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 | 4,232,523,681 | I_kwDOQb6kR878Ry-h | 63,751 | https://github.com/openclaw/openclaw/issues/63751 | https://api.github.com/repos/openclaw/openclaw/issues/63751 | [Bug]: models.mode: "replace" does not suppress Bedrock auto-discovered models | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Setting models.mode to "replace" does not suppress Amazon Bedrock auto-discovered models from the model selector.
### Steps to reproduce
1. Set "models.mode": "replace" in openclaw.json 2. Configure only two providers (ton... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-09T13:18:19Z | 2026-04-30T04:42:48Z | 2026-04-26T04:42:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | gaoletian737-byte | 242,212,193 | U_kgDODm_dYQ | User | false |
openclaw/openclaw | 4,232,617,384 | I_kwDOQb6kR878SJ2o | 63,756 | https://github.com/openclaw/openclaw/issues/63756 | https://api.github.com/repos/openclaw/openclaw/issues/63756 | [Windows] qmd memory backend unavailable — .cmd wrapper not resolved | ## Environment
- OS: Windows 11 (native, no WSL2)
- OpenClaw version: latest
- Node.js: v24.x
- QMD: @tobilu/qmd@2.1.0 installed via npm to D:\nvm\global-packages\
- PATH includes D:\nvm\global-packages where qmd.cmd lives
## Config
json
memory: {
backend: qmd,
qmd: {
searchMode: query,
command: qmd
}
}
... | closed | completed | true | 1 | [] | [] | 2026-04-09T13:33:36Z | 2026-04-30T04:42:47Z | 2026-04-26T04:42:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | SkyFragments | 150,218,224 | U_kgDOCPQl8A | User | false |
openclaw/openclaw | 4,232,653,882 | I_kwDOQb6kR878SSw6 | 63,761 | https://github.com/openclaw/openclaw/issues/63761 | https://api.github.com/repos/openclaw/openclaw/issues/63761 | HTTP 412 (Account Suspended) does not trigger model fallback | ## Bug
When a provider returns HTTP 412 (e.g., Fireworks "Account suspended due to spending limit"), the gateway does NOT trigger model fallback. The bot simply stops responding instead of falling back to the next model in the fallback chain.
## Root Cause
In `classifyFailoverClassificationFromHttpStatus()`, HTTP 41... | closed | completed | true | 1 | [] | [] | 2026-04-09T13:39:48Z | 2026-04-30T04:42:46Z | 2026-04-26T04:42:21Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | superhello2099 | 190,775,223 | U_kgDOC17_tw | User | false |
openclaw/openclaw | 4,232,665,058 | I_kwDOQb6kR878SVfi | 63,763 | https://github.com/openclaw/openclaw/issues/63763 | https://api.github.com/repos/openclaw/openclaw/issues/63763 | Plugin CLI commands for lossless-claw not registering despite plugin loading and allowlist permitting it | ## Summary
OpenClaw loads the `lossless-claw` plugin successfully and compaction works, but the documented native/plugin CLI commands are not being registered on the root CLI surface.
In this environment:
- lossless-claw is enabled
- `plugins.allow` explicitly includes `lossless-claw`
- the context engine slot is set... | closed | not_planned | true | 1 | [] | [] | 2026-04-09T13:41:36Z | 2026-04-30T04:42:45Z | 2026-04-26T04:42:25Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | brokemac79 | 255,583,030 | U_kgDODzvjNg | User | false |
openclaw/openclaw | 4,232,989,997 | I_kwDOQb6kR878Tk0t | 63,784 | https://github.com/openclaw/openclaw/issues/63784 | https://api.github.com/repos/openclaw/openclaw/issues/63784 | Session lock file not released after LLM idle timeout (no gateway restart) | ## Bug Description
Session lock files (`.jsonl.lock`) are not released when an LLM idle timeout occurs, even when the gateway process itself is not restarted. This causes the session to become permanently stuck — the lock file remains on disk and the session never recovers.
## Steps to Reproduce
1. Agent receives a ... | closed | completed | true | 1 | [] | [] | 2026-04-09T14:36:12Z | 2026-04-30T04:42:45Z | 2026-04-26T04:42:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ZaneChen76 | 70,149,446 | MDQ6VXNlcjcwMTQ5NDQ2 | User | false |
openclaw/openclaw | 4,233,555,899 | I_kwDOQb6kR878Vu-7 | 63,830 | https://github.com/openclaw/openclaw/issues/63830 | https://api.github.com/repos/openclaw/openclaw/issues/63830 | Plugin services registered/loaded after runtime start may never receive host service.start() | ## Symptom | 症状
In a Linux + systemd deployment, plugin services that are registered after runtime startup can miss the normal host `service.start()` call entirely.
Observed symptom with `memos-local-openclaw-plugin@1.0.8`:
```text
memos-local: service.start() not called by host, self-starting viewer...
```
The plu... | closed | completed | true | 1 | [] | [] | 2026-04-09T16:18:49Z | 2026-04-30T04:42:43Z | 2026-04-26T04:42:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | starm0on | 134,861,587 | U_kgDOCAnTEw | User | false |
openclaw/openclaw | 4,233,692,892 | I_kwDOQb6kR878WQbc | 63,854 | https://github.com/openclaw/openclaw/issues/63854 | https://api.github.com/repos/openclaw/openclaw/issues/63854 | WhatsApp watchdog messageTimeoutMs should be configurable (30min hardcoded timeout causes cascading 499/408 errors) | ## Summary
The WhatsApp connection watchdog has a hardcoded 30-minute idle timeout (`MESSAGE_TIMEOUT_MS = 1800 * 1e3`). When no messages arrive for 30 minutes (normal for low-traffic accounts), the watchdog force-restarts the connection, generating cascading 408/499 errors in logs and causing unnecessary healthcheck w... | closed | not_planned | true | 1 | [] | [] | 2026-04-09T16:47:05Z | 2026-04-30T04:42:42Z | 2026-04-26T04:42:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | djimit | 19,958,832 | MDQ6VXNlcjE5OTU4ODMy | User | false |
openclaw/openclaw | 4,233,767,277 | I_kwDOQb6kR878Wilt | 63,859 | https://github.com/openclaw/openclaw/issues/63859 | https://api.github.com/repos/openclaw/openclaw/issues/63859 | [Bug]: ACP non-interactive sessions fail with exec due to permissionMode not propagating from global approvals config | ## Environment
- **OpenClaw version**: 2026.4.8 (9ece252)
- **Platform**: macOS Apple Silicon (M1/M4)
- **Channel**: Telegram → Enoch agent → Codex agent via ACP
## Problem
When an agent (Enoch) dispatches an ACP oneshot task to another agent (Codex) that requires `exec` tool usage, the ACP session fails with:
```
... | closed | completed | true | 1 | [] | [] | 2026-04-09T17:04:16Z | 2026-04-30T04:42:42Z | 2026-04-26T04:42:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | baochen10luo | 262,308,901 | U_kgDOD6KEJQ | User | false |
openclaw/openclaw | 4,233,851,680 | I_kwDOQb6kR878W3Mg | 63,865 | https://github.com/openclaw/openclaw/issues/63865 | https://api.github.com/repos/openclaw/openclaw/issues/63865 | Critical: Pre-compaction flush prompts replace user messages in web UI | ## Summary
Pre-compaction memory flush prompts are **replacing** user messages in the OpenClaw web interface, causing messages to disappear from the conversation display.
## Steps to Reproduce
1. Send a message via OpenClaw web chat
2. Trigger occurs while approaching compaction threshold
3. Observe that user message ... | closed | not_planned | true | 1 | [] | [] | 2026-04-09T17:21:27Z | 2026-04-30T04:42:41Z | 2026-04-26T04:42:53Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | bmilne1981 | 138,170,065 | U_kgDOCDxO0Q | User | false |
openclaw/openclaw | 4,234,886,073 | I_kwDOQb6kR878azu5 | 63,922 | https://github.com/openclaw/openclaw/issues/63922 | https://api.github.com/repos/openclaw/openclaw/issues/63922 | Edit tool: add file-level locking for concurrent workspace file edits | ## Problem
When multiple agents (or a main agent + spawned subagents) edit the same workspace file concurrently, the Edit tool's `oldText` match can fail because another agent modified the file between the read and the write. This causes transient `"Could not find the exact text"` errors.
Common collision targets in ... | closed | completed | true | 1 | [] | [] | 2026-04-09T21:01:58Z | 2026-04-30T04:42:38Z | 2026-04-26T04:43:05Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | willfivestar | 212,644,549 | U_kgDODKyyxQ | User | false |
openclaw/openclaw | 4,235,322,385 | I_kwDOQb6kR878ceQR | 63,955 | https://github.com/openclaw/openclaw/issues/63955 | https://api.github.com/repos/openclaw/openclaw/issues/63955 | Agent "analysis paralysis" — repeatedly stalls at analysis→execution transition | ### Description
The agent gets stuck in a loop where it completes analysis/review of a task but fails to transition to actual execution. It repeatedly produces "I'll do it now" intent statements without outputting the work itself.
### Reproduction Pattern
1. Agent reads and analyzes code/files (multiple tool calls)
... | closed | completed | true | 2 | [] | [] | 2026-04-09T22:43:48Z | 2026-04-30T04:42:37Z | 2026-04-26T04:43:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | amdhelper | 232,188,533 | U_kgDODdbqdQ | User | false |
openclaw/openclaw | 4,236,220,790 | I_kwDOQb6kR878f5l2 | 64,039 | https://github.com/openclaw/openclaw/issues/64039 | https://api.github.com/repos/openclaw/openclaw/issues/64039 | Telegram: some normal final replies disappear while explicit message sends persist | ## Summary
Some normal assistant replies in a Telegram DM appear briefly and then disappear. Explicit `message` tool sends remain visible and stable.
## Environment
- OpenClaw 2026.4.5
- Main session on Telegram DM
- Gateway healthy
- Telegram provider healthy in `openclaw status`
- Session remained healthy during inc... | closed | completed | true | 1 | [] | [] | 2026-04-10T02:56:02Z | 2026-04-30T04:42:35Z | 2026-04-26T04:43:25Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | PaulDillingham | 72,823,541 | MDQ6VXNlcjcyODIzNTQx | User | false |
openclaw/openclaw | 4,329,935,700 | I_kwDOQb6kR88AAAABAhWTVA | 71,927 | https://github.com/openclaw/openclaw/issues/71927 | https://api.github.com/repos/openclaw/openclaw/issues/71927 | WebChat: Assistant messages rendered twice (duplicate bubbles) with streaming | **Describe the bug**
When receiving a streaming response in WebChat, the assistant message is displayed twice — once during streaming and again after the response completes, resulting in duplicate message bubbles.
**To Reproduce**
1. Open WebChat
2. Send any message to the assistant
3. Observe the assistant reply appe... | closed | completed | true | 2 | [] | [] | 2026-04-26T03:33:40Z | 2026-04-29T04:42:26Z | 2026-04-26T07:06:14Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | xinlelights123-design | 226,178,426 | U_kgDODXs1eg | User | false |
openclaw/openclaw | 4,046,296,870 | I_kwDOQb6kR87xLZcm | 41,193 | https://github.com/openclaw/openclaw/issues/41193 | https://api.github.com/repos/openclaw/openclaw/issues/41193 | v2026.3.8: Crash loop on startup — plugin errors, SyntaxErrors, V8 JIT crashes | ## Summary
After upgrading from v2026.2.26 to v2026.3.8, the gateway enters a crash loop on startup. **32 crashes in ~50 minutes.** v2026.2.26 was stable (19 crashes over 4 days, mostly cold-start JIT). The gateway is essentially unusable on 2026.3.8.
## Environment
- **OS:** Ubuntu 24.04.4 LTS
- **Kernel:** 6.17.0-... | closed | not_planned | true | 5 | [
"cannot-reproduce",
"stale"
] | [] | 2026-03-09T15:34:05Z | 2026-04-30T04:42:34Z | 2026-04-26T04:44:39Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | russellweiss | 44,983,860 | MDQ6VXNlcjQ0OTgzODYw | User | false |
openclaw/openclaw | 4,198,359,788 | I_kwDOQb6kR876PeLs | 60,083 | https://github.com/openclaw/openclaw/issues/60083 | https://api.github.com/repos/openclaw/openclaw/issues/60083 | Cron startup catch-up fires duplicate when previous run was itself a late catch-up | ## Bug Description
The cron scheduler's startup catch-up logic (`planStartupCatchup` → `isRunnableJob`) has a structural flaw where a job that ran as a late catch-up can be incorrectly classified as "missed" on the next gateway restart, causing a duplicate execution.
## Steps to Reproduce
1. A cron job is scheduled ... | closed | not_planned | true | 1 | [] | [] | 2026-04-03T05:01:37Z | 2026-04-30T04:42:21Z | 2026-04-26T04:46:34Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | tob-kai | 267,897,187 | U_kgDOD_fJYw | User | false |
openclaw/openclaw | 4,079,667,684 | I_kwDOQb6kR87zKsnk | 47,721 | https://github.com/openclaw/openclaw/issues/47721 | https://api.github.com/repos/openclaw/openclaw/issues/47721 | [Bug]: Please update dependencies to fix deprecation warnings (glob, node-domexception) | ### Bug type
Regression (worked before, now fails)
### Summary
npm warn deprecated node-domexception@1.0.0: Use your platform's native DOMException instead
npm warn deprecated glob@10.5.0: Old versions of glob are not supported, and contain widely publicized security vulnerabilities, which have been fixed in the cur... | closed | completed | true | 2 | [
"bug",
"regression"
] | [
"vincentkoc"
] | 2026-03-16T01:07:19Z | 2026-04-30T04:42:18Z | 2026-04-26T04:46:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 18205322580-star | 243,345,932 | U_kgDODoEqDA | User | false |
openclaw/openclaw | 4,237,147,053 | I_kwDOQb6kR878jbut | 64,144 | https://github.com/openclaw/openclaw/issues/64144 | https://api.github.com/repos/openclaw/openclaw/issues/64144 | [Feature]: Exclude node_modules from extensions in backup archive | ## Summary
Exclude node_modules directories inside extensions/ when creating backup archives to reduce backup size significantly.
## Problem to solve
openclaw backup create archives the entire state directory including ~/.openclaw/extensions/*/node_modules/. These dependency trees are large (often hundreds of MB) an... | closed | not_planned | true | 1 | [] | [] | 2026-04-10T06:41:47Z | 2026-04-30T04:42:18Z | 2026-04-26T04:46:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BrilliantWang | 120,538,413 | U_kgDOBy9FLQ | User | false |
openclaw/openclaw | 4,330,026,583 | I_kwDOQb6kR88AAAABAhb2Vw | 71,963 | https://github.com/openclaw/openclaw/issues/71963 | https://api.github.com/repos/openclaw/openclaw/issues/71963 | Cron jobs with sessionTarget=isolated produce false `lost` errors after successful runs | ## Summary
Cron jobs with `sessionTarget: "isolated"` are flagged as `lost` errors in `openclaw tasks audit` even after the cron execution completes successfully (`lastRunStatus: ok`). This causes monotonically growing `lost` error counts on long-running deployments and creates false alarms in audit reports.
## Repro
... | closed | completed | true | 2 | [] | [] | 2026-04-26T04:34:42Z | 2026-04-29T04:37:17Z | 2026-04-26T10:45:08Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Tommy3808 | 79,340,377 | MDQ6VXNlcjc5MzQwMzc3 | User | false |
openclaw/openclaw | 4,237,559,069 | I_kwDOQb6kR878lAUd | 64,183 | https://github.com/openclaw/openclaw/issues/64183 | https://api.github.com/repos/openclaw/openclaw/issues/64183 | status: memory may show unavailable while Memoria plugin health is ok | ## Summary
`openclaw status` can report memory as unavailable even when the Memoria plugin and backend are healthy.
## Observed behavior
On a live installation:
- `openclaw status` showed `Memory unavailable`
- `thememoria` was loaded successfully
- `openclaw memoria health` returned `status: ok`
- `openclaw ltm list ... | closed | not_planned | true | 1 | [] | [] | 2026-04-10T07:56:52Z | 2026-04-30T04:42:17Z | 2026-04-26T04:48:19Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | MatrixAdventurer | 85,048,713 | MDQ6VXNlcjg1MDQ4NzEz | User | false |
openclaw/openclaw | 4,237,616,650 | I_kwDOQb6kR878lOYK | 64,190 | https://github.com/openclaw/openclaw/issues/64190 | https://api.github.com/repos/openclaw/openclaw/issues/64190 | [Feature]: Batch ack reactions when inbound debounce is active | ## Problem
When `messages.inbound.debounceMs` is configured (e.g. 3000ms), rapid text messages from the same sender are correctly bundled into a single agent turn. However, ack reactions (`messages.ackReaction`) fire immediately on each individual message as it arrives at the gateway, before the debounce window closes... | closed | completed | true | 1 | [] | [] | 2026-04-10T08:08:01Z | 2026-04-30T04:42:16Z | 2026-04-26T04:48:24Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | M-Lietz | 222,629,290 | U_kgDODUUNqg | User | false |
openclaw/openclaw | 4,237,900,727 | I_kwDOQb6kR878mTu3 | 64,204 | https://github.com/openclaw/openclaw/issues/64204 | https://api.github.com/repos/openclaw/openclaw/issues/64204 | [Bug]: browser.* config reload rule triggers unnecessary full gateway restart for profile-only changes | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The config hot-reload matrix classifies the entire `browser.*` prefix as `kind: "restart"`, meaning any change under `browser.profiles` — including adding a new named browser profile — triggers a full gateway p... | closed | not_planned | true | 1 | [] | [] | 2026-04-10T08:53:01Z | 2026-04-30T04:42:15Z | 2026-04-26T04:48:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | justin-vin | 270,847,143 | U_kgDOECTMpw | User | false |
openclaw/openclaw | 4,330,023,000 | I_kwDOQb6kR88AAAABAhboWA | 71,962 | https://github.com/openclaw/openclaw/issues/71962 | https://api.github.com/repos/openclaw/openclaw/issues/71962 | Feature Request: Support WeChat Group Chat (群聊支持) | ## 需求描述 / Description
希望 OpenClaw 微信插件支持群聊功能：
1. **能够被拉入微信群** - 目前只支持私聊 (chatTypes: ["direct"])
2. **能够收到群内的 @ 提及** - 当有人在群里 @ 机器人时能够响应
3. **能够在群内回复** - 机器人可以在群里发送消息
## 使用场景 / Use Cases
- 作为客服助手加入业务相关的微信群
- 在多个群里提供不同领域的专业问答服务
- 家庭群、校友群等场景的 AI 助手
## 当前限制 / Current Limitation
查看 channel.ts 发现 capabilities 硬编码为：
``... | closed | not_planned | true | 3 | [] | [] | 2026-04-26T04:32:19Z | 2026-04-29T04:37:18Z | 2026-04-26T10:45:03Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | zhanguo711022 | 214,910,977 | U_kgDODM9IAQ | User | false |
openclaw/openclaw | 4,049,534,761 | I_kwDOQb6kR87xXv8p | 41,797 | https://github.com/openclaw/openclaw/issues/41797 | https://api.github.com/repos/openclaw/openclaw/issues/41797 | [Bug]: Install script (install.ps1) forcefully exits PowerShell process on systems without winget (e.g., Windows LTSC) | ### Bug type
Crash (process/app exits or hangs)
### Summary
**Problem:** The install.ps1 script uses exit 1 in the Install-Node function when a package manager is not found.
## Specific Snippet:
```powershell
# Line 122 in the provided script
Write-Host "Or install winget (App Installer) from the Microsoft Store." ... | closed | duplicate | true | 3 | [
"bug",
"stale",
"bug:crash"
] | [] | 2026-03-10T05:38:57Z | 2026-04-30T04:41:32Z | 2026-04-26T04:54:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | itshusky01 | 79,142,465 | MDQ6VXNlcjc5MTQyNDY1 | User | false |
openclaw/openclaw | 4,330,059,141 | I_kwDOQb6kR88AAAABAhd1hQ | 71,972 | https://github.com/openclaw/openclaw/issues/71972 | https://api.github.com/repos/openclaw/openclaw/issues/71972 | fix(bonjour): CIAO PROBING CANCELLED unhandled promise rejection crashes gateway on Windows (4.24) | ## Summary
On Windows, upgrading to 4.24 (and 4.23) causes the gateway to crash shortly after startup with an unhandled promise rejection from the Bonjour/mDNS advertiser.
## Error
\\\
[openclaw] Unhandled promise rejection: CIAO PROBING CANCELLED
[openclaw] wrote stability bundle: openclaw-stability-2026-04-26T04-4... | closed | completed | true | 3 | [] | [] | 2026-04-26T04:57:01Z | 2026-04-29T04:31:08Z | 2026-04-26T06:18:40Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | andyk-ms | 91,510,251 | U_kgDOBXRV6w | User | false |
openclaw/openclaw | 4,330,019,416 | I_kwDOQb6kR88AAAABAhbaWA | 71,958 | https://github.com/openclaw/openclaw/issues/71958 | https://api.github.com/repos/openclaw/openclaw/issues/71958 | Bonjour gateway discovery can trigger unhandled CIAO PROBING/ANNOUNCEMENT CANCELLED rejections and cause gateway restart loops | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Gateway startup can enter repeated restart loops when the Bonjour discovery plugin hits CIAO probing/announcement cancellation failures.
### Steps to reproduce
1. Start OpenClaw 2026.4.24 with Bonjour gateway... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-04-26T04:29:49Z | 2026-04-30T04:38:17Z | 2026-04-26T06:18:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jihoon-ernesto | 41,365,514 | MDQ6VXNlcjQxMzY1NTE0 | User | false |
openclaw/openclaw | 4,046,114,298 | I_kwDOQb6kR87xKs36 | 41,164 | https://github.com/openclaw/openclaw/issues/41164 | https://api.github.com/repos/openclaw/openclaw/issues/41164 | Feature: default cron delivery channel/target in config | # Feature Request: Default cron delivery channel/target in config
## Problem
When creating new cron jobs, `delivery.channel` defaults to `"last"` (most recently active conversation). This causes cron output to be delivered to unintended recipients — e.g., if the user was chatting with a contact via iMessage, the next ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-09T14:59:56Z | 2026-04-30T04:41:09Z | 2026-04-26T04:59:39Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | timharris707 | 263,555,515 | U_kgDOD7WJuw | User | false |
openclaw/openclaw | 4,046,643,845 | I_kwDOQb6kR87xMuKF | 41,264 | https://github.com/openclaw/openclaw/issues/41264 | https://api.github.com/repos/openclaw/openclaw/issues/41264 | [Bug]: Slack multi-account public-channel root mentions still fail for non-default personas (DM works) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In a Slack multi-account setup on OpenClaw 2026.3.7, root-channel explicit mentions for bound public personas (for example `@ceo`, `@manager`) do not start a persona session/run, while the default assistant persona in the same channel still s... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-09T16:37:50Z | 2026-04-30T04:41:08Z | 2026-04-26T04:59:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | babutree | 21,240,045 | MDQ6VXNlcjIxMjQwMDQ1 | User | false |
openclaw/openclaw | 4,047,077,988 | I_kwDOQb6kR87xOYJk | 41,329 | https://github.com/openclaw/openclaw/issues/41329 | https://api.github.com/repos/openclaw/openclaw/issues/41329 | Bug: Heartbeat tool_use blocks leak into parent session transcript, corrupting Anthropic sessions | ## Description
When a heartbeat agent runs on a non-Anthropic model (e.g. Grok 4.1 Fast via OpenRouter) and produces an alert that gets delivered to the parent session channel, the heartbeat's `tool_use` blocks leak into the parent session's transcript **without their corresponding `tool_result` blocks**. The next Ant... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-09T18:01:23Z | 2026-04-30T04:41:07Z | 2026-04-26T04:59:48Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BooRy95 | 257,679,890 | U_kgDOD1viEg | User | false |
openclaw/openclaw | 4,048,143,599 | I_kwDOQb6kR87xScTv | 41,472 | https://github.com/openclaw/openclaw/issues/41472 | https://api.github.com/repos/openclaw/openclaw/issues/41472 | tailscale.mode: serve kills WebSocket connections (incompatible with iOS clients) | ## Summary
`tailscale.mode: "serve"` (which proxies the gateway through Tailscale's HTTPS reverse proxy) kills WebSocket connections after a few seconds. This forces users to choose between Tailscale serve (HTTPS dashboard URL, BlueBubbles webhook compatibility) and working WebSocket connections for iOS clients like [... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-09T22:03:22Z | 2026-04-30T04:41:06Z | 2026-04-26T04:59:56Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | brandonshoreai | 260,288,022 | U_kgDOD4OuFg | User | false |
openclaw/openclaw | 4,048,271,948 | I_kwDOQb6kR87xS7pM | 41,487 | https://github.com/openclaw/openclaw/issues/41487 | https://api.github.com/repos/openclaw/openclaw/issues/41487 | providerOverride/modelOverride is not consistently honored at final execution | ## Summary
In some execution paths, a plugin can return `providerOverride` / `modelOverride`, but the final execution model still appears to differ from the override.
This makes request-time routing unreliable for plugins that depend on deterministic model selection.
## Why this matters
The current plugin API alrea... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-09T22:34:43Z | 2026-04-30T04:41:05Z | 2026-04-26T05:00:01Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | sparrowzhou | 21,187,527 | MDQ6VXNlcjIxMTg3NTI3 | User | false |
openclaw/openclaw | 4,048,493,537 | I_kwDOQb6kR87xTxvh | 41,530 | https://github.com/openclaw/openclaw/issues/41530 | https://api.github.com/repos/openclaw/openclaw/issues/41530 | Regression: subagent archiveAfterMinutes (default 60m) silently deletes session transcripts with no migration or changelog entry | ## Bug Report
### Version
- **Affected:** 2026.3.8
- **Previous (working):** 2026.3.7
### Description
The 2026.3.8 release introduced a subagent session auto-archive sweeper with a **60-minute default** (`archiveAfterMinutes ?? 60`). This feature:
1. **Silently deletes completed subagent session transcripts** (`del... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-09T23:36:12Z | 2026-04-30T04:41:04Z | 2026-04-26T05:00:05Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | AFunLS | 190,305,908 | U_kgDOC1fWdA | User | false |
openclaw/openclaw | 4,048,902,659 | I_kwDOQb6kR87xVVoD | 41,607 | https://github.com/openclaw/openclaw/issues/41607 | https://api.github.com/repos/openclaw/openclaw/issues/41607 | 飞书 Interactive Card 消息内容解析缺失 | ## 问题描述
飞书 Interactive Card 的内容在消息事件中未完整解析。
当前问题：
- OpenClaw 收到飞书卡片消息时，仅收到类型标识（如 ）
- 卡片内的具体内容（按钮文字、选项、表单数据等）未包含在推送事件中
- 需要配置卡片回调事件才能获取完整内容
## 期望行为
能够解析并展示 Interactive Card 的完整内容，包括：
- 卡片内的文本内容
- 按钮名称和值
- 下拉选项
- 表单输入值
## 技术说明
飞书 Interactive Card 通过消息事件回调推送，需要：
1. 配置卡片回调事件（）
2. 解析回调 payload 中的具体内容
## 影响范围
- 飞书消息处... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-10T02:01:56Z | 2026-04-30T04:41:04Z | 2026-04-26T05:00:09Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | JasonFang1993 | 55,202,939 | MDQ6VXNlcjU1MjAyOTM5 | User | false |
openclaw/openclaw | 4,048,983,052 | I_kwDOQb6kR87xVpQM | 41,642 | https://github.com/openclaw/openclaw/issues/41642 | https://api.github.com/repos/openclaw/openclaw/issues/41642 | Model fallback starvation: primary model timeout consumes entire session budget | ## Problem
When the primary model hangs (e.g. a third-party provider behind DashScope's model marketplace never responds), the model fallback loop in `runWithModelFallback` waits indefinitely for the primary attempt to fail. By the time the timeout error surfaces, the session-level timeout (`agents.defaults.timeoutSec... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-10T02:31:19Z | 2026-04-30T04:41:03Z | 2026-04-26T05:00:14Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | liuxiaoyusky | 49,766,325 | MDQ6VXNlcjQ5NzY2MzI1 | User | false |
openclaw/openclaw | 4,049,032,033 | I_kwDOQb6kR87xV1Nh | 41,654 | https://github.com/openclaw/openclaw/issues/41654 | https://api.github.com/repos/openclaw/openclaw/issues/41654 | Cron subagent API errors (e.g. thinking block rejection) leak as visible channel messages | ## Bug Description
When a cron job or subagent session encounters an LLM provider API error, the raw error message is delivered to the configured Discord channel as a regular visible message instead of being handled internally.
## Steps to Reproduce
1. Spawn a subagent with `thinking=high` using Anthropic Claude Son... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-10T02:48:42Z | 2026-04-30T04:41:02Z | 2026-04-26T05:00:18Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | CountofGlamorgan | 59,939,121 | MDQ6VXNlcjU5OTM5MTIx | User | false |
openclaw/openclaw | 4,049,505,460 | I_kwDOQb6kR87xXoy0 | 41,788 | https://github.com/openclaw/openclaw/issues/41788 | https://api.github.com/repos/openclaw/openclaw/issues/41788 | Google Chat DM threading broken - each response creates new thread | # Google Chat DM threading broken - each response creates new thread
## Description
In Google Chat DMs, OpenClaw responses create new threads instead of staying in the existing conversation thread. Each assistant reply spawns a separate thread, making conversations fragmented and hard to follow.
## Steps to Reproduce... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-10T05:31:21Z | 2026-04-30T04:41:01Z | 2026-04-26T05:00:22Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | shonbayer | 206,165,712 | U_kgDODEnW0A | User | false |
openclaw/openclaw | 4,049,632,312 | I_kwDOQb6kR87xYHw4 | 41,820 | https://github.com/openclaw/openclaw/issues/41820 | https://api.github.com/repos/openclaw/openclaw/issues/41820 | [Bug]: Axios requests fail with 400 error when HTTPS_PROXY is set (Plain HTTP request was sent to HTTPS port) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When system proxy environment variables (HTTP_PROXY, HTTPS_PROXY, ALL_PROXY) are configured, OpenClaw fails to probe the Feishu channel (and potentially other HTTPS endpoints). The request is incorrectly downgraded, causing the target serve... | closed | not_planned | true | 4 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-10T06:05:14Z | 2026-04-30T04:41:00Z | 2026-04-26T05:00:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | EterMT | 69,464,788 | MDQ6VXNlcjY5NDY0Nzg4 | User | false |
openclaw/openclaw | 4,226,678,960 | I_kwDOQb6kR8777gCw | 63,294 | https://github.com/openclaw/openclaw/issues/63294 | https://api.github.com/repos/openclaw/openclaw/issues/63294 | Custom providers cannot use Responses API reasoning params (multiple hardcoded guards) | ## Summary
When using a custom provider with `api: "openai-responses"`, reasoning parameters (`params.thinking: "high"`) are silently ignored. The request reaches the provider but with `reasoning: { effort: "none" }` instead of `{ effort: "high" }`.
This affects any custom provider (e.g. load balancers like litellm, ... | closed | completed | true | 1 | [] | [] | 2026-04-08T18:20:01Z | 2026-04-30T04:41:00Z | 2026-04-26T05:00:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | geegewu | 129,384,489 | U_kgDOB7ZAKQ | User | false |
openclaw/openclaw | 4,330,065,237 | I_kwDOQb6kR88AAAABAheNVQ | 71,973 | https://github.com/openclaw/openclaw/issues/71973 | https://api.github.com/repos/openclaw/openclaw/issues/71973 | Optimize prompt caching by moving volatile current-time injection out of cache-sensitive prompt prefixes | ## Summary
OpenClaw appears to inject a volatile `Current time: ...` line into heartbeat, cron, and post-compaction prompt content in a way that likely reduces prompt-cache reuse for those turns.
The main concern is not timezone metadata, but exact wall-clock time being inserted into cache-sensitive prompt prefixes.
... | open | null | false | 1 | [] | [] | 2026-04-26T05:01:17Z | 2026-05-03T02:15:15Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | williamcodes | 4,542,879 | MDQ6VXNlcjQ1NDI4Nzk= | User | false |
openclaw/openclaw | 4,330,071,027 | I_kwDOQb6kR88AAAABAhej8w | 71,974 | https://github.com/openclaw/openclaw/issues/71974 | https://api.github.com/repos/openclaw/openclaw/issues/71974 | Bug: WhatsApp channel health JSON reports running=false/connected=false while status --deep shows OK/LINKED | ## Summary
`openclaw health --json` reports the WhatsApp channel as:
- `statusState: linked`
- `running: false`
- `connected: false`
- `lastError: null`
At the same time, `openclaw status --deep` consistently reports the same channel as:
- `WhatsApp: ON / OK`
- `Health: LINKED`
- fresh auth age / linked state
This... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T05:05:02Z | 2026-04-29T04:37:17Z | 2026-04-26T10:45:33Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | helgekrause007 | 271,625,953 | U_kgDOEDCu4Q | User | false |
openclaw/openclaw | 4,330,072,427 | I_kwDOQb6kR88AAAABAhepaw | 71,975 | https://github.com/openclaw/openclaw/issues/71975 | https://api.github.com/repos/openclaw/openclaw/issues/71975 | [Feature]: Add LTX-Video support to video_generate | ### Summary
First-class support for Lightricks' LTX-Video models in the shared `video_generate` capability.
### Problem to solve
LTX-Video is one of the few open-weight video models with near-real-time generation (DiT-based, ~4–5s for a short clip on a single GPU) and strong text/image-to-video quality. It's current... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T05:05:55Z | 2026-04-29T04:35:04Z | 2026-04-26T16:17:19Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | alBronko | 19,271,745 | MDQ6VXNlcjE5MjcxNzQ1 | User | false |
openclaw/openclaw | 4,330,083,943 | I_kwDOQb6kR88AAAABAhfWZw | 71,977 | https://github.com/openclaw/openclaw/issues/71977 | https://api.github.com/repos/openclaw/openclaw/issues/71977 | [Feishu] Tool-call-heavy replies create multiple cards after streaming completes | ## Bug
When using Feishu with `streaming: true` and `renderMode: "card"`, replies that involve multiple tool calls produce multiple cards (one streaming card + N additional cards).
### Environment
- OpenClaw: 2026.4.24 (cbcfdf6)
- Feishu channel, WebSocket mode
- Config: `streaming: true`, `renderMode: "card"`, `bloc... | closed | completed | true | 1 | [] | [] | 2026-04-26T05:13:13Z | 2026-04-29T08:47:22Z | 2026-04-26T19:00:07Z | NONE | null | 20260429T090020Z | 2026-04-29T09:00:20Z | TwinsLee | 1,460,694 | MDQ6VXNlcjE0NjA2OTQ= | User | false |
openclaw/openclaw | 4,330,080,259 | I_kwDOQb6kR88AAAABAhfIAw | 71,976 | https://github.com/openclaw/openclaw/issues/71976 | https://api.github.com/repos/openclaw/openclaw/issues/71976 | Memory Dreaming: Light sleep sort prioritizes recency over recallCount, hiding real recall data | ## Bug 1: Light dreaming sort order (dreaming-Dmn9KbNB.js)
The light dreaming phase sorts entries by lastRecalledAt DESC first, then
ecallCount DESC. Because ingestDailyMemorySignals and ingestSessionTranscriptSignals refresh lastRecalledAt daily for all entries (with signalType: "daily"), the daily-injected entries ... | open | null | false | 3 | [] | [] | 2026-04-26T05:10:55Z | 2026-05-03T01:37:54Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | CSZ-w | 219,657,130 | U_kgDODRezqg | User | false |
openclaw/openclaw | 4,330,100,569 | I_kwDOQb6kR88AAAABAhgXWQ | 71,979 | https://github.com/openclaw/openclaw/issues/71979 | https://api.github.com/repos/openclaw/openclaw/issues/71979 | [Bug]: Custom avatar shows broken image in Control UI after v2026.4.24 (CSP blob: URL blocked) | ## Bug type
Control UI / avatar rendering (regression in v2026.4.24)
## Beta release blocker
No
## Summary
Avatar image shows as broken placeholder in Control UI after updating to v2026.4.24. The 401 error is fixed, but a new CSP (Content Security Policy) bug blocks the avatar from displaying.
## Steps to reproduce
... | closed | completed | true | 1 | [] | [] | 2026-04-26T05:23:15Z | 2026-04-29T04:37:16Z | 2026-04-26T10:45:38Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | wangzihe843-hash | 266,487,497 | U_kgDOD-JGyQ | User | false |
openclaw/openclaw | 4,207,216,477 | I_kwDOQb6kR876xQdd | 61,295 | https://github.com/openclaw/openclaw/issues/61295 | https://api.github.com/repos/openclaw/openclaw/issues/61295 | [Feature]: [Bug] --log-file and logging.file config ignored, logs always default to /tmp/openclaw/ | ### Summary
Type: Bug Report / Technical Analysis
Description:
Attempts to change the log file path via CLI arguments (--log-file), environment variables (OPENCLAW_LOG_FILE), or the openclaw.json configuration file are unsuccessful. The application consistently writes logs to the default path: /tmp/openclaw/openclaw-Y... | closed | completed | true | 2 | [
"enhancement"
] | [
"vincentkoc"
] | 2026-04-05T10:41:50Z | 2026-04-30T04:39:02Z | 2026-04-26T05:23:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zsjlovelike | 64,449,443 | MDQ6VXNlcjY0NDQ5NDQz | User | false |
openclaw/openclaw | 4,046,027,578 | I_kwDOQb6kR87xKXs6 | 41,146 | https://github.com/openclaw/openclaw/issues/41146 | https://api.github.com/repos/openclaw/openclaw/issues/41146 | [Bug]:Instalation hangs when there is no nodejs and linux build tools installed | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
It simply wont install on brand new Ubuntu 24.04 LTS
sudo apt update
sudo apt upgrade
curl -fsSL https://openclaw.ai/install.sh | bash
and it hangs forever on:
[1/3] Preparing environment
· Node.js not found, installing it now
· Installin... | closed | completed | true | 6 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-09T14:44:54Z | 2026-04-30T04:39:01Z | 2026-04-26T05:24:14Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | iht76 | 38,973,674 | MDQ6VXNlcjM4OTczNjc0 | User | false |
openclaw/openclaw | 4,330,110,784 | I_kwDOQb6kR88AAAABAhg_QA | 71,984 | https://github.com/openclaw/openclaw/issues/71984 | https://api.github.com/repos/openclaw/openclaw/issues/71984 | Server-side rendered status footer (model can't lie about runtime state) | ## Problem
LLM agents asked to append a status footer to every chat message (e.g. `📚 X% (Xk/200k) · 🧹 N compactions · 🧠 model`) cannot self-report runtime state reliably across long sessions. The model writes whatever number it last saw, last estimated, or last hallucinated. No prompt rule fixes this - we have thre... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T05:29:36Z | 2026-04-29T04:37:15Z | 2026-04-26T10:45:42Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | wearegoodthanks | 192,329,194 | U_kgDOC3a16g | User | false |
openclaw/openclaw | 4,330,093,661 | I_kwDOQb6kR88AAAABAhf8XQ | 71,978 | https://github.com/openclaw/openclaw/issues/71978 | https://api.github.com/repos/openclaw/openclaw/issues/71978 | Bundled runtime deps can recursively stage from the canonical stage root, generating openclaw-unknown-* and split-stage runtime resolution on 2026.4.24 | ## Summary
On a packaged 2026.4.24 install, bundled runtime dependency staging can deterministically generate an anomalous secondary stage directory like `~/.openclaw/plugin-runtime-deps/openclaw-unknown-a44f41aef101`.
This is not stale residue: the running Gateway can actively resolve modules from that `unknown` stag... | closed | completed | true | 1 | [] | [] | 2026-04-26T05:19:31Z | 2026-04-30T04:38:39Z | 2026-04-26T05:34:39Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xiao398008 | 211,473,177 | U_kgDODJrTGQ | User | false |
openclaw/openclaw | 4,330,109,424 | I_kwDOQb6kR88AAAABAhg58A | 71,982 | https://github.com/openclaw/openclaw/issues/71982 | https://api.github.com/repos/openclaw/openclaw/issues/71982 | cli-backend dispatch regression in 2026.4.24: claude-cli/* models route through synthetic-OAuth API transport instead of spawning subprocess | ## Summary
In OpenClaw 2026.4.24, `agents.defaults.model.primary = "claude-cli/claude-sonnet-4-6"` no longer triggers the cli-backend subprocess (`claude -p`). Instead the request is routed through the regular Anthropic provider transport using synthetic-OAuth auth (reading `~/.claude/.credentials.json` and calling Ant... | closed | completed | true | 3 | [] | [] | 2026-04-26T05:28:39Z | 2026-04-29T04:30:52Z | 2026-04-26T05:36:06Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | darrenfon | 209,860,188 | U_kgDODII2XA | User | false |
openclaw/openclaw | 4,329,994,264 | I_kwDOQb6kR88AAAABAhZ4GA | 71,946 | https://github.com/openclaw/openclaw/issues/71946 | https://api.github.com/repos/openclaw/openclaw/issues/71946 | agent: --thinking minimal rejected by openai-codex provider; default + retry causes per-turn latency | ## Summary
`openclaw agent` defaults thinking to `minimal` (and `--thinking minimal` is in `--help`), but the openai-codex responses API rejects it for `gpt-5.5` / `gpt-5.4-mini`:
```
Unsupported value: 'minimal' is not supported with the 'gpt-5.5' model.
Supported values are: 'none', 'low', 'medium', 'high', and 'xh... | closed | completed | true | 0 | [] | [] | 2026-04-26T04:11:59Z | 2026-04-30T04:38:36Z | 2026-04-26T05:40:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 1OAKDesign | 143,361,737 | U_kgDOCIuGyQ | User | false |
openclaw/openclaw | 4,330,129,522 | I_kwDOQb6kR88AAAABAhiIcg | 71,990 | https://github.com/openclaw/openclaw/issues/71990 | https://api.github.com/repos/openclaw/openclaw/issues/71990 | [Bug]: device.token.revoke skips caller-scope containment check, allowing any PAIRING_SCOPE client to revoke tokens for any device | ## Severity Assessment
### CVSS Assessment
| Metric | v3.1 | v4.0 |
|--------|------|------|
| **Score** | 8.1 / 10.0 | 7.2 / 10.0 |
| **Severity** | High | High |
| **Vector** | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:H | CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:H/VA:H/SC:N/SI:N/SA:N |
| **Calculator** | [CVSS v... | closed | completed | true | 1 | [] | [] | 2026-04-26T05:42:10Z | 2026-04-29T04:40:53Z | 2026-04-26T08:41:46Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | coygeek | 65,363,919 | MDQ6VXNlcjY1MzYzOTE5 | User | false |
openclaw/openclaw | 4,093,236,636 | I_kwDOQb6kR87z-dWc | 49,556 | https://github.com/openclaw/openclaw/issues/49556 | https://api.github.com/repos/openclaw/openclaw/issues/49556 | [Bug]: Installer script does not respect nvm Node version (detects system Node instead) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
I encountered an issue when installing OpenClaw on macOS where the installer script detects the system Node.js version instead of the nvm-managed version.
This causes the installer to incorrectly think that Node.js is outdated, even when a... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-18T07:15:00Z | 2026-04-30T04:38:35Z | 2026-04-26T05:43:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | heavenlxj | 4,516,605 | MDQ6VXNlcjQ1MTY2MDU= | User | false |
openclaw/openclaw | 4,330,139,279 | I_kwDOQb6kR88AAAABAhiujw | 71,992 | https://github.com/openclaw/openclaw/issues/71992 | https://api.github.com/repos/openclaw/openclaw/issues/71992 | [Bug]: Control UI webchat duplicates every assistant reply on 2026.4.21 — regression from #5964/#39469 | ### Bug description
Every assistant reply in the Control UI webchat appears **exactly twice** in the chat view. Full duplicate, not streaming fragments.
- Fresh install of OpenClaw 2026.4.21 via npm
- `/debug system-events off` is disabled by default and cannot be toggled
- Clearing Local Storage + hard refresh tempo... | open | null | false | 6 | [] | [] | 2026-04-26T05:47:47Z | 2026-05-02T23:57:47Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | rzhnrhjr6j-cloud | 277,363,469 | U_kgDOEIg7DQ | User | false |
openclaw/openclaw | 4,135,963,836 | I_kwDOQb6kR872hcy8 | 54,517 | https://github.com/openclaw/openclaw/issues/54517 | https://api.github.com/repos/openclaw/openclaw/issues/54517 | Bug: systemctl enable fails on Ubuntu Server - should use systemctl --user enable for user services | # Bug Report: `systemctl enable` fails on Ubuntu Server - should use `systemctl --user enable` for user services
## Bug Description
When installing OpenClaw Gateway on Ubuntu Server using the default installation method, the service installation fails with the following error:
```
Gateway service install failed: sys... | closed | completed | true | 2 | [] | [] | 2026-03-25T14:41:30Z | 2026-04-30T04:38:27Z | 2026-04-26T05:47:58Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 43622283 | 42,956,557 | MDQ6VXNlcjQyOTU2NTU3 | User | false |
openclaw/openclaw | 4,101,679,975 | I_kwDOQb6kR870eqtn | 50,443 | https://github.com/openclaw/openclaw/issues/50443 | https://api.github.com/repos/openclaw/openclaw/issues/50443 | Gateway ECONNREFUSED on macOS 12.7 + Node 22.x - node-macos-network-fix package not found | ## Environment
- **OS**: macOS 12.7.6 (21H1320)
- **Hardware**: Mac mini 7.1
- **Node.js**: 22.22.0 (managed by `n`)
- **OpenClaw**: 2026.3.7 (latest version not working)
## Problem
Getting `Gateway ECONNREFUSED` error when trying to connect to OpenClaw Gateway.
## Known Workaround
Found a community case (Discord use... | closed | not_planned | true | 3 | [] | [] | 2026-03-19T13:26:54Z | 2026-04-30T04:38:26Z | 2026-04-26T05:56:48Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jascharli | 259,474,958 | U_kgDOD3dGDg | User | false |
openclaw/openclaw | 4,182,641,485 | I_kwDOQb6kR875TgtN | 58,583 | https://github.com/openclaw/openclaw/issues/58583 | https://api.github.com/repos/openclaw/openclaw/issues/58583 | Log rotation: implement proper rotation instead of suppression at 500MB | ## Problem
The current log file cap is 500MB with suppression (stops writing) when reached. During an infinite retry loop (e.g. from an unresolvable model), logs fill to 500MB within minutes. Once suppressed, the gateway loses all diagnostic output and eventually crashes — making it impossible to debug what went wrong... | closed | completed | true | 1 | [] | [
"vincentkoc"
] | 2026-03-31T22:01:10Z | 2026-04-30T04:38:25Z | 2026-04-26T05:57:04Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jpeghead | 7,033,691 | MDQ6VXNlcjcwMzM2OTE= | User | false |
openclaw/openclaw | 4,216,327,509 | I_kwDOQb6kR877UA1V | 62,381 | https://github.com/openclaw/openclaw/issues/62381 | https://api.github.com/repos/openclaw/openclaw/issues/62381 | [Bug]: Subsystem File Logger Writes to Stale Date-Rolled Log File | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Subsystem File Logger Writes to Stale Date-Rolled Log File。
Root cause: createSubsystemLogger() caches its fileLogger (a tslog child logger) in a closure variable. Once assigned on first use, it is never refres... | closed | completed | true | 3 | [
"bug"
] | [
"vincentkoc"
] | 2026-04-07T08:05:11Z | 2026-04-30T04:38:25Z | 2026-04-26T05:57:16Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zhaoleink | 140,932,057 | U_kgDOCGZz2Q | User | false |
openclaw/openclaw | 4,330,107,559 | I_kwDOQb6kR88AAAABAhgypw | 71,981 | https://github.com/openclaw/openclaw/issues/71981 | https://api.github.com/repos/openclaw/openclaw/issues/71981 | [Bug]: Feishu channel fails to dispatch messages after upgrading from v2026.4.5 to v2026.4.24 — createScopedPairingAccess is not a function | ## Bug Description
After upgrading OpenClaw from v2026.4.5 to v2026.4.24, the Feishu channel fails to dispatch any messages. Both private messages (DM) and group messages are received but cannot be replied to.
## Error Message
```
feishu[default]: failed to dispatch message: TypeError: (0 , _feishu.createScopedPairi... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T05:27:22Z | 2026-04-30T04:38:24Z | 2026-04-26T05:57:33Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 8Sign24 | 73,635,964 | MDQ6VXNlcjczNjM1OTY0 | User | false |
openclaw/openclaw | 4,330,154,277 | I_kwDOQb6kR88AAAABAhjpJQ | 71,994 | https://github.com/openclaw/openclaw/issues/71994 | https://api.github.com/repos/openclaw/openclaw/issues/71994 | Gateway startup probes bundled WhatsApp runtime deps and can hang/delay restarts even when channels.whatsapp is not configured | ## Summary
On a packaged Linux/systemd install, Gateway startup can still probe the bundled WhatsApp channel runtime-deps path and invoke npm/git resolution **even when `channels.whatsapp` is absent from `openclaw.json`**.
Operationally, this makes restart look hung or badly delayed when the host cannot satisfy the Wh... | closed | completed | true | 1 | [] | [] | 2026-04-26T05:58:10Z | 2026-04-30T04:38:20Z | 2026-04-26T06:09:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xiao398008 | 211,473,177 | U_kgDODJrTGQ | User | false |
openclaw/openclaw | 4,330,113,557 | I_kwDOQb6kR88AAAABAhhKFQ | 71,986 | https://github.com/openclaw/openclaw/issues/71986 | https://api.github.com/repos/openclaw/openclaw/issues/71986 | [Bug]: 2026.4.24 is a huge messs! on vps gpt5.5 high. whatsapp and telegram gone | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
[Bug]: 2026.4.24 is a huge messs! on vps gpt5.5 high. whatsapp and telegram gone
Update goes in wrong folder. doctor failes, repeats install plugins, does nothing, gateway crashing.
### Steps to reproduce
yea just update t... | closed | not_planned | true | 10 | [
"bug",
"bug:crash"
] | [] | 2026-04-26T05:31:36Z | 2026-04-29T04:30:55Z | 2026-04-26T10:45:47Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jokedul | 54,840,321 | MDQ6VXNlcjU0ODQwMzIx | User | false |
openclaw/openclaw | 4,330,168,760 | I_kwDOQb6kR88AAAABAhkhuA | 71,996 | https://github.com/openclaw/openclaw/issues/71996 | https://api.github.com/repos/openclaw/openclaw/issues/71996 | image_generate ignores agents.defaults.imageGenerationModel.primary, uses dialog model provider instead | ## Bug
`image_generate` tool ignores `agents.defaults.imageGenerationModel.primary` config and selects provider based on the **dialog model**'s provider instead.
## Reproduction
Config:
```json
{
"agents": {
"defaults": {
"model": { "primary": "zai/glm-5-turbo" },
"imageGenerationModel": { "primary... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T06:06:56Z | 2026-04-29T04:37:14Z | 2026-04-26T10:46:02Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Gracker | 2,181,004 | MDQ6VXNlcjIxODEwMDQ= | User | false |
openclaw/openclaw | 4,330,184,274 | I_kwDOQb6kR88AAAABAhleUg | 71,999 | https://github.com/openclaw/openclaw/issues/71999 | https://api.github.com/repos/openclaw/openclaw/issues/71999 | Make typing indicator TTL configurable and increase default to 5 minutes | ## Summary
Long-running Telegram tasks can appear to have stopped because the typing indicator stops after about 2 minutes. Please consider increasing the default typing TTL to 5 minutes and making it configurable.
## Current behavior
Telegram `sendChatAction("typing")` indicators expire after a few seconds, so Open... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T06:16:10Z | 2026-04-29T04:37:14Z | 2026-04-26T10:47:22Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | suntp | 20,786,998 | MDQ6VXNlcjIwNzg2OTk4 | User | false |
openclaw/openclaw | 4,330,190,581 | I_kwDOQb6kR88AAAABAhl29Q | 72,002 | https://github.com/openclaw/openclaw/issues/72002 | https://api.github.com/repos/openclaw/openclaw/issues/72002 | WhatsApp per-peer sessions missing buildDirectChatContext delivery context injection | ## Summary
WhatsApp (and likely Signal/Telegram) per-peer direct sessions are systematically missing the `buildDirectChatContext` delivery context injection that Slack sessions receive. This causes the model to leak internal process narration into messages sent directly to contacts.
## Expected behavior
WhatsApp per... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T06:20:07Z | 2026-04-29T04:36:16Z | 2026-04-26T12:08:15Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | redasadki | 5,021,946 | MDQ6VXNlcjUwMjE5NDY= | User | false |
openclaw/openclaw | 4,199,606,126 | I_kwDOQb6kR876UOdu | 60,207 | https://github.com/openclaw/openclaw/issues/60207 | https://api.github.com/repos/openclaw/openclaw/issues/60207 | [Bug] Heartbeat and normal replies use parallel outbound paths without shared cross-source dedupe, causing duplicate / near-duplicate user-visible messages | ## Summary
OpenClaw can send duplicate or near-duplicate user-visible messages when a normal assistant reply and a heartbeat-generated reply happen around the same topic/time.
This is not just a config issue. The root cause appears to be a **structural outbound architecture gap**:
- `heartbeat` replies and `normal` ... | closed | completed | true | 1 | [] | [] | 2026-04-03T09:45:26Z | 2026-04-30T04:38:16Z | 2026-04-26T06:26:44Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | KimGLee | 150,593,189 | U_kgDOCPnepQ | User | false |
openclaw/openclaw | 4,330,191,129 | I_kwDOQb6kR88AAAABAhl5GQ | 72,003 | https://github.com/openclaw/openclaw/issues/72003 | https://api.github.com/repos/openclaw/openclaw/issues/72003 | WhatsApp per-peer sessions missing buildDirectChatContext delivery context injection | ## Summary
WhatsApp (and likely Signal/Telegram) per-peer direct sessions are systematically missing the `buildDirectChatContext` delivery context injection that Slack sessions receive. This causes the model to leak internal process narration into messages sent directly to contacts.
## Expected behavior
WhatsApp per... | closed | not_planned | true | 2 | [] | [] | 2026-04-26T06:20:29Z | 2026-04-29T04:36:16Z | 2026-04-26T12:08:19Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | redasadki | 5,021,946 | MDQ6VXNlcjUwMjE5NDY= | User | false |
openclaw/openclaw | 4,195,983,002 | I_kwDOQb6kR876GZ6a | 59,828 | https://github.com/openclaw/openclaw/issues/59828 | https://api.github.com/repos/openclaw/openclaw/issues/59828 | feat: Add YandexGPT (Yandex AI Studio) as native provider | ## 🚀 Feature Request: YandexGPT as Native Provider
**Author & Initiator:** [@Bullwinkle](https://github.com/Bullwinkle) — this idea, implementation, and PR are my initiative.
### What
Add [Yandex AI Studio](https://aistudio.yandex.ru/) (YandexGPT) as a first-class bundled provider in OpenClaw, alongside existing pro... | closed | completed | true | 1 | [
"r: third-party-extension"
] | [] | 2026-04-02T17:31:00Z | 2026-04-30T04:38:13Z | 2026-04-26T06:29:03Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | bullwinkle | 3,613,558 | MDQ6VXNlcjM2MTM1NTg= | User | false |
openclaw/openclaw | 4,330,223,044 | I_kwDOQb6kR88AAAABAhn1xA | 72,013 | https://github.com/openclaw/openclaw/issues/72013 | https://api.github.com/repos/openclaw/openclaw/issues/72013 | ACP startup identity reconcile warns on terminal one-shot sessions | ## Summary
Gateway startup currently tries to reconcile persisted ACP session identities even when those records are terminal one-shot ACP sessions. If the stored identity remains `pending`, startup logs a scary warning like:
```text
[gateway] acp startup identity reconcile (renderer=v1): checked=7 resolved=0 failed=... | open | null | false | 1 | [] | [] | 2026-04-26T06:39:07Z | 2026-05-02T20:43:17Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | scipher888 | 263,496,593 | U_kgDOD7SjkQ | User | false |
openclaw/openclaw | 4,330,225,634 | I_kwDOQb6kR88AAAABAhn_4g | 72,015 | https://github.com/openclaw/openclaw/issues/72015 | https://api.github.com/repos/openclaw/openclaw/issues/72015 | Reliability: active-memory blocks replies and QMD boot initialization can overload multi-agent gateways | # Reliability suggestion: active-memory blocks replies and QMD boot initialization can overload gateway in 2026.4.24
## Summary
On a multi-agent OpenClaw gateway, enabling the official `active-memory` plugin can make normal replies slow or unreliable. The main issue is that `active-memory` currently runs a full embed... | open | null | false | 5 | [] | [] | 2026-04-26T06:40:41Z | 2026-05-03T02:15:17Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | 0xCNAI | 210,723,999 | U_kgDODI9knw | User | false |
openclaw/openclaw | 4,330,228,695 | I_kwDOQb6kR88AAAABAhoL1w | 72,016 | https://github.com/openclaw/openclaw/issues/72016 | https://api.github.com/repos/openclaw/openclaw/issues/72016 | [Feature]: doctor api/extendability | ### Summary
extensibility on the doctors checks we run so that we can do custom environment specific checks to identify issues with the setup before hand and have it protentially fixed as well by a --fix command.
### Problem to solve
was looking to get a doctor extensibility thing so i can add custom checks on the o... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-04-26T06:42:02Z | 2026-05-02T22:59:01Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | pranitl | 73,791,278 | MDQ6VXNlcjczNzkxMjc4 | User | false |
openclaw/openclaw | 4,057,747,204 | I_kwDOQb6kR87x3E8E | 43,150 | https://github.com/openclaw/openclaw/issues/43150 | https://api.github.com/repos/openclaw/openclaw/issues/43150 | Config Write Race Condition Causes Lost Updates | ## Summary
When multiple config.patch calls are made in quick succession (within ~2 seconds), later writes can overwrite earlier writes because each patch is calculated based on a stale config snapshot.
## Impact
Severity: High - Configuration changes are silently lost
Affected Operations:
- gateway action=config.p... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T11:47:02Z | 2026-04-30T04:38:08Z | 2026-04-26T06:48:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | liuky1980 | 266,408,168 | U_kgDOD-EQ6A | User | false |
openclaw/openclaw | 4,160,885,899 | I_kwDOQb6kR874AhSL | 56,525 | https://github.com/openclaw/openclaw/issues/56525 | https://api.github.com/repos/openclaw/openclaw/issues/56525 | [Feature]: Support Browser View Port Config | ### Summary
The brower tools viewport defaults to 1280×720 and apparently there is no way to overwrite it persistently
### Problem to solve
I would like the browser to use at least Full HD 1920x1080 resolution
### Proposed solution
Expose it as a config variable via `browser.viewport` or similar
### Alternatives ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-28T17:08:41Z | 2026-04-30T04:38:07Z | 2026-04-26T06:50:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jaschaio | 11,058,666 | MDQ6VXNlcjExMDU4NjY2 | User | false |
openclaw/openclaw | 4,165,380,773 | I_kwDOQb6kR874Rqql | 57,228 | https://github.com/openclaw/openclaw/issues/57228 | https://api.github.com/repos/openclaw/openclaw/issues/57228 | Memory indexing fails with mxbai-embed-large on corpus >150KB | ## Problem
Memory indexing fails when corpus exceeds ~150KB with mxbai-embed-large embedding model.
## Error
```
Memory index failed (main): Ollama embeddings HTTP 500: {"error":"the input length exceeds the context length"}
```
## Steps to Reproduce
1. Accumulate memory files totaling >150KB across `memory/topics... | closed | completed | true | 1 | [] | [] | 2026-03-29T18:12:13Z | 2026-04-30T04:38:05Z | 2026-04-26T06:51:52Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | darrenbrain78 | 261,775,758 | U_kgDOD5phjg | User | false |
openclaw/openclaw | 4,330,261,864 | I_kwDOQb6kR88AAAABAhqNaA | 72,018 | https://github.com/openclaw/openclaw/issues/72018 | https://api.github.com/repos/openclaw/openclaw/issues/72018 | Control UI may resolve short model IDs to the wrong provider unless alias is defined | When selecting a model in Control UI, short model IDs without `alias` may be resolved to the wrong provider.
## Repro
My config includes models such as:
- `openai-codex/gpt-5.3-codex`
- `openai-codex/gpt-5.4`
The default provider/model path in my setup is under `minimax-portal`.
If I define:
```json
"openai-codex/g... | closed | completed | true | 1 | [] | [] | 2026-04-26T06:59:26Z | 2026-04-29T04:37:13Z | 2026-04-26T10:47:35Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | iamzhaoxin | 51,520,135 | MDQ6VXNlcjUxNTIwMTM1 | User | false |
openclaw/openclaw | 4,077,230,410 | I_kwDOQb6kR87zBZlK | 46,683 | https://github.com/openclaw/openclaw/issues/46683 | https://api.github.com/repos/openclaw/openclaw/issues/46683 | Slack interactive reply button clicks not routed to agent sessions | ## Summary
`[[slack_buttons:]]` and `[[slack_select:]]` directives now render correctly as Slack Block Kit elements (after fixes for #46647). However, when a user clicks a button, the interaction reaches the gateway but is **not dispatched to the originating agent session**.
## Steps to Reproduce
1. Enable `channels... | closed | not_planned | true | 3 | [] | [] | 2026-03-15T00:56:49Z | 2026-04-29T04:42:30Z | 2026-04-26T07:03:41Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jeremykoerber | 2,133,982 | MDQ6VXNlcjIxMzM5ODI= | User | false |
openclaw/openclaw | 4,330,202,508 | I_kwDOQb6kR88AAAABAhmljA | 72,008 | https://github.com/openclaw/openclaw/issues/72008 | https://api.github.com/repos/openclaw/openclaw/issues/72008 | feat: add Portkey AI gateway provider extension | ## Summary
Add a new bundled provider extension for [Portkey](https://portkey.ai), an AI gateway that provides a unified OpenAI-compatible API to 250+ LLM providers with observability, caching, fallbacks, load balancing, and cost tracking.
## Motivation
Portkey is a widely used AI gateway (similar to LiteLLM and Ope... | closed | completed | true | 1 | [
"r: third-party-extension"
] | [
"vincentkoc"
] | 2026-04-26T06:26:40Z | 2026-04-29T04:42:30Z | 2026-04-26T07:05:33Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Priyank295 | 67,854,862 | MDQ6VXNlcjY3ODU0ODYy | User | false |
openclaw/openclaw | 4,330,217,524 | I_kwDOQb6kR88AAAABAhngNA | 72,012 | https://github.com/openclaw/openclaw/issues/72012 | https://api.github.com/repos/openclaw/openclaw/issues/72012 | Bug: WebChat transcript prompt rewrite duplicates assistant/tool entries in 2026.4.24 | # Bug report: WebChat transcript prompt rewrite duplicates assistant/tool entries in 2026.4.24
## Summary
After upgrading to OpenClaw `2026.4.24`, WebChat / Control UI sessions can persist duplicate assistant responses in the JSONL transcript.
The duplicate is not a second model generation. The duplicated assistant ... | closed | completed | true | 1 | [] | [] | 2026-04-26T06:35:58Z | 2026-04-29T04:42:25Z | 2026-04-26T07:06:16Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | 0xCNAI | 210,723,999 | U_kgDODI9knw | User | false |
openclaw/openclaw | 4,208,539,206 | I_kwDOQb6kR8762TZG | 61,511 | https://github.com/openclaw/openclaw/issues/61511 | https://api.github.com/repos/openclaw/openclaw/issues/61511 | Feature request: native Sentry DSN support for gateway + agents | ## Problem
Operators running OpenClaw in production (Docker, VPS, WSL) have no visibility into crashes of the `openclaw` binary / `openclaw-gateway` process. When the gateway segfaults or throws an unhandled exception, the container restarts (docker `--restart=always`) but the stack trace is lost with the previous con... | closed | not_planned | true | 1 | [] | [] | 2026-04-05T20:44:50Z | 2026-04-29T04:42:24Z | 2026-04-26T07:07:28Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | spencerht24 | 160,431,832 | U_kgDOCY_-2A | User | false |
openclaw/openclaw | 4,330,239,926 | I_kwDOQb6kR88AAAABAho3tg | 72,017 | https://github.com/openclaw/openclaw/issues/72017 | https://api.github.com/repos/openclaw/openclaw/issues/72017 | [Bug]: Node 24 + Windows: dynamic `import()` of absolute paths fails with `ERR_UNSUPPORTED_ESM_URL_SCHEME | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On Windows under **Node.js v24**, multiple OpenClaw subsystems fail to load with:
```
Only URLs with a scheme in: file, data, and node are supported by the default ESM loader.
On Windows, absolute paths must be valid file:/... | closed | completed | true | 2 | [
"bug",
"duplicate",
"close:duplicate",
"regression"
] | [] | 2026-04-26T06:47:22Z | 2026-04-29T04:42:15Z | 2026-04-26T07:10:53Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | aemeathpa | 269,386,194 | U_kgDOEA6B0g | User | false |
openclaw/openclaw | 4,330,186,482 | I_kwDOQb6kR88AAAABAhlm8g | 72,000 | https://github.com/openclaw/openclaw/issues/72000 | https://api.github.com/repos/openclaw/openclaw/issues/72000 | Windows: ESM loader fails with absolute paths (Only URLs with scheme file, data, node are supported) | ## Bug Description
On Windows, OpenClaw 2026.4.24 fails to start channel plugins (e.g., feishu, discord) due to an ESM loader path issue. The error occurs when the dynamic \import()\ function receives a Windows absolute path starting with a drive letter (e.g., \E:\OpenClawData\...\).
## Error Message
\\\
Error [ERR_... | closed | completed | true | 1 | [
"duplicate",
"close:duplicate"
] | [] | 2026-04-26T06:17:27Z | 2026-04-29T04:42:12Z | 2026-04-26T07:11:05Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | int7899 | 258,010,246 | U_kgDOD2Dshg | User | false |
openclaw/openclaw | 4,330,181,930 | I_kwDOQb6kR88AAAABAhlVKg | 71,998 | https://github.com/openclaw/openclaw/issues/71998 | https://api.github.com/repos/openclaw/openclaw/issues/71998 | DeepSeek: thinking mode still broken in v4.24 (content[].thinking must be passed back) | ## Bug description
OpenClaw 2026.4.24 changelog states:
> DeepSeek thinking/replay behavior is fixed for follow-up tool-call turns.
However, the content[].thinking passing error persists. When DeepSeek V4 Pro (or V4 Flash) is used with think mode enabled, the provider rejects the request with a 400 error, causing a s... | closed | completed | true | 1 | [] | [] | 2026-04-26T06:14:43Z | 2026-04-29T04:41:52Z | 2026-04-26T07:18:26Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | CSZ-w | 219,657,130 | U_kgDODRezqg | User | false |
openclaw/openclaw | 4,330,297,950 | I_kwDOQb6kR88AAAABAhsaXg | 72,020 | https://github.com/openclaw/openclaw/issues/72020 | https://api.github.com/repos/openclaw/openclaw/issues/72020 | [Bug]: Signal status reactions do not run when messages.statusReactions.enabled is true | ## Bug type
Behavior bug (incorrect output/state without crash)
## Beta release blocker
No
## Summary
Signal does not emit lifecycle status reactions when `messages.statusReactions.enabled` is true because the Signal inbound handler never wires the status reaction controller into dispatch callbacks.
## Steps to r... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T07:18:41Z | 2026-04-29T04:35:03Z | 2026-04-26T16:17:23Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Hua688 | 5,764,877 | MDQ6VXNlcjU3NjQ4Nzc= | User | false |
openclaw/openclaw | 4,330,298,498 | I_kwDOQb6kR88AAAABAhscgg | 72,021 | https://github.com/openclaw/openclaw/issues/72021 | https://api.github.com/repos/openclaw/openclaw/issues/72021 | Short-term promotion: signalCount mixes daily/session signals with real recalls | ## Bug description
Short-term memory promotion (`rankShortTermPromotionCandidates`) uses `totalSignalCountForEntry` to gate minRecallCount, but that function sums recallCount + dailyCount + groundedCount. Daily and session signals (from ingestDailyMemorySignals / ingestSessionTranscriptSignals) are counted equally wit... | open | null | false | 1 | [] | [] | 2026-04-26T07:19:01Z | 2026-05-03T06:07:33Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | CSZ-w | 219,657,130 | U_kgDODRezqg | User | false |
openclaw/openclaw | 4,330,338,012 | I_kwDOQb6kR88AAAABAhu23A | 72,034 | https://github.com/openclaw/openclaw/issues/72034 | https://api.github.com/repos/openclaw/openclaw/issues/72034 | [Bug] Feishu TTS auto:inbound not triggering - missing inboundAudio check in dispatch | ## Summary
When `messages.tts.auto` is set to `"inbound"` on Feishu channel, the system should respond with voice when receiving audio messages, but it responds with text instead. No TTS synthesis attempt is made.
## Environment
- OpenClaw: 2026.4.24
- Channel: Feishu (Lark)
- TTS config: `messages.tts.auto = "inbou... | closed | completed | true | 2 | [] | [] | 2026-04-26T07:43:02Z | 2026-04-29T04:37:12Z | 2026-04-26T10:47:48Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | motor820 | 36,178,147 | MDQ6VXNlcjM2MTc4MTQ3 | User | false |
openclaw/openclaw | 4,330,357,973 | I_kwDOQb6kR88AAAABAhwE1Q | 72,042 | https://github.com/openclaw/openclaw/issues/72042 | https://api.github.com/repos/openclaw/openclaw/issues/72042 | [Bug]: Gateway startup fails after update to 2026.4.24 — ERR_MODULE_NOT_FOUND, 1617 dist files missing from installed package (postinstall pruning bug) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After updating to OpenClaw 2026.4.24 (from 2026.4.23), the gateway fails to start with `ERR_MODULE_NOT_FOUND`. Investigation reveals that the npm tarball contains 4116 `.js` files in `dist/`, but after installation only 2499... | closed | completed | true | 1 | [] | [] | 2026-04-26T07:55:04Z | 2026-04-29T04:37:10Z | 2026-04-26T10:48:04Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jwong-art | 269,971,191 | U_kgDOEBdu9w | User | false |
openclaw/openclaw | 4,330,368,020 | I_kwDOQb6kR88AAAABAhwsFA | 72,043 | https://github.com/openclaw/openclaw/issues/72043 | https://api.github.com/repos/openclaw/openclaw/issues/72043 | feat: add gateway.systemEvents.recoveryNotice config knob to suppress recovery prompt for customer-facing bots | ## Summary
After a whole-file `last-known-good` config recovery, the Gateway pushes
`Gateway recovered automatically after a failed config change and restored the last known good configuration.`
into the next agent turn (as a `System:` prefix via `enqueueSystemEvent`) AND delivers it out-of-band on the originating cha... | open | null | false | 1 | [] | [] | 2026-04-26T08:01:13Z | 2026-05-03T07:04:53Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | MaxK000 | 96,078,504 | U_kgDOBboKqA | User | false |
openclaw/openclaw | 4,330,370,636 | I_kwDOQb6kR88AAAABAhw2TA | 72,044 | https://github.com/openclaw/openclaw/issues/72044 | https://api.github.com/repos/openclaw/openclaw/issues/72044 | DeepSeek-v4-pro thinking mode breaks on multi-turn tool-call flows (4.24 fix is incomplete) | ## Summary
DeepSeek-v4-pro with thinking-level `high` consistently fails on multi-turn tool-call flows (e.g. `sessions_spawn` subagent paths) with a 400 error from the DeepSeek server. The agent then falls back to the configured fallback model (e.g. `openai-codex/gpt-5.4`), so the user never gets DeepSeek-quality outp... | closed | completed | true | 4 | [] | [] | 2026-04-26T08:02:41Z | 2026-05-03T04:35:31Z | 2026-04-30T18:45:22Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | yuxiaoyang2007-prog | 258,562,502 | U_kgDOD2lZxg | User | false |
openclaw/openclaw | 4,330,377,784 | I_kwDOQb6kR88AAAABAhxSOA | 72,048 | https://github.com/openclaw/openclaw/issues/72048 | https://api.github.com/repos/openclaw/openclaw/issues/72048 | memory-core: narrative session cleanup fails with 'missing scope: operator.admin' | # Bug Report: memory-core narrative session cleanup fails due to missing operator.admin scope
## Environment
- **OpenClaw Version**: 2026.4.23 (a979721)
- **Node.js**: v22.22.2
- **OS**: Linux 6.8.0-110-generic (x64)
- **Plugin**: memory-core (bundled with OpenClaw)
## Bug Description
The `memory-core` plugin's drea... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T08:06:39Z | 2026-04-29T04:37:10Z | 2026-04-26T10:48:17Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | HankSU7889 | 32,285,831 | MDQ6VXNlcjMyMjg1ODMx | User | false |
openclaw/openclaw | 4,144,588,744 | I_kwDOQb6kR873CWfI | 55,215 | https://github.com/openclaw/openclaw/issues/55215 | https://api.github.com/repos/openclaw/openclaw/issues/55215 | [Feature]: openclaw webchat can't view image, when AI response one image url with in markdown style | ### Summary
让我再次尝试直接显示图片：
<img width="690" height="160" alt="Image" src="https://github.com/user-attachments/assets/5381a12d-5eb4-461d-b4c9-9618a712747f" />
<img width="1212" height="998" alt="Image" src="https:... | closed | completed | true | 5 | [
"enhancement"
] | [] | 2026-03-26T14:39:59Z | 2026-04-29T04:41:35Z | 2026-04-26T08:18:02Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | wuweiqi1993 | 32,814,990 | MDQ6VXNlcjMyODE0OTkw | User | false |
openclaw/openclaw | 4,145,158,658 | I_kwDOQb6kR873EhoC | 55,252 | https://github.com/openclaw/openclaw/issues/55252 | https://api.github.com/repos/openclaw/openclaw/issues/55252 | [Feature]: suppressAssistantText config for Slack channels | ### Summary
Add a `suppressAssistantText` boolean config option to suppress all regular assistant text output in Slack, leaving only explicit `message` tool sends visible.
### Problem to solve
In interactive Slack sessions, agents narrate their intermediate reasoning ("Let me check...", "Got it.", "Now let me create... | closed | not_planned | true | 1 | [] | [] | 2026-03-26T15:45:34Z | 2026-04-29T04:41:34Z | 2026-04-26T08:18:07Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | SahilAujla | 83,442,423 | MDQ6VXNlcjgzNDQyNDIz | User | false |
openclaw/openclaw | 4,147,888,664 | I_kwDOQb6kR873O8IY | 55,409 | https://github.com/openclaw/openclaw/issues/55409 | https://api.github.com/repos/openclaw/openclaw/issues/55409 | Nostr channel: WebSocket subscriptions fail with timeout/connection refused | ## Bug Description
The Nostr channel plugin fails to maintain WebSocket subscriptions. It connects to relays successfully, but subscriptions close immediately with timeout/connection errors, causing an infinite restart loop.
## Environment
- OpenClaw version: 2026.3.2
- Plugin: @openclaw/nostr@2026.3.2
- nostr-tools... | closed | not_planned | true | 1 | [] | [] | 2026-03-26T23:04:50Z | 2026-04-29T04:41:33Z | 2026-04-26T08:18:17Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | luminexo | 268,188,294 | U_kgDOD_w6hg | User | false |
openclaw/openclaw | 4,168,409,678 | I_kwDOQb6kR874dOJO | 57,569 | https://github.com/openclaw/openclaw/issues/57569 | https://api.github.com/repos/openclaw/openclaw/issues/57569 | [Feature Request] health-monitor: channel-level reconnect should preserve Anthropic prompt cache | ## Problem
When a channel (Discord, Telegram, etc.) goes stale, the OpenClaw health-monitor correctly restarts only the channel connector (via `stopChannel` + `startChannel`). However, this still **invalidates the Anthropic prompt cache**, which is process-level.
### Real-world impact
- Normal operation: ~500 new to... | closed | not_planned | true | 2 | [] | [] | 2026-03-30T07:52:25Z | 2026-04-29T04:41:32Z | 2026-04-26T08:20:31Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | serhiizghama | 20,826,225 | MDQ6VXNlcjIwODI2MjI1 | User | false |
openclaw/openclaw | 4,330,401,514 | I_kwDOQb6kR88AAAABAhyu6g | 72,052 | https://github.com/openclaw/openclaw/issues/72052 | https://api.github.com/repos/openclaw/openclaw/issues/72052 | Telegram restarts should not emit silent-reply filler; expose staged readiness | ## Summary
When OpenClaw restarts the gateway from inside an active Telegram direct-message turn, the turn can be terminated before a normal final response is delivered. Telegram delivery then treats the turn as having ended without a visible final response, creates a `NO_REPLY` fallback, and the default direct-chat s... | closed | completed | true | 1 | [] | [] | 2026-04-26T08:19:47Z | 2026-04-29T04:41:31Z | 2026-04-26T08:20:36Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | mikekarnj | 85,970 | MDQ6VXNlcjg1OTcw | User | false |
openclaw/openclaw | 4,175,530,515 | I_kwDOQb6kR8744YoT | 58,117 | https://github.com/openclaw/openclaw/issues/58117 | https://api.github.com/repos/openclaw/openclaw/issues/58117 | [Feature]: Mistral Voxtral TTS provider support | ### Summary
Add Mistral Voxtral as a supported TTS provider
### Problem to solve
Currently OpenClaw only supports elevenlabs, openai, and edge
as TTS providers. Mistral released Voxtral TTS on March 26 2026
which is 4-7x cheaper than ElevenLabs ($16/M vs $60-120/M chars)
and supports zero-shot voice cloning from ... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-31T03:51:27Z | 2026-04-29T04:41:32Z | 2026-04-26T08:20:36Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | scottbrackstone | 235,456,102 | U_kgDODgjGZg | User | false |
openclaw/openclaw | 4,195,095,167 | I_kwDOQb6kR876DBJ_ | 59,739 | https://github.com/openclaw/openclaw/issues/59739 | https://api.github.com/repos/openclaw/openclaw/issues/59739 | Feature: option to route specific group chats to main agent session | ## Problem
Currently, all Telegram group messages are routed to isolated sessions, separate from the main DM session. There is no way to opt specific groups into the main session context.
This means the agent in a group chat has no access to the conversation history from DMs (and vice-versa), even when the group is a... | closed | not_planned | true | 2 | [] | [] | 2026-04-02T14:45:02Z | 2026-04-29T04:41:30Z | 2026-04-26T08:20:52Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | ikenraf | 214,547,488 | U_kgDODMm8IA | User | false |
openclaw/openclaw | 4,199,344,679 | I_kwDOQb6kR876TOon | 60,188 | https://github.com/openclaw/openclaw/issues/60188 | https://api.github.com/repos/openclaw/openclaw/issues/60188 | CLI --profile flag does not propagate gateway.port to RPC probe/service discovery | ## Bug Description
When running `openclaw --profile matt <command>`, the CLI correctly sets `OPENCLAW_CONFIG_PATH` and `OPENCLAW_STATE_DIR` to the profile's directories, but gateway-interacting commands (`gateway status`, `pairing list`, `pairing approve`, etc.) still probe the **main gateway** on the default port (18... | closed | completed | true | 2 | [] | [] | 2026-04-03T08:56:16Z | 2026-04-29T04:41:29Z | 2026-04-26T08:20:56Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | sovereignmedia | 262,008,899 | U_kgDOD53wQw | User | false |
openclaw/openclaw | 4,330,344,102 | I_kwDOQb6kR88AAAABAhvOpg | 72,037 | https://github.com/openclaw/openclaw/issues/72037 | https://api.github.com/repos/openclaw/openclaw/issues/72037 | [Bug]: Codex ACP falls back to `npm exec` (`acpx` bundle issue) | ## Bug type
Bug (packaging regression in `acpx`?)
## Beta release blocker
No
## Summary
On my openclaw 2026.4.24, every Codex ACP startup shells out through `npm exec` rather than running the package's own binary. `@zed-industries/codex-acp` dep is not present so the npx fallback fires and adds significant SD Card... | closed | completed | true | 4 | [] | [] | 2026-04-26T07:46:42Z | 2026-04-30T04:31:59Z | 2026-04-28T04:03:27Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jasonftl | 175,325,999 | U_kgDOCnNDLw | User | false |
openclaw/openclaw | 4,135,232,606 | I_kwDOQb6kR872eqRe | 54,467 | https://github.com/openclaw/openclaw/issues/54467 | https://api.github.com/repos/openclaw/openclaw/issues/54467 | Bonjour: no config key for per-instance mDNS service name (multi-gateway same-host) | ## Description
Running two gateway instances on the same macOS host (e.g., two AI agents sharing a Mac Mini). Both register as `<ComputerName> (OpenClaw)` via Bonjour — the second gets renamed to `(2)` on every restart.
## Root Cause
`startGatewayBonjourAdvertiser()` derives the service name from `scutil --get Compu... | closed | not_planned | true | 2 | [] | [] | 2026-03-25T12:56:36Z | 2026-04-29T04:38:17Z | 2026-04-26T09:49:50Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | markthebest12 | 6,416,874 | MDQ6VXNlcjY0MTY4NzQ= | User | false |
openclaw/openclaw | 4,221,622,876 | I_kwDOQb6kR877oNpc | 62,839 | https://github.com/openclaw/openclaw/issues/62839 | https://api.github.com/repos/openclaw/openclaw/issues/62839 | [Bug]: openclaw update persists HTTP_PROXY into systemd service, breaking Feishu integration | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading via `openclaw update`, Feishu integration stops working. Gateway requests are routed through a local HTTP proxy due to persisted HTTP_PROXY in systemd service.
This is likely not Feishu-specific. Any integra... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-08T02:09:40Z | 2026-04-30T04:35:17Z | 2026-04-27T09:46:57Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | richardjung-R | 52,922,907 | MDQ6VXNlcjUyOTIyOTA3 | User | false |
openclaw/openclaw | 4,330,351,006 | I_kwDOQb6kR88AAAABAhvpng | 72,040 | https://github.com/openclaw/openclaw/issues/72040 | https://api.github.com/repos/openclaw/openclaw/issues/72040 | [Bug]: Telegram channel crash on Windows 11 — ERR_UNSUPPORTED_ESM_URL_SCHEME (Received protocol 'c:') in 2026.4.24, grammY deps missing out-of-the-box | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On Windows 11, the Telegram channel provider crashes immediately on startup in 2026.4.24 with `ERR_UNSUPPORTED_ESM_URL_SCHEME` (`Received protocol 'c:'`), making the channel completely unusable. Additionally, grammY and rela... | closed | completed | true | 6 | [
"bug",
"regression"
] | [] | 2026-04-26T07:50:57Z | 2026-04-29T04:35:54Z | 2026-04-26T10:48:00Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | centralpc | 59,361,530 | MDQ6VXNlcjU5MzYxNTMw | User | false |
openclaw/openclaw | 4,330,414,535 | I_kwDOQb6kR88AAAABAhzhxw | 72,057 | https://github.com/openclaw/openclaw/issues/72057 | https://api.github.com/repos/openclaw/openclaw/issues/72057 | [Bug]: proxy not working on google llm | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When I update openclaw container from ver 4.05 to 4.23, I find my google llm not work and the following networking error.
I have setup HTTPS_PROXY and https_proxy in the compose file also works on curl https://generativelang... | closed | duplicate | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-26T08:27:16Z | 2026-04-30T04:35:50Z | 2026-04-27T07:53:15Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | robbiewu003 | 28,888,023 | MDQ6VXNlcjI4ODg4MDIz | User | false |
openclaw/openclaw | 4,330,341,201 | I_kwDOQb6kR88AAAABAhvDUQ | 72,036 | https://github.com/openclaw/openclaw/issues/72036 | https://api.github.com/repos/openclaw/openclaw/issues/72036 | Feishu plugin v2026.4.20 fails to start: bot ping API returns 404 or timeout | ## Problem
After upgrading to OpenClaw 4.24, the Feishu channel plugin (v2026.4.20) fails to initialize. The gateway starts but all Feishu accounts fail with:
```
AxiosError: timeout of 10000ms exceeded
url: 'https://open.feishu.cn/open-apis/bot/v1/openclaw_bot/ping'
```
or
```
404 Not Found
url: 'https://open... | closed | completed | true | 2 | [] | [] | 2026-04-26T07:44:55Z | 2026-04-29T04:37:11Z | 2026-04-26T10:47:51Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | fvsgj | 154,893,352 | U_kgDOCTt8KA | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.