:
## Problem
Failures are visible, but the recovery path is often ad... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T03:58:19Z | 2026-04-22T04:31:26Z | 2026-04-19T04:25:09Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Yuzc-001 | 199,756,267 | U_kgDOC-gJ6w | User | false |
openclaw/openclaw | 4,019,890,642 | I_kwDOQb6kR87vmqnS | 33,883 | https://github.com/openclaw/openclaw/issues/33883 | https://api.github.com/repos/openclaw/openclaw/issues/33883 | [Bug]: Feishu blockStreaming causes 30+ duplicate message display on mobile client | ## Bug Report
### Summary
When `blockStreaming: true` is enabled in the Feishu channel config, a single agent reply is displayed **30+ times** on the Feishu **mobile** client. The OpenClaw TUI and desktop Feishu show the message correctly (once). Disabling `blockStreaming` immediately fixes the issue.
### Steps to R... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-03-04T03:59:27Z | 2026-04-22T04:31:25Z | 2026-04-19T04:25:11Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | wgx112358 | 97,893,937 | U_kgDOBdW-MQ | User | false |
openclaw/openclaw | 4,019,905,614 | I_kwDOQb6kR87vmuRO | 33,892 | https://github.com/openclaw/openclaw/issues/33892 | https://api.github.com/repos/openclaw/openclaw/issues/33892 | llm-task plugin: Cannot find module pi-embedded-runner.js in npm build (3.2) | ## Description
The `llm-task` plugin fails on OpenClaw 2026.3.2 (npm install) with:
```
Cannot find module '../../../src/agents/pi-embedded-runner.js'
Require stack:
- /opt/homebrew/lib/node_modules/openclaw/extensions/llm-task/src/llm-task-tool.ts
```
## Root Cause
`llm-task-tool.ts` line 17 and 28 both import fro... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T04:05:41Z | 2026-04-22T04:31:24Z | 2026-04-19T04:25:12Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | sbonafe | 60,677,766 | MDQ6VXNlcjYwNjc3NzY2 | User | false |
openclaw/openclaw | 4,019,919,214 | I_kwDOQb6kR87vmxlu | 33,902 | https://github.com/openclaw/openclaw/issues/33902 | https://api.github.com/repos/openclaw/openclaw/issues/33902 | Multi-profile gateways share the same log file — no per-profile log isolation | ## Summary
When running multiple OpenClaw profiles on the same host (e.g. `--profile xv` alongside the default profile), both gateway instances write to the same log file `/tmp/openclaw/openclaw-YYYY-MM-DD.log`. This makes logs extremely difficult to read — entries from both gateways are interleaved with no way to dis... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T04:11:01Z | 2026-04-22T04:31:23Z | 2026-04-19T04:25:14Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | SiyiLi | 6,840,081 | MDQ6VXNlcjY4NDAwODE= | User | false |
openclaw/openclaw | 4,019,929,236 | I_kwDOQb6kR87vm0CU | 33,906 | https://github.com/openclaw/openclaw/issues/33906 | https://api.github.com/repos/openclaw/openclaw/issues/33906 | DiscordMessageListener blocks for 120s causing silent event drop | ## Bug Description
When a Discord message arrives while `DiscordMessageListener` is blocked processing a previous event, the incoming message is silently dropped — no session is created, no error is surfaced to the user.
## Steps to Reproduce
1. Trigger a sequence of API overload / rate limit errors causing a sessio... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T04:15:13Z | 2026-04-22T04:31:23Z | 2026-04-19T04:25:16Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | p3nchan | 5,032,148 | MDQ6VXNlcjUwMzIxNDg= | User | false |
openclaw/openclaw | 4,019,967,274 | I_kwDOQb6kR87vm9Uq | 33,926 | https://github.com/openclaw/openclaw/issues/33926 | https://api.github.com/repos/openclaw/openclaw/issues/33926 | RFC: Skills placeholders + context: fork execution | ## Context
We just added two related Skills UX features:
1) **Dynamic placeholders** in SKILL.md when invoked via `/skill …` or `/<skill> …`:
- `{{CWD}}`, `{{ARGS}}`, `{{SELECTION}}`
- defaults `{{KEY|default}}`
- conditionals `{{#if KEY}} … {{/if}}`
2) **Fork execution** via SKILL.md frontmatter:
```yaml
co... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T04:31:30Z | 2026-04-22T04:31:22Z | 2026-04-19T04:25:18Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | zhangjunmengyang | 49,865,968 | MDQ6VXNlcjQ5ODY1OTY4 | User | false |
openclaw/openclaw | 4,020,000,084 | I_kwDOQb6kR87vnFVU | 33,940 | https://github.com/openclaw/openclaw/issues/33940 | https://api.github.com/repos/openclaw/openclaw/issues/33940 | [Bug]: Executing a Daily Cron Task Manually Changes the Timing of the Execution of the Cron Task | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
If you execute a daily cron task manually, it will change the timing of the cron task.
### Steps to reproduce
Suppose you create a cron task that gives you a morning affirmation every day at 7am.
```
ID ... | closed | not_planned | true | 5 | [
"bug",
"stale"
] | [] | 2026-03-04T04:43:50Z | 2026-04-22T04:31:21Z | 2026-04-19T04:25:19Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | arjunaskykok | 32,124,593 | MDQ6VXNlcjMyMTI0NTkz | User | false |
openclaw/openclaw | 4,020,008,346 | I_kwDOQb6kR87vnHWa | 33,943 | https://github.com/openclaw/openclaw/issues/33943 | https://api.github.com/repos/openclaw/openclaw/issues/33943 | [Bug] Feishu channel: Commands work but regular messages get no response | # [Bug] Feishu channel: Commands work but regular messages get no response
## Description
Feishu bot receives messages and processes commands (like `/status`) successfully, but regular text messages produce no response. Messages appear to enter the main session but replies are not sent back to Feishu.
## Environment
... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T04:47:01Z | 2026-04-22T04:31:20Z | 2026-04-19T04:25:21Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | ifdog | 1,519,790 | MDQ6VXNlcjE1MTk3OTA= | User | false |
openclaw/openclaw | 4,020,009,313 | I_kwDOQb6kR87vnHlh | 33,944 | https://github.com/openclaw/openclaw/issues/33944 | https://api.github.com/repos/openclaw/openclaw/issues/33944 | [Feature]: Add model whitelist to filter sidebar model picker | ---
name: Feature request
about: Propose a new capability or product improvement.
title: "[Feature]: Add model whitelist to filter sidebar model picker"
labels: enhancement
---
## Summary
Add a configuration option to filter which models appear in the sidebar model picker dropdown, allowing users to show only their con... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T04:47:24Z | 2026-04-22T04:31:20Z | 2026-04-19T04:25:23Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Ethan-Pier | 249,102,092 | U_kgDODtj_DA | User | false |
openclaw/openclaw | 4,021,010,842 | I_kwDOQb6kR87vq8Ga | 34,241 | https://github.com/openclaw/openclaw/issues/34241 | https://api.github.com/repos/openclaw/openclaw/issues/34241 | [Bug]: Intermittent no-tool execution after switching to azure-openai-responses/gpt-5.3-codex (assistant says it will act but emits no tool call) | ### Bug type
Regression (worked before, now fails)
### Summary
After switching from `azure-openai/gpt-5.2` to `azure-openai-responses/gpt-5.3-codex`, the agent intermittently says it will execute a task but returns plain text without issuing any tool call.
### Steps to reproduce
1. Use OpenClaw with `azure-openai-... | closed | not_planned | true | 7 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T09:31:48Z | 2026-04-22T04:31:19Z | 2026-04-19T04:25:25Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | dustookk | 1,394,913 | MDQ6VXNlcjEzOTQ5MTM= | User | false |
openclaw/openclaw | 4,021,015,233 | I_kwDOQb6kR87vq9LB | 34,243 | https://github.com/openclaw/openclaw/issues/34243 | https://api.github.com/repos/openclaw/openclaw/issues/34243 | [Bug]: Cannot access bundled skills in sandbox, path escape error occurs | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
- 在 Sandbox 环境中运行的 Agent 无法加载 OpenClaw 预装的默认 skills(例如：`skill-creator`)，尝试读取技能文件时出现 "Path escapes sandbox root" 错误。
- Agents running in Sandbox environment cannot load OpenClaw's pre-installed default skills( such as: `skill-creator`), a... | closed | not_planned | true | 4 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-04T09:32:49Z | 2026-04-22T04:31:18Z | 2026-04-19T04:25:27Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Spirit597 | 38,465,227 | MDQ6VXNlcjM4NDY1MjI3 | User | false |
openclaw/openclaw | 4,021,032,665 | I_kwDOQb6kR87vrBbZ | 34,247 | https://github.com/openclaw/openclaw/issues/34247 | https://api.github.com/repos/openclaw/openclaw/issues/34247 | [Bug]: Telegram channel fails with 404 - $VARIABLE references in botToken not resolved (v2026.3.2) | ## Description
Telegram channel fails to start with repeated `404: Not Found` errors on all Bot API methods (`deleteWebhook`, `setMyCommands`, `deleteMyCommands`, `sendMessage`). The channel enters a retry loop (up to 10/10) and never recovers.
## Root Cause
When `channels.telegram.botToken` is set to a `$VARIABLE` ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T09:36:58Z | 2026-04-22T04:31:17Z | 2026-04-19T04:25:29Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | FmBlueSystem | 185,002,238 | U_kgDOCwbo_g | User | false |
openclaw/openclaw | 4,021,036,864 | I_kwDOQb6kR87vrCdA | 34,249 | https://github.com/openclaw/openclaw/issues/34249 | https://api.github.com/repos/openclaw/openclaw/issues/34249 | [Question]: Why hasn't boot been configured after creating a new agent using the command line? | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="1848" height="567" alt="Image" src="https://github.com/user-attachments/assets/b852eeac-381e-4162-92aa-1c7b409b5ba3" />
### Steps to reproduce
openclaw agents add test-agent --workspace ~/.openclaw/workspace-test-agent
### Expected behavio... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T09:37:58Z | 2026-04-22T04:31:16Z | 2026-04-19T04:25:31Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | wangGit-W | 102,007,391 | U_kgDOBhSCXw | User | false |
openclaw/openclaw | 4,021,051,893 | I_kwDOQb6kR87vrGH1 | 34,253 | https://github.com/openclaw/openclaw/issues/34253 | https://api.github.com/repos/openclaw/openclaw/issues/34253 | [Bug] Feishu Multi-Agent Routing Failure - Binding with peer.match not working | # Bug Report: Feishu Multi-Agent Routing Failure
## Summary
Feishu channel's `bindings` with `peer` matching fails to route messages to the correct agent. Messages from configured users are incorrectly rejected with "not in allowlist" error, even when the user ID is present in the allowlist.
## Environment
- **OpenCl... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T09:41:26Z | 2026-04-22T04:31:16Z | 2026-04-19T04:25:32Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | shaozhengkun123 | 224,505,443 | U_kgDODWGuYw | User | false |
openclaw/openclaw | 4,021,067,618 | I_kwDOQb6kR87vrJ9i | 34,255 | https://github.com/openclaw/openclaw/issues/34255 | https://api.github.com/repos/openclaw/openclaw/issues/34255 | Control UI Chat: allow pinning a default session (avoid auto-selecting agent:guest:main) | ### Problem
When using multi-agent routing with a restricted **default agent** (e.g. `guest`) to handle unknown WhatsApp DMs, the Control UI **/chat** session selector shows `agent:guest:main` as the first option. Clicking the left sidebar **Chat** entry (navigates to `/chat`) consistently lands the UI in the first ses... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T09:45:14Z | 2026-04-22T04:31:15Z | 2026-04-19T04:25:34Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | jeffrey4341 | 77,844,634 | MDQ6VXNlcjc3ODQ0NjM0 | User | false |
openclaw/openclaw | 4,289,915,466 | I_kwDOQb6kR87_supK | 68,818 | https://github.com/openclaw/openclaw/issues/68818 | https://api.github.com/repos/openclaw/openclaw/issues/68818 | [Bug] MCP diagram-generator tools not exposed | ## Bug Description
MCP diagram-generator tools are not appearing in available tools list despite correct configuration.
## Steps to Reproduce
1. Install diagram-generator skill from ClawHub
2. Configure MCP server: `openclaw mcp set mcp-diagram-generator '{"command":"npx","args":["-y","mcp-diagram-generator"]}'`
3. Re... | closed | completed | true | 1 | [] | [] | 2026-04-19T04:30:43Z | 2026-04-25T04:16:07Z | 2026-04-22T23:48:01Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | mexonzzh | 274,827,405 | U_kgDOEGGIjQ | User | false |
openclaw/openclaw | 4,032,100,104 | I_kwDOQb6kR87wVPcI | 37,148 | https://github.com/openclaw/openclaw/issues/37148 | https://api.github.com/repos/openclaw/openclaw/issues/37148 | 支持 24 小时时间制显示 | ## 问题描述
当前 OpenClaw 在显示时间时使用的是 12 小时制（AM/PM），这在某些场景下容易造成混淆，尤其是对于习惯使用 24 小时制的用户。
## 建议方案
1. **默认使用 24 小时制**：更清晰，避免 AM/PM 混淆
2. **可配置**：允许用户在配置文件中选择时间格式
- 24 小时制（默认）
- 12 小时制（AM/PM）
## 影响范围
- 心跳提醒时间显示
- 定时任务时间显示
- 日志时间戳
- 其他需要显示时间的地方
## 参考
用户反馈：12 小时制在跨时区或快速阅读时容易产生误解，尤其是在中文环境下 24 小时制更为普遍。
---
**作者**: @ITHA... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T04:07:01Z | 2026-04-27T04:33:33Z | 2026-04-25T04:15:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ITHACAJASON | 199,630,942 | U_kgDOC-YgXg | User | false |
openclaw/openclaw | 4,032,101,542 | I_kwDOQb6kR87wVPym | 37,150 | https://github.com/openclaw/openclaw/issues/37150 | https://api.github.com/repos/openclaw/openclaw/issues/37150 | Feature Request: Display current model in TUI status bar/header | ## Summary
It would be helpful to display the currently connected model in the OpenClaw TUI interface (status bar, header, or similar location).
## Use Case
When using or switching models via , the model is shown in the output. However, users would like to see the current model displayed persistently in the TUI its... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T04:07:33Z | 2026-04-27T05:02:56Z | 2026-04-24T19:25:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | joemcurry | 1,192,832 | MDQ6VXNlcjExOTI4MzI= | User | false |
openclaw/openclaw | 4,032,131,515 | I_kwDOQb6kR87wVXG7 | 37,172 | https://github.com/openclaw/openclaw/issues/37172 | https://api.github.com/repos/openclaw/openclaw/issues/37172 | Feature: Separate context visibility from response triggering in group chats (allowFrom drops pending history) | ## Summary
In group chats, `allowFrom` / `groupAllowFrom` drops messages from non-listed senders entirely — they never enter the pending history buffer. This makes it impossible to configure "bot sees all messages for context, but only responds to specific users."
## Problem
When running a bot in a Telegram group (a... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T04:19:15Z | 2026-04-27T04:33:31Z | 2026-04-25T04:15:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mmto-io | 182,759,316 | U_kgDOCuSvlA | User | false |
openclaw/openclaw | 4,032,156,738 | I_kwDOQb6kR87wVdRC | 37,177 | https://github.com/openclaw/openclaw/issues/37177 | https://api.github.com/repos/openclaw/openclaw/issues/37177 | [SEC] OPENCLAW_ALLOW_INSECURE_PRIVATE_WS currently permits any ws:// hostname (not only private/loopback) | ## Summary
When OPENCLAW_ALLOW_INSECURE_PRIVATE_WS=1 is enabled, current URL validation accepts any ws:// hostname, not only private/loopback targets.
## Why this matters
The flag is positioned as break-glass for trusted private-network overlays. Current behavior effectively disables host safety for non-TLS websocket ... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-06T04:29:25Z | 2026-04-27T04:33:31Z | 2026-04-25T04:15:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sxu75374 | 72,003,735 | MDQ6VXNlcjcyMDAzNzM1 | User | false |
openclaw/openclaw | 4,032,221,661 | I_kwDOQb6kR87wVtHd | 37,198 | https://github.com/openclaw/openclaw/issues/37198 | https://api.github.com/repos/openclaw/openclaw/issues/37198 | [Bug]: Cron job with "gateway restart" in message causes infinite Gateway restart loop when manually triggered after scheduled time | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Cron job with "gateway restart" in message causes infinite Gateway restart loop when manually triggered after scheduled time
### Steps to reproduce
- 1. Create a one-shot cron job that restarts the gateway itself:
```bash
openclaw cr... | closed | completed | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T04:57:06Z | 2026-04-27T05:02:55Z | 2026-04-24T19:25:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Narlea | 163,425,703 | U_kgDOCb2tpw | User | false |
openclaw/openclaw | 4,032,270,348 | I_kwDOQb6kR87wV5AM | 37,220 | https://github.com/openclaw/openclaw/issues/37220 | https://api.github.com/repos/openclaw/openclaw/issues/37220 | Telegram send: guard non-numeric replyToMessageId to avoid 400 reply-not-found | ## Summary
Telegram sends can fail with `400: Bad Request: message to be replied not found` when a non-numeric `replyToMessageId` (for example a UUID from cross-surface reply tags) is passed into Telegram send helpers.
## Impact
- Affected path can happen in cross-surface/cross-channel flows (e.g. TUI/WebChat session ... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-06T05:15:18Z | 2026-04-27T05:02:55Z | 2026-04-24T19:25:09Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | bread22 | 17,005,481 | MDQ6VXNlcjE3MDA1NDgx | User | false |
openclaw/openclaw | 4,032,282,890 | I_kwDOQb6kR87wV8EK | 37,225 | https://github.com/openclaw/openclaw/issues/37225 | https://api.github.com/repos/openclaw/openclaw/issues/37225 | Feature Request: Preserve sessionKey in archived .reset.* JSONL files after /new | ## Problem
When a user triggers `/new` (or `/reset`), OpenClaw:
1. Renames the current session JSONL file to `<sessionId>.jsonl.reset.<timestamp>`
2. Updates `sessions.json` to point to a new `sessionId`
The old session record in `sessions.json` is **overwritten** — the new `sessionId` replaces the old one. The archi... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T05:19:09Z | 2026-04-27T05:02:54Z | 2026-04-24T19:25:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yulin7645 | 253,032,811 | U_kgDODxT5aw | User | false |
openclaw/openclaw | 4,032,289,813 | I_kwDOQb6kR87wV9wV | 37,230 | https://github.com/openclaw/openclaw/issues/37230 | https://api.github.com/repos/openclaw/openclaw/issues/37230 | OpenAI Responses WS path can hang on upstream 502/SSE close and fails hard on stale rs_* (store=false) refs after /new | ## Summary
In OpenClaw runtime (OpenAI Responses/WebSocket path), we observed two failure patterns that can lead to long perceived hangs or unrecoverable errors:
1. 404 `Item with id 'rs_*' not found` after `/new` (stale response item reference from previous context while `store=false`).
2. Upstream transient failure... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T05:21:07Z | 2026-04-27T05:02:53Z | 2026-04-24T19:25:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | llw2011 | 258,054,251 | U_kgDOD2GYaw | User | false |
openclaw/openclaw | 4,032,304,015 | I_kwDOQb6kR87wWBOP | 37,232 | https://github.com/openclaw/openclaw/issues/37232 | https://api.github.com/repos/openclaw/openclaw/issues/37232 | TTL-aware status layer: prevent stale facts in USER.md/MEMORY.md from being treated as current truth | Title: TTL-aware status layer: prevent stale facts in USER.md/MEMORY.md from being treated as current truth
Problem
OpenClaw encourages storing user context in workspace files (e.g. USER.md, MEMORY.md). These files are timeless (flat Markdown). In real usage, they often contain time-sensitive state like “departs tomor... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T05:24:57Z | 2026-04-27T04:33:30Z | 2026-04-25T04:15:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | LindalyX-Lee | 256,247,180 | U_kgDOD0YFjA | User | false |
openclaw/openclaw | 4,032,309,669 | I_kwDOQb6kR87wWCml | 37,235 | https://github.com/openclaw/openclaw/issues/37235 | https://api.github.com/repos/openclaw/openclaw/issues/37235 | [Bug] Media download returns cached/wrong image - all images have same MD5 | ## Description
When receiving images through any channel (Feishu, WebChat), the downloaded images are incorrect. All downloaded images have the same MD5 hash, regardless of the actual image sent.
## Reproduction
1. Send image A via Feishu/WebChat
2. Send image B via Feishu/WebChat
3. Check downloaded files in `~/.open... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T05:26:30Z | 2026-04-27T05:02:52Z | 2026-04-24T19:25:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | xiaodaihaoguai | 26,459,744 | MDQ6VXNlcjI2NDU5NzQ0 | User | false |
openclaw/openclaw | 4,032,324,566 | I_kwDOQb6kR87wWGPW | 37,238 | https://github.com/openclaw/openclaw/issues/37238 | https://api.github.com/repos/openclaw/openclaw/issues/37238 | Dashboard dropdown shows all historical subagent sessions (completed/killed) with no cleanup option | ## Bug Description
The OpenClaw Control UI dashboard dropdown displays **all session records** including completed, killed, and timed-out subagent sessions. These persist indefinitely even after the subagents are no longer running.
## Current Behavior
- `sessions_list` API returns all sessions from the transcript st... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T05:30:52Z | 2026-04-27T04:33:29Z | 2026-04-25T04:15:29Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mizoz | 56,456,861 | MDQ6VXNlcjU2NDU2ODYx | User | false |
openclaw/openclaw | 4,032,331,859 | I_kwDOQb6kR87wWIBT | 37,239 | https://github.com/openclaw/openclaw/issues/37239 | https://api.github.com/repos/openclaw/openclaw/issues/37239 | Internal reminder note leaks into chat output | When no reminder/cron is scheduled in a turn, the system appends:
> Note: I did not schedule a reminder in this turn, so this will not trigger automatically.
This internal annotation is leaking into user-visible chat messages on Discord (and likely other surfaces). It should be stripped before delivery.
**Repro**: A... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T05:32:57Z | 2026-04-27T04:33:29Z | 2026-04-25T04:15:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Defying | 2,173,281 | MDQ6VXNlcjIxNzMyODE= | User | false |
openclaw/openclaw | 4,032,401,767 | I_kwDOQb6kR87wWZFn | 37,250 | https://github.com/openclaw/openclaw/issues/37250 | https://api.github.com/repos/openclaw/openclaw/issues/37250 | [Feature]: Sub-agent partial result persistence on timeout | When a sub-agent times out, all output is lost. No partial results, no intermediate log. Especially painful for long-running SSH/build tasks.
Expected: Incremental persistence + sessions_history should still return completed tool calls. | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T05:55:16Z | 2026-04-27T05:02:52Z | 2026-04-24T19:25:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | biohazardxxx | 1,370,362 | MDQ6VXNlcjEzNzAzNjI= | User | false |
openclaw/openclaw | 4,032,402,621 | I_kwDOQb6kR87wWZS9 | 37,252 | https://github.com/openclaw/openclaw/issues/37252 | https://api.github.com/repos/openclaw/openclaw/issues/37252 | Telegram outbound: enforce strict numeric parsing for reply_to_message_id | ## Summary
Telegram outbound reply threading currently accepts partially numeric `replyTo` values (for example, `"123abc"`) because parsing uses permissive integer conversion.
## Problem
When invalid `replyTo` values are partially parsed, OpenClaw may send an unintended `reply_to_message_id`, which can lead to Telegra... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-06T05:55:32Z | 2026-04-27T05:02:51Z | 2026-04-24T19:25:30Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | shuofengzhang | 24,763,026 | MDQ6VXNlcjI0NzYzMDI2 | User | false |
openclaw/openclaw | 4,032,428,372 | I_kwDOQb6kR87wWflU | 37,256 | https://github.com/openclaw/openclaw/issues/37256 | https://api.github.com/repos/openclaw/openclaw/issues/37256 | [PERF] notification-service: /api/v1/alim-talks/matching p95=7.25s | ## 🐢 성능 병목 요약
- **서비스**: notification-service
- **엔드포인트**: POST /api/v1/alim-talks/matching
- **환경**: PROD
- **P95 Latency**: 7.25s
- **요청량**: 0.05 req/s
## 📊 메트릭
- P95: 7.25s
## 🔍 근본 원인 분석
### Tempo 트레이스
- POST /api/v1/alim-talks/matching 의 전체 소요 시간은 약 6.9s.
- DB 조회 등은 수십 밀리초 내로 매우 빠름.
- `KakaoAlimTalkSendUseCas... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T06:03:38Z | 2026-04-27T05:02:50Z | 2026-04-24T19:25:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 12OneTwo12 | 105,261,146 | U_kgDOBkYoWg | User | false |
openclaw/openclaw | 4,032,493,113 | I_kwDOQb6kR87wWvY5 | 37,284 | https://github.com/openclaw/openclaw/issues/37284 | https://api.github.com/repos/openclaw/openclaw/issues/37284 | Feature Request: Add right-click and mouse events to browser tool | **Is your feature request related to a problem? Please describe.**
The current OpenClaw browser tool supports click, double click, hover, drag, etc. However, it lacks:
1. Right-click (context menu) - No way to trigger context menu on elements
2. Mouse down/move/up sequence - No way to do individual mouse button events... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T06:21:35Z | 2026-04-27T04:33:28Z | 2026-04-25T04:15:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kanfred | 124,653,365 | U_kgDOB24PNQ | User | false |
openclaw/openclaw | 4,032,534,181 | I_kwDOQb6kR87wW5al | 37,299 | https://github.com/openclaw/openclaw/issues/37299 | https://api.github.com/repos/openclaw/openclaw/issues/37299 | [Bug]: cron list/run commands throw TypeError in OpenClaw 2026.3.2 | ### Bug type
Regression (worked before, now fails)
### Summary
cron list/run commands throw TypeError in CLI
### Steps to reproduce
1. Have cron jobs configured (e.g., daily-file-cleanup, daily-file-check)
2. Run `openclaw cron list` or `openclaw cron run daily-file-cleanup`
### Expected behavior
List cron jobs ... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T06:34:11Z | 2026-04-27T04:33:27Z | 2026-04-25T04:15:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | blizzardetna-ship-it | 264,942,934 | U_kgDOD8q1Vg | User | false |
openclaw/openclaw | 4,032,538,853 | I_kwDOQb6kR87wW6jl | 37,301 | https://github.com/openclaw/openclaw/issues/37301 | https://api.github.com/repos/openclaw/openclaw/issues/37301 | Bug: partial Telegram topic history in active context causes wrong operational decisions | ## Summary
In long Telegram topic workflows, the active runtime context can miss parts of prior messages/history. This leads to wrong operational decisions (re-asking for already provided secrets/parameters, repeated steps, wrong assumptions).
## Impact
- Agent does not see previously provided critical data (e.g., tok... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T06:35:37Z | 2026-04-27T04:33:26Z | 2026-04-25T04:15:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,032,538,909 | I_kwDOQb6kR87wW6kd | 37,302 | https://github.com/openclaw/openclaw/issues/37302 | https://api.github.com/repos/openclaw/openclaw/issues/37302 | Feature/Safety: enforce hard preflight guardrails for bot-token and target isolation | ## Summary
There are no hard preflight guardrails for bot-token/channel-target isolation before applying integration changes. In multi-account/multi-chat operations, this can allow accidental token application to the wrong target context.
## Impact
- Risk of cross-target misconfiguration (wrong bot/channel/topic)
- Po... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T06:35:38Z | 2026-04-27T04:33:26Z | 2026-04-25T04:15:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,032,539,306 | I_kwDOQb6kR87wW6qq | 37,303 | https://github.com/openclaw/openclaw/issues/37303 | https://api.github.com/repos/openclaw/openclaw/issues/37303 | [Bug]: Onboarding crashes when using exec secret provider for channel tokens (telegram, slack, discord) | ### Bug type
Regression (worked before, now fails)
### Summary
When configuring a channel (Telegram, Slack, or Discord) during `openclaw onboard` and choosing an **exec secret provider** for the bot token, the onboarding wizard crashes immediately after the "Reference validated" step with:
```
Error: channels.teleg... | closed | completed | true | 10 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T06:35:46Z | 2026-04-27T05:02:48Z | 2026-04-24T19:25:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | spotshare-nick | 5,923,070 | MDQ6VXNlcjU5MjMwNzA= | User | false |
openclaw/openclaw | 4,032,544,260 | I_kwDOQb6kR87wW74E | 37,306 | https://github.com/openclaw/openclaw/issues/37306 | https://api.github.com/repos/openclaw/openclaw/issues/37306 | [Test] API提交bug report测试 | 这是一条测试消息，用于验证 GitHub Token 是否可以自动提交 bug | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T06:37:24Z | 2026-04-27T05:02:48Z | 2026-04-24T19:25:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | blizzardetna-ship-it | 264,942,934 | U_kgDOD8q1Vg | User | false |
openclaw/openclaw | 4,032,544,263 | I_kwDOQb6kR87wW74H | 37,307 | https://github.com/openclaw/openclaw/issues/37307 | https://api.github.com/repos/openclaw/openclaw/issues/37307 | [Bug]: 定时任务调用模型mistral-large-latest报错 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
2026.3.2版本 配置mistral/mistral-large-latest模型后 通过定时任务推送到telegram机器人信息时报错
### Steps to reproduce
2026.3.2版本 配置mistral/mistral-large-latest模型后 通过定时任务推送到telegram机器人信息时报错
### Expected behavior
一切正常
### Actual behavior
基本聊天是正常的
还是有问题
报错提示也不... | closed | completed | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T06:37:24Z | 2026-04-27T05:02:47Z | 2026-04-24T19:25:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aiYShen | 58,358,565 | MDQ6VXNlcjU4MzU4NTY1 | User | false |
openclaw/openclaw | 4,032,575,051 | I_kwDOQb6kR87wXDZL | 37,315 | https://github.com/openclaw/openclaw/issues/37315 | https://api.github.com/repos/openclaw/openclaw/issues/37315 | Cron delivered status shows as error - false positive when message actually sent successfully | ## Description
When using cron jobs with announce delivery mode to send messages via Feishu, the cron status shows error even when the message was actually delivered successfully.
## Steps to Reproduce
1. Set up a cron job with delivery mode announce to Feishu
2. Wait for cron to trigger
3. Check cron status - shows ... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-06T06:46:45Z | 2026-04-27T05:02:46Z | 2026-04-24T19:25:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cattear1120 | 244,715,720 | U_kgDODpYQyA | User | false |
openclaw/openclaw | 4,032,596,376 | I_kwDOQb6kR87wXImY | 37,322 | https://github.com/openclaw/openclaw/issues/37322 | https://api.github.com/repos/openclaw/openclaw/issues/37322 | [Bug]: tui and web return `run erro: 404 status code (no body)` | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="1235" height="271" alt="Image" src="https://github.com/user-attachments/assets/4f2730d4-d53a-4810-a162-83efcab06e81" />
I change model to bailian and using aliyun/coding-plan, while I for all model I got error.
### Steps to reproduce
<im... | closed | completed | true | 5 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T06:52:48Z | 2026-04-27T05:02:45Z | 2026-04-24T19:26:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | linshengli | 19,223,302 | MDQ6VXNlcjE5MjIzMzAy | User | false |
openclaw/openclaw | 4,032,596,745 | I_kwDOQb6kR87wXIsJ | 37,323 | https://github.com/openclaw/openclaw/issues/37323 | https://api.github.com/repos/openclaw/openclaw/issues/37323 | [Bug]: --json mode CLI stdout is polluted by plugin/bootstrap logs, breaking machine-readable output | ## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
Commands invoked with `--json` can emit plugin/bootstrap logs to stdout before the JSON payload, which breaks machine-readable CLI output and causes direct `jq` parsing to fail.
## Steps to reproduce
1. Use a setup where plugins are enabled an... | closed | completed | true | 4 | [
"stale"
] | [] | 2026-03-06T06:52:54Z | 2026-04-27T05:02:45Z | 2026-04-24T19:26:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | furedericca-lab | 263,020,793 | U_kgDOD61g-Q | User | false |
openclaw/openclaw | 4,032,622,662 | I_kwDOQb6kR87wXPBG | 37,330 | https://github.com/openclaw/openclaw/issues/37330 | https://api.github.com/repos/openclaw/openclaw/issues/37330 | Feature: Multi-agent reply stagger for natural collaboration | ## Problem
When multiple agents are bound to the same Slack channel (or any group chat), they all receive an inbound message simultaneously and start processing in parallel. This means:
1. **No agent can read another agent's reply before composing its own** — they are all thinking at the same time.
2. **Multiple agen... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T07:00:36Z | 2026-04-27T04:33:24Z | 2026-04-25T04:15:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | alexei-led | 1,898,375 | MDQ6VXNlcjE4OTgzNzU= | User | false |
openclaw/openclaw | 4,032,655,669 | I_kwDOQb6kR87wXXE1 | 37,344 | https://github.com/openclaw/openclaw/issues/37344 | https://api.github.com/repos/openclaw/openclaw/issues/37344 | [Bug]: Discord messages not being sent - "fetch failed" despite proxy configuration | ### Bug type
Regression (worked before, now fails)
### Summary
OpenClaw fails to send reply messages to Discord, showing "TypeError: fetch failed" in logs, even though the Discord proxy is properly configured and enabled.
- OpenClaw version: 2026.3.2
- Node.js: 22.22.0
- Platform: Linux (systemd service)
- Proxy: H... | closed | completed | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T07:10:35Z | 2026-04-27T05:02:44Z | 2026-04-24T19:26:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dechace5-hash | 265,997,837 | U_kgDOD9rODQ | User | false |
openclaw/openclaw | 4,032,668,909 | I_kwDOQb6kR87wXaTt | 37,361 | https://github.com/openclaw/openclaw/issues/37361 | https://api.github.com/repos/openclaw/openclaw/issues/37361 | memory_search intermittently fails with ENOENT during embedding model finalize (.ipull -> .gguf) | ## Summary
`memory_search` can fail with an `ENOENT` while trying to finalize/download the local embedding model, making semantic memory unavailable until the model appears later.
## Observed error
```
ENOENT: no such file or directory, rename '<model-cache>/...Q8_0.gguf.ipull' -> '<model-cache>/...Q8_0.gguf'
```
## ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T07:14:17Z | 2026-04-27T04:33:23Z | 2026-04-25T04:15:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | alf-oc | 264,007,320 | U_kgDOD7xumA | User | false |
openclaw/openclaw | 4,032,694,314 | I_kwDOQb6kR87wXggq | 37,388 | https://github.com/openclaw/openclaw/issues/37388 | https://api.github.com/repos/openclaw/openclaw/issues/37388 | [Bug]: Log file does not rotate automatically on date change (stuck in previous day's file) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Log file does not rotate automatically on date change (stuck in previous day's file)
### Steps to reproduce
- Gateway runs continuously without restart across midnight.
- After midnight, a new log file (e.g., openclaw-2026-03-06.log) is c... | closed | not_planned | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T07:22:11Z | 2026-04-27T04:33:23Z | 2026-04-25T04:15:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Narlea | 163,425,703 | U_kgDOCb2tpw | User | false |
openclaw/openclaw | 4,032,700,035 | I_kwDOQb6kR87wXh6D | 37,401 | https://github.com/openclaw/openclaw/issues/37401 | https://api.github.com/repos/openclaw/openclaw/issues/37401 | feat: load workspace .env per-agent automatically | ## Problem
When running multiple agents with separate workspaces (e.g. `workspace-rcoding`, `workspace-home`), each workspace can have its own `.env` file. However, the Gateway only loads `.env` from the **current working directory at startup** and the global `~/.openclaw/.env` — it does not load the per-workspace `.e... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T07:23:54Z | 2026-04-27T04:33:22Z | 2026-04-25T04:15:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | quantumppulse | 200,023,130 | U_kgDOC-wcWg | User | false |
openclaw/openclaw | 4,032,812,370 | I_kwDOQb6kR87wX9VS | 37,426 | https://github.com/openclaw/openclaw/issues/37426 | https://api.github.com/repos/openclaw/openclaw/issues/37426 | Reply media dedupe should normalize Windows local paths and file:// URLs | ## Summary
Media dedupe in reply payload handling can miss duplicates on Windows when the same file is represented as a local path (`C:\...` or `C:/...`) versus a `file://` URL (`file:///C:/...`).
## Problem
`filterMessagingToolMediaDuplicates` previously compared path forms too literally, so equivalent Windows path v... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T07:54:53Z | 2026-04-27T04:33:21Z | 2026-04-25T04:15:46Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | shuofengzhang | 24,763,026 | MDQ6VXNlcjI0NzYzMDI2 | User | false |
openclaw/openclaw | 4,032,815,895 | I_kwDOQb6kR87wX-MX | 37,428 | https://github.com/openclaw/openclaw/issues/37428 | https://api.github.com/repos/openclaw/openclaw/issues/37428 | delivery-recovery burns retries before WhatsApp listener is ready after restart | ## Problem
After gateway restart, delivery-recovery runs immediately and retries queued WhatsApp outbound deliveries before the WhatsApp Web listener is ready. This produces errors like:
`No active WhatsApp Web listener (account: default)`
Those attempts currently consume retryCount and can push entries to `delivery-... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T07:55:46Z | 2026-04-27T04:33:21Z | 2026-04-25T04:15:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jeffrey4341 | 77,844,634 | MDQ6VXNlcjc3ODQ0NjM0 | User | false |
openclaw/openclaw | 4,032,817,660 | I_kwDOQb6kR87wX-n8 | 37,430 | https://github.com/openclaw/openclaw/issues/37430 | https://api.github.com/repos/openclaw/openclaw/issues/37430 | Message tool: error notification sent after successful retry causes confusion | ## Summary
When the `message` tool fails on a first attempt (e.g., file path not in allowed directory) and then succeeds on a retry, the error notification for the first attempt is still delivered to the user **after** the successful delivery, creating confusion.
## Steps to Reproduce
1. Use the `message` tool to se... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T07:56:11Z | 2026-04-27T04:33:20Z | 2026-04-25T04:15:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cwmine | 4,034,274 | MDQ6VXNlcjQwMzQyNzQ= | User | false |
openclaw/openclaw | 4,032,820,273 | I_kwDOQb6kR87wX_Qx | 37,431 | https://github.com/openclaw/openclaw/issues/37431 | https://api.github.com/repos/openclaw/openclaw/issues/37431 | [Bug]: 网关自带的web应用总是出现吐字卡住 | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="2694" height="504" alt="Image" src="https://github.com/user-attachments/assets/02ebc4a4-0fb8-4bc4-ab2a-e46af8f74176" />
### Steps to reproduce
吐字经常卡住
### Expected behavior
吐字经常卡住
### Actual behavior
吐字经常卡住
### OpenClaw version
一直都是
#... | closed | completed | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T07:56:54Z | 2026-04-27T05:02:43Z | 2026-04-24T19:26:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Jeesica2025 | 239,816,661 | U_kgDODktP1Q | User | false |
openclaw/openclaw | 4,032,825,671 | I_kwDOQb6kR87wYAlH | 37,432 | https://github.com/openclaw/openclaw/issues/37432 | https://api.github.com/repos/openclaw/openclaw/issues/37432 | [Bug] Duplicate identical replies in TUI/gateway-client interface | In the TUI/gateway-client interface, the assistant sometimes sends the exact same response multiple times in a row.
**Reproduction:**
- Occurs intermittently with normal conversational replies
- Same message text is repeated (2-8 times observed)
- Does not appear to happen (or much less) on Signal and other normal cha... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T07:58:24Z | 2026-04-27T05:02:42Z | 2026-04-24T19:26:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jemmel | 153,074 | MDQ6VXNlcjE1MzA3NA== | User | false |
openclaw/openclaw | 4,032,859,684 | I_kwDOQb6kR87wYI4k | 37,446 | https://github.com/openclaw/openclaw/issues/37446 | https://api.github.com/repos/openclaw/openclaw/issues/37446 | Sub-agent timeout recovery creates duplicate API posts (idempotency gap) | ## Problem
When a sub-agent makes a POST request that succeeds (HTTP 201) but the TCP connection drops before the client receives the response, the retry path re-POSTs the same content, creating duplicate comments/actions.
## Evidence
Observed in multi-agent ClawInstitute engagement sessions: duplicate comments on p... | open | null | false | 4 | [] | [] | 2026-03-06T08:07:35Z | 2026-05-04T06:14:57Z | null | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | gasvn | 20,515,144 | MDQ6VXNlcjIwNTE1MTQ0 | User | false |
openclaw/openclaw | 4,032,884,027 | I_kwDOQb6kR87wYO07 | 37,452 | https://github.com/openclaw/openclaw/issues/37452 | https://api.github.com/repos/openclaw/openclaw/issues/37452 | feat(ui): Dashboard 消息来源标识 | ## 需求
飞书发送的消息在 Dashboard 看不到，但回复看得到。需要改进消息显示，区分消息来源。
## 功能
1. 在消息气泡上显示来源标签（飞书 / Dashboard）
2. 更新 CHANNEL_LABELS 添加 eishu: 飞书
3. 不同来源的消息有视觉区分
## 相关文件
- \ui/src/ui/chat/grouped-render.ts\ - 消息渲染
- \ui/src/ui/views/chat.ts\ - 聊天视图
- \ui/src/ui/app-render.helpers.ts\ - Channel 标签
## 验收标准
- [ ] 消息显示来源标签
- [ ] 飞书消息有飞... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-06T08:14:13Z | 2026-04-27T04:33:19Z | 2026-04-25T04:15:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Uniranger | 12,763,309 | MDQ6VXNlcjEyNzYzMzA5 | User | false |
openclaw/openclaw | 4,032,682,104 | I_kwDOQb6kR87wXdh4 | 37,374 | https://github.com/openclaw/openclaw/issues/37374 | https://api.github.com/repos/openclaw/openclaw/issues/37374 | feat: 飞书群聊多 Agent 支持 | ## 目标
为飞书群聊实现多 Agent 同时回复功能，类似 WhatsApp broadcast 模式。
## 需求
1. 飞书群聊支持多个 agent 同时回复（头脑风暴模式）
2. 配置方式：
\\\json5
{
broadcast: {
strategy: 'parallel', // or 'sequential'
'oc_xxx': ['creative-agent', 'critic-agent', 'practical-agent']
}
}
\\\
## 参考
- WhatsApp broadcast 实现: src/web/auto-reply/monitor/broadcast.... | closed | completed | true | 4 | [] | [] | 2026-03-06T07:18:16Z | 2026-04-27T05:02:43Z | 2026-04-24T19:26:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Uniranger | 12,763,309 | MDQ6VXNlcjEyNzYzMzA5 | User | false |
openclaw/openclaw | 4,032,918,473 | I_kwDOQb6kR87wYXPJ | 37,457 | https://github.com/openclaw/openclaw/issues/37457 | https://api.github.com/repos/openclaw/openclaw/issues/37457 | [Bug]: Auto-created agent dirs ignore OPENCLAW_STATE_DIR, fallback to ~/.openclaw/agents/ | ## Bug Report
### Environment
- **OpenClaw version:** 2026.3.2
- **OS:** macOS (arm64), Darwin 25.3.0
- **Install method:** Homebrew (npm global)
- **Service:** LaunchAgent
### Description
When `OPENCLAW_STATE_DIR` is set (e.g. `~/.openclaw/state`), explicitly configured agents correctly use `$OPENCLAW_STATE_DIR/age... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-06T08:22:45Z | 2026-04-27T05:02:41Z | 2026-04-24T19:26:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mantisai-bot | 259,690,819 | U_kgDOD3qRQw | User | false |
openclaw/openclaw | 4,032,971,642 | I_kwDOQb6kR87wYkN6 | 37,465 | https://github.com/openclaw/openclaw/issues/37465 | https://api.github.com/repos/openclaw/openclaw/issues/37465 | [Bug]: k2p5 model generates tool calls in incompatible format causing 41 validation errors | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The kimi-coding/k2p5 model generates tool calls in a format that is incompatible with OpenAI-compatible API, causing 41 validation errors. The k2p5 model outputs tools without the required "function" wrapper, while glm-5 works correctly.
#... | closed | not_planned | true | 4 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T08:36:19Z | 2026-04-27T05:02:40Z | 2026-04-24T19:26:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Dylan18 | 39,365,266 | MDQ6VXNlcjM5MzY1MjY2 | User | false |
openclaw/openclaw | 4,032,985,664 | I_kwDOQb6kR87wYnpA | 37,469 | https://github.com/openclaw/openclaw/issues/37469 | https://api.github.com/repos/openclaw/openclaw/issues/37469 | feat(workspace): schema enforcement for markdown workspace files (MEMORY.md, AGENTS.md, etc.) | ## Problem
Workspace files (\`MEMORY.md\`, \`AGENTS.md\`, \`HEARTBEAT.md\`, etc.) are free-form markdown. Agents write whatever structure they want; readers get no guarantees. This causes:
- Silent corruption: an agent overwrites a list section with prose, breaking downstream readers
- Drift: required sections go mis... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T08:39:55Z | 2026-04-27T04:33:18Z | 2026-04-25T04:15:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aronchick | 51,317 | MDQ6VXNlcjUxMzE3 | User | false |
openclaw/openclaw | 4,033,051,000 | I_kwDOQb6kR87wY3l4 | 37,494 | https://github.com/openclaw/openclaw/issues/37494 | https://api.github.com/repos/openclaw/openclaw/issues/37494 | [Bug]: MAC install issue SHARP_IGNOR_GLOBAL_LIBVIPS=1 | ### Bug type
Crash (process/app exits or hangs)
### Summary
cannot install always appear SHARP_IGNOR_GLOBAL_LIBVIPS=1 error. try use SHARP for skip it and also sudo. but still not work. assumme 3.2 issue!!!!!! need help!!!!
systm: mac os
device: mini m4
### Steps to reproduce
stop in first step
### Expected beh... | closed | not_planned | true | 2 | [
"bug",
"stale",
"bug:crash"
] | [] | 2026-03-06T08:57:43Z | 2026-04-27T04:33:18Z | 2026-04-25T04:15:54Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chandler0714 | 33,366,056 | MDQ6VXNlcjMzMzY2MDU2 | User | false |
openclaw/openclaw | 4,033,055,995 | I_kwDOQb6kR87wY4z7 | 37,497 | https://github.com/openclaw/openclaw/issues/37497 | https://api.github.com/repos/openclaw/openclaw/issues/37497 | delivery-recovery retries permanent failures indefinitely (message too long, reply target not found) | ## Bug Report
### Summary
The `delivery-recovery` system retries failed Telegram message deliveries indefinitely, even for **permanent, non-recoverable errors** that will never succeed on retry.
### Affected Error Types
1. **`400: Bad Request: message to be replied not found`**
Triggered when a reply references ... | closed | completed | true | 3 | [] | [] | 2026-03-06T08:58:52Z | 2026-04-27T05:02:38Z | 2026-04-24T19:26:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mood5168-wq | 252,055,911 | U_kgDODwYRZw | User | false |
openclaw/openclaw | 4,033,057,300 | I_kwDOQb6kR87wY5IU | 37,499 | https://github.com/openclaw/openclaw/issues/37499 | https://api.github.com/repos/openclaw/openclaw/issues/37499 | Context compaction causes conversation to freeze/hang | ## Bug Description
After context compaction (when context exceeds limit), the conversation sometimes freezes and the assistant stops responding. This happens consistently after multiple compaction cycles.
## Steps to Reproduce
1. Have a long conversation that triggers multiple context compressions (context > 138%)
2... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T08:59:09Z | 2026-04-27T05:02:37Z | 2026-04-24T19:26:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rootwhois | 53,112,648 | MDQ6VXNlcjUzMTEyNjQ4 | User | false |
openclaw/openclaw | 4,033,064,042 | I_kwDOQb6kR87wY6xq | 37,500 | https://github.com/openclaw/openclaw/issues/37500 | https://api.github.com/repos/openclaw/openclaw/issues/37500 | [Control UI] Chat auto-scroll overrides user scroll during streaming | ## Problem
When the assistant is streaming a response in the Control UI chat, the page continuously auto-scrolls to the bottom. If the user scrolls up to review previous messages during streaming, they get pulled back down almost immediately.
This makes it impossible to reference earlier messages while the assistant ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T09:00:43Z | 2026-04-27T04:33:16Z | 2026-04-25T04:15:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | liyuefan0214 | 40,573,335 | MDQ6VXNlcjQwNTczMzM1 | User | false |
openclaw/openclaw | 4,033,076,196 | I_kwDOQb6kR87wY9vk | 37,505 | https://github.com/openclaw/openclaw/issues/37505 | https://api.github.com/repos/openclaw/openclaw/issues/37505 | Cron job timeout aborts entire model fallback chain via shared AbortController | ## Summary
When a cron job's timeout fires, the fallback model chain never executes. The abort signal from the cron timeout is shared with all fallback provider attempts, causing them to fail instantly (~100-200ms) without making a network request.
## Version
OpenClaw 2026.3.2 (85377a2), stable channel
## Steps to ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T09:03:29Z | 2026-04-27T04:33:15Z | 2026-04-25T04:15:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | n3moxyz | 254,834,596 | U_kgDODzB3pA | User | false |
openclaw/openclaw | 4,033,098,448 | I_kwDOQb6kR87wZDLQ | 37,515 | https://github.com/openclaw/openclaw/issues/37515 | https://api.github.com/repos/openclaw/openclaw/issues/37515 | Telegram voice notes can inject known Whisper hallucinations like "Субтитры сделал DimaTorzok" as trusted transcript text | ## Summary
Telegram voice notes can be transcribed into known Whisper hallucination strings such as:
- `Субтитры сделал DimaTorzok`
- other subtitle / watermark-like credits
OpenClaw currently forwards that transcript as trusted user text into the agent prompt, which can poison the conversation and make the agent re... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T09:08:42Z | 2026-04-27T04:33:14Z | 2026-04-25T04:15:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aleksesipenko | 252,942,994 | U_kgDODxOakg | User | false |
openclaw/openclaw | 4,033,124,530 | I_kwDOQb6kR87wZJiy | 37,544 | https://github.com/openclaw/openclaw/issues/37544 | https://api.github.com/repos/openclaw/openclaw/issues/37544 | Slack message tool: agent loops on message_not_found / fileId errors, burns API quota | ## Problem
When the Slack message tool encounters `message_not_found` or `fileId required` errors, the agent retries in a tight loop rather than failing fast. This burns through API rate limits and cascades into failures for all other sessions (cron jobs, heartbeats, etc).
## What happened
A Slack session got stuck ... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T09:15:08Z | 2026-04-27T05:02:37Z | 2026-04-24T19:26:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | PamelaFernandez2 | 129,841,443 | U_kgDOB705Iw | User | false |
openclaw/openclaw | 4,033,139,364 | I_kwDOQb6kR87wZNKk | 37,546 | https://github.com/openclaw/openclaw/issues/37546 | https://api.github.com/repos/openclaw/openclaw/issues/37546 | [Bug]: Message ordering conflict - please try again. If this persists, use /new to start a fresh session. | ### Bug type
Regression (worked before, now fails)
### Summary
No matter what message I send, I get the same reply, and /new doesn't work either; it's always the same response.Message ordering conflict - please try again. If this persists, use /new to start a fresh session.
### Steps to reproduce
No matter what me... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T09:18:12Z | 2026-04-27T04:33:12Z | 2026-04-25T04:16:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yunxilyf | 44,936,564 | MDQ6VXNlcjQ0OTM2NTY0 | User | false |
openclaw/openclaw | 4,033,163,783 | I_kwDOQb6kR87wZTIH | 37,550 | https://github.com/openclaw/openclaw/issues/37550 | https://api.github.com/repos/openclaw/openclaw/issues/37550 | macOS: OpenClaw.app v2026.3.2 crashes on launch (Fatal access conflict in TalkOverlayView) | ## Summary
OpenClaw.app `v2026.3.2` crashes on launch on macOS (menu bar app does not stay alive).
## Environment
- OpenClaw.app: `2026.3.2` (build `2026030290`)
- OpenClaw CLI: `2026.3.2`
- macOS: `26.4 (25E5223i)`
- Hardware: Apple Silicon (`Mac16,10`)
## Repro steps
1. Install DMG release `OpenClaw-2026.3.2.dmg`.
... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-06T09:23:41Z | 2026-04-27T04:33:12Z | 2026-04-25T04:16:03Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lamkan0210 | 65,824,253 | MDQ6VXNlcjY1ODI0MjUz | User | false |
openclaw/openclaw | 4,033,212,886 | I_kwDOQb6kR87wZfHW | 37,563 | https://github.com/openclaw/openclaw/issues/37563 | https://api.github.com/repos/openclaw/openclaw/issues/37563 | [Bug]: `openclaw plugins install` fails on Windows when Node.js path contains spaces (e.g. `C:\Program Files\nodejs`) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
openclaw plugins install <npm-spec> fails on Windows when Node.js is installed in a path containing spaces (e.g. C:\Program Files\nodejs\). The npm pack step produces garbled output and exits non-zero because the resolved npm-cli.js path i... | closed | not_planned | true | 4 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T09:35:26Z | 2026-04-27T04:33:11Z | 2026-04-25T04:16:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ChrisJaunes | 28,571,176 | MDQ6VXNlcjI4NTcxMTc2 | User | false |
openclaw/openclaw | 4,033,215,676 | I_kwDOQb6kR87wZfy8 | 37,564 | https://github.com/openclaw/openclaw/issues/37564 | https://api.github.com/repos/openclaw/openclaw/issues/37564 | [Bug]: {"subsystem":"gateway/channels/feishu"} feishu[default]: dispatch complete (queuedFinal=false, replies=0) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
[Bug]: {"subsystem":"gateway/channels/feishu"} feishu[default]: dispatch complete (queuedFinal=false, replies=0)
### Steps to reproduce
<img width="1656" height="270" alt="Image" src="https://github.com/user-attachments/assets/da5157bb-4... | closed | completed | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T09:36:03Z | 2026-04-27T05:02:34Z | 2026-04-24T19:27:06Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 13281846650-lab | 251,552,073 | U_kgDODv5hSQ | User | false |
openclaw/openclaw | 4,033,217,494 | I_kwDOQb6kR87wZgPW | 37,565 | https://github.com/openclaw/openclaw/issues/37565 | https://api.github.com/repos/openclaw/openclaw/issues/37565 | Telegram: metadata leak in DMs | Repro: Messages prefixed with "Human: Conversation info (untrusted metadata): {". Config: dmPolicy=pairing, streaming=partial→off. v2026.3.2, Linux. Status clean. Screenshot repro if needed. | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-06T09:36:28Z | 2026-04-27T05:02:34Z | 2026-04-24T19:27:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | charlesamartinez | 71,293,971 | MDQ6VXNlcjcxMjkzOTcx | User | false |
openclaw/openclaw | 4,033,219,231 | I_kwDOQb6kR87wZgqf | 37,566 | https://github.com/openclaw/openclaw/issues/37566 | https://api.github.com/repos/openclaw/openclaw/issues/37566 | [Bug]: TUI 不显示推送消息 | ### Bug type
Regression (worked before, now fails)
### Summary
### 环境
- **OpenClaw 版本**: 2026.3.3
- **终端**: Windows Terminal (WSL2)
- **Gateway**: 127.0.0.1:18789
### 问题描述
通过 TUI 发送消息后，助手回复不会自动显示，需要手动刷新才能看到。
### 复现步骤
1. 在 Windows Terminal 打开 TUI
2. 发送一条消息
3. 等待助手回复
4. 回复不会自动显示
###... | closed | completed | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T09:36:51Z | 2026-04-27T04:55:04Z | 2026-04-24T22:05:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Rainorr | 133,183,615 | U_kgDOB_A4fw | User | false |
openclaw/openclaw | 4,033,238,323 | I_kwDOQb6kR87wZlUz | 37,571 | https://github.com/openclaw/openclaw/issues/37571 | https://api.github.com/repos/openclaw/openclaw/issues/37571 | [Bug]: Custom anthropic-messages providers fail when proxy omits SSE event: field | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Custom providers using `api: "anthropic-messages"` with a third-party proxy that returns SSE streams without `event:` field lines fail with `"request ended without sending any chunks"`. The `@anthropic-ai/sdk` SSE parser requires the `event... | closed | completed | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T09:41:23Z | 2026-04-27T04:55:03Z | 2026-04-24T22:05:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | FuzzyFade | 25,416,941 | MDQ6VXNlcjI1NDE2OTQx | User | false |
openclaw/openclaw | 4,033,262,091 | I_kwDOQb6kR87wZrIL | 37,579 | https://github.com/openclaw/openclaw/issues/37579 | https://api.github.com/repos/openclaw/openclaw/issues/37579 | [Bug] Compaction safeguard 频繁触发导致消息丢失 | ## 问题描述
当对话历史较长时，UI 显示 "context compacted"，但实际 compaction 被 safeguard 取消。在此期间，用户发送的消息无法被 AI 接收/响应。
## 日志证据
```
WARN: Compaction safeguard: cancelling compaction with no real conversation messages to summarize
```
这个警告在短时间内频繁出现（每分钟多次），说明：
1. compact 流程被触发
2. safeguard 判断"没有真正的对话消息"后取消
3. 但 UI 可能已经显示 compacting/compa... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-06T09:47:03Z | 2026-04-27T05:02:33Z | 2026-04-24T19:27:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ahsen198911-debug | 263,568,933 | U_kgDOD7W-JQ | User | false |
openclaw/openclaw | 4,024,910,924 | I_kwDOQb6kR87v50RM | 35,095 | https://github.com/openclaw/openclaw/issues/35095 | https://api.github.com/repos/openclaw/openclaw/issues/35095 | [Bug]: Google Chat inbound webhook returns 401 Unauthorized despite matching audience | ### Bug type
Regression (worked before, now fails)
### Summary
Google Chat inbound webhooks reach /googlechat but are rejected with 401 Unauthorized before pairing, even when JWT aud exactly matches configured app-url audience.
### Steps to reproduce
1. Configure channels.googlechat with audienceType=app-url and a... | closed | not_planned | true | 7 | [
"bug",
"regression"
] | [] | 2026-03-05T00:22:12Z | 2026-04-28T04:43:52Z | 2026-04-26T01:23:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JakeDred81 | 206,614,992 | U_kgDODFCx0A | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.