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 | 4,196,500,282 | I_kwDOQb6kR876IYM6 | 59,871 | https://github.com/openclaw/openclaw/issues/59871 | https://api.github.com/repos/openclaw/openclaw/issues/59871 | [Bug]: [Bug] TUI renders internal command envelopes to operator; async completion events inject verbose agent instructions into session transcript | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Two related but distinct bugs were found by inspecting the compiled TUI bundle (dist/tui-Da-djW_y.js) and the strip-inbound-meta module.
## Bug 1 — TUI Display Layer: Command envelopes rendered to operator
**... | closed | not_planned | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-02T19:16:45Z | 2026-04-28T04:34:18Z | 2026-04-26T04:19:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Ncross97 | 273,191,801 | U_kgDOEEiTeQ | User | false |
openclaw/openclaw | 4,196,522,704 | I_kwDOQb6kR876IdrQ | 59,876 | https://github.com/openclaw/openclaw/issues/59876 | https://api.github.com/repos/openclaw/openclaw/issues/59876 | [Feature]: Include retry/reset time in rate limit error messages | ### Summary
Rate limit errors should show the reset/wait time instead of just "try again later."
### Problem to solve
When the upstream model provider (e.g. Anthropic subscription) returns a rate limit error, the user-facing message is:
:warning: API rate limit reached. Please try again later.
OpenClaw already t... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-02T19:22:10Z | 2026-04-28T04:34:17Z | 2026-04-26T04:19:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | arnagar | 122,839,331 | U_kgDOB1JhIw | User | false |
openclaw/openclaw | 4,196,890,333 | I_kwDOQb6kR876J3bd | 59,905 | https://github.com/openclaw/openclaw/issues/59905 | https://api.github.com/repos/openclaw/openclaw/issues/59905 | [Bug]: Intermittent execution mismatch after update: main agent acknowledges actions but does not execute (often after browser/CDP contention) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After update, the main agent intermittently enters a broken state where it acknowledges requested actions but does not actually execute them (or does not return concrete execution output).
This tends to appear after browser/... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-02T20:44:17Z | 2026-04-28T04:34:16Z | 2026-04-26T04:19:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | castorle7-rgb | 263,298,182 | U_kgDOD7Gchg | User | false |
openclaw/openclaw | 4,197,076,908 | I_kwDOQb6kR876Kk-s | 59,917 | https://github.com/openclaw/openclaw/issues/59917 | https://api.github.com/repos/openclaw/openclaw/issues/59917 | CI/CD: tools-automation cron job failed to send build status to Discord | The tools-automation cron job failed to send build status to Discord. The error was:
```
{"message": "Unknown Webhook", "code": 10015}
``` | closed | not_planned | true | 1 | [] | [] | 2026-04-02T21:29:46Z | 2026-04-28T04:34:16Z | 2026-04-26T04:19:25Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | dboone323 | 65,318,437 | MDQ6VXNlcjY1MzE4NDM3 | User | false |
openclaw/openclaw | 4,197,082,331 | I_kwDOQb6kR876KmTb | 59,918 | https://github.com/openclaw/openclaw/issues/59918 | https://api.github.com/repos/openclaw/openclaw/issues/59918 | [Bug]: CLI JSONL fallback parser drops final-answer semantics and can surface commentary instead | ### Summary
In a fallback run using an OpenAI/Codex-style CLI backend with `output: "jsonl"`, OpenClaw can surface commentary/intermediate text instead of the terminal answer.
This looks adjacent to the broader “text between tool calls leaks to messaging channels” issue family, but narrower: in the captured fallback ... | closed | not_planned | true | 2 | [] | [] | 2026-04-02T21:31:01Z | 2026-04-28T04:34:15Z | 2026-04-26T04:19:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mySebbe | 14,988,278 | MDQ6VXNlcjE0OTg4Mjc4 | User | false |
openclaw/openclaw | 4,197,369,615 | I_kwDOQb6kR876LscP | 59,942 | https://github.com/openclaw/openclaw/issues/59942 | https://api.github.com/repos/openclaw/openclaw/issues/59942 | status/doctor reports memory unavailable while memory-lancedb-pro appears registered at runtime | ## Summary
After upgrading to OpenClaw 2026.4.2, the runtime logs suggest `memory-lancedb-pro` is successfully loaded and registered, but CLI health/status commands still report memory as unavailable/inactive.
This may be a mismatch between runtime plugin registration and the logic used by `openclaw status` / `opencl... | closed | not_planned | true | 1 | [] | [] | 2026-04-02T22:58:09Z | 2026-04-28T04:34:14Z | 2026-04-26T04:19:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lamkan0210 | 65,824,253 | MDQ6VXNlcjY1ODI0MjUz | User | false |
openclaw/openclaw | 4,198,737,375 | I_kwDOQb6kR876Q6Xf | 60,111 | https://github.com/openclaw/openclaw/issues/60111 | https://api.github.com/repos/openclaw/openclaw/issues/60111 | [Bug]: `generic_repeat` loop detector never escalates to blocking — `criticalThreshold` has no effect | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The `generic_repeat` loop detector only checks `warningThreshold` and always returns `level: "warning"`. It never checks `criticalThreshold`, so the most common loop pattern (same tool, same arguments, repeated... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-03T06:38:16Z | 2026-04-28T04:34:12Z | 2026-04-26T04:21:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lykeion-dev | 140,231,077 | U_kgDOCFvBpQ | User | false |
openclaw/openclaw | 4,199,066,931 | I_kwDOQb6kR876SK0z | 60,145 | https://github.com/openclaw/openclaw/issues/60145 | https://api.github.com/repos/openclaw/openclaw/issues/60145 | [Bug]: security audit symlink_escape probe uses wrong boundary (workspace root vs skills directory) | ## Description
The `skills.workspace.symlink_escape` audit probe checks whether skill file realpaths escape the **workspace root** (`~/.openclaw/workspace/`), but the skill loader (`resolveContainedSkillPath` in `src/agents/skills/workspace.ts`) checks against the **skills directory root** (`~/.openclaw/workspace/skil... | closed | not_planned | true | 1 | [] | [] | 2026-04-03T07:53:42Z | 2026-04-28T04:34:12Z | 2026-04-26T04:21:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | menhguin | 17,287,724 | MDQ6VXNlcjE3Mjg3NzI0 | User | false |
openclaw/openclaw | 4,199,259,218 | I_kwDOQb6kR876S5xS | 60,165 | https://github.com/openclaw/openclaw/issues/60165 | https://api.github.com/repos/openclaw/openclaw/issues/60165 | [功能建议] exec 工具应自动加载用户的 shell 配置文件（如 ~/.zshrc） | ## 问题描述
OpenClaw 的 exec 工具在执行命令时，shell 环境没有自动加载用户的 `~/.zshrc` 配置文件，导致 nvm/rbenv/pyenv 等版本管理器管理的工具无法被正确识别。
## 复现步骤
1. 用户通过 nvm 安装了 node（如 v20.19.5）
2. 在用户自己的终端中 `node --version` 正常输出
3. OpenClaw 的 exec 工具执行 `which node` 返回「未找到」
4. 需要手动 `source ~/.nvm/nvm.sh` 或硬编码 PATH 才能使用
## 期望行为
exec 工具应该以「登录 shell」方式启动（类似 `zsh -... | closed | completed | true | 1 | [] | [] | 2026-04-03T08:38:39Z | 2026-04-28T04:34:11Z | 2026-04-26T04:21:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yuelei118 | 75,592,721 | MDQ6VXNlcjc1NTkyNzIx | User | false |
openclaw/openclaw | 4,199,461,553 | I_kwDOQb6kR876TrKx | 60,195 | https://github.com/openclaw/openclaw/issues/60195 | https://api.github.com/repos/openclaw/openclaw/issues/60195 | [openclaw-gmail] health-monitor restarts entire plugin when "default" account key exists in accounts config | ## Bug Description
The `openclaw-gmail` channel plugin is silently killed every 5 minutes by the health-monitor when a `"default"` key exists in `channels.openclaw-gmail.accounts`. This is a common config pattern created by the onboarding wizard or early setup.
## Root Cause
The health-monitor checks `openclaw-gmail... | closed | completed | true | 2 | [] | [] | 2026-04-03T09:17:41Z | 2026-04-28T04:34:10Z | 2026-04-26T04:21:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sabo961 | 191,984 | MDQ6VXNlcjE5MTk4NA== | User | false |
openclaw/openclaw | 4,200,212,634 | I_kwDOQb6kR876Wiia | 60,261 | https://github.com/openclaw/openclaw/issues/60261 | https://api.github.com/repos/openclaw/openclaw/issues/60261 | [Bug]: Xiaomi/MiMo models output to reasoning_content instead of content field | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
## Problem
When using Xiaomi MiMo models (mimo-v2-pro, mimo-v2-omni) via the Token Plan API, the model outputs text to the `reasoning_content` field instead of the standard `content` field in both streaming and non-streaming re... | closed | not_planned | true | 1 | [
"bug",
"bug:crash"
] | [] | 2026-04-03T11:48:05Z | 2026-04-28T04:34:10Z | 2026-04-26T04:21:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Uitatsu | 128,948,791 | U_kgDOB6-aNw | User | false |
openclaw/openclaw | 4,200,574,647 | I_kwDOQb6kR876X663 | 60,312 | https://github.com/openclaw/openclaw/issues/60312 | https://api.github.com/repos/openclaw/openclaw/issues/60312 | OpenClaw Subagent Spawning Failure - Gateway/Pairing Connectivity Issue | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
Subagent Spawning Failure - Gateway/Pairing Connectivity
**Priority:** Medium
## ISSUE SUMMARY
Attempted to spawn a persistent subagent session named "Emma Clarke" but encountered a gateway/pairing connectivity issue that pre... | closed | completed | true | 1 | [
"bug",
"bug:crash"
] | [] | 2026-04-03T13:31:05Z | 2026-04-28T04:34:09Z | 2026-04-26T04:21:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | shearamariz | 79,451,126 | MDQ6VXNlcjc5NDUxMTI2 | User | false |
openclaw/openclaw | 4,200,649,170 | I_kwDOQb6kR876YNHS | 60,321 | https://github.com/openclaw/openclaw/issues/60321 | https://api.github.com/repos/openclaw/openclaw/issues/60321 | [Feature]: [Windows] gateway start resets Scheduled Task to interactive mode (console window pops up) | ### Summary
`openclaw gateway start` on Windows should run the gateway in background mode without popping up a console window
### Problem to solve
On Windows, `openclaw gateway start` registers a Scheduled Task with `LogonType=InteractiveToken` (interactive only mode). This causes a console window to pop up every ti... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-03T13:52:09Z | 2026-04-28T04:34:08Z | 2026-04-26T04:21:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | loco777bot | 273,363,354 | U_kgDOEEsxmg | User | false |
openclaw/openclaw | 4,201,030,797 | I_kwDOQb6kR876ZqSN | 60,377 | https://github.com/openclaw/openclaw/issues/60377 | https://api.github.com/repos/openclaw/openclaw/issues/60377 | [Bug]: Google Chat 401 after update — crash-loop + missing appPrincipal (two compounding issues) | ## Environment
- OpenClaw version: 2026.4.2
- OS: Arch Linux 6.19.6-arch1-1 (x64)
- Node: v25.7.0
- Tunnel: ngrok (static domain, paid plan)
## Related
Builds on / extends #22362 — same `ambiguous webhook target` symptom, but this report covers a second compounding issue that appears once the crash-loop is resolved.... | closed | completed | true | 2 | [] | [] | 2026-04-03T15:21:50Z | 2026-04-28T04:34:07Z | 2026-04-26T04:21:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JethroLifeLab | 258,327,894 | U_kgDOD2XFVg | User | false |
openclaw/openclaw | 4,201,165,543 | I_kwDOQb6kR876aLLn | 60,396 | https://github.com/openclaw/openclaw/issues/60396 | https://api.github.com/repos/openclaw/openclaw/issues/60396 | [Feature Request] Add fallback token calculation for providers that don't return usage data | # [Feature Request] Add fallback token calculation for providers that don't return usage data
## Problem Description
Some model providers (e.g., `bailian/glm-5`) do not return token usage data in their API responses. This causes the TUI footer to display `tokens ?/203k (0%)` instead of actual token counts.
### Curre... | closed | completed | true | 3 | [] | [] | 2026-04-03T15:59:49Z | 2026-04-28T04:34:06Z | 2026-04-26T04:21:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | HosheaLi | 102,918,582 | U_kgDOBiJptg | User | false |
openclaw/openclaw | 4,201,468,023 | I_kwDOQb6kR876bVB3 | 60,446 | https://github.com/openclaw/openclaw/issues/60446 | https://api.github.com/repos/openclaw/openclaw/issues/60446 | iMessage plugin: surface tapback/reaction events to agent | ## Feature Request
**Channel:** iMessage plugin
**Description:**
iMessage tapback reactions (heart, thumbs up, thumbs down, ha ha, !!, ?) are currently dropped by the iMessage plugin and never delivered to the agent. The agent has no awareness of user reactions to its messages.
**Desired behavior:**
Tapback events s... | closed | not_planned | true | 1 | [] | [] | 2026-04-03T17:23:58Z | 2026-04-28T04:34:06Z | 2026-04-26T04:21:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,201,616,092 | I_kwDOQb6kR876b5Lc | 60,458 | https://github.com/openclaw/openclaw/issues/60458 | https://api.github.com/repos/openclaw/openclaw/issues/60458 | CLI shows http:// and ws:// URLs when gateway.tls.enabled is true (bind: lan) | ### Description
When `gateway.tls.enabled: true` with custom cert/key and `gateway.bind: "lan"`, the CLI (`openclaw onboard`, `openclaw status`) displays `http://` and `ws://` URLs instead of `https://` and `wss://`.
### Environment
- OpenClaw: 2026.4.2
- OS: macOS (arm64)
- Config:
+Y```yaml
+gateway:
++port: 18789... | closed | completed | true | 1 | [] | [] | 2026-04-03T18:02:57Z | 2026-04-28T04:34:05Z | 2026-04-26T04:22:01Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lhtpluto | 132,045,861 | U_kgDOB97cJQ | User | false |
openclaw/openclaw | 4,201,706,795 | I_kwDOQb6kR876cPUr | 60,472 | https://github.com/openclaw/openclaw/issues/60472 | https://api.github.com/repos/openclaw/openclaw/issues/60472 | [Bug]: Telegram media download fails when proxy is socks5 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When channels.telegram.proxy is set to socks5://127.0.0.1:1082, Telegram media file downloads fail with "Explicit proxy URL must use http or https". Regular bot API requests succeed.
### Steps to reproduce
1.... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-03T18:29:12Z | 2026-04-28T04:34:04Z | 2026-04-26T04:22:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | heichaowo | 22,045,978 | MDQ6VXNlcjIyMDQ1OTc4 | User | false |
openclaw/openclaw | 4,202,169,042 | I_kwDOQb6kR876eALS | 60,530 | https://github.com/openclaw/openclaw/issues/60530 | https://api.github.com/repos/openclaw/openclaw/issues/60530 | WebChat: Add human-readable session aliases in sidebar | ## Problem
The WebChat sidebar displays raw session keys like:
```
telegram:g-leaf · telegram:group:-1003789128818:topic:167
telegram:g-leaf · telegram:group:-1003789128818:topic:158
webchat:g-agent-main-main
```
These are unreadable for end users. It's nearly impossible to tell which session is which without memori... | closed | completed | true | 1 | [] | [] | 2026-04-03T20:55:37Z | 2026-04-28T04:34:04Z | 2026-04-26T04:22:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yoonow-gdi | 211,648,917 | U_kgDODJ2BlQ | User | false |
openclaw/openclaw | 4,202,496,986 | I_kwDOQb6kR876fQPa | 60,580 | https://github.com/openclaw/openclaw/issues/60580 | https://api.github.com/repos/openclaw/openclaw/issues/60580 | [Bug]: Agent repeats own content across turns (4.2) | After upgrading to 4.2, agents repeat content they've already said, as if prior responses aren't in their conversation history. Intermittent, worse in longer sessions.
Two separate instances, both Anthropic Opus 4.6, Telegram streaming off, compaction safeguard. Started after 3.31 → 4.2 upgrade.
Not delivery duplicat... | closed | completed | true | 2 | [] | [] | 2026-04-03T22:58:31Z | 2026-04-28T04:34:03Z | 2026-04-26T04:22:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | robrodriguezjr | 117,524,017 | U_kgDOBwFGMQ | User | false |
openclaw/openclaw | 4,203,912,952 | I_kwDOQb6kR876kp74 | 60,727 | https://github.com/openclaw/openclaw/issues/60727 | https://api.github.com/repos/openclaw/openclaw/issues/60727 | Support input_type for asymmetric embeddings in built-in memorySearch | ### Summary
Built-in memorySearch does not currently support passing input_type for OpenAI-compatible embedding providers.
This blocks asymmetric embedding models that require input_type (for example query vs passage).
### Current Behavior
- memorySearch embedding requests are sent without input_type.
- Asymmetric mo... | closed | not_planned | true | 1 | [] | [] | 2026-04-04T06:49:35Z | 2026-04-28T04:34:01Z | 2026-04-26T04:22:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | prospect1314521 | 135,647,789 | U_kgDOCBXSLQ | User | false |
openclaw/openclaw | 4,204,100,935 | I_kwDOQb6kR876lX1H | 60,747 | https://github.com/openclaw/openclaw/issues/60747 | https://api.github.com/repos/openclaw/openclaw/issues/60747 | fix(gateway): stale device repair request blocks all RPC operations (1008 pairing required) | file:///root/.openclaw/workspace/memory/tmp-issue-body.md | closed | completed | true | 1 | [] | [] | 2026-04-04T07:51:19Z | 2026-04-28T04:34:00Z | 2026-04-26T04:22:21Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | EronFan | 50,734,013 | MDQ6VXNlcjUwNzM0MDEz | User | false |
openclaw/openclaw | 4,204,424,197 | I_kwDOQb6kR876mmwF | 60,792 | https://github.com/openclaw/openclaw/issues/60792 | https://api.github.com/repos/openclaw/openclaw/issues/60792 | [Feature]: Community hardware reference: DGX Spark + vLLM local setup | ### Summary
Link to a tested, open-source DGX Spark + vLLM + OpenClaw reference setup in the local hardware docs.
### Problem to solve
DGX Spark (128GB Blackwell, ~$4K-$5K) is shipping to consumers and there's no
documented reference for running OpenClaw on this hardware with vLLM. Users
buying a Spark for private l... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-04T09:52:10Z | 2026-04-28T04:33:59Z | 2026-04-26T04:22:26Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | thatwonguy | 78,534,460 | MDQ6VXNlcjc4NTM0NDYw | User | false |
openclaw/openclaw | 4,204,566,732 | I_kwDOQb6kR876nJjM | 60,810 | https://github.com/openclaw/openclaw/issues/60810 | https://api.github.com/repos/openclaw/openclaw/issues/60810 | [Feature]: Real-time agent activity dashboard — execution tree, timeline, and live metrics | ## Summary
Real-time agent activity dashboard showing what the agent is doing behind the scenes — execution tree, tool calls, thinking steps, and timeline — replacing the current black-box experience.
## Problem to solve
The current Control UI provides post-hoc views (logs, sessions, usage) but no live visibility in... | closed | not_planned | true | 3 | [] | [] | 2026-04-04T10:41:13Z | 2026-04-28T04:33:58Z | 2026-04-26T04:22:30Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | LeVarez | 76,524,987 | MDQ6VXNlcjc2NTI0OTg3 | User | false |
openclaw/openclaw | 4,204,579,548 | I_kwDOQb6kR876nMrc | 60,813 | https://github.com/openclaw/openclaw/issues/60813 | https://api.github.com/repos/openclaw/openclaw/issues/60813 | fix(line): upgrade hardcoded error fallback to typed error routing | ## Problem
LINE is the only non-Telegram channel that sends a user-visible fallback message on dispatch failure, but it uses a hardcoded generic string via a bare try/catch rather than the typed error routing introduced for Telegram.
Current LINE fallback (src/line/monitor.ts):
```
"Sorry, I encountered an error proc... | closed | not_planned | true | 1 | [] | [] | 2026-04-04T10:47:58Z | 2026-04-28T04:33:58Z | 2026-04-26T04:22:34Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | garnetlyx | 12,513,503 | MDQ6VXNlcjEyNTEzNTAz | User | false |
openclaw/openclaw | 4,329,974,709 | I_kwDOQb6kR88AAAABAhYrtQ | 71,938 | https://github.com/openclaw/openclaw/issues/71938 | https://api.github.com/repos/openclaw/openclaw/issues/71938 | [Bug]: openclaw onboard take too long | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrade to 2026.4.24 openclaw onboard take too long for each step.
### Steps to reproduce
1. Update to openclaw 2026.4.24
2. Run openclaw onboard
3. It take too long for each step (5 to 10 minutes on my mac mini m4 ... | closed | completed | true | 6 | [
"bug",
"regression"
] | [] | 2026-04-26T03:58:28Z | 2026-04-29T04:35:53Z | 2026-04-26T10:44:39Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | nolunxzxc22 | 267,327,614 | U_kgDOD-8Yfg | User | false |
openclaw/openclaw | 4,186,317,478 | I_kwDOQb6kR875hiKm | 58,987 | https://github.com/openclaw/openclaw/issues/58987 | https://api.github.com/repos/openclaw/openclaw/issues/58987 | [Bug]: upgrade fail | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
latest update gives me a Internal Server Error
### Steps to reproduce
latest update gives me a Internal Server Error
### Expected behavior
latest update gives me a Internal Server Error
### Actual behavior
latest updat... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-04-01T11:52:10Z | 2026-04-30T21:58:51Z | null | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | DragonBtc93 | 36,767,505 | MDQ6VXNlcjM2NzY3NTA1 | User | false |
openclaw/openclaw | 4,204,653,418 | I_kwDOQb6kR876netq | 60,837 | https://github.com/openclaw/openclaw/issues/60837 | https://api.github.com/repos/openclaw/openclaw/issues/60837 | Chat response auto-follow should pause immediately on any manual scroll input | ### Summary
Chat auto-follow should pause immediately on any manual scroll input:
In the chat view, manual scrolling does not reliably pause response-follow mode, at least not when the view is still considered “near bottom.” Right now, it might be possible to interrupt follow mode only if I scroll far enough or persi... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-04T11:30:19Z | 2026-04-28T04:33:57Z | 2026-04-26T04:24:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | BBMconsulting | 174,531,100 | U_kgDOCmciHA | User | false |
openclaw/openclaw | 4,205,028,193 | I_kwDOQb6kR876o6Nh | 60,893 | https://github.com/openclaw/openclaw/issues/60893 | https://api.github.com/repos/openclaw/openclaw/issues/60893 | Multi-tool responses dispatched as separate messages instead of coalesced | ## Bug Report
### Summary
When an agent response involves multiple sequential tool calls, each tool result or intermediate message is dispatched to the chat as a separate message, sometimes 3-5 messages in under 500ms. These should be coalesced into a single response.
### Steps to Reproduce
1. Ask a question that req... | closed | completed | true | 1 | [] | [] | 2026-04-04T14:10:09Z | 2026-04-28T04:33:56Z | 2026-04-26T04:24:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | devtony42 | 16,560,790 | MDQ6VXNlcjE2NTYwNzkw | User | false |
openclaw/openclaw | 4,205,042,306 | I_kwDOQb6kR876o9qC | 60,897 | https://github.com/openclaw/openclaw/issues/60897 | https://api.github.com/repos/openclaw/openclaw/issues/60897 | Cache hit rate shows 28.5% in Control UI after upgrading to 2026.4.2 | After upgrading to OpenClaw 2026.4.2, the cache hit rate displayed in the Control UI shows only 28.5%, which seems unusually low.\n\n## Environment\n- OpenClaw version: 2026.4.2\n- Platform: Windows\n- Model: minimax/MiniMax-M2.7-highspeed\n\n## Expected behavior\nCache hit rate should be significantly higher (previous... | closed | completed | true | 1 | [] | [] | 2026-04-04T14:16:48Z | 2026-04-28T04:33:56Z | 2026-04-26T04:24:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wangzihe843-hash | 266,487,497 | U_kgDOD-JGyQ | User | false |
openclaw/openclaw | 4,205,089,685 | I_kwDOQb6kR876pJOV | 60,908 | https://github.com/openclaw/openclaw/issues/60908 | https://api.github.com/repos/openclaw/openclaw/issues/60908 | ClawHub capability tags are false positives: Crypto, Can make purchases, Requires OAuth | **Skill:** earlvanze/renatus-icm (v2.1.3)
## Problem
The capability signals panel on the skill page shows three tags that are false positives:
| Detected Tag | Why It is Wrong |
|---|---|
| **Crypto** | Triggered by `accessToken()`, `xsrfToken()` — these are OAuth/access tokens for the Renatus back office API, not c... | closed | not_planned | true | 1 | [] | [] | 2026-04-04T14:44:11Z | 2026-04-28T04:33:55Z | 2026-04-26T04:24:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | earlvanze | 3,733,346 | MDQ6VXNlcjM3MzMzNDY= | User | false |
openclaw/openclaw | 4,205,152,915 | I_kwDOQb6kR876pYqT | 60,919 | https://github.com/openclaw/openclaw/issues/60919 | https://api.github.com/repos/openclaw/openclaw/issues/60919 | Feature: derive model fallback candidates from agents.defaults.models | ## Summary
OpenClaw already supports model fallback through `agents.defaults.model.fallbacks`, but operators who keep multiple models in `agents.defaults.models` still have to duplicate that catalog by hand in `fallbacks`.
If the configured primary model becomes unavailable, the runtime currently ignores the rest of ... | closed | not_planned | true | 1 | [] | [] | 2026-04-04T15:16:58Z | 2026-04-28T04:33:54Z | 2026-04-26T04:24:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 08820048 | 71,597,859 | MDQ6VXNlcjcxNTk3ODU5 | User | false |
openclaw/openclaw | 4,205,185,206 | I_kwDOQb6kR876pgi2 | 60,924 | https://github.com/openclaw/openclaw/issues/60924 | https://api.github.com/repos/openclaw/openclaw/issues/60924 | [Bug]: sessions_send into reused fixed workbench succeeds, but timeout/announce can return stale previous-round results | # OpenClaw Issue Draft: fixed workbench reuse succeeds, but timeout / announce can return stale previous-round results
## Suggested title
[Bug]: sessions_send into reused fixed workbench succeeds and executes in-place, but timeout / announce may surface stale previous-round results
## Summary
When reusing an existing... | closed | completed | true | 1 | [] | [] | 2026-04-04T15:33:07Z | 2026-04-28T04:33:53Z | 2026-04-26T04:24:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ChenYongJie666 | 33,951,512 | MDQ6VXNlcjMzOTUxNTEy | User | false |
openclaw/openclaw | 4,205,307,234 | I_kwDOQb6kR876p-Vi | 60,952 | https://github.com/openclaw/openclaw/issues/60952 | https://api.github.com/repos/openclaw/openclaw/issues/60952 | [Feature]: MS Teams plugin should forward messageBack card actions (Action.Submit) to the agent | ### Summary
Forward Action.Submit messageBack activities (type="message" with empty text + value object) to the agent instead of dropping them as empty messages.
### Problem to solve
When a user clicks an Action.Submit button on an Adaptive Card in Teams, the click is sent as a messageBack activity (type="message" w... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-04T16:33:51Z | 2026-04-28T04:33:53Z | 2026-04-26T04:24:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ndholakia | 9,597,297 | MDQ6VXNlcjk1OTcyOTc= | User | false |
openclaw/openclaw | 4,205,467,109 | I_kwDOQb6kR876qlXl | 60,987 | https://github.com/openclaw/openclaw/issues/60987 | https://api.github.com/repos/openclaw/openclaw/issues/60987 | MiniMax OAuth onboard writes empty models array, blocking model resolution | ## Bug
### Summary
The MiniMax OAuth onboard flow (`openclaw models auth login --provider minimax-portal --method oauth` or `openclaw onboard --auth-choice minimax-global-oauth`) writes `"models": []` into `openclaw.json` under `models.providers.minimax-portal`. This empty array tells the gateway the provider has zero... | closed | completed | true | 1 | [] | [] | 2026-04-04T17:53:16Z | 2026-04-28T04:33:52Z | 2026-04-26T04:24:31Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | chenxin-yan | 71,162,231 | MDQ6VXNlcjcxMTYyMjMx | User | false |
openclaw/openclaw | 4,205,551,153 | I_kwDOQb6kR876q54x | 61,003 | https://github.com/openclaw/openclaw/issues/61003 | https://api.github.com/repos/openclaw/openclaw/issues/61003 | [Feature]: Control UI chat should allow creating a new session directly from the session switcher | ### Summary
Add a visible way to create a new session in Control UI chat, instead of requiring the ?session=... URL workaround.
### Problem to solve
Control UI chat can switch between existing sessions, but there is no obvious way to create a new one from the UI.
Right now, the workaround is to manually edit the UR... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-04T18:32:17Z | 2026-04-28T04:33:51Z | 2026-04-26T04:24:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wxzhoulf | 68,071,555 | MDQ6VXNlcjY4MDcxNTU1 | User | false |
openclaw/openclaw | 4,205,740,011 | I_kwDOQb6kR876rn_r | 61,026 | https://github.com/openclaw/openclaw/issues/61026 | https://api.github.com/repos/openclaw/openclaw/issues/61026 | [Bug]: Manual session stop triggers false timeout error and cascading model failover | ## Bug Description
When a user manually stops or pauses a session (User Interrupt), OpenClaw incorrectly interprets this action as a **timeout error**. This triggers the model failover mechanism, causing the system to attempt switching to fallback models unnecessarily.
## Observed Behavior
1. User clicks "Stop" or i... | closed | completed | true | 2 | [] | [] | 2026-04-04T20:20:00Z | 2026-04-28T04:33:50Z | 2026-04-26T04:24:39Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | prospect1314521 | 135,647,789 | U_kgDOCBXSLQ | User | false |
openclaw/openclaw | 4,206,230,964 | I_kwDOQb6kR876tf20 | 61,135 | https://github.com/openclaw/openclaw/issues/61135 | https://api.github.com/repos/openclaw/openclaw/issues/61135 | Agent often said tool but not use tool | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
after upgrade to 2026.4.2,in some reason the agent said tools command but not use tools
### Steps to reproduce
1.upgrade 2026.4.2
2.start gateway
3.like what i said
### Expected behavior
useing tool
### Ac... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-05T02:06:02Z | 2026-04-28T04:33:49Z | 2026-04-26T04:24:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | cuiJY-still-in-school | 261,708,640 | U_kgDOD5lbYA | User | false |
openclaw/openclaw | 4,206,275,699 | I_kwDOQb6kR876tqxz | 61,153 | https://github.com/openclaw/openclaw/issues/61153 | https://api.github.com/repos/openclaw/openclaw/issues/61153 | Discord DM channel stops responding after WebSocket disconnect (code 1005) | ## Issue Summary
The Discord bot stops responding to Direct Messages (DMs) after the gateway encounters WebSocket disconnects. The bot appears online in Discord but ignores all DM messages. The issue occurs multiple times per day and requires a full gateway restart to recover.
## Severity
High — Blocks primary communi... | closed | completed | true | 4 | [] | [] | 2026-04-05T02:41:05Z | 2026-04-28T04:33:49Z | 2026-04-26T04:24:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | LIGHTSPEED1699 | 20,879,328 | MDQ6VXNlcjIwODc5MzI4 | User | false |
openclaw/openclaw | 4,206,409,779 | I_kwDOQb6kR876uLgz | 61,179 | https://github.com/openclaw/openclaw/issues/61179 | https://api.github.com/repos/openclaw/openclaw/issues/61179 | [Bug]: Windows: repeated gateway/ws 1008 device-required from cli probe on loopback (closed before connect spam) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
#### Summary
On Windows (local Scheduled Task gateway), gateway logs repeatedly show:
closed before connect ... code=1008 reason=connect failed
structured context shows cause=device-required, client=cli, mode=... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-05T04:10:03Z | 2026-04-28T04:33:48Z | 2026-04-26T04:24:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | aemeathpa | 269,386,194 | U_kgDOEA6B0g | User | false |
openclaw/openclaw | 4,206,448,739 | I_kwDOQb6kR876uVBj | 61,185 | https://github.com/openclaw/openclaw/issues/61185 | https://api.github.com/repos/openclaw/openclaw/issues/61185 | bug(ollama/wsl2): Ollama systemd autostart causes WSL2 crash loop via Hyper-V balloon driver conflict | ## Bug Report: WSL2 + Ollama systemd autostart causes crash loop on boot
### Summary
When OpenClaw runs on WSL2 with Ollama configured as a local inference provider, the default Ollama systemd service configuration causes a crash loop that makes WSL2 unusable. The gateway restarts every 20–60 seconds indefinitely unt... | closed | not_planned | true | 2 | [] | [] | 2026-04-05T04:26:09Z | 2026-04-28T04:30:39Z | 2026-04-26T04:24:55Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yhyatt | 10,474,956 | MDQ6VXNlcjEwNDc0OTU2 | User | false |
openclaw/openclaw | 4,206,503,983 | I_kwDOQb6kR876uigv | 61,193 | https://github.com/openclaw/openclaw/issues/61193 | https://api.github.com/repos/openclaw/openclaw/issues/61193 | Bug: exec approvals allowlist glob pattern not matching commands | **Bug: exec approvals allowlist glob pattern not matching commands**
## Description
The `exec approvals` allowlist glob patterns are not matching commands correctly, even when the pattern should match according to standard glob rules.
## Steps to Reproduce
1. Set up exec approvals with a glob pattern:
```bash
... | closed | completed | true | 4 | [] | [] | 2026-04-05T04:53:24Z | 2026-04-28T04:33:47Z | 2026-04-26T04:24:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kevin-lucifer | 20,728,002 | MDQ6VXNlcjIwNzI4MDAy | User | false |
openclaw/openclaw | 4,206,834,024 | I_kwDOQb6kR876vzFo | 61,229 | https://github.com/openclaw/openclaw/issues/61229 | https://api.github.com/repos/openclaw/openclaw/issues/61229 | [Bug]: stale exec node display name survives rebinding and triggers "exec node not allowed" | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
In a long-lived Telegram topic session, `exec` kept sending `node:"Logan’s Mac mini"` after the bound exec node changed, and every attempt failed with `exec node not allowed`.
### Steps to reproduce
1. Pair two n... | closed | not_planned | true | 1 | [] | [] | 2026-04-05T07:25:24Z | 2026-04-28T04:33:46Z | 2026-04-26T04:25:03Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | loganprit | 72,722,788 | MDQ6VXNlcjcyNzIyNzg4 | User | false |
openclaw/openclaw | 4,206,862,736 | I_kwDOQb6kR876v6GQ | 61,235 | https://github.com/openclaw/openclaw/issues/61235 | https://api.github.com/repos/openclaw/openclaw/issues/61235 | fix: claude-cli backend config guide — 3 pitfalls that prevent CLI dispatch | ## Context
Trying to use `claude -p` (pipe mode) as a CLI backend in OpenClaw v2026.4.2 via `cliBackends` config. Related to #57326.
After extensive debugging, found **3 independent config-level pitfalls** that each silently prevent CLI backend dispatch. Documenting here for others.
## Pitfall 1: `agents.list[].mode... | closed | completed | true | 3 | [] | [] | 2026-04-05T07:39:05Z | 2026-04-28T04:33:45Z | 2026-04-26T04:25:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | geegewu | 129,384,489 | U_kgDOB7ZAKQ | User | false |
openclaw/openclaw | 4,207,137,738 | I_kwDOQb6kR876w9PK | 61,273 | https://github.com/openclaw/openclaw/issues/61273 | https://api.github.com/repos/openclaw/openclaw/issues/61273 | [Bug]: Telegram channel not initializing on 2026.4.2 (valid bot token, no telegram runtime logs) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
## Summary
Telegram channel does not initialize on OpenClaw `2026.4.2` even though:
- Bot token is valid via Telegram API `getMe`
- Config writes succeed
- Gateway restarts cleanly
- Agent model is valid (`op... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-04-05T09:46:38Z | 2026-04-28T04:33:45Z | 2026-04-26T04:25:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | spongtaneous69-dev | 273,757,259 | U_kgDOEFE0Sw | User | false |
openclaw/openclaw | 4,207,187,669 | I_kwDOQb6kR876xJbV | 61,287 | https://github.com/openclaw/openclaw/issues/61287 | https://api.github.com/repos/openclaw/openclaw/issues/61287 | [Bug]: openclaw sessions shows inflated context (>100%) for claude-cli backend — cumulative cacheRead summed across turns | [Bug]: openclaw sessions shows inflated context (>100%) for claude-cli backend — cumulative cacheRead summed across turns
## Summary
`openclaw sessions` reports `1189k/1000k (119%)` for the active direct session when using the `claude-cli` backend, even though the actual in-context tokens on the last turn are ~82k. T... | closed | completed | true | 1 | [] | [] | 2026-04-05T10:21:10Z | 2026-04-28T04:33:44Z | 2026-04-26T04:25:15Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mantisai-bot | 259,690,819 | U_kgDOD3qRQw | User | false |
openclaw/openclaw | 4,207,278,002 | I_kwDOQb6kR876xfey | 61,312 | https://github.com/openclaw/openclaw/issues/61312 | https://api.github.com/repos/openclaw/openclaw/issues/61312 | iMessage echo loop in self-chat: U+FFFD from attributedBody breaks echo cache text matching | ## Description
In a self-chat scenario (Mac's iMessage handle matches the DM contact), the echo cache fails to detect reflected bot messages due to a Unicode replacement character (U+FFFD / `�`) prepended during `attributedBody` text extraction.
## Steps to Reproduce
1. Set up iMessage channel where the Mac's Apple ... | closed | not_planned | true | 1 | [] | [] | 2026-04-05T11:19:54Z | 2026-04-28T04:33:43Z | 2026-04-26T04:25:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | emmet-bot | 258,447,653 | U_kgDOD2eZJQ | User | false |
openclaw/openclaw | 4,116,385,689 | I_kwDOQb6kR871Ww-Z | 52,343 | https://github.com/openclaw/openclaw/issues/52343 | https://api.github.com/repos/openclaw/openclaw/issues/52343 | [Bug]: openclaw-weixin replies can appear out of order in DM conversations | ### Bug type
Behavior bug
### Summary
During local Weixin testing on OpenClaw `2026.3.14`, replies in the same Weixin DM appeared out of order. Gateway logs from the same session show the plugin receiving a new inbound Weixin message from the same user immediately after a successful outbound send, which suggests the r... | closed | not_planned | true | 1 | [] | [] | 2026-03-22T15:53:35Z | 2026-04-28T04:33:42Z | 2026-04-26T04:26:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mist-wu | 184,920,539 | U_kgDOCwWp2w | User | false |
openclaw/openclaw | 4,207,314,084 | I_kwDOQb6kR876xoSk | 61,318 | https://github.com/openclaw/openclaw/issues/61318 | https://api.github.com/repos/openclaw/openclaw/issues/61318 | Fallback model configuration not loading after gateway restart | Changed `agents.defaults.model.fallbacks` in openclaw.json from `[anthropic/claude-haiku-4-5]` to `[google/gemini-3.1-flash-lite-preview, anthropic/claude-haiku-4-5, anthropic/claude-opus-4-6]`.
Config file shows the correct value when read directly. After three separate `openclaw gateway restart` commands and multipl... | closed | completed | true | 1 | [] | [] | 2026-04-05T11:40:08Z | 2026-04-28T04:33:41Z | 2026-04-26T04:27:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | AIFaithLab | 218,689,467 | U_kgDODQjvuw | User | false |
openclaw/openclaw | 4,207,521,814 | I_kwDOQb6kR876ybAW | 61,350 | https://github.com/openclaw/openclaw/issues/61350 | https://api.github.com/repos/openclaw/openclaw/issues/61350 | Gemini thought-block leak with "...94>thought CRITICAL INSTRUCTION" format | The regex filter catches standard <thinking> and <thought> tags but misses this variant seen multiple times in the April 4 Gemini 3.1 session:
'...94>thought CRITICAL INSTRUCTION 1:...'
It leaks into Telegram even on 'silent' responses. Reproducible in heartbeat/silent paths.
**Steps to reproduce:** Run a Gemini 3.... | closed | not_planned | true | 4 | [] | [] | 2026-04-05T13:17:02Z | 2026-04-28T04:33:40Z | 2026-04-26T04:27:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | bo-blue | 257,455,029 | U_kgDOD1hztQ | User | false |
openclaw/openclaw | 4,207,723,487 | I_kwDOQb6kR876zMPf | 61,379 | https://github.com/openclaw/openclaw/issues/61379 | https://api.github.com/repos/openclaw/openclaw/issues/61379 | Bug: `kimi` provider config defaults to `anthropic-messages` instead of `openai-completions` | **Describe the bug**
When using OpenClaw's configuration modification feature to add/configure `kimi`, it incorrectly sets the API format to `anthropic-messages` in `openclaw.json`.
**Expected behavior**
Kimi's API format is OpenAI-compatible, so it should default to `openai-completions` (or `openai-messages`).
**Cur... | closed | completed | true | 7 | [] | [] | 2026-04-05T14:43:00Z | 2026-04-28T04:33:39Z | 2026-04-26T04:27:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Jageri | 14,195,779 | MDQ6VXNlcjE0MTk1Nzc5 | User | false |
openclaw/openclaw | 4,207,728,531 | I_kwDOQb6kR876zNeT | 61,380 | https://github.com/openclaw/openclaw/issues/61380 | https://api.github.com/repos/openclaw/openclaw/issues/61380 | sortToolsMessageItems uses toSorted() which is not stable | **Describe the bug*
In `dist/pi-embedded-BYdcxQ5A.js`, `sortToolsMessageItems` uses `toSorted`:
```javascript
function sortToolsMessageItems(items) {
return items.toSorted((a, b) => a.name.localeCompare(b.name));
}
```
`toSorted` (ES2024) is not guaranteed stable. Claude Code explicitly avoids it with this comment... | closed | not_planned | true | 2 | [] | [] | 2026-04-05T14:45:34Z | 2026-04-28T04:33:38Z | 2026-04-26T04:27:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | revinwang-git | 265,145,880 | U_kgDOD83OGA | User | false |
openclaw/openclaw | 4,208,062,023 | I_kwDOQb6kR8760e5H | 61,434 | https://github.com/openclaw/openclaw/issues/61434 | https://api.github.com/repos/openclaw/openclaw/issues/61434 | [Bug]: Signal channel SSE connection fails persistently with fetch failed every 10 seconds | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Signal channel SSE connection fails persistently with fetch failed every 10 seconds
Info: Occurs with both JVM and native signal-cli binaries on high-end hardware (Fedora 43, native binary, IPv4 confirmed)
##... | closed | not_planned | true | 1 | [
"bug"
] | [] | 2026-04-05T16:42:18Z | 2026-04-28T04:33:37Z | 2026-04-26T04:27:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ctohastek | 9,472,197 | MDQ6VXNlcjk0NzIxOTc= | User | false |
openclaw/openclaw | 4,208,066,268 | I_kwDOQb6kR8760f7c | 61,436 | https://github.com/openclaw/openclaw/issues/61436 | https://api.github.com/repos/openclaw/openclaw/issues/61436 | [Feature] Path-specific skill loading (like Claude Code `paths:` frontmatter) | ## Problem
Currently, OpenClaw loads ALL skills on every request, consuming context tokens regardless of whether the skill is relevant to the current task. This wastes tokens and increases response times.
## Proposed Solution
Add a `paths:` field to SKILL.md frontmatter (like Claude Code's implementation) that allow... | closed | not_planned | true | 1 | [] | [] | 2026-04-05T16:44:20Z | 2026-04-28T04:33:36Z | 2026-04-26T04:27:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Donmeusi | 109,555,289 | U_kgDOBoeuWQ | User | false |
openclaw/openclaw | 4,208,349,274 | I_kwDOQb6kR8761lBa | 61,480 | https://github.com/openclaw/openclaw/issues/61480 | https://api.github.com/repos/openclaw/openclaw/issues/61480 | web_fetch ignores HTTP_PROXY environment variables - dispatcherPolicy defaults to direct | ## Bug Description
In OpenClaw 2026.4.2, the `web_fetch` tool does not respect `HTTP_PROXY`/`HTTPS_PROXY` environment variables even when they are correctly configured in the systemd service.
## Root Cause
In `http-client-COwW1KRl.js` line 111:
```javascript
dispatcher = createPinnedDispatcher(
await resolvePinne... | closed | not_planned | true | 1 | [] | [] | 2026-04-05T19:01:19Z | 2026-04-28T04:33:34Z | 2026-04-26T04:27:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | dt-todtom | 252,163,579 | U_kgDODwe1-w | User | false |
openclaw/openclaw | 4,209,063,763 | I_kwDOQb6kR8764TdT | 61,607 | https://github.com/openclaw/openclaw/issues/61607 | https://api.github.com/repos/openclaw/openclaw/issues/61607 | Add DM trigger gating separate from DM access policy | ## Problem
> **Note:** Claude and GPT were used to help scope this problem, describe the issue, and outline the implementation approach.
---
OpenClaw currently has two layers for **group chats**:
1. **Access control** — `groupPolicy` / allowlists determine if a message is permitted at all
2. **Trigger gating** — `re... | closed | not_planned | true | 1 | [] | [] | 2026-04-06T01:46:58Z | 2026-04-28T04:33:32Z | 2026-04-26T04:27:55Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mubbie | 66,747,577 | MDQ6VXNlcjY2NzQ3NTc3 | User | false |
openclaw/openclaw | 4,209,105,793 | I_kwDOQb6kR8764duB | 61,611 | https://github.com/openclaw/openclaw/issues/61611 | https://api.github.com/repos/openclaw/openclaw/issues/61611 | [Feature]: OAuth authentication for remote MCP servers | ### Summary
Wire up the MCP SDK's existing `authProvider` support on `StreamableHTTPClientTransport` and `SSEClientTransport` so openclaw can authenticate to OAuth-protected remote MCP servers.
### Problem to solve
Remote MCP servers that require OAuth authentication (per the https://modelcontextprotocol.io/specifi... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-06T02:07:16Z | 2026-04-28T04:33:31Z | 2026-04-26T04:27:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nrcrews | 33,267,791 | MDQ6VXNlcjMzMjY3Nzkx | User | false |
openclaw/openclaw | 4,209,301,352 | I_kwDOQb6kR8765Ndo | 61,653 | https://github.com/openclaw/openclaw/issues/61653 | https://api.github.com/repos/openclaw/openclaw/issues/61653 | [Bug]: # Bug Report: exec approval 的命令哈希机制导致 allow-always 无法有效覆盖常用命令 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
# Bug Report: exec approval 的命令哈希机制导致 allow-always 无法有效覆盖常用命令
**版本**：OpenClaw 2026.4.1 (da64a97)
**问题描述**
每次 agent 执行命令时，即使操作类型完全相同（如运行 Python 脚本），只要命令参数有任何细微差异，就会生成新的命令哈希（=command:xxxxxxxx），触发新的 exec 审批弹窗。用户点击「Always allo... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-06T03:12:55Z | 2026-04-28T04:33:30Z | 2026-04-26T04:28:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | heronerar | 128,018,749 | U_kgDOB6FpPQ | User | false |
openclaw/openclaw | 4,209,908,364 | I_kwDOQb6kR8767hqM | 61,707 | https://github.com/openclaw/openclaw/issues/61707 | https://api.github.com/repos/openclaw/openclaw/issues/61707 | [Feature]: tiered slash-command auth — session reset vs commands.allowFrom | ### Summary
Tier slash-command authorization so session commands (`/new`, `/reset`) can remain available to channel-authorized chatters while `commands.allowFrom` restricts only elevated/admin-style commands.
### Problem to solve
When `commands.allowFrom` is set, it is the **only** authorization source for slash com... | closed | completed | true | 2 | [] | [] | 2026-04-06T06:01:44Z | 2026-04-28T04:33:29Z | 2026-04-26T04:28:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | eightHundreds | 18,695,431 | MDQ6VXNlcjE4Njk1NDMx | User | false |
openclaw/openclaw | 4,209,932,237 | I_kwDOQb6kR8767nfN | 61,712 | https://github.com/openclaw/openclaw/issues/61712 | https://api.github.com/repos/openclaw/openclaw/issues/61712 | Telegram polling fails to start after gateway restart (LaunchAgent) | ## Bug Description
Telegram polling fails to start after gateway restart on macOS with LaunchAgent.
## Environment
- OpenClaw: 2026.4.2
- OS: macOS
- Service: LaunchAgent (launchd)
## Steps to Reproduce
1. Configure channels.telegram with botToken in openclaw.json
2. Run `launchctl stop ai.openclaw.gateway && sleep 3... | closed | completed | true | 1 | [] | [] | 2026-04-06T06:09:29Z | 2026-04-28T04:33:29Z | 2026-04-26T04:28:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | longxiahsiao-sketch | 270,969,766 | U_kgDOECarpg | User | false |
openclaw/openclaw | 4,210,333,828 | I_kwDOQb6kR8769JiE | 61,741 | https://github.com/openclaw/openclaw/issues/61741 | https://api.github.com/repos/openclaw/openclaw/issues/61741 | [Bug]: Race condition in subagent/session cleanup causes late child stdout to hit cleared active run, leading to announce timeout, missing-session-entry, and orphaned child processes | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
We are seeing a recurring lifecycle failure in OpenClaw where a subagent or exec child still emits stdout/update events after the parent run/session has already been ended, patched away, or deleted.
The first hard error is:
`... | closed | completed | true | 5 | [
"bug",
"bug:crash"
] | [] | 2026-04-06T07:42:07Z | 2026-04-28T04:33:28Z | 2026-04-26T04:28:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jsompis | 6,247,785 | MDQ6VXNlcjYyNDc3ODU= | User | false |
openclaw/openclaw | 4,210,379,569 | I_kwDOQb6kR8769Usx | 61,744 | https://github.com/openclaw/openclaw/issues/61744 | https://api.github.com/repos/openclaw/openclaw/issues/61744 | [Bug]: WhatsApp links successfully, but peers/groups discovery is empty and the gateway reconnects repeatedly with status 499 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
After linking WhatsApp successfully in OpenClaw, the channel appears configured and connects, but WhatsApp discovery is not usable: openclaw directory peers list --channel whatsapp only returns self, openclaw d... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-06T07:51:04Z | 2026-04-28T04:33:27Z | 2026-04-26T04:28:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rubencolomer | 25,428,013 | MDQ6VXNlcjI1NDI4MDEz | User | false |
openclaw/openclaw | 4,210,555,823 | I_kwDOQb6kR8769_uv | 61,762 | https://github.com/openclaw/openclaw/issues/61762 | https://api.github.com/repos/openclaw/openclaw/issues/61762 | [Bug]: `cron add` fails with "pairing required" on local loopback mode (macOS) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
# [Bug]: `cron add` fails with "pairing required" on local loopback mode (macOS)
## Bug type
Regression (worked before, now fails)
## Summary
The `cron` tool's `add` action consistently fails with `gateway closed (1008): ... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-06T08:34:15Z | 2026-04-28T04:33:25Z | 2026-04-26T04:30:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | vbachele | 62,295,364 | MDQ6VXNlcjYyMjk1MzY0 | User | false |
openclaw/openclaw | 4,210,614,436 | I_kwDOQb6kR876-OCk | 61,771 | https://github.com/openclaw/openclaw/issues/61771 | https://api.github.com/repos/openclaw/openclaw/issues/61771 | spawning of secondary startup fails. | https://github.com/openclaw/openclaw/blob/f7068a1349af466e68ebf6b4ceaaa2ac21e9b05b/scripts/write-cli-startup-metadata.ts
The OpenClaw build process includes a script (write-cli-startup-metadata.ts) that spawns a secondary Node.js process to pre-render the command-line help text (root-help.ts). This subprocess (spawnSy... | closed | completed | true | 2 | [] | [] | 2026-04-06T08:47:31Z | 2026-04-28T04:33:24Z | 2026-04-26T04:30:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sskassamali | 269,874,435 | U_kgDOEBX1Aw | User | false |
openclaw/openclaw | 4,211,531,178 | I_kwDOQb6kR877Bt2q | 61,852 | https://github.com/openclaw/openclaw/issues/61852 | https://api.github.com/repos/openclaw/openclaw/issues/61852 | Compaction max_tokens exceeds model output limit, crashes gateway | ## Bug Description
When compaction is triggered on long conversations, the gateway calculates `max_tokens = Math.ceil(targetLength / 2)`, which can produce values far exceeding the model's maximum output token limit. This causes the API to return a 400 error and crashes the gateway.
## Steps to Reproduce
1. Run a bo... | closed | not_planned | true | 2 | [] | [] | 2026-04-06T12:09:06Z | 2026-04-30T04:43:57Z | 2026-04-26T04:30:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Methux | 60,614,869 | MDQ6VXNlcjYwNjE0ODY5 | User | false |
openclaw/openclaw | 4,211,791,397 | I_kwDOQb6kR877CtYl | 61,886 | https://github.com/openclaw/openclaw/issues/61886 | https://api.github.com/repos/openclaw/openclaw/issues/61886 | [Bug]: logging.file config setting is ignored — logger always writes to /tmp/openclaw | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The `logging.file` setting in `openclaw.json` is accepted by the config schema and validated without error, but has no effect on the gateway's structured log output. Logs always write to `/tmp/openclaw/openclaw... | closed | not_planned | true | 2 | [] | [] | 2026-04-06T13:06:13Z | 2026-04-30T04:43:56Z | 2026-04-26T04:30:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | neo-is-1 | 274,007,627 | U_kgDOEFUGSw | User | false |
openclaw/openclaw | 4,211,945,915 | I_kwDOQb6kR877DTG7 | 61,904 | https://github.com/openclaw/openclaw/issues/61904 | https://api.github.com/repos/openclaw/openclaw/issues/61904 | embedded: false-positive "read tool called without path" when tool args use file/filePath aliases | ### Summary
The gateway logs frequently show:
`[agent/embedded] read tool called without path: toolCallId=... argsType=object`
even when the `read` tool call is **valid** and **succeeds**.
### Root cause (observed in released npm package)
On `tool_execution_start`, `handleToolExecutionStart` only treats `path` and... | closed | not_planned | true | 2 | [] | [] | 2026-04-06T13:38:53Z | 2026-04-30T04:43:55Z | 2026-04-26T04:30:37Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | withoutbuff | 29,725,977 | MDQ6VXNlcjI5NzI1OTc3 | User | false |
openclaw/openclaw | 4,211,977,576 | I_kwDOQb6kR877Da1o | 61,910 | https://github.com/openclaw/openclaw/issues/61910 | https://api.github.com/repos/openclaw/openclaw/issues/61910 | [Feature]: Extend before_model_resolve for runtime auth override and request headers | ## Summary
Extend `before_model_resolve` so plugins can provide runtime auth override plus request-header overrides during model resolution.
## Problem to solve
I have an external plugin integration that already works through normal OpenClaw plugin/config surfaces for almost everything. The remaining gap is model reso... | closed | not_planned | true | 2 | [] | [] | 2026-04-06T13:45:10Z | 2026-04-30T04:43:55Z | 2026-04-26T04:30:42Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | supersantux | 145,187 | MDQ6VXNlcjE0NTE4Nw== | User | false |
openclaw/openclaw | 4,212,045,880 | I_kwDOQb6kR877Drg4 | 61,922 | https://github.com/openclaw/openclaw/issues/61922 | https://api.github.com/repos/openclaw/openclaw/issues/61922 | 2026.4.5: `openclaw dashboard` crashes with Maximum call stack size exceeded | ## Environment
- **OpenClaw version:** 2026.4.5 (3e72c03)
- **Node.js:** 22.21.1
- **OS:** Windows 11 (win32 x64)
## Bug 1: `openclaw dashboard` crashes with stack overflow
Running `openclaw dashboard --no-open` immediately crashes:
```
RangeError: Maximum call stack size exceeded
```
**Sanitized stack trace** (use... | closed | completed | true | 2 | [] | [] | 2026-04-06T13:59:20Z | 2026-04-30T04:43:54Z | 2026-04-26T04:30:46Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | vdruts | 20,264,016 | MDQ6VXNlcjIwMjY0MDE2 | User | false |
openclaw/openclaw | 4,330,020,842 | I_kwDOQb6kR88AAAABAhbf6g | 71,960 | https://github.com/openclaw/openclaw/issues/71960 | https://api.github.com/repos/openclaw/openclaw/issues/71960 | openclaw doctor state-integrity orphan transcript detection appears to ignore active *.trajectory.jsonl sidecars | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
openclaw doctor can report an active session trajectory sidecar as an orphan transcript and say it is safe to archive.
### Steps to reproduce
1. Run OpenClaw with an active session that has a live *.trajectory-path.json po... | closed | not_planned | true | 1 | [
"bug"
] | [] | 2026-04-26T04:30:48Z | 2026-04-29T04:37:19Z | 2026-04-26T10:44:58Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jihoon-ernesto | 41,365,514 | MDQ6VXNlcjQxMzY1NTE0 | User | false |
openclaw/openclaw | 4,212,230,091 | I_kwDOQb6kR877EYfL | 61,941 | https://github.com/openclaw/openclaw/issues/61941 | https://api.github.com/repos/openclaw/openclaw/issues/61941 | Gateway does not dispatch LLM hooks to plugin-registered services (opik traces never fire) | ## Description
User-installed plugins that register services via `api.registerService()` load successfully in the gateway, but their LLM hooks (`llm_input`, `llm_output`, `agent_end`) are never dispatched during agent runs. The plugin's exporter shows zero flush attempts at shutdown.
This was verified with a minimal ... | closed | completed | true | 1 | [] | [] | 2026-04-06T14:37:32Z | 2026-04-30T04:43:52Z | 2026-04-26T04:30:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mathaix | 155,020 | MDQ6VXNlcjE1NTAyMA== | User | false |
openclaw/openclaw | 4,212,322,903 | I_kwDOQb6kR877EvJX | 61,951 | https://github.com/openclaw/openclaw/issues/61951 | https://api.github.com/repos/openclaw/openclaw/issues/61951 | [Bug] memory-core Dreaming causes RangeError: Maximum call stack size exceeded during JSON Schema compilation (AJV) | ## Bug Description
When running `openclaw memory promote --apply` or when the Dreaming cron job triggers, memory-core attempts to load google/minimax plugins, which in turn triggers compilation of a very large JSON Schema (AJV). This causes `RangeError: Maximum call stack size exceeded` during schema validation code g... | closed | completed | true | 1 | [] | [] | 2026-04-06T14:57:07Z | 2026-04-30T04:43:52Z | 2026-04-26T04:30:55Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | glorydu-byte | 265,040,332 | U_kgDOD8wxzA | User | false |
openclaw/openclaw | 4,212,399,979 | I_kwDOQb6kR877FB9r | 61,957 | https://github.com/openclaw/openclaw/issues/61957 | https://api.github.com/repos/openclaw/openclaw/issues/61957 | openclaw status: false lossless-claw provenance warnings, missing cfg in memory-search path | ## Summary
\`openclaw status\` and \`openclaw status --deep\` emit false trust/provenance warnings for the \`lossless-claw\` plugin even though \`openclaw plugins list --verbose\` and \`openclaw plugins inspect lossless-claw\` are clean.
The false warnings observed were:
- \`plugins.allow is empty\`
- \`loaded withou... | closed | completed | true | 3 | [] | [] | 2026-04-06T15:10:46Z | 2026-04-30T04:43:51Z | 2026-04-26T04:31:00Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jeades | 11,338,478 | MDQ6VXNlcjExMzM4NDc4 | User | false |
openclaw/openclaw | 4,212,601,741 | I_kwDOQb6kR877FzON | 61,981 | https://github.com/openclaw/openclaw/issues/61981 | https://api.github.com/repos/openclaw/openclaw/issues/61981 | [Bug] Context compaction preserves invalid file read offsets | ## Bug Description | Bug 描述
When the OpenClaw context is compacted (compressed), file read positions with invalid offsets are preserved, causing subsequent read operations to fail.
## Steps to Reproduce | 复现步骤
1. Use the `read` tool with a specific `offset` parameter to read a file
2. Context compaction occurs (trig... | closed | not_planned | true | 1 | [] | [] | 2026-04-06T15:51:13Z | 2026-04-30T04:43:50Z | 2026-04-26T04:31:05Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | chinazll | 266,670,425 | U_kgDOD-URWQ | User | false |
openclaw/openclaw | 4,212,677,368 | I_kwDOQb6kR877GFr4 | 61,994 | https://github.com/openclaw/openclaw/issues/61994 | https://api.github.com/repos/openclaw/openclaw/issues/61994 | [Feishu] Group chat session abnormally ends (status: done) causing no response to group messages, DM works fine | ## Description
In Feishu channel, group chat sessions abnormally enter `done` status, causing the bot to not respond to group messages. DM sessions remain `running` and work correctly.
**Core symptoms:**
- DM session: `status: running` ✅
- Group session: `status: done` ❌ (ends shortly after conversation, no idle time... | closed | completed | true | 5 | [] | [] | 2026-04-06T16:06:09Z | 2026-04-30T04:43:49Z | 2026-04-26T04:31:09Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Suidge | 34,770,064 | MDQ6VXNlcjM0NzcwMDY0 | User | false |
openclaw/openclaw | 4,212,685,915 | I_kwDOQb6kR877GHxb | 61,997 | https://github.com/openclaw/openclaw/issues/61997 | https://api.github.com/repos/openclaw/openclaw/issues/61997 | LM Studio local model timeout with Invalid type for input error | ## Description
使用 LM Studio 本地部署模型时经常遇到请求超时问题，返回 HTTP 400: Invalid type for `input`.
即使调整了 `timeoutSeconds` 配置仍然会出现此问题。
## Environment
- **OpenClaw**: v2026.4.5
- **Model**: lm-studio/qwen3.5-35b-a3b
- **API**: http://127.0.0.1:1234/v1
## Steps to Reproduce
1. 配置 LM Studio 作为本地模型后端
2. 通过 OpenClaw 发起模型请求
3. 经常遇到超时和 ... | closed | completed | true | 2 | [] | [] | 2026-04-06T16:08:03Z | 2026-04-30T04:43:49Z | 2026-04-26T04:31:14Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | q7793527 | 139,299,979 | U_kgDOCE2Miw | User | false |
openclaw/openclaw | 4,213,130,272 | I_kwDOQb6kR877H0Qg | 62,047 | https://github.com/openclaw/openclaw/issues/62047 | https://api.github.com/repos/openclaw/openclaw/issues/62047 | [Bug]: fallover loop | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
I testing gemma4:26b on my OC.
I have OC continuously fallover to another model and it starts to loop around the models. Also it starts from the last one that is a qwen3.5:4b on the fallover list.
These are the messages I ... | closed | completed | true | 6 | [
"bug",
"regression"
] | [] | 2026-04-06T17:43:10Z | 2026-04-30T04:43:46Z | 2026-04-26T04:31:19Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xeonai44 | 261,166,030 | U_kgDOD5ETzg | User | false |
openclaw/openclaw | 4,213,161,641 | I_kwDOQb6kR877H76p | 62,051 | https://github.com/openclaw/openclaw/issues/62051 | https://api.github.com/repos/openclaw/openclaw/issues/62051 | v2026.4.5 regression: worker processes load all plugins, causing CPU saturation | ## Bug Report: v2026.4.5 Plugin Loading Regression
### Summary
Upgrading from v2026.4.2 to v2026.4.5 causes every worker child process (`openclaw`, `openclaw-cron`, `openclaw-approvals`) to independently register all plugins (BlockRun, partner tools, etc.), creating massive CPU overhead. On a Mac Mini with 8-10 cores... | closed | completed | true | 2 | [] | [] | 2026-04-06T17:50:19Z | 2026-04-30T04:43:45Z | 2026-04-26T04:31:23Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | bryangauvin | 36,140,319 | MDQ6VXNlcjM2MTQwMzE5 | User | false |
openclaw/openclaw | 4,213,313,469 | I_kwDOQb6kR877Ig-9 | 62,065 | https://github.com/openclaw/openclaw/issues/62065 | https://api.github.com/repos/openclaw/openclaw/issues/62065 | [Feature]: Feature Request: Add MiniMax as memory search embedding provider | ### Summary
## Feature Request: Add MiniMax as memory search embedding provider
### Problem
MiniMax has an official Embeddings API, but OpenClaw's memory search
doesn't support MiniMax as an embedding provider.
### Proposed Solution
Add MiniMax to the list of supported embedding providers in memory search.
### Altern... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-06T18:20:49Z | 2026-04-30T04:43:45Z | 2026-04-26T04:31:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | wordhelo | 227,120,888 | U_kgDODYmW-A | User | false |
openclaw/openclaw | 4,214,541,768 | I_kwDOQb6kR877NM3I | 62,182 | https://github.com/openclaw/openclaw/issues/62182 | https://api.github.com/repos/openclaw/openclaw/issues/62182 | Config validation rejects `bootstrapMaxCharsPerFile` as unrecognized key (2026.4.5) | ## Bug Report
### Version
- OpenClaw: 2026.4.5
### Description
Setting `bootstrapMaxCharsPerFile` in `openclaw.json` causes config validation to fail:
```
Config invalid
File: ~/.openclaw/openclaw.json
Problem:
- <root>: Unrecognized key "bootstrapMaxCharsPerFile"
```
This key was previously accepted (or at leas... | closed | completed | true | 1 | [] | [] | 2026-04-06T23:18:02Z | 2026-04-30T04:43:44Z | 2026-04-26T04:31:33Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mjamiv | 142,179,942 | U_kgDOCHl-Zg | User | false |
openclaw/openclaw | 4,214,554,128 | I_kwDOQb6kR877NP4Q | 62,185 | https://github.com/openclaw/openclaw/issues/62185 | https://api.github.com/repos/openclaw/openclaw/issues/62185 | feat: preHook gate for cron and heartbeat pre-run script checks | ## Problem
Operators running OpenClaw on resource-constrained hosts (Raspberry Pi, shared VMs, metered cloud) have no native way to gate cron or heartbeat agent runs on system preconditions — disk space, network reachability, API availability, or dependent service health. Everyone rebuilds custom wrapper logic indepen... | closed | not_planned | true | 2 | [] | [] | 2026-04-06T23:22:39Z | 2026-04-30T04:43:43Z | 2026-04-26T04:31:37Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | scottgl9 | 10,623,288 | MDQ6VXNlcjEwNjIzMjg4 | User | false |
openclaw/openclaw | 4,214,923,145 | I_kwDOQb6kR877Op-J | 62,226 | https://github.com/openclaw/openclaw/issues/62226 | https://api.github.com/repos/openclaw/openclaw/issues/62226 | [Bug]: TypeError: undefined is not iterable (cannot read property Symbol(Symbol.iterator)) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
version 4.5, when input the something inside the chat, something wrong is under the following image.
<img width="1180" height="240" alt="Image" src="https://github.com/user-attachments/assets/2f4c0e08-70f5-43b0-8a42-cd8a0b8... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-07T01:35:35Z | 2026-04-30T04:43:42Z | 2026-04-26T04:31:42Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | TonyLv | 2,041,151 | MDQ6VXNlcjIwNDExNTE= | User | false |
openclaw/openclaw | 4,214,940,639 | I_kwDOQb6kR877OuPf | 62,229 | https://github.com/openclaw/openclaw/issues/62229 | https://api.github.com/repos/openclaw/openclaw/issues/62229 | Google Chat inbound webhook returns 401 after upgrade, outbound still works | After upgrading OpenClaw, Google Chat inbound messages stopped working. Outbound sending still works, but messages sent from users to the bot no longer make it through normal handling. On the Google side, the webhook request is returning HTTP 401 Unauthorized.
This integration was working before the upgrade. No intent... | closed | not_planned | true | 1 | [] | [] | 2026-04-07T01:42:33Z | 2026-04-30T04:43:41Z | 2026-04-26T04:33:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | lxkk7230-kk | 273,990,251 | U_kgDOEFTCaw | User | false |
openclaw/openclaw | 4,215,037,765 | I_kwDOQb6kR877PF9F | 62,245 | https://github.com/openclaw/openclaw/issues/62245 | https://api.github.com/repos/openclaw/openclaw/issues/62245 | [Bug]: 微信插件不支持技能自动触发 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
主题：微信插件不支持技能自动触发
问题描述：
在Telegram中，file-processor技能配置auto: true正常工作，助理完全静默。
在微信中，相同配置不工作，技能从不自动触发，助理只能自述回复。
影响：
微信中文件处理功能失效，用户体验差。
已尝试：
- 升级到最新版本2026.4.5
- 各种技能配置修改
- 规则文件强制行为
- 多次重启
请求：
调查微信插件是否支持技能自动触发，如不支... | closed | not_planned | true | 1 | [
"bug"
] | [] | 2026-04-07T02:18:33Z | 2026-04-30T04:43:41Z | 2026-04-26T04:33:24Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xinbinhan | 50,063,263 | MDQ6VXNlcjUwMDYzMjYz | User | false |
openclaw/openclaw | 4,215,170,451 | I_kwDOQb6kR877PmWT | 62,268 | https://github.com/openclaw/openclaw/issues/62268 | https://api.github.com/repos/openclaw/openclaw/issues/62268 | [Bug]: Gateway entered restart loop when Slack enabled | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
I've set up Telegram for 4 agents, and works well. But Slack always killed the gateway. Gateway crashed, then reopen, and crashed again, and looping... It got back to normal only if Slack is disabled..
### Steps to reproduce... | closed | completed | true | 1 | [
"bug",
"bug:crash"
] | [] | 2026-04-07T03:08:56Z | 2026-04-30T04:43:40Z | 2026-04-26T04:33:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | stuartxu2 | 15,028,997 | MDQ6VXNlcjE1MDI4OTk3 | User | false |
openclaw/openclaw | 4,330,013,031 | I_kwDOQb6kR88AAAABAhbBZw | 71,956 | https://github.com/openclaw/openclaw/issues/71956 | https://api.github.com/repos/openclaw/openclaw/issues/71956 | memory-core: narrative session cleanup fails with "missing scope: operator.admin" during dreaming phases | ## Description
The memory-core plugin's dreaming promotion cycle fails to clean up narrative sessions for all three phases (light, rem, deep). The cleanup step errors with `missing scope: operator.admin`, suggesting the internal API call lacks the required permission scope.
## Environment
- OpenClaw: 2026.4.21 (f788... | closed | not_planned | true | 2 | [] | [] | 2026-04-26T04:25:20Z | 2026-04-29T04:37:20Z | 2026-04-26T10:44:54Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | kobewang1234 | 20,372,811 | MDQ6VXNlcjIwMzcyODEx | User | false |
openclaw/openclaw | 4,215,497,334 | I_kwDOQb6kR877Q2J2 | 62,304 | https://github.com/openclaw/openclaw/issues/62304 | https://api.github.com/repos/openclaw/openclaw/issues/62304 | Feishu reply routed to webchat instead of Feishu in 2026.4.5 | **Describe the bug**
After upgrading to OpenClaw 2026.4.5, Feishu replies are sent to webchat instead of Feishu channel. Messages are received from Feishu but replies are not delivered back to Feishu.
**To Reproduce**
1. Send a message from Feishu to the bot
2. The bot receives the message correctly
3. Reply appears i... | closed | completed | true | 1 | [] | [] | 2026-04-07T04:53:15Z | 2026-04-30T04:43:39Z | 2026-04-26T04:33:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rogerwengch | 209,658,749 | U_kgDODH8jfQ | User | false |
openclaw/openclaw | 4,215,717,312 | I_kwDOQb6kR877Rr3A | 62,321 | https://github.com/openclaw/openclaw/issues/62321 | https://api.github.com/repos/openclaw/openclaw/issues/62321 | [Bug] OOM crash when running large parallel subagent tasks — no concurrency cap or memory pressure circuit breaker | ## Bug Report: OOM System Crash Due to Unbounded Parallel Subagents
### Environment
- **OpenClaw version:** 2026.4.5 (3e72c03)
- **OS:** Ubuntu 24.04, Linux 6.8.0-90-generic (x64)
- **RAM:** 8GB (no swap)
- **Node:** v22.22.0
### Summary
Running a complex subagent task that internally spawns multiple parallel sub-s... | closed | completed | true | 1 | [] | [] | 2026-04-07T05:46:54Z | 2026-04-30T04:43:38Z | 2026-04-26T04:33:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | JerryliDe | 127,085,793 | U_kgDOB5Ms4Q | User | false |
openclaw/openclaw | 4,216,203,985 | I_kwDOQb6kR877TirR | 62,371 | https://github.com/openclaw/openclaw/issues/62371 | https://api.github.com/repos/openclaw/openclaw/issues/62371 | [Bug]: | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
In OpenClaw Control UI, dragging and dropping images into the chat window does not send the image to the session. The image appears to be accepted by the UI (shows drop indicator) but is never attached to the message sent to... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-07T07:39:18Z | 2026-04-30T04:43:38Z | 2026-04-26T04:33:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kongyuanming | 8,285,171 | MDQ6VXNlcjgyODUxNzE= | User | false |
openclaw/openclaw | 4,216,233,414 | I_kwDOQb6kR877Tp3G | 62,373 | https://github.com/openclaw/openclaw/issues/62373 | https://api.github.com/repos/openclaw/openclaw/issues/62373 | [Feature]: Option to use Codex OAuth billing with local gateway tool execution (decouple billing from Responses API) | ### Summary
Allow openai-codex/ models to use Codex OAuth authentication for billing while routing tool execution through OpenClaw's native gateway tools (exec, write, read) instead of the Codex Responses API's server-side execution.
### Problem to solve
When using `openai-codex/gpt-5.4` as the primary model in Open... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-07T07:45:42Z | 2026-04-30T04:43:37Z | 2026-04-26T04:33:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | islandpreneur007 | 258,422,617 | U_kgDOD2c3WQ | User | false |
openclaw/openclaw | 4,330,015,913 | I_kwDOQb6kR88AAAABAhbMqQ | 71,957 | https://github.com/openclaw/openclaw/issues/71957 | https://api.github.com/repos/openclaw/openclaw/issues/71957 | Gateway startup race: channels fire before anthropic plugin registers claude-cli harness | ## Summary
The gateway declares itself "ready" and fires channel startup before the `anthropic` plugin (which provides the `claude-cli` agent harness) has finished initializing. This causes immediate channel startup failure on every boot:
```
[gateway] ready (2 plugins: memory-core, memory-wiki; 3.6s)
[gateway/channe... | closed | completed | true | 4 | [] | [] | 2026-04-26T04:27:26Z | 2026-04-29T04:36:17Z | 2026-04-26T12:08:11Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | WolvenRA | 35,547,932 | MDQ6VXNlcjM1NTQ3OTMy | User | false |
openclaw/openclaw | 4,216,730,442 | I_kwDOQb6kR877VjNK | 62,412 | https://github.com/openclaw/openclaw/issues/62412 | https://api.github.com/repos/openclaw/openclaw/issues/62412 | [Bug]: New sessions inherit authProfileOverride from cooldown/rate-limited backup profile instead of using default | ## Bug type
Behavior bug (incorrect state without crash)
## Summary
New sessions are created with `authProfileOverride` set to a non-default auth profile (e.g. `anthropic:backup1`) even when:
1. The `lastGood` field in `auth-profiles.json` correctly points to `anthropic:default`
2. The backup profile is **in cooldo... | closed | not_planned | true | 1 | [] | [] | 2026-04-07T09:25:45Z | 2026-04-30T04:43:36Z | 2026-04-26T04:33:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zeynalnia | 19,180,856 | MDQ6VXNlcjE5MTgwODU2 | User | false |
openclaw/openclaw | 4,216,858,971 | I_kwDOQb6kR877WClb | 62,418 | https://github.com/openclaw/openclaw/issues/62418 | https://api.github.com/repos/openclaw/openclaw/issues/62418 | [Bug] WebChat Control UI displays internal exec notifications in user-visible chat | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
WebChat Control UI shows internal system messages (exec completion notifications, session startup messages) directly in the user-visible chat transcript, which should not be exposed to end users.
### Steps to reproduce
1. ... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-07T09:50:04Z | 2026-04-30T04:43:35Z | 2026-04-26T04:33:53Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | isaone | 17,076,890 | MDQ6VXNlcjE3MDc2ODkw | User | false |
openclaw/openclaw | 4,217,008,336 | I_kwDOQb6kR877WnDQ | 62,430 | https://github.com/openclaw/openclaw/issues/62430 | https://api.github.com/repos/openclaw/openclaw/issues/62430 | [Bug]: openclaw tui crashes with Maximum call stack size exceeded during schema compilation | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
openclaw tui crashes immediately on startup with RangeError: Maximum call stack size exceeded during JSON schema compilation
### Steps to reproduce
1. Run `openclaw tui` on Windows 11
2. Observe immediate crash with "Maximum ... | closed | completed | true | 3 | [
"bug",
"bug:crash"
] | [] | 2026-04-07T10:20:07Z | 2026-04-30T04:43:34Z | 2026-04-26T04:33:57Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | liuqiyong2026 | 273,964,362 | U_kgDOEFRdSg | User | false |
openclaw/openclaw | 4,218,881,604 | I_kwDOQb6kR877dwZE | 62,577 | https://github.com/openclaw/openclaw/issues/62577 | https://api.github.com/repos/openclaw/openclaw/issues/62577 | openclaw tasks cancel fails with 'ACP metadata is missing' - stuck task cannot be cancelled | ## Bug Description
\`openclaw tasks cancel\` fails with error "ACP metadata is missing for [session-key]" when trying to cancel a stuck ACP task. The task appears to be in a zombie state - it shows as \"running\" in task list but is actually stale (has been running for 26+ hours with no progress).
## Steps to Reprodu... | closed | completed | true | 1 | [] | [] | 2026-04-07T15:46:20Z | 2026-04-30T04:43:33Z | 2026-04-26T04:34:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | nullvoid42 | 269,955,406 | U_kgDOEBcxTg | User | false |
openclaw/openclaw | 4,218,931,743 | I_kwDOQb6kR877d8of | 62,583 | https://github.com/openclaw/openclaw/issues/62583 | https://api.github.com/repos/openclaw/openclaw/issues/62583 | Expose health-monitor checkIntervalMs in ChannelHealthMonitorSchema (currently hardcoded 5min) | ## Summary
The gateway health-monitor polling interval is hardcoded at 5 minutes (`DEFAULT_CHECK_INTERVAL_MS = 5 * 6e4`) and there is no config path to override it. The Zod schema for per-channel `healthMonitor` is `.strict()` and only accepts `{ enabled: boolean }`, so any attempt to set an interval via `openclaw.jso... | closed | completed | true | 1 | [] | [] | 2026-04-07T15:55:25Z | 2026-04-30T04:43:32Z | 2026-04-26T04:34:10Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | chrislro | 7,277,502 | MDQ6VXNlcjcyNzc1MDI= | User | false |
openclaw/openclaw | 4,219,018,087 | I_kwDOQb6kR877eRtn | 62,587 | https://github.com/openclaw/openclaw/issues/62587 | https://api.github.com/repos/openclaw/openclaw/issues/62587 | /new or /reset does not clear session-level authProfileOverride and modelOverride between sessions | ## Bug Report
### Problem
When a user runs `/new` or `/reset` in Telegram (or triggers session reset via the gateway), the new session inherits the previous session's `authProfileOverride` and `modelOverride` from the `sessions.json` store. If a prior session ran on Anthropic (Opus/Sonnet), the new session starts on A... | closed | completed | true | 1 | [] | [] | 2026-04-07T16:10:41Z | 2026-04-30T04:43:31Z | 2026-04-26T04:34:14Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | drinkwithrapha | 261,669,649 | U_kgDOD5jDEQ | User | false |
openclaw/openclaw | 4,219,255,640 | I_kwDOQb6kR877fLtY | 62,603 | https://github.com/openclaw/openclaw/issues/62603 | https://api.github.com/repos/openclaw/openclaw/issues/62603 | [Bug] Exec tool silently corrupts PowerShell pipeline variables on Windows (2026.4.5) | ## Summary
The exec tool silently corrupts PowerShell pipeline variables (`$_`, `$_.Property`) on Windows by stripping the `$` before PowerShell receives the command. This is the same bug reported in #6443 (locked), still present in 2026.4.5.
## Reproduction
Any inline PowerShell command using `$_` in a pipeline:
`... | closed | completed | true | 1 | [] | [] | 2026-04-07T16:51:00Z | 2026-04-30T04:43:30Z | 2026-04-26T04:34:18Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Hag-Fish | 271,590,354 | U_kgDOEDAj0g | User | false |
openclaw/openclaw | 4,219,649,152 | I_kwDOQb6kR877gryA | 62,652 | https://github.com/openclaw/openclaw/issues/62652 | https://api.github.com/repos/openclaw/openclaw/issues/62652 | 2026.4.5: Bonjour/mDNS watchdog causes repeated SIGKILL on production VPS under load | ## Summary
On a heavily-used production VPS, upgrading from 2026.3.22 → 2026.4.5 causes the gateway to be repeatedly restarted and eventually SIGKILLed by systemd. The root trigger appears to be the bonjour/mDNS watchdog entering a stuck state, but the process then fails to shut down cleanly, leading to systemd issuin... | closed | completed | true | 1 | [] | [] | 2026-04-07T18:02:10Z | 2026-04-30T04:43:29Z | 2026-04-26T04:34:23Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jtpcap | 258,234,293 | U_kgDOD2RXtQ | User | false |
openclaw/openclaw | 4,329,955,937 | I_kwDOQb6kR88AAAABAhXiYQ | 71,934 | https://github.com/openclaw/openclaw/issues/71934 | https://api.github.com/repos/openclaw/openclaw/issues/71934 | Active Memory (Dreaming) embedded runs frequently timeout with Opus model | ## Description
Active Memory plugin's embedded runs (Light/REM dreaming phases) consistently timeout when using Claude Opus 4.6 via Amazon Bedrock. The gateway logs show repeated `surface_error` failover decisions with `reason=timeout`.
## Environment
- OpenClaw: 2026.4.21 (f788c88)
- Node: v22.22.2
- OS: Ubuntu 24.... | closed | completed | true | 2 | [] | [] | 2026-04-26T03:48:06Z | 2026-04-29T04:37:21Z | 2026-04-26T10:44:34Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | kobewang1234 | 20,372,811 | MDQ6VXNlcjIwMzcyODEx | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.