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 | 4,321,171,330 | I_kwDOQb6kR88AAAABAY_Xgg | 70,982 | https://github.com/openclaw/openclaw/issues/70982 | https://api.github.com/repos/openclaw/openclaw/issues/70982 | [Bug]: Ollama remote host autodiscovery impossible — schema validation contradicts documentation | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The documentation states autodiscovery activates when models.providers.ollama is not defined, but for remote Ollama hosts the schema validator rejects any config without a provider block, making autodiscovery s... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-24T07:00:00Z | 2026-04-29T04:34:09Z | 2026-04-26T21:09:44Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | cptdrew | 7,881,770 | MDQ6VXNlcjc4ODE3NzA= | User | false |
openclaw/openclaw | 4,321,202,490 | I_kwDOQb6kR88AAAABAZBROg | 70,986 | https://github.com/openclaw/openclaw/issues/70986 | https://api.github.com/repos/openclaw/openclaw/issues/70986 | memory-wiki lint: wikilink extractor is not markdown-aware; resolver never matches frontmatter titles | ## Summary
`openclaw wiki lint` produces two classes of false/unfixable "Broken wikilink target" warnings that are rooted in the **wikilink extractor** and **link resolver** implementations themselves, not in vault content. After a live vault audit (~150 pages, 84 warnings), roughly all warnings fall into these two bu... | open | null | false | 4 | [] | [] | 2026-04-24T07:05:19Z | 2026-05-02T06:31:56Z | null | NONE | null | 20260502T090030Z | 2026-05-02T09:00:30Z | richardmqq | 14,178,267 | MDQ6VXNlcjE0MTc4MjY3 | User | false |
openclaw/openclaw | 4,321,214,944 | I_kwDOQb6kR88AAAABAZCB4A | 70,988 | https://github.com/openclaw/openclaw/issues/70988 | https://api.github.com/repos/openclaw/openclaw/issues/70988 | Feishu channel crashes during plugin-runtime-deps hot-rebuild: Cannot find package openclaw | ## Bug Description
When a plugin (e.g., Microsoft plugin with `node-edge-tts`) installs new dependencies, the `plugin-runtime-deps` directory gets hot-rebuilt. During this rebuild, the Feishu channel crashes repeatedly with the following error and fails to auto-recover across multiple restart attempts. Only a full Gat... | closed | completed | true | 1 | [] | [] | 2026-04-24T07:07:41Z | 2026-04-28T04:47:07Z | 2026-04-25T19:30:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | WS-Q0758 | 274,219,387 | U_kgDOEFhBew | User | false |
openclaw/openclaw | 4,321,227,299 | I_kwDOQb6kR88AAAABAZCyIw | 70,989 | https://github.com/openclaw/openclaw/issues/70989 | https://api.github.com/repos/openclaw/openclaw/issues/70989 | Cron job messages not accessible from agent's main session context | ## Problem
When a cron job runs, it executes in an `isolated` session. Any information the agent discovers during that cron execution (e.g., scanning emails, finding comments, checking statuses) is not visible to the agent when operating in the main session.
This creates a disconnect: the agent can find information v... | closed | completed | true | 1 | [] | [] | 2026-04-24T07:10:00Z | 2026-04-28T04:47:06Z | 2026-04-25T19:30:22Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fleurytian | 157,778,017 | U_kgDOCWeAYQ | User | false |
openclaw/openclaw | 4,321,114,934 | I_kwDOQb6kR88AAAABAY77Ng | 70,979 | https://github.com/openclaw/openclaw/issues/70979 | https://api.github.com/repos/openclaw/openclaw/issues/70979 | Beta blocker: ollama-provider - plugin appends /v1 to baseUrl in models.json overriding correct config | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
Yes
### Summary
The bundled Ollama provider plugin writes baseUrl: http://host:11434/v1 to models.json regardless of the correct baseUrl: http://host:11434 specified in openclaw.json, silently breaking native Ollama tool call... | closed | completed | true | 1 | [
"bug",
"bug:behavior",
"beta-blocker"
] | [] | 2026-04-24T06:49:14Z | 2026-04-29T04:34:08Z | 2026-04-26T21:09:53Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | cptdrew | 7,881,770 | MDQ6VXNlcjc4ODE3NzA= | User | false |
openclaw/openclaw | 4,321,090,941 | I_kwDOQb6kR88AAAABAY6dfQ | 70,975 | https://github.com/openclaw/openclaw/issues/70975 | https://api.github.com/repos/openclaw/openclaw/issues/70975 | Beta blocker: ollama - qwen3.5:9b produces garbage output under full agent system prompt | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
Yes
### Summary
When OpenClaw sends a full agent system prompt (~42,000 chars) to ollama/qwen3.5:9b, the model produces repetition loops instead of a coherent response; confirmed across multiple sessions on v2026.4.12 and v20... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior",
"beta-blocker"
] | [] | 2026-04-24T06:44:49Z | 2026-04-30T04:35:16Z | 2026-04-27T09:50:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | cptdrew | 7,881,770 | MDQ6VXNlcjc4ODE3NzA= | User | false |
openclaw/openclaw | 4,321,175,417 | I_kwDOQb6kR88AAAABAY_neQ | 70,983 | https://github.com/openclaw/openclaw/issues/70983 | https://api.github.com/repos/openclaw/openclaw/issues/70983 | Feature: Pass Android device GPS location to agents | ## Summary
The OpenClaw Android app has a location sharing setting, but GPS coordinates are not surfaced to agents in message metadata or any queryable API.
## Expected behavior
When a user enables location sharing in the Android app, agents should have access to the device's GPS coordinates — either in inbound mess... | closed | not_planned | true | 1 | [] | [] | 2026-04-24T07:00:40Z | 2026-04-27T05:09:17Z | 2026-04-24T07:14:54Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | NullArbitrage | 263,110,202 | U_kgDOD66-Og | User | false |
openclaw/openclaw | 4,321,187,913 | I_kwDOQb6kR88AAAABAZAYSQ | 70,984 | https://github.com/openclaw/openclaw/issues/70984 | https://api.github.com/repos/openclaw/openclaw/issues/70984 | [Bug]: gpt-image-2 image generation does not use existing openai-codex OAuth and still requires OPENAI_API_KEY | ## Summary
`openai/gpt-image-2` image generation does not appear to reuse an existing `openai-codex` OAuth profile in my setup, even though the docs say it should.
This seems **different from #70885**.
- #70885 is about `missing image scope`
- my case is that OpenClaw appears to go down the **direct OpenAI API-key p... | closed | completed | true | 1 | [] | [] | 2026-04-24T07:02:57Z | 2026-04-27T05:09:16Z | 2026-04-24T07:15:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | eddy3268 | 22,745,821 | MDQ6VXNlcjIyNzQ1ODIx | User | false |
openclaw/openclaw | 4,321,269,249 | I_kwDOQb6kR88AAAABAZFWAQ | 70,992 | https://github.com/openclaw/openclaw/issues/70992 | https://api.github.com/repos/openclaw/openclaw/issues/70992 | [Bug]: `openclaw matrix verify status` throws "Matrix runtime client requires a resolved runtime config" | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The `openclaw matrix verify status` command fails with an error indicating that the runtime config is not passed to the Matrix client resolver, despite the config being loaded internally.
### Steps to reproduc... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-24T07:16:59Z | 2026-04-27T04:30:05Z | 2026-04-24T19:44:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jerry-yuan | 13,380,647 | MDQ6VXNlcjEzMzgwNjQ3 | User | false |
openclaw/openclaw | 4,321,259,716 | I_kwDOQb6kR88AAAABAZEwxA | 70,991 | https://github.com/openclaw/openclaw/issues/70991 | https://api.github.com/repos/openclaw/openclaw/issues/70991 | [Bug]: connection drops between gateway and OpenCL Control + unable to stop running task with /stop and button disapear | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
The connection between the gateway and the OpenClaw Control page drops during long tasks and prevents stopping the model, even when using /stop
### Steps to reproduce
start a long coding task in OpenCLaw
let it run (often ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-24T07:15:15Z | 2026-05-05T04:26:53Z | 2026-05-02T15:41:28Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | alexandre-leng | 72,550,137 | MDQ6VXNlcjcyNTUwMTM3 | User | false |
openclaw/openclaw | 4,158,062,876 | I_kwDOQb6kR8731wEc | 56,225 | https://github.com/openclaw/openclaw/issues/56225 | https://api.github.com/repos/openclaw/openclaw/issues/56225 | [WeChat] TTS should send voice messages instead of audio file attachments | ## Problem
When TTS is enabled with the WeChat (openclaw-weixin) channel, generated audio is sent as a **file attachment** rather than a native **voice message** (语音条).
On WeChat, there's a clear UX difference:
- **Voice message (type 3)**: Plays inline in the chat as a voice note, same as how friends send voice mess... | closed | not_planned | true | 3 | [] | [] | 2026-03-28T06:26:48Z | 2026-04-28T05:02:43Z | 2026-04-25T15:49:26Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yihan1944 | 34,528,525 | MDQ6VXNlcjM0NTI4NTI1 | User | false |
openclaw/openclaw | 4,321,355,176 | I_kwDOQb6kR88AAAABAZKlqA | 70,996 | https://github.com/openclaw/openclaw/issues/70996 | https://api.github.com/repos/openclaw/openclaw/issues/70996 | lossless-claw + thinking + tool-call triggers 400 invalid_request_error, causing fallback | **Version:** OpenClaw 2026.4.21, lossless-claw (bundled)
**Description:**
When `plugins.slots.contextEngine = "lossless-claw"` and the agent has thinking enabled, executing tasks that involve tool calls triggers a fallback to the secondary model.
**Error:**
```
400 invalid_request_error: "thinking is enabled but reas... | closed | not_planned | true | 1 | [] | [] | 2026-04-24T07:31:59Z | 2026-04-27T04:34:45Z | 2026-04-25T03:55:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Lunareith | 262,595,192 | U_kgDOD6bieA | User | false |
openclaw/openclaw | 4,044,760,776 | I_kwDOQb6kR87xFibI | 40,868 | https://github.com/openclaw/openclaw/issues/40868 | https://api.github.com/repos/openclaw/openclaw/issues/40868 | Cron isolated sessions timing out after 2026.3.8 update | ## Description
After updating to OpenClaw 2026.3.8, cron jobs with `sessionTarget: "isolated"` consistently time out. The same tasks work fine when run manually in the main session.
## Timeline
- **2026.3.8 released:** March 9, 07:44 GMT
- **Last successful cron run:** March 8, 09:00 (before update)
- **First timeou... | closed | not_planned | true | 11 | [
"dedupe:parent"
] | [] | 2026-03-09T10:31:59Z | 2026-04-27T04:42:30Z | 2026-04-25T01:21:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | andifischer | 75,562,528 | MDQ6VXNlcjc1NTYyNTI4 | User | false |
openclaw/openclaw | 4,321,397,765 | I_kwDOQb6kR88AAAABAZNMBQ | 70,999 | https://github.com/openclaw/openclaw/issues/70999 | https://api.github.com/repos/openclaw/openclaw/issues/70999 | [Ballast] E2E product pipeline — run remaining 11 T1/T2 SKUs | ## Context
\`run_product_pipeline.mjs\` is proven end-to-end (Glycine 3g + Melatonin 0.5mg both ran successfully on the live Ballast store on 2026-04-24). The pipeline: research bundle → external sources → Shopify draft → Imagen 4 image → Telegram HITL approval with photo.
## Open work
13 T1/T2 SKUs passed vetting in ... | closed | not_planned | true | 1 | [
"invalid"
] | [] | 2026-04-24T07:39:17Z | 2026-04-27T05:07:44Z | 2026-04-24T18:17:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dbental | 16,681,993 | MDQ6VXNlcjE2NjgxOTkz | User | false |
openclaw/openclaw | 4,321,398,697 | I_kwDOQb6kR88AAAABAZNPqQ | 71,000 | https://github.com/openclaw/openclaw/issues/71000 | https://api.github.com/repos/openclaw/openclaw/issues/71000 | [Ballast] Set up CJ Dropshipping credentials to enable supplier wiring | ## Context
\`lib/shopify/cj_wiring.mjs\` (B.6) is fully built and tested. It searches CJ Dropshipping for each SKU and patches the Shopify variant's SKU to \`cjPid:cjVid\` format so orders route automatically. However, \`CJ_API_KEY\` and \`CJ_EMAIL\` are not set in \`.env\`, so supplier wiring is skipped on every pipel... | closed | not_planned | true | 0 | [
"invalid"
] | [] | 2026-04-24T07:39:29Z | 2026-04-27T05:07:46Z | 2026-04-24T18:17:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dbental | 16,681,993 | MDQ6VXNlcjE2NjgxOTkz | User | false |
openclaw/openclaw | 4,321,399,919 | I_kwDOQb6kR88AAAABAZNUbw | 71,001 | https://github.com/openclaw/openclaw/issues/71001 | https://api.github.com/repos/openclaw/openclaw/issues/71001 | [Ballast] Activate hero animation in Shopify theme editor | ## Context
The molecular network animation (nodes + edges floating behind headline text) was designed as the Ballast homepage hero background. The JS asset and Liquid snippet have been uploaded to the store:
- \`assets/ballast-network.js\` — Three.js network animation
- \`snippets/ballast-hero-animation.liquid\` — Liqu... | closed | not_planned | true | 1 | [
"invalid"
] | [] | 2026-04-24T07:39:42Z | 2026-04-27T05:07:44Z | 2026-04-24T18:17:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dbental | 16,681,993 | MDQ6VXNlcjE2NjgxOTkz | User | false |
openclaw/openclaw | 4,321,401,459 | I_kwDOQb6kR88AAAABAZNacw | 71,002 | https://github.com/openclaw/openclaw/issues/71002 | https://api.github.com/repos/openclaw/openclaw/issues/71002 | [Ballast] How We Vet page — customer-facing content with real pass/fail examples | ## Context
Issue #69865 (original stub) is still open. The \`/pages/how-we-vet\` page exists on the store (created unpublished by \`legal_pages.mjs\`) but contains only a template placeholder. The page needs real content.
## What to write
The page should explain the T1–T4 vetting rubric in plain language and show real... | closed | not_planned | true | 1 | [
"invalid"
] | [] | 2026-04-24T07:39:58Z | 2026-04-27T05:07:45Z | 2026-04-24T18:17:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dbental | 16,681,993 | MDQ6VXNlcjE2NjgxOTkz | User | false |
openclaw/openclaw | 4,321,402,768 | I_kwDOQb6kR88AAAABAZNfkA | 71,003 | https://github.com/openclaw/openclaw/issues/71003 | https://api.github.com/repos/openclaw/openclaw/issues/71003 | [Ballast] B.12 evidence expander — ship complete | Completed 2026-04-24. See commit bfb7a79. evidence_expander.mjs + run_evidence_expander.mjs + research_bundle.mjs external sources block. 23 tests green. | closed | completed | true | 1 | [] | [] | 2026-04-24T07:40:12Z | 2026-04-27T05:09:12Z | 2026-04-24T07:40:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dbental | 16,681,993 | MDQ6VXNlcjE2NjgxOTkz | User | false |
openclaw/openclaw | 4,321,460,566 | I_kwDOQb6kR88AAAABAZRBVg | 71,006 | https://github.com/openclaw/openclaw/issues/71006 | https://api.github.com/repos/openclaw/openclaw/issues/71006 | [Feature]: DeepSeek V4 Support | ### Summary
Please add DeepSeek V4 OpenRouter Support
### Problem to solve
Newest DeepSeek V4 Support
### Proposed solution
DeepSeek V4 Support
### Alternatives considered
_No response_
### Impact
DeepSeek V4 Support
### Evidence/examples
_No response_
### Additional information
_No response_ | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-24T07:50:07Z | 2026-04-27T05:09:07Z | 2026-04-24T15:24:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | wbe7 | 43,054,302 | MDQ6VXNlcjQzMDU0MzAy | User | false |
openclaw/openclaw | 4,321,462,693 | I_kwDOQb6kR88AAAABAZRJpQ | 71,007 | https://github.com/openclaw/openclaw/issues/71007 | https://api.github.com/repos/openclaw/openclaw/issues/71007 | [Ballast] White Label Phase 2 — branded supplement sourcing workflow | ## Overview
Phase 2 of the Ballast product strategy: switch validated SKUs from CJ dropship to Ballast-branded white label manufacturing once sales data confirms demand.
**Full plan:** `data/scripts/PLAN_WHITE_LABEL.md`
## The model
- **Phase 1 (now):** CJ dropship — zero inventory risk, ~35–45% margin, generic pac... | closed | not_planned | true | 1 | [
"invalid"
] | [] | 2026-04-24T07:50:29Z | 2026-04-27T05:07:46Z | 2026-04-24T18:17:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dbental | 16,681,993 | MDQ6VXNlcjE2NjgxOTkz | User | false |
openclaw/openclaw | 4,321,509,872 | I_kwDOQb6kR88AAAABAZUB8A | 71,010 | https://github.com/openclaw/openclaw/issues/71010 | https://api.github.com/repos/openclaw/openclaw/issues/71010 | [Feature]: When will WeiXin channel plugin be included in OpenClaw's official plugin management? | ### Summary
When will WeiXin channel plugin be included in OpenClaw's official plugin management? Currently, it can still only be installed via third-party methods.
### Problem to solve
WeiXin channel plugin can still only be installed via third-party methods.
### Proposed solution
WeiXin channel plugin be include... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-24T07:58:53Z | 2026-04-29T04:41:45Z | 2026-04-26T07:27:25Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | yourfei | 13,514,881 | MDQ6VXNlcjEzNTE0ODgx | User | false |
openclaw/openclaw | 4,321,525,742 | I_kwDOQb6kR88AAAABAZU_7g | 71,014 | https://github.com/openclaw/openclaw/issues/71014 | https://api.github.com/repos/openclaw/openclaw/issues/71014 | Enhancement: Missing Agent Status Visibility in Webchat During Tool Execution | # Issue: Missing Agent Status Visibility in Webchat During Tool Execution
## Summary
The webchat interface has no way to indicate the agent is actively working during tool calls, making it impossible for users to distinguish between "agent is working" and "agent is dead/hung."
## Current Behavior
1. When the agent in... | closed | completed | true | 1 | [] | [] | 2026-04-24T08:02:28Z | 2026-04-28T04:47:05Z | 2026-04-25T19:30:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jingram-jpg | 264,282,605 | U_kgDOD8Ch7Q | User | false |
openclaw/openclaw | 4,321,546,045 | I_kwDOQb6kR88AAAABAZWPPQ | 71,017 | https://github.com/openclaw/openclaw/issues/71017 | https://api.github.com/repos/openclaw/openclaw/issues/71017 | [Bug]: [Feishu] Multi-account: all sub-accounts resolve to default account's bot open_id | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When configuring multiple Feishu accounts under `channels.feishu.accounts`,
all sub-accounts (j--ceo, j--cto, j--coo) resolve to the same bot open_id
as the default/main account. As a result, all agent replie... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-24T08:06:41Z | 2026-04-28T04:47:04Z | 2026-04-25T19:30:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jiaoxiaoxiao2025-dot | 250,653,582 | U_kgDODvCrjg | User | false |
openclaw/openclaw | 4,321,560,452 | I_kwDOQb6kR88AAAABAZXHhA | 71,018 | https://github.com/openclaw/openclaw/issues/71018 | https://api.github.com/repos/openclaw/openclaw/issues/71018 | openclaw-lark 插件工具执行步骤不显示（工具執行区域始终显示'暂无工具步骤'） | ## 问题描述
升级到 OpenClaw 2026.4.22 后，飞书消息卡片中的**「工具執行」区域始终显示「暂无工具步骤」**，即使已设置 。
## 环境信息
- **OpenClaw 版本**: 2026.4.22
- **openclaw-lark 插件版本**: 2026.4.7
- **Node.js**: v24.15.0
- **系统**: macOS arm64
- **飞书渠道**: openclaw-lark (官方插件)
## 已尝试的修复
1. ✅ 设置 - 配置已生效
2. ✅ 修改 添加 参数支持 - 未解决问题
3. ✅ 修改 传入全局配置 - 未解决问题
4. ✅ 添加调试日志到 ... | closed | completed | true | 1 | [] | [] | 2026-04-24T08:09:51Z | 2026-04-29T04:33:12Z | 2026-04-27T01:05:13Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | linanzhi | 39,115,668 | MDQ6VXNlcjM5MTE1NjY4 | User | false |
openclaw/openclaw | 4,321,577,004 | I_kwDOQb6kR88AAAABAZYILA | 71,019 | https://github.com/openclaw/openclaw/issues/71019 | https://api.github.com/repos/openclaw/openclaw/issues/71019 | [Bug] Built-in feishu plugin setup fails: @larksuiteoapi/node-sdk not resolved in config CLI context (2026.4.22) | **Description** (问题描述)
When running `openclaw config --section feishu` to configure the built-in Feishu channel, the setup wizard fails with an error that the `@larksuiteoapi/node-sdk` package cannot be found in the environment. This prevents the guided setup from completing.
**Steps to Reproduce** (复现步骤)
1. Run `... | closed | completed | true | 1 | [] | [] | 2026-04-24T08:13:18Z | 2026-04-27T05:07:49Z | 2026-04-24T18:06:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rongtianhua | 45,085,618 | MDQ6VXNlcjQ1MDg1NjE4 | User | false |
openclaw/openclaw | 4,321,579,404 | I_kwDOQb6kR88AAAABAZYRjA | 71,020 | https://github.com/openclaw/openclaw/issues/71020 | https://api.github.com/repos/openclaw/openclaw/issues/71020 | [Bug]: Codex app-server runs ignore agent thinkingDefault unless thinking is explicitly set | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On the Codex app-server run path, agent `thinkingDefault` is ignored. `/status` shows `Think: medium` even when `agentCfg.thinkingDefault = "xhigh"` and `agents.defaults.models[codex/gpt-5.5].params.thinking = ... | closed | completed | true | 1 | [] | [] | 2026-04-24T08:13:47Z | 2026-04-27T05:02:24Z | 2026-04-24T19:28:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | richardmqq | 14,178,267 | MDQ6VXNlcjE0MTc4MjY3 | User | false |
openclaw/openclaw | 4,321,536,979 | I_kwDOQb6kR88AAAABAZVr0w | 71,015 | https://github.com/openclaw/openclaw/issues/71015 | https://api.github.com/repos/openclaw/openclaw/issues/71015 | [Feature]: Pass Android device GPS location to agents | ## Summary
Surface Android companion app GPS coordinates to agents via inbound message metadata when the user has opted in via the existing location sharing toggle.
## Problem to solve
The Android app has a location sharing toggle (Off / While Using with Precise Location ON), but GPS data is never delivered to agent... | closed | duplicate | true | 1 | [] | [] | 2026-04-24T08:04:44Z | 2026-04-27T05:09:12Z | 2026-04-24T08:17:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | NullArbitrage | 263,110,202 | U_kgDOD66-Og | User | false |
openclaw/openclaw | 4,321,423,752 | I_kwDOQb6kR88AAAABAZOxiA | 71,004 | https://github.com/openclaw/openclaw/issues/71004 | https://api.github.com/repos/openclaw/openclaw/issues/71004 | RFC: Contract-first Pi/Codex agent runtime rewrite | ## Summary
This RFC tracks a contract-first path for stabilizing the Pi/Codex agent runtime boundary.
Important scope calibration: this issue is **not** a commitment to do a large rewrite immediately. It is an umbrella that explains why the Phase 1 contract-test suite exists, what those tests protect, and what the sm... | closed | completed | true | 7 | [] | [] | 2026-04-24T07:43:58Z | 2026-04-29T04:41:45Z | 2026-04-26T07:27:20Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,321,666,005 | I_kwDOQb6kR88AAAABAZdj1Q | 71,023 | https://github.com/openclaw/openclaw/issues/71023 | https://api.github.com/repos/openclaw/openclaw/issues/71023 | [Bug]: Slack exec approval buttons do not resolve approvals — commands time out despite user clicking Allow Once | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Slack exec approval interactive buttons (Allow Once / Always Allow / Deny) are displayed correctly via Block Kit, and the gateway acknowledges the button click, but the approval decision is never forwarded to `... | closed | completed | true | 1 | [] | [] | 2026-04-24T08:30:46Z | 2026-04-27T04:40:50Z | 2026-04-25T00:03:10Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | marusan03 | 22,651,034 | MDQ6VXNlcjIyNjUxMDM0 | User | false |
openclaw/openclaw | 4,321,704,972 | I_kwDOQb6kR88AAAABAZf8DA | 71,026 | https://github.com/openclaw/openclaw/issues/71026 | https://api.github.com/repos/openclaw/openclaw/issues/71026 | Auth token sync: claude-cli OAuth token not refreshed in auth-profiles.json | ## Summary
When Claude CLI refreshes its OAuth token (stored in `~/.claude/.credentials.json`), OpenClaw's copy in `~/.openclaw/agents/main/agent/auth-profiles.json` is not updated. This causes the Control UI status page to report `anthropic:claude-cli: expired` even though the underlying CLI auth is valid and working... | closed | completed | true | 3 | [] | [] | 2026-04-24T08:39:17Z | 2026-04-29T04:29:13Z | 2026-04-26T07:27:30Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | WolvenRA | 35,547,932 | MDQ6VXNlcjM1NTQ3OTMy | User | false |
openclaw/openclaw | 4,321,751,132 | I_kwDOQb6kR88AAAABAZiwXA | 71,030 | https://github.com/openclaw/openclaw/issues/71030 | https://api.github.com/repos/openclaw/openclaw/issues/71030 | Dreaming deep promotion: candidates ranked but never promoted due to diff-format snippets | ## Bug Description
The dreaming deep promotion phase ranks candidates for promotion to `MEMORY.md` but **never actually promotes any**, even when candidates clearly meet all thresholds.
## Reproduction
1. Enable dreaming with deep phase (default config works)
2. Have memory entries that get recalled multiple times w... | closed | completed | true | 1 | [] | [] | 2026-04-24T08:48:22Z | 2026-04-29T04:36:54Z | 2026-04-26T10:53:50Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | bobgitmcgrath | 10,177,172 | MDQ6VXNlcjEwMTc3MTcy | User | false |
openclaw/openclaw | 4,321,753,615 | I_kwDOQb6kR88AAAABAZi6Dw | 71,031 | https://github.com/openclaw/openclaw/issues/71031 | https://api.github.com/repos/openclaw/openclaw/issues/71031 | [Bug] session-memory hook saves raw model output (chat-template tokens, NO_REPLY) to memory files, progressively degrading agent responsiveness | ## Description
The bundled `session-memory` hook saves **raw model output** to memory files without any sanitization. This includes:
- Chat-template control tokens (e.g. `[INST]`, `[/INST]`, `<|im_start|>`, `<|im_end|>`)
- `NO_REPLY` and other housekeeping conventions
- System instructions embedded in assistant respon... | closed | not_planned | true | 1 | [] | [] | 2026-04-24T08:48:54Z | 2026-04-29T04:41:44Z | 2026-04-26T07:27:36Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | SweetSophia | 44,297,511 | MDQ6VXNlcjQ0Mjk3NTEx | User | false |
openclaw/openclaw | 4,321,815,134 | I_kwDOQb6kR88AAAABAZmqXg | 71,037 | https://github.com/openclaw/openclaw/issues/71037 | https://api.github.com/repos/openclaw/openclaw/issues/71037 | Bug: DeepSeek models with reasoning enabled return 400 on multi-turn tool-calling conversations | ## Bug Description
When using DeepSeek V3/V4 models (e.g. `deepseek-v4-pro`) via `openai-completions` API with `reasoning: true`, multi-turn conversations that involve **tool calls** fail with HTTP 400:
```
HTTP 400 invalid_request_error: The content[].thinking in the thinking mode must be passed back to the API.
```... | closed | completed | true | 3 | [] | [] | 2026-04-24T09:00:35Z | 2026-04-27T05:09:03Z | 2026-04-24T15:24:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Tyduss | 34,669,075 | MDQ6VXNlcjM0NjY5MDc1 | User | false |
openclaw/openclaw | 4,321,901,470 | I_kwDOQb6kR88AAAABAZr7ng | 71,041 | https://github.com/openclaw/openclaw/issues/71041 | https://api.github.com/repos/openclaw/openclaw/issues/71041 | [Feature]: Prompt Safety Sandbox / Risk Guardrail before agent execution | ### Summary
Add an optional prompt safety sandbox for pre-execution risk checks in OpenClaw workflows.
### Problem to solve
As an OpenClaw user, I noticed another agent product (QClaw) provides a “Lobster Butler - AI Security Sandbox” style capability that performs prompt safety checks before sensitive actions. In O... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-24T09:16:39Z | 2026-04-28T04:47:03Z | 2026-04-25T19:30:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zhangzixuan961128-blip | 247,313,522 | U_kgDODr20cg | User | false |
openclaw/openclaw | 4,321,926,980 | I_kwDOQb6kR88AAAABAZtfRA | 71,043 | https://github.com/openclaw/openclaw/issues/71043 | https://api.github.com/repos/openclaw/openclaw/issues/71043 | [bug] Gemini / Mistral fallback responses prefixed with ~25 [object Object] in Telegram channel |
**Package** : `openclaw` (npm) · via Hostinger wrapper image `ghcr.io/hostinger/hvps-openclaw:latest`
**Version observed** : `OpenClaw 2026.4.21 (f788c88)`
**Environment** : self-hosted Docker, Ubuntu 24.04 ARM64 VPS (Hostinger KVM)
**Agent** : single-agent setup, default model `openai-codex/gpt-5.4` (OAuth ChatGPT Pl... | closed | completed | true | 1 | [] | [] | 2026-04-24T09:20:54Z | 2026-04-28T04:47:02Z | 2026-04-25T19:30:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wilsonsemedo92-png | 268,731,174 | U_kgDOEASDJg | User | false |
openclaw/openclaw | 4,321,935,313 | I_kwDOQb6kR88AAAABAZt_0Q | 71,045 | https://github.com/openclaw/openclaw/issues/71045 | https://api.github.com/repos/openclaw/openclaw/issues/71045 | Silent max_tokens=8192 fallback when model isn't declared in models.json causes reasoning models to truncate mid-response | ## Summary
When `agents.defaults.model` (or `agents.list[].model`) points to a provider model that is NOT declared in that agent's `models.json` (e.g., `openrouter/openai/gpt-oss-120b` on an agent whose `providers.openrouter.models` doesn't list that id), OpenClaw resolves the model metadata dynamically from the provi... | closed | completed | true | 1 | [] | [] | 2026-04-24T09:22:19Z | 2026-04-28T04:47:02Z | 2026-04-25T19:30:55Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | souliane | 21,343,492 | MDQ6VXNlcjIxMzQzNDky | User | false |
openclaw/openclaw | 4,148,434,137 | I_kwDOQb6kR873RBTZ | 55,455 | https://github.com/openclaw/openclaw/issues/55455 | https://api.github.com/repos/openclaw/openclaw/issues/55455 | [Bug]: Non-compliant | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading to 2026.3.24, agent tool calls no longer dispatch — gateway logs show only sendMessage entries with zero tool execution activity despite tools being listed as available.
### Steps to reproduce
Upgrade to 20... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-03-27T01:29:24Z | 2026-04-28T05:23:56Z | 2026-04-25T11:07:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | marnimelrose | 83,539,591 | MDQ6VXNlcjgzNTM5NTkx | User | false |
openclaw/openclaw | 4,172,215,606 | I_kwDOQb6kR874rvU2 | 57,826 | https://github.com/openclaw/openclaw/issues/57826 | https://api.github.com/repos/openclaw/openclaw/issues/57826 | [Bug]: [Regression 2026.3.28] AI agent systematically omits required tool parameters | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Bug Description
After updating to OpenClaw 2026.3.28, the AI agent systematically omits required parameters when calling tools, particularly `exec`, `process`, and `read`.
## Error Pattern
### Steps to reproduce
Valida... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-03-30T16:31:12Z | 2026-04-28T04:58:43Z | 2026-04-25T17:06:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rcm700311-hub | 263,833,027 | U_kgDOD7nFww | User | false |
openclaw/openclaw | 4,322,000,858 | I_kwDOQb6kR88AAAABAZx_2g | 71,049 | https://github.com/openclaw/openclaw/issues/71049 | https://api.github.com/repos/openclaw/openclaw/issues/71049 | feat(msteams): add option to disable Copilot-style informative stream status in DMs | ## Summary
The MS Teams channel plugin currently hardcodes the **informative stream status** (Copilot-style "Checking the details..." with gradient progress bar and Stop button) for DM conversations. There is no config option to use the standard Teams typing indicator (`isTyping`) for DMs instead.
The type definition... | open | null | false | 1 | [] | [] | 2026-04-24T09:34:01Z | 2026-05-02T20:26:23Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | software-ec | 229,284,822 | U_kgDODaqb1g | User | false |
openclaw/openclaw | 4,321,993,830 | I_kwDOQb6kR88AAAABAZxkZg | 71,047 | https://github.com/openclaw/openclaw/issues/71047 | https://api.github.com/repos/openclaw/openclaw/issues/71047 | Model switch via /model does not take effect in Telegram topic sessions | ## Bug Description
When switching models using `/model` inside a Telegram **topic (forum) session**, the switch appears to succeed (system message confirms the change), but the actual model remains unchanged.
## Steps to Reproduce
1. Open a Telegram topic session (e.g., a forum group with topics)
2. Send `/model dee... | closed | completed | true | 2 | [] | [] | 2026-04-24T09:32:35Z | 2026-04-27T05:08:08Z | 2026-04-24T17:37:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | nicolas-hey | 258,840,845 | U_kgDOD22ZDQ | User | false |
openclaw/openclaw | 4,322,097,421 | I_kwDOQb6kR88AAAABAZ35DQ | 71,053 | https://github.com/openclaw/openclaw/issues/71053 | https://api.github.com/repos/openclaw/openclaw/issues/71053 | [Feature]: Add explicit web_fetch env-proxy toggle | ### Summary
In proxy-restricted intranet/corporate environments, `web_fetch` can become unusable without explicit env-proxy control;
I want add a option to make `web_fetch` should honor `HTTP_PROXY` / `HTTPS_PROXY`.
### Problem to solve
In proxy-restricted intranet/corporate environments, `web_fetch` can become un... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-24T09:50:59Z | 2026-04-29T04:36:53Z | 2026-04-26T10:53:55Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | xialonglee | 22,994,703 | MDQ6VXNlcjIyOTk0NzAz | User | false |
openclaw/openclaw | 4,322,098,742 | I_kwDOQb6kR88AAAABAZ3-Ng | 71,054 | https://github.com/openclaw/openclaw/issues/71054 | https://api.github.com/repos/openclaw/openclaw/issues/71054 | [Bug]: Discord thread-bound native subagent sessions fail before binding without runtime config | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Discord `sessions_spawn` native subagent requests with `thread: true` and `mode: "session"` can fail before child startup with `Unable to create or bind a Discord thread for this subagent session. Session mode ... | closed | completed | true | 0 | [] | [] | 2026-04-24T09:51:12Z | 2026-04-27T04:49:19Z | 2026-04-24T22:55:50Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jai | 1,943,284 | MDQ6VXNlcjE5NDMyODQ= | User | false |
openclaw/openclaw | 4,322,054,956 | I_kwDOQb6kR88AAAABAZ1TLA | 71,052 | https://github.com/openclaw/openclaw/issues/71052 | https://api.github.com/repos/openclaw/openclaw/issues/71052 | Control UI shows 'Attachment unavailable' for valid image_generate MEDIA paths in ~/.openclaw/media/ | ## Summary
Generated images from `image_generate` tool (saved to `~/.openclaw/media/tool-image-generation/`) show as "Attachment unavailable" in Control UI, even though:
1. The file exists and is readable
2. The path is within an allowed `localMediaPreviewRoot`
3. The backend serves the file correctly via `/__openclaw... | closed | completed | true | 2 | [] | [] | 2026-04-24T09:43:30Z | 2026-04-30T04:44:48Z | 2026-04-25T07:09:04Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | modellflieger | 14,354,291 | MDQ6VXNlcjE0MzU0Mjkx | User | false |
openclaw/openclaw | 4,322,206,500 | I_kwDOQb6kR88AAAABAZ-jJA | 71,059 | https://github.com/openclaw/openclaw/issues/71059 | https://api.github.com/repos/openclaw/openclaw/issues/71059 | [Bug]: Gateway restart strips channel credentials during config hydration | ## Description
When running `openclaw restart`, the gateway process truncates channel config objects, losing credentials (bot tokens, app tokens, etc.) during the initial config write.
A channel configured with full credentials:
```json
{"slack": {"enabled": true, "botToken": "xoxb-...", "appToken": "xapp-1-..."}}
``... | closed | completed | true | 1 | [] | [] | 2026-04-24T10:11:03Z | 2026-04-28T04:46:59Z | 2026-04-25T19:31:15Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Jerzyfilatow | 78,585,933 | MDQ6VXNlcjc4NTg1OTMz | User | false |
openclaw/openclaw | 4,322,177,171 | I_kwDOQb6kR88AAAABAZ8wkw | 71,058 | https://github.com/openclaw/openclaw/issues/71058 | https://api.github.com/repos/openclaw/openclaw/issues/71058 | [Feature]: Support for multiple Azure/Teams bots on a single Openclaw Gateway | ### Summary
Add support for mutliple Teams bots for single Openclaw gateway.
### Problem to solve
Currently channels.msteams is a single object (additionalProperties: false). One Azure App Registration, one Teams bot identity for the entire OpenClaw instance. ... | open | null | false | 4 | [
"enhancement"
] | [] | 2026-04-24T10:05:50Z | 2026-05-02T21:25:50Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | ashukla147 | 211,003,448 | U_kgDODJOoOA | User | false |
openclaw/openclaw | 4,322,280,015 | I_kwDOQb6kR88AAAABAaDCTw | 71,060 | https://github.com/openclaw/openclaw/issues/71060 | https://api.github.com/repos/openclaw/openclaw/issues/71060 | gateway stop/restart can leave LaunchAgent unloaded (KeepAlive bypassed) when drain exceeds 1s | ### Summary
On macOS, `openclaw gateway stop` (and by extension `restart`) can leave the LaunchAgent fully unloaded from launchd rather than merely stopped. Once unloaded, `KeepAlive=true` is inert; the service stays down until someone manually bootstraps it back. In one observed case, the gateway remained down for sev... | closed | completed | true | 2 | [] | [] | 2026-04-24T10:24:55Z | 2026-05-03T04:37:25Z | 2026-04-30T13:50:14Z | CONTRIBUTOR | null | 20260503T060030Z | 2026-05-03T06:00:30Z | vishutdhar | 68,405,187 | MDQ6VXNlcjY4NDA1MTg3 | User | false |
openclaw/openclaw | 4,322,303,846 | I_kwDOQb6kR88AAAABAaEfZg | 71,061 | https://github.com/openclaw/openclaw/issues/71061 | https://api.github.com/repos/openclaw/openclaw/issues/71061 | Bug title | **Reporting Agent**: [Agent Name]
**Session ID**: [Session ID]
**Component**: [e.g. heartbeat, whatsapp, channels, gateway, ui, lossless-claw]
**Severity**: [Low / Medium / High / Critical]
### Description
[Clear, concise description of the bug or feature request]
### Reproducible Steps
1.
2.
3.
### Expected Behavio... | closed | not_planned | true | 0 | [
"invalid"
] | [] | 2026-04-24T10:29:38Z | 2026-04-27T05:07:57Z | 2026-04-24T17:59:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | waynegault | 39,455,578 | MDQ6VXNlcjM5NDU1NTc4 | User | false |
openclaw/openclaw | 4,212,196,834 | I_kwDOQb6kR877EQXi | 61,937 | https://github.com/openclaw/openclaw/issues/61937 | https://api.github.com/repos/openclaw/openclaw/issues/61937 | Audio transcription fails with HTTP 400: undici FormData drops fields when dispatcher is present | ## Description
`transcribeOpenAiCompatibleAudio()` fails with HTTP 400: `"you must provide a model parameter"` when sending multipart FormData through `fetchWithSsrFGuard`. The `model` field is silently dropped from the multipart payload.
## Root Cause
`fetchWithSsrFGuard` creates a pinned DNS dispatcher via `create... | closed | completed | true | 5 | [] | [] | 2026-04-06T14:31:04Z | 2026-04-30T04:44:57Z | 2026-04-25T06:58:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BaraBonc | 219,340,109 | U_kgDODRLdTQ | User | false |
openclaw/openclaw | 4,201,998,334 | I_kwDOQb6kR876dWf- | 60,503 | https://github.com/openclaw/openclaw/issues/60503 | https://api.github.com/repos/openclaw/openclaw/issues/60503 | Telegram exec approval inline buttons fail due to callback_query sequentialization timeout | ## Bug Description
Telegram exec approval inline buttons (Allow Once / Allow Always / Deny) consistently fail with `❌ Failed to submit approval` when clicked by the user, even immediately after the approval card is sent.
## Root Cause Analysis
The Telegram bot uses grammY's `sequentialize` middleware, which processe... | closed | completed | true | 4 | [] | [] | 2026-04-03T19:57:00Z | 2026-04-28T05:10:56Z | 2026-04-25T13:39:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yyxiaohui | 235,090,347 | U_kgDODgMxqw | User | false |
openclaw/openclaw | 4,079,399,583 | I_kwDOQb6kR87zJrKf | 47,677 | https://github.com/openclaw/openclaw/issues/47677 | https://api.github.com/repos/openclaw/openclaw/issues/47677 | Feature request: first-class Telegram reaction triggers for agent wake-up and execution | ## Summary
Feature request: make Telegram reactions a first-class control/input surface for OpenClaw.
Today, reaction-driven workflows can be very useful operationally, but they do not feel like a fully native, durable capability. In practice, reaction-only inputs should be able to wake the agent reliably and support... | open | null | false | 5 | [] | [] | 2026-03-15T23:02:33Z | 2026-05-04T01:22:27Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | 100menotu001 | 64,228,916 | MDQ6VXNlcjY0MjI4OTE2 | User | false |
openclaw/openclaw | 4,322,487,203 | I_kwDOQb6kR88AAAABAaProw | 71,066 | https://github.com/openclaw/openclaw/issues/71066 | https://api.github.com/repos/openclaw/openclaw/issues/71066 | Telegram subsystem: getUpdates polling silently non-functional despite reachable API | # Telegram subsystem: getUpdates polling silently non-functional despite reachable API
## Summary
The Telegram channel's long-polling never consumes updates from Telegram's Bot API, even though:
- The bot token is valid (Telegram confirms `getMe` returns the bot profile)
- `api.telegram.org` is fully reachable from t... | open | null | false | 3 | [] | [
"vincentkoc"
] | 2026-04-24T11:07:15Z | 2026-05-02T23:12:42Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | uttungmalkan-source | 264,367,401 | U_kgDOD8HtKQ | User | false |
openclaw/openclaw | 4,322,522,808 | I_kwDOQb6kR88AAAABAaR2uA | 71,069 | https://github.com/openclaw/openclaw/issues/71069 | https://api.github.com/repos/openclaw/openclaw/issues/71069 | [Bug]: Gemma4-26b-a4-it-gguf override is rejected and reverts to gpt-4o | ## Summary
When I try to use the model `Gemma4-26b-a4-it-gguf` with OpenClaw, the model override is rejected and automatically reverted to `github-copilot/gpt-4o`.
## Error message
```
System: [2026-04-24 08:03:10 ADT] Model override not allowed for this agent; reverted to github-copilot/gpt-4o.
```
## Steps to repro... | closed | completed | true | 2 | [] | [] | 2026-04-24T11:14:47Z | 2026-05-01T04:41:19Z | 2026-04-28T22:33:51Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | CyberRaccoonTeam | 195,688,984 | U_kgDOC6n6GA | User | false |
openclaw/openclaw | 4,322,111,441 | I_kwDOQb6kR88AAAABAZ4v0Q | 71,055 | https://github.com/openclaw/openclaw/issues/71055 | https://api.github.com/repos/openclaw/openclaw/issues/71055 | [Windows] exec tool: Multiple shell escaping issues with PowerShell -Command wrapper | ## Summary
The `exec` tool on Windows wraps commands with `PowerShell -Command ...`, causing multiple quoting/escaping issues that lead to command failures or corrupted output.
---
## Bug Details
- Severity: 🔴 High
- Component: exec tool
### BUG-001: Command execution fails with path arguments
**Steps to reproduce:... | closed | completed | true | 2 | [] | [] | 2026-04-24T09:53:21Z | 2026-04-28T04:47:01Z | 2026-04-25T19:31:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | frank998877 | 278,582,903 | U_kgDOEJrWdw | User | false |
openclaw/openclaw | 4,322,155,958 | I_kwDOQb6kR88AAAABAZ7dtg | 71,057 | https://github.com/openclaw/openclaw/issues/71057 | https://api.github.com/repos/openclaw/openclaw/issues/71057 | [WeChat] delivered status field false positive - unreliable delivery confirmation | ## Summary
The WeChat channel's message `delivered` field has false positives: messages show as delivered even when the user did not actually receive them.
---
## Bug Details
- Severity: 🔴 High
- Component: WeChat channel
**Steps to reproduce:**
1. Configure the WeChat channel
2. Send a message
3. Observe the `deli... | closed | not_planned | true | 2 | [] | [] | 2026-04-24T10:01:45Z | 2026-04-28T04:47:00Z | 2026-04-25T19:31:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | frank998877 | 278,582,903 | U_kgDOEJrWdw | User | false |
openclaw/openclaw | 4,322,496,462 | I_kwDOQb6kR88AAAABAaQPzg | 71,067 | https://github.com/openclaw/openclaw/issues/71067 | https://api.github.com/repos/openclaw/openclaw/issues/71067 | Ensure each OpenClaw agent uses the correct GitHub identity when performing GitHub actions | Opened in the wrong repository by mistake.
This tracking issue has been moved to the correct repository owned by simondsmason. Closing this public issue immediately to avoid further confusion. | closed | completed | true | 0 | [] | [] | 2026-04-24T11:09:09Z | 2026-04-27T05:09:11Z | 2026-04-24T11:18:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mason-loki | 272,005,004 | U_kgDOEDZ3jA | User | false |
openclaw/openclaw | 4,322,542,722 | I_kwDOQb6kR88AAAABAaTEgg | 71,070 | https://github.com/openclaw/openclaw/issues/71070 | https://api.github.com/repos/openclaw/openclaw/issues/71070 | [Bug]: 4.22 plugin loader performs runtime npm install into ~/.npm/_cacache, fails under systemd ProtectHome=true | Bug Type
Regression (worked before, now fails)
Summary
On OpenClaw 2026.4.22, the bundled plugin loader attempts runtime npm install of channel plugin dependencies (@slack/bolt, @grammyjs/runner) into /home/<serviceuser>/.npm/_cacache/ during service startup. On systemd units that enforce ProtectHome=true (a standard h... | closed | completed | true | 2 | [] | [
"vincentkoc"
] | 2026-04-24T11:19:00Z | 2026-04-28T04:39:12Z | 2026-04-26T03:21:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tconwellopenclaw | 262,652,437 | U_kgDOD6fCFQ | User | false |
openclaw/openclaw | 4,322,567,636 | I_kwDOQb6kR88AAAABAaUl1A | 71,071 | https://github.com/openclaw/openclaw/issues/71071 | https://api.github.com/repos/openclaw/openclaw/issues/71071 | [Bug]: Stability bundle writer drops error message and stack on startup failures | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When the gateway fails to start due to a config validation error, the stability bundle written to ~/.openclaw/logs/stability/openclaw-stability-*-gateway.startup_failed.json contains {"error": {"name": "Error"}... | closed | not_planned | true | 2 | [
"bug"
] | [] | 2026-04-24T11:24:07Z | 2026-04-29T04:35:59Z | 2026-04-26T12:21:43Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jlewisoakes | 38,225,466 | MDQ6VXNlcjM4MjI1NDY2 | User | false |
openclaw/openclaw | 4,322,601,367 | I_kwDOQb6kR88AAAABAaWplw | 71,074 | https://github.com/openclaw/openclaw/issues/71074 | https://api.github.com/repos/openclaw/openclaw/issues/71074 | Incomplete-turn recovery (reasoning-only, empty, planning-only) not triggered for Google Gemini models | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When using Google Gemini (e.g. gemini-2.5-pro, gemini-3.1-pro, flash, etc.), agent turns that end with reasoning/thinking blocks but no visible user answer do not trigger a retry or nudge, resulting in silent f... | closed | completed | true | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-04-24T11:31:31Z | 2026-04-27T04:36:42Z | 2026-04-25T03:21:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ahmedsaed | 37,080,003 | MDQ6VXNlcjM3MDgwMDAz | User | false |
openclaw/openclaw | 4,322,013,175 | I_kwDOQb6kR88AAAABAZyv9w | 71,050 | https://github.com/openclaw/openclaw/issues/71050 | https://api.github.com/repos/openclaw/openclaw/issues/71050 | DeepSeek V4 thinking mode fails with 400: reasoning_content not passed back in multi-turn conversations | ## Summary
When using DeepSeek V4 Pro or V4 Flash with thinking/reasoning mode enabled, multi-turn conversations fail with:
```
400 The reasoning_content in the thinking mode must be passed back to the API
```
## Environment
- OpenClaw version: 2026.4.22
- Models affected: `deepseek/deepseek-v4-pro`, `deepseek/deep... | closed | completed | true | 2 | [] | [] | 2026-04-24T09:36:30Z | 2026-04-27T05:09:06Z | 2026-04-24T15:24:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rayrage213 | 261,946,476 | U_kgDOD5z8bA | User | false |
openclaw/openclaw | 4,322,610,336 | I_kwDOQb6kR88AAAABAaXMoA | 71,077 | https://github.com/openclaw/openclaw/issues/71077 | https://api.github.com/repos/openclaw/openclaw/issues/71077 | [Bug]: 2026.4.22 regression: async heartbeat prompt resets active heartbeat session, then later periodic heartbeats continue in a different branch | ### Bug type
Regression (worked before, now behaves differently)
### Summary
After upgrading to `OpenClaw 2026.4.22`, async heartbeat/exec-event prompts no longer collapse back into the same recycled heartbeat session branch.
Observed runtime behavior:
- a heartbeat session was actively running in Control UI
- at ... | closed | completed | true | 1 | [] | [] | 2026-04-24T11:33:28Z | 2026-04-28T04:46:57Z | 2026-04-25T19:31:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ismael-81 | 273,866,131 | U_kgDOEFLdkw | User | false |
openclaw/openclaw | 4,322,615,533 | I_kwDOQb6kR88AAAABAaXg7Q | 71,078 | https://github.com/openclaw/openclaw/issues/71078 | https://api.github.com/repos/openclaw/openclaw/issues/71078 | Observability gap: `verifyGoogleChatRequest` reject reasons are swallowed; missing `appPrincipal` presents as opaque 401 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When the Google Chat plugin rejects an inbound webhook, the structured `reason` from `verifyGoogleChatRequest` is discarded by `isAuthorizedGoogleChatTarget` and never logged. An `audienceType: "app-url"` account missing or ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-24T11:34:36Z | 2026-04-27T05:03:32Z | 2026-04-24T19:10:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sacc54 | 258,008,972 | U_kgDOD2DnjA | User | false |
openclaw/openclaw | 4,322,626,917 | I_kwDOQb6kR88AAAABAaYNZQ | 71,079 | https://github.com/openclaw/openclaw/issues/71079 | https://api.github.com/repos/openclaw/openclaw/issues/71079 | [Feature]: Make Feishu card header/footer (resolveCardHeader / resolveCardNote) configurable | ### Summary
Currently every Feishu card sent by OpenClaw is rendered with:
- A blue header bar containing `<emoji> <agent-name>` (from `resolveCardHeader`)
- A footer note `Agent: <id> | Model: <model> | Provider: <provider>` (from `resolveCardNote`)
### Problem to solve
These are hardcoded in `dist/extensions/feis... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-04-24T11:37:08Z | 2026-04-29T04:41:43Z | 2026-04-26T07:27:41Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | JinxQuQ | 54,772,349 | MDQ6VXNlcjU0NzcyMzQ5 | User | false |
openclaw/openclaw | 4,322,473,232 | I_kwDOQb6kR88AAAABAaO1EA | 71,065 | https://github.com/openclaw/openclaw/issues/71065 | https://api.github.com/repos/openclaw/openclaw/issues/71065 | [Bug]: 2026.4.22 regression: heartbeat embedded runs with xiaomi/mimo-v2-pro abort with TypeError reading 'input' | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to `OpenClaw 2026.4.22 (00bd2cf)`, periodic heartbeat runs using `xiaomi/mimo-v2-pro` consistently fail in embedded mode with:
`TypeError: Cannot read properties of undefined (reading 'input')`
This is **not** an async-event prompt cont... | closed | completed | true | 3 | [] | [] | 2026-04-24T11:04:26Z | 2026-04-28T04:46:58Z | 2026-04-25T19:31:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ismael-81 | 273,866,131 | U_kgDOEFLdkw | User | false |
openclaw/openclaw | 4,322,675,128 | I_kwDOQb6kR88AAAABAabJuA | 71,080 | https://github.com/openclaw/openclaw/issues/71080 | https://api.github.com/repos/openclaw/openclaw/issues/71080 | Make CLAUDE_LIVE_MAX_STDOUT_BUFFER_CHARS configurable — 256KB cap causes deterministic session aborts on large MCP/tool configs | ### Problem
Every Claude CLI session aborts before the first reply with:
```
Embedded agent failed before reply: Claude CLI stdout buffer exceeded limit.
```
The failure is deterministic — same error hash on every retry (`sha256:b2802736e694` in my case). All channels are affected (webchat, Telegram, WhatsApp), beca... | closed | completed | true | 2 | [] | [] | 2026-04-24T11:47:22Z | 2026-04-28T04:42:49Z | 2026-04-26T02:36:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | shivamgrover21 | 39,154,072 | MDQ6VXNlcjM5MTU0MDcy | User | false |
openclaw/openclaw | 4,322,703,156 | I_kwDOQb6kR88AAAABAac3NA | 71,082 | https://github.com/openclaw/openclaw/issues/71082 | https://api.github.com/repos/openclaw/openclaw/issues/71082 | Bundled plugin runtime deps install swarm despite `plugins.entries.<id>.enabled=false` via `includeConfiguredChannels: true` in doctor-health-contributions | ## Summary
In OpenClaw `2026.4.22`, setting `plugins.entries.<channel-id>.enabled=false` in `~/.openclaw/openclaw.json` does **not** prevent `doctor-bundled-plugin-runtime-deps` from installing that channel's runtime dependencies. As a result, any command path that triggers health contributions (e.g. `openclaw tasks m... | closed | completed | true | 3 | [] | [] | 2026-04-24T11:53:23Z | 2026-04-27T05:08:07Z | 2026-04-24T17:51:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hyoseop1231 | 126,223,759 | U_kgDOB4YFjw | User | false |
openclaw/openclaw | 4,322,729,243 | I_kwDOQb6kR88AAAABAaedGw | 71,083 | https://github.com/openclaw/openclaw/issues/71083 | https://api.github.com/repos/openclaw/openclaw/issues/71083 | Slack double-reply bug: draftStream.flush() fires after deliverNormally when streaming defaults to "partial" | ## Summary
When a Slack account has no explicit `streaming` config, OpenClaw defaults to `streaming: "partial"` (via `resolveSlackStreamingMode` returning `"partial"` as fallback). This initialises a `draftStream` for every channel message. The `draftStream.flush()` call in the `finally` block of `dispatchPreparedSlac... | closed | completed | true | 2 | [] | [] | 2026-04-24T11:58:31Z | 2026-04-27T04:47:28Z | 2026-04-24T23:43:09Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ella-zanette | 273,837,350 | U_kgDOEFJtJg | User | false |
openclaw/openclaw | 4,322,739,189 | I_kwDOQb6kR88AAAABAafD9Q | 71,084 | https://github.com/openclaw/openclaw/issues/71084 | https://api.github.com/repos/openclaw/openclaw/issues/71084 | Question: should CLI-backed agents be allowed to trigger memoryFlush / preflightCompaction? | ## Context
\`src/auto-reply/reply/agent-runner-memory.ts\` currently short-circuits memory flush and preflight compaction for CLI-backed agents:
- \`runPreflightCompactionIfNeeded\` returns early if \`isCli\`
- \`runMemoryFlushIfNeeded\` has \`canAttemptFlush = ... && !isCli\` and \`shouldFlushMemory = !isCli && ...\... | closed | not_planned | true | 1 | [] | [] | 2026-04-24T12:00:35Z | 2026-04-29T04:33:12Z | 2026-04-27T01:05:22Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | irchelper | 38,723,383 | MDQ6VXNlcjM4NzIzMzgz | User | false |
openclaw/openclaw | 4,322,806,597 | I_kwDOQb6kR88AAAABAajLRQ | 71,087 | https://github.com/openclaw/openclaw/issues/71087 | https://api.github.com/repos/openclaw/openclaw/issues/71087 | Test issue from API | Testing GitHub API access | closed | completed | true | 0 | [] | [] | 2026-04-24T12:12:35Z | 2026-04-27T05:09:10Z | 2026-04-24T12:13:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | le910n7 | 278,591,303 | U_kgDOEJr3Rw | User | false |
openclaw/openclaw | 4,071,433,545 | I_kwDOQb6kR87yrSVJ | 45,178 | https://github.com/openclaw/openclaw/issues/45178 | https://api.github.com/repos/openclaw/openclaw/issues/45178 | macOS: config-triggered gateway restart can leave LaunchAgent 'not loaded / not installed'; doctor --repair recovers | ## Summary
On macOS, config-triggered gateway restarts can sometimes leave the gateway LaunchAgent in a broken state where it appears **not loaded / not installed**.
Typical symptoms:
- `openclaw gateway status` reports the LaunchAgent as not loaded
- `launchctl` says it cannot find the service
- `openclaw gateway re... | closed | completed | true | 5 | [
"dedupe:parent"
] | [] | 2026-03-13T14:23:31Z | 2026-04-29T04:29:11Z | 2026-04-25T05:27:07Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | olmirr | 30,650,652 | MDQ6VXNlcjMwNjUwNjUy | User | false |
openclaw/openclaw | 4,322,808,777 | I_kwDOQb6kR88AAAABAajTyQ | 71,088 | https://github.com/openclaw/openclaw/issues/71088 | https://api.github.com/repos/openclaw/openclaw/issues/71088 | Ollama thinking models (Qwen 3.5) always think in heartbeat despite params.think: false | ## Bug
When using an Ollama thinking model (e.g. `qwen3.5:2b-q4_K_M`) for heartbeat, the model always generates thinking tokens despite `params.think: false` in the model config.
## Config
```json
{
"agents": {
"defaults": {
"heartbeat": {
"model": "ollama/qwen3.5:2b-q4_K_M",
"lightContex... | closed | completed | true | 1 | [] | [] | 2026-04-24T12:13:02Z | 2026-04-27T05:09:10Z | 2026-04-24T12:18:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | le910n7 | 278,591,303 | U_kgDOEJr3Rw | User | false |
openclaw/openclaw | 4,322,872,656 | I_kwDOQb6kR88AAAABAanNUA | 71,095 | https://github.com/openclaw/openclaw/issues/71095 | https://api.github.com/repos/openclaw/openclaw/issues/71095 | Bug: i18n-tasks crashes on main with ActiveSupport Logger NameError | ## Summary
Running `i18n-tasks` on `main` currently crashes with a `NameError` inside ActiveSupport logger setup.
## Error
```text
i18n-tasks
/Users/wtom/.rvm/gems/ruby-3.2.5@popometer/gems/activesupport-7.0.8.7/lib/active_support/logger_thread_safe_level.rb:12:in `<module:LoggerThreadSafeLevel>': uninitialized consta... | closed | not_planned | true | 0 | [
"invalid"
] | [] | 2026-04-24T12:24:27Z | 2026-04-27T05:07:48Z | 2026-04-24T18:09:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | wtom | 1,482,002 | MDQ6VXNlcjE0ODIwMDI= | User | false |
openclaw/openclaw | 4,322,911,834 | I_kwDOQb6kR88AAAABAapmWg | 71,097 | https://github.com/openclaw/openclaw/issues/71097 | https://api.github.com/repos/openclaw/openclaw/issues/71097 | [Feature] Add denylist mode to exec.security for balanced security | ### Summary
Add a denylist exec security mode that allows all commands by default and only blocks/requires approval for explicitly dangerous patterns.
### Problem to solve
The current exec.security has three modes: deny (blocks everything), allowlist (blocks unknown commands until manually approved), and full (allow... | open | null | false | 6 | [
"enhancement",
"security"
] | [] | 2026-04-24T12:31:32Z | 2026-04-29T08:20:46Z | null | NONE | null | 20260429T090020Z | 2026-04-29T09:00:20Z | flamezealot | 11,676,414 | MDQ6VXNlcjExNjc2NDE0 | User | false |
openclaw/openclaw | 4,322,917,267 | I_kwDOQb6kR88AAAABAap7kw | 71,098 | https://github.com/openclaw/openclaw/issues/71098 | https://api.github.com/repos/openclaw/openclaw/issues/71098 | [Bug]: dashboard command fails to open browser on Windows (Silent Failure with False Positive) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
Yes
### Summary
**The openclaw dashboard command on Windows triggers a false-positive success message in the terminal while failing to actually launch the system browser due to a logic error in the openUrl function.**
### Steps to reprodu... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-24T12:32:36Z | 2026-04-27T04:30:35Z | 2026-04-25T02:24:24Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | xinhanliu0216-droid | 260,815,317 | U_kgDOD4u51Q | User | false |
openclaw/openclaw | 4,322,922,234 | I_kwDOQb6kR88AAAABAaqO-g | 71,099 | https://github.com/openclaw/openclaw/issues/71099 | https://api.github.com/repos/openclaw/openclaw/issues/71099 | Telegram channel: ByteString encoding error on Unicode characters (◆ U+25C6) | ## Description
Telegram channel messages fail with a ByteString encoding error whenever the embedded agent's response contains the `◆` (U+25C6, BLACK DIAMOND) character. This character appears to be used internally by Claude CLI for output formatting.
## Error
```
Cannot convert argument to a ByteString because the ... | closed | completed | true | 2 | [] | [] | 2026-04-24T12:33:31Z | 2026-05-03T04:35:32Z | 2026-04-30T18:45:14Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | dooream84-boop | 271,959,582 | U_kgDOEDXGHg | User | false |
openclaw/openclaw | 4,322,927,783 | I_kwDOQb6kR88AAAABAaqkpw | 71,100 | https://github.com/openclaw/openclaw/issues/71100 | https://api.github.com/repos/openclaw/openclaw/issues/71100 | Plugin registry cache hit does not restore command and interactive handler registries | ## Summary
When OpenClaw serves a plugin registry from the loader cache, the cached `PluginRegistry` is restored and activated, but `pluginCommands` and `pluginInteractiveHandlers` are not restored into their global registries. As a result, plugins can appear as loaded while their commands or interactive callback hand... | closed | completed | true | 1 | [] | [] | 2026-04-24T12:34:34Z | 2026-04-27T04:55:11Z | 2026-04-24T21:49:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | BomBastikDE | 48,631,963 | MDQ6VXNlcjQ4NjMxOTYz | User | false |
openclaw/openclaw | 4,087,959,014 | I_kwDOQb6kR87zqU3m | 48,959 | https://github.com/openclaw/openclaw/issues/48959 | https://api.github.com/repos/openclaw/openclaw/issues/48959 | CLI WebSocket handshake timeout - all CLI commands fail to connect gateway | ### Bug type
Regression (worked before, now fails)
### Summary
All CLI commands (cron, config, doctor, etc.) fail to connect to the gateway via WebSocket with handshake timeout. The gateway itself runs normally - channel messaging (Feishu) and browser dashboard both work. Only CLI-gateway RPC connections are broken.... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-17T11:31:41Z | 2026-04-27T05:09:09Z | 2026-04-24T12:42:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | c-amg | 83,049,593 | MDQ6VXNlcjgzMDQ5NTkz | User | false |
openclaw/openclaw | 4,323,068,745 | I_kwDOQb6kR88AAAABAazLSQ | 71,103 | https://github.com/openclaw/openclaw/issues/71103 | https://api.github.com/repos/openclaw/openclaw/issues/71103 | fix(gateway): isLocalDirectRequest ignores `trustedProxies`, breaks WS pairing for Tailscale Serve / reverse-proxy loopback setups | ### Summary
After v2026.4.22 (specifically commit dc2c3a4920 `fix(gateway): harden WS pairing locality`), `isLocalDirectRequest()` in `src/gateway/auth.ts` classifies any request carrying `X-Forwarded-For` / `X-Forwarded-Host` / `X-Forwarded-Proto` / `X-Real-IP` / `Forwarded` headers as non-local — regardless of socke... | closed | not_planned | true | 3 | [] | [] | 2026-04-24T12:59:31Z | 2026-04-29T04:41:42Z | 2026-04-26T07:27:46Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | haymovie | 198,532,158 | U_kgDOC9VcPg | User | false |
openclaw/openclaw | 4,323,105,025 | I_kwDOQb6kR88AAAABAa1ZAQ | 71,106 | https://github.com/openclaw/openclaw/issues/71106 | https://api.github.com/repos/openclaw/openclaw/issues/71106 | Bundled MCP runtimes leaked per-turn by ephemeral embedded sub-agents (active-memory, llm-slug-generator, llm-task, skill-workshop, list.probe) | # Bundled MCP runtimes leaked per-turn by ephemeral embedded sub-agents
**Version:** 2026.4.22 (00bd2cf), installed via `npm i -g openclaw` on Linux (Node 22.22.2)
## Summary
`openclaw gateway` accumulates orphan stdio MCP server processes continuously during normal conversational use. Under a typical configuration ... | closed | completed | true | 2 | [] | [] | 2026-04-24T13:05:59Z | 2026-04-30T04:44:42Z | 2026-04-25T07:15:42Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | LaFleurAdvertising | 14,811,991 | MDQ6VXNlcjE0ODExOTkx | User | false |
openclaw/openclaw | 4,113,739,170 | I_kwDOQb6kR871Mq2i | 51,860 | https://github.com/openclaw/openclaw/issues/51860 | https://api.github.com/repos/openclaw/openclaw/issues/51860 | gateway install/status hardcodes ~/Library/LaunchAgents — prevents boot-without-login on macOS | ## Problem
On macOS, `~/Library/LaunchAgents/` requires a user login to load. This means the gateway **does not start automatically after a power outage or reboot** unless the user physically logs in at the Mac. For a home server or always-on assistant setup, this is a critical failure — the user can be away and have ... | open | null | false | 4 | [] | [] | 2026-03-21T20:20:53Z | 2026-05-04T20:23:18Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | felpsanchez | 268,795,311 | U_kgDOEAV9rw | User | false |
openclaw/openclaw | 4,322,823,012 | I_kwDOQb6kR88AAAABAakLZA | 71,089 | https://github.com/openclaw/openclaw/issues/71089 | https://api.github.com/repos/openclaw/openclaw/issues/71089 | params.think not forwarded to Ollama provider for heartbeat models | ## Bug
When `params.think: false` is configured in the model config (e.g. for Qwen 3.5 thinking models used in heartbeat), the parameter is read from config but **not forwarded** to the Ollama provider API.
## Expected behavior
`params.think: false` should disable thinking mode in the Ollama API request (`options.th... | closed | completed | true | 2 | [] | [] | 2026-04-24T12:15:46Z | 2026-04-27T04:41:58Z | 2026-04-25T01:33:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | le910n7 | 278,591,303 | U_kgDOEJr3Rw | User | false |
openclaw/openclaw | 4,323,129,923 | I_kwDOQb6kR88AAAABAa26Qw | 71,109 | https://github.com/openclaw/openclaw/issues/71109 | https://api.github.com/repos/openclaw/openclaw/issues/71109 | [Bug]: /models lists non-openai-codex fallbacks under openai-codex prefix in chat UI | ### Summary
In chat/UI model listing, `/models` and provider-scoped model list output can incorrectly show fallback models from other providers as if they belong to `openai-codex`.
Example incorrect output:
```text
Models (openai-codex · 🔑 oauth (openai-codex:daczaragosa@gmail.com (daczaragosa@gmail.com))) — showin... | closed | completed | true | 1 | [] | [] | 2026-04-24T13:10:35Z | 2026-04-27T05:08:16Z | 2026-04-24T17:16:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Lunareith | 262,595,192 | U_kgDOD6bieA | User | false |
openclaw/openclaw | 4,323,140,457 | I_kwDOQb6kR88AAAABAa3jaQ | 71,110 | https://github.com/openclaw/openclaw/issues/71110 | https://api.github.com/repos/openclaw/openclaw/issues/71110 | [Bug]: Session MCP runtime leaks (no idle TTL / no dispose on run-end) — exhausts host memory via persistent SSE streams | ### Summary
`createSessionMcpRuntimeManager` in `dist/pi-bundle-mcp-runtime-*.js` creates one MCP runtime per OpenClaw `sessionId` and stores it in `runtimesBySessionId` with **no idle TTL and no automatic disposal when the owning agent run ends**. Each runtime holds a persistent `StreamableHTTPClientTransport` GET SSE... | closed | completed | true | 1 | [] | [] | 2026-04-24T13:12:36Z | 2026-04-30T04:44:41Z | 2026-04-25T07:15:50Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | bradfreels | 179,389,909 | U_kgDOCrFF1Q | User | false |
openclaw/openclaw | 4,323,160,699 | I_kwDOQb6kR88AAAABAa4yew | 71,111 | https://github.com/openclaw/openclaw/issues/71111 | https://api.github.com/repos/openclaw/openclaw/issues/71111 | [Bug]: Global model fallbacks not inherited by session runner (fallbackConfigured: false) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Description:
Despite having agents.defaults.model.fallbacks correctly configured in openclaw.json, the system fails to trigger the failover mechanism when the primary model returns a server error (e.g., HTTP 500). Instead of... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-24T13:16:09Z | 2026-04-27T04:44:51Z | 2026-04-25T00:43:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | amitylabs | 152,884,216 | U_kgDOCRzT-A | User | false |
openclaw/openclaw | 4,323,245,940 | I_kwDOQb6kR88AAAABAa9_dA | 71,114 | https://github.com/openclaw/openclaw/issues/71114 | https://api.github.com/repos/openclaw/openclaw/issues/71114 | [Bug]: Openai gpt-image-2 does not work with LLM proxy in Docker container | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When I use the `gpt` image generation model from the devcontainer, it does not work, firstly because the `OPENCLAW_QA_ALLOW_LOCAL_IMAGE_PROVIDER` env var is not respected by the image-generation-provider, and s... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-24T13:31:01Z | 2026-04-27T04:55:28Z | 2026-04-24T21:28:26Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kvokka | 15,954,013 | MDQ6VXNlcjE1OTU0MDEz | User | false |
openclaw/openclaw | 4,323,302,276 | I_kwDOQb6kR88AAAABAbBbhA | 71,117 | https://github.com/openclaw/openclaw/issues/71117 | https://api.github.com/repos/openclaw/openclaw/issues/71117 | testuppgift marcel | Skapa upp en ny uppgift på kanban boarden för att testa att du kan skapa nya uppgifter | closed | not_planned | true | 0 | [
"invalid"
] | [] | 2026-04-24T13:39:59Z | 2026-04-27T05:07:48Z | 2026-04-24T18:09:09Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pikkenix | 1,273,617 | MDQ6VXNlcjEyNzM2MTc= | User | false |
openclaw/openclaw | 4,323,306,047 | I_kwDOQb6kR88AAAABAbBqPw | 71,118 | https://github.com/openclaw/openclaw/issues/71118 | https://api.github.com/repos/openclaw/openclaw/issues/71118 | [Bug]: /v1/chat/completions returns usage: {0, 0, 0} across providers, despite llm_output hook receiving valid usage | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
The OpenAI-compatible /v1/chat/completions HTTP endpoint returns usage: { prompt_tokens: 0, completion_tokens: 0, total_tokens: 0 } for every backend provider, even though llm_output plugin hooks receive correctly populated ... | closed | completed | true | 0 | [
"bug",
"regression"
] | [] | 2026-04-24T13:40:35Z | 2026-04-27T04:55:10Z | 2026-04-24T21:54:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | FieldForceCyber | 278,258,168 | U_kgDOEJXh-A | User | false |
openclaw/openclaw | 4,323,345,150 | I_kwDOQb6kR88AAAABAbEC_g | 71,120 | https://github.com/openclaw/openclaw/issues/71120 | https://api.github.com/repos/openclaw/openclaw/issues/71120 | [Bug]: HTTP 429 from GitHub Copilot causes 10-minute silent hang instead of immediate failure | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When a primary LLM call to GitHub Copilot's `/responses` endpoint returns **HTTP 429** (e.g. weekly rate limit exhausted), OpenClaw does **not** detect the rate-limit response. The embedded run silently waits the full config... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-24T13:46:54Z | 2026-04-27T04:45:26Z | 2026-04-25T00:31:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zzl360 | 100,331,525 | U_kgDOBfrwBQ | User | false |
openclaw/openclaw | 4,323,438,850 | I_kwDOQb6kR88AAAABAbJxAg | 71,125 | https://github.com/openclaw/openclaw/issues/71125 | https://api.github.com/repos/openclaw/openclaw/issues/71125 | [Bug]: 2026.4.21 active-memory embedded runs with BigModel glm-4-flash-250414 intermittently fail with TypeError reading 'input', strongly correlated with Telegram approval context | ## Bug type
Regression / runtime bug
## Summary
On OpenClaw **2026.4.21 (f788c88)**, active-memory embedded runs using:
- provider: **bigmodel**
- model: **glm-4-flash-250414**
intermittently fail with:
```txt
TypeError: Cannot read properties of undefined (reading 'input')
```
This is **not random across the day*... | closed | completed | true | 1 | [] | [] | 2026-04-24T14:02:14Z | 2026-04-29T04:41:42Z | 2026-04-26T07:27:56Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | xiao398008 | 211,473,177 | U_kgDODJrTGQ | User | false |
openclaw/openclaw | 4,323,448,826 | I_kwDOQb6kR88AAAABAbKX-g | 71,126 | https://github.com/openclaw/openclaw/issues/71126 | https://api.github.com/repos/openclaw/openclaw/issues/71126 | [Bug]: Gemini model usage can lock the gateway/session loop until gateway restart | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Using a Gemini model in OpenClaw can cause the system to enter a stuck/looped state that only clears after a gateway restart.
### Steps to reproduce
Not fully minimized yet, but the issue appears when switching to / runnin... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-24T14:03:56Z | 2026-04-28T04:46:56Z | 2026-04-25T19:31:39Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | f23783 | 95,570,518 | U_kgDOBbJKVg | User | false |
openclaw/openclaw | 4,323,436,073 | I_kwDOQb6kR88AAAABAbJmKQ | 71,123 | https://github.com/openclaw/openclaw/issues/71123 | https://api.github.com/repos/openclaw/openclaw/issues/71123 | [Bug]: openclaw status shows "Tailscale: off" even when Tailscale client is connected | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On macOS, `openclaw status` shows `Tailscale: off` while the local Tailscale CLI is installed and reports an active tailnet connection.
### Steps to reproduce
1. On macOS, install and connect Tailscale.
2. ... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-04-24T14:01:43Z | 2026-04-28T04:42:53Z | 2026-04-26T02:27:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | texaiassistant-gif | 278,991,328 | U_kgDOEKER4A | User | false |
openclaw/openclaw | 4,323,484,703 | I_kwDOQb6kR88AAAABAbMkHw | 71,128 | https://github.com/openclaw/openclaw/issues/71128 | https://api.github.com/repos/openclaw/openclaw/issues/71128 | [Bug]: chat.abort cannot stop active agent runs; only works for chat.send runs | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
### Summary
`chat.abort` currently cannot stop active runs started via the public
`agent` RPC.
It only aborts runs tracked in `chatAbortControllers`, which are created by
`chat.send`.
This makes ... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-24T14:09:58Z | 2026-04-27T05:03:20Z | 2026-04-24T19:15:57Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | xiaomomo | 2,564,689 | MDQ6VXNlcjI1NjQ2ODk= | User | false |
openclaw/openclaw | 4,323,478,537 | I_kwDOQb6kR88AAAABAbMMCQ | 71,127 | https://github.com/openclaw/openclaw/issues/71127 | https://api.github.com/repos/openclaw/openclaw/issues/71127 | [Bug]: Stuck processing sessions are detected but never aborted — gateway requires external restart to recover | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
The diagnostic subsystem detects stuck processing sessions and emits a `stuck session` WARN, but the runtime has no recovery action wired to that detection, so affected sessions remain frozen indefinitely until the gateway proc... | open | null | false | 7 | [
"bug",
"bug:crash"
] | [] | 2026-04-24T14:08:58Z | 2026-05-05T03:27:17Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | EmpireCreator | 12,735,079 | MDQ6VXNlcjEyNzM1MDc5 | User | false |
openclaw/openclaw | 4,323,565,088 | I_kwDOQb6kR88AAAABAbReIA | 71,130 | https://github.com/openclaw/openclaw/issues/71130 | https://api.github.com/repos/openclaw/openclaw/issues/71130 | [Bug]: 2026.4.21/22 — agent/embedded sends empty messages array to OpenAI Responses API, gets 400 (2013), and surface_error path leaks error text into user channels (Telegram) | # [Bug]: 2026.4.21/22 — agent/embedded sends empty messages array to OpenAI Responses API, gets 400 (2013), and surface_error path leaks error text into user channels (Telegram)
## Bug type
Behavior bug + UX bug (user-visible side effect)
## Beta release blocker
No
## Summary
On OpenClaw **2026.4.21 (f788c88)** (an... | closed | completed | true | 2 | [] | [] | 2026-04-24T14:23:49Z | 2026-04-27T04:45:02Z | 2026-04-25T00:39:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | MagnaAI | 265,843,828 | U_kgDOD9h0dA | User | false |
openclaw/openclaw | 3,911,341,541 | I_kwDOQb6kR87pIlXl | 11,483 | https://github.com/openclaw/openclaw/issues/11483 | https://api.github.com/repos/openclaw/openclaw/issues/11483 | Feature: Per-agent TTS voice configuration | ## Summary
Allow configuring a distinct TTS voice per agent in `openclaw.json`, so each agent in a multi-agent setup can have its own voice identity.
## Current behavior
- `messages.tts.elevenlabs.voiceId` sets a **global** voice for all agents.
- The `[[tts:voiceId=...]]` model-driven override exists, but in practi... | closed | completed | true | 7 | [
"enhancement"
] | [] | 2026-02-07T21:52:24Z | 2026-04-28T04:43:01Z | 2026-04-26T02:16:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ImWally26 | 260,066,078 | U_kgDOD4BLHg | User | false |
openclaw/openclaw | 4,323,601,528 | I_kwDOQb6kR88AAAABAbTseA | 71,133 | https://github.com/openclaw/openclaw/issues/71133 | https://api.github.com/repos/openclaw/openclaw/issues/71133 | memory-core: narrative session cleanup fails with missing scope: operator.admin | **Description**
memory-core plugin's narrative session cleanup fails with 'missing scope: operator.admin' error every minute during light/rem/deep phase cleanup cycles.
**Steps to reproduce**
1. Enable memory-core with dreaming feature
2. Wait for narrative generation to complete (light/rem/deep phases)
3. Check logs:... | closed | not_planned | true | 3 | [
"clownfish"
] | [] | 2026-04-24T14:30:37Z | 2026-04-28T04:32:05Z | 2026-04-25T19:53:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | daxiali | 24,525,239 | MDQ6VXNlcjI0NTI1MjM5 | User | false |
openclaw/openclaw | 4,323,372,588 | I_kwDOQb6kR88AAAABAbFuLA | 71,121 | https://github.com/openclaw/openclaw/issues/71121 | https://api.github.com/repos/openclaw/openclaw/issues/71121 | [Bug] DeepSeek V4: reasoning_content must be passed back to API on multi-turn conversations | ## Bug Summary
When using `deepseek-v4-flash` (or other DeepSeek V4 models) with thinking disabled, multi-turn conversations fail with a 400 error after the second user message.
**Error:**
```
400 The reasoning_content in the thinking mode must be passed back to the API.
```
## Steps to Reproduce
1. Configure `deep... | closed | completed | true | 4 | [] | [] | 2026-04-24T13:51:12Z | 2026-04-27T05:09:05Z | 2026-04-24T15:24:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Jason-zhang-0417 | 266,283,621 | U_kgDOD98qZQ | User | false |
openclaw/openclaw | 4,323,645,639 | I_kwDOQb6kR88AAAABAbWYxw | 71,134 | https://github.com/openclaw/openclaw/issues/71134 | https://api.github.com/repos/openclaw/openclaw/issues/71134 | Cron jobs with sessionTarget: isolated leak orphan .jsonl transcripts | ## Summary
Cron jobs configured with `sessionTarget: "isolated"` create a new `.jsonl` session transcript on every run, but the transcripts are never registered in `sessions.json` and `openclaw sessions cleanup` ignores them. This leaks transcript files indefinitely.
## Environment
- OpenClaw `2026.4.22` (commit `00... | closed | duplicate | true | 2 | [
"close:duplicate",
"dedupe:child"
] | [] | 2026-04-24T14:38:40Z | 2026-04-27T04:42:50Z | 2026-04-25T01:20:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ketwi15 | 20,687,695 | MDQ6VXNlcjIwNjg3Njk1 | User | false |
openclaw/openclaw | 4,323,660,861 | I_kwDOQb6kR88AAAABAbXUPQ | 71,136 | https://github.com/openclaw/openclaw/issues/71136 | https://api.github.com/repos/openclaw/openclaw/issues/71136 | Use minimum of agent config, model config, and model discovery for contextTokens/maxTokens | ## Problem
Currently, the context token resolution in OpenClaw follows a strict priority order:
1. Agent config (`agents.defaults.contextTokens` or `agents.list[].contextTokens`)
2. Provider model config (`models.providers.*.models[].contextWindow`)
3. Model discovery cache (`models.json`)
4. Hardcoded default (200,0... | open | null | false | 1 | [] | [] | 2026-04-24T14:41:23Z | 2026-05-03T04:59:44Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | lykeion-dev | 140,231,077 | U_kgDOCFvBpQ | User | false |
openclaw/openclaw | 4,323,674,271 | I_kwDOQb6kR88AAAABAbYInw | 71,138 | https://github.com/openclaw/openclaw/issues/71138 | https://api.github.com/repos/openclaw/openclaw/issues/71138 | [Bug]: Telegram media attachments silently dropped — "Path escapes sandbox root" blocks ~/.openclaw/media/outbound/ | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When an agent sends a file to Telegram using a MEDIA:/workspace/<file> tag, the gateway copies the file to ~/.openclaw/media/outbound/ and then drops it with a path validation error. The attachment is never del... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-24T14:43:53Z | 2026-04-27T04:35:40Z | 2026-04-25T03:39:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mayor686 | 39,762,982 | MDQ6VXNlcjM5NzYyOTgy | User | false |
openclaw/openclaw | 4,323,685,271 | I_kwDOQb6kR88AAAABAbYzlw | 71,139 | https://github.com/openclaw/openclaw/issues/71139 | https://api.github.com/repos/openclaw/openclaw/issues/71139 | Gateway agent harness fails with `spawn codex ENOENT` on Windows (Node can't resolve .cmd wrappers without shell:true) | ### Summary
On Windows, the OpenClaw agent harness spawns `codex` bare-named via `child_process.spawn('codex', ...)`, which returns `ENOENT` because Node's `CreateProcessW` does not honor `PATHEXT` unless `shell: true` is passed. The `codex` CLI installed via `npm install -g @openai/codex` creates `.cmd` and `.ps1` wr... | closed | completed | true | 1 | [] | [] | 2026-04-24T14:45:38Z | 2026-04-27T05:02:23Z | 2026-04-24T19:28:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | haymovie | 198,532,158 | U_kgDOC9VcPg | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.