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 | 4,031,248,834 | I_kwDOQb6kR87wR_nC | 36,871 | https://github.com/openclaw/openclaw/issues/36871 | https://api.github.com/repos/openclaw/openclaw/issues/36871 | [Bug]: CLI and config path parser incorrectly splits custom provider keys containing periods (e.g. llama.cpp) | **Describe the bug**
When configuring custom local models that use periods in their names (such as `llama.cpp` or `vllm.ai`), the config engine silently shatters the provider key into deeply nested objects instead of interpreting it as a single string. This prevents the provider from resolving correctly at runtime, cau... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T23:32:23Z | 2026-04-26T04:30:48Z | 2026-04-23T04:24:57Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | guanyu-zhang | 63,548,771 | MDQ6VXNlcjYzNTQ4Nzcx | User | false |
openclaw/openclaw | 4,031,321,741 | I_kwDOQb6kR87wSRaN | 36,882 | https://github.com/openclaw/openclaw/issues/36882 | https://api.github.com/repos/openclaw/openclaw/issues/36882 | Webchat: HEARTBEAT_OK and heartbeat prompt visible to user instead of being suppressed | ## Bug Report
**OpenClaw version:** 2026.3.2
**Surface:** Webchat
## Describe the bug
On the webchat surface, both the heartbeat prompt message and the \HEARTBEAT_OK\ acknowledgement reply are visible to the user as regular chat messages. They should be silently suppressed / not displayed.
On Telegram this works cor... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-05T23:54:32Z | 2026-04-26T04:30:47Z | 2026-04-23T04:24:59Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | AxelrodAI | 254,786,013 | U_kgDODy-53Q | User | false |
openclaw/openclaw | 4,031,447,154 | I_kwDOQb6kR87wSwBy | 36,911 | https://github.com/openclaw/openclaw/issues/36911 | https://api.github.com/repos/openclaw/openclaw/issues/36911 | Control UI: reading indicator stuck after assistant response completes (runId mismatch) | ## Bug Description
The Control UI webchat shows the reading indicator (three dots / typing animation) permanently after the assistant completes its response. Users must manually refresh the page to see the completed response.
## Steps to Reproduce
1. Open the Control UI webchat
2. Send any message
3. Wait for the as... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T00:29:56Z | 2026-04-26T04:30:46Z | 2026-04-23T04:25:01Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | harshmoney123 | 19,524,701 | MDQ6VXNlcjE5NTI0NzAx | User | false |
openclaw/openclaw | 4,031,535,845 | I_kwDOQb6kR87wTFrl | 36,935 | https://github.com/openclaw/openclaw/issues/36935 | https://api.github.com/repos/openclaw/openclaw/issues/36935 | [Slack] Duplicate messages when using reply_to tags in DMs | ## Description
When the agent uses `[[reply_to_current]]` tags in Slack DM replies, duplicate messages appear in the channel. The first message shows an "(edited)" indicator, and a second identical copy appears immediately after.
## Steps to Reproduce
1. User sends a message in a Slack DM session
2. Agent responds w... | closed | completed | true | 5 | [
"stale"
] | [] | 2026-03-06T00:57:59Z | 2026-04-24T04:26:03Z | 2026-04-22T01:34:32Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | jwilhelmy | 13,839,961 | MDQ6VXNlcjEzODM5OTYx | User | false |
openclaw/openclaw | 4,031,553,895 | I_kwDOQb6kR87wTKFn | 36,940 | https://github.com/openclaw/openclaw/issues/36940 | https://api.github.com/repos/openclaw/openclaw/issues/36940 | Daily Lint Check: Found 1 warnings or errors | - | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T01:03:31Z | 2026-04-26T04:30:46Z | 2026-04-23T04:25:03Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | lzl-mt | 114,667,476 | U_kgDOBtWv1A | User | false |
openclaw/openclaw | 4,031,606,259 | I_kwDOQb6kR87wTW3z | 36,964 | https://github.com/openclaw/openclaw/issues/36964 | https://api.github.com/repos/openclaw/openclaw/issues/36964 | Long-session stickiness: short turn budgets + late compaction/pruning (runbook + config proposal) | ## Summary
Long, tool-heavy sessions can become sticky/unresponsive when turn time budgets are short and compaction/pruning happen too late.
## Environment
- OpenClaw: 2026.2.26
- Model: `openai-codex/gpt-5.3-codex`
- Channel: Telegram (direct)
- Host: local gateway
## Symptom
- Session becomes "stuck" during long ar... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-06T01:20:57Z | 2026-04-26T04:30:45Z | 2026-04-23T04:25:04Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | tadbitrusty | 195,634,316 | U_kgDOC6kkjA | User | false |
openclaw/openclaw | 4,031,647,131 | I_kwDOQb6kR87wTg2b | 36,973 | https://github.com/openclaw/openclaw/issues/36973 | https://api.github.com/repos/openclaw/openclaw/issues/36973 | Companion re-pair after macOS point update should auto-approve when isRepair=true and deviceId+publicKey match | ## Summary
After a macOS point update (26.3.0 → 26.3.1), the OpenClaw companion app sends a re-pair request to the gateway with `isRepair: true`, the same `deviceId`, and the same `publicKey`. The only change is the `platform` string.
The gateway does not auto-approve this request. Instead, the companion shows a "pai... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T01:33:16Z | 2026-04-26T04:30:44Z | 2026-04-23T04:25:06Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bo-blue | 257,455,029 | U_kgDOD1hztQ | User | false |
openclaw/openclaw | 4,031,683,041 | I_kwDOQb6kR87wTpnh | 36,982 | https://github.com/openclaw/openclaw/issues/36982 | https://api.github.com/repos/openclaw/openclaw/issues/36982 | [Bug]: Qwen OAuth refresh token expired or invalid — requires frequent re-authentication since v2026.3.2 | ## Environment
- **Version:** 2026.3.2 (and previous version)
- **OS:** Linux 6.8.0-71-generic
- **Provider:** Qwen-Portal
- **Models affected:** qwen-portal/coder-model, qwen-portal/vision-model
## Describe the bug
Starting from version 2026.3.2 (and the version before), Qwen OAuth tokens frequently expire and the a... | closed | not_planned | true | 10 | [
"stale"
] | [] | 2026-03-06T01:42:23Z | 2026-04-26T04:30:44Z | 2026-04-23T04:25:08Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | WuGc1999 | 91,599,344 | U_kgDOBXWx8A | User | false |
openclaw/openclaw | 4,031,688,654 | I_kwDOQb6kR87wTq_O | 36,983 | https://github.com/openclaw/openclaw/issues/36983 | https://api.github.com/repos/openclaw/openclaw/issues/36983 | [Bug] macOS app crashes on launch when Talk mode was previously enabled (Swift exclusivity violation in TalkOverlayView) | **Environment**
- OpenClaw version: 2026.3.2 (build 2026030290)
- macOS: 26.3 (25D125)
- Hardware: Mac16,13 (MacBook Pro, Apple Silicon ARM-64)
**Description**
When Talk mode is enabled via the menu bar and the app is restarted, it crashes immediately on launch before any UI is visible.
The only workaround is deleting... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T01:43:55Z | 2026-04-26T04:30:43Z | 2026-04-23T04:25:10Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | MarceloVasquez | 28,541,885 | MDQ6VXNlcjI4NTQxODg1 | User | false |
openclaw/openclaw | 4,031,703,356 | I_kwDOQb6kR87wTuk8 | 36,985 | https://github.com/openclaw/openclaw/issues/36985 | https://api.github.com/repos/openclaw/openclaw/issues/36985 | Migrate legacy gateway.bind config on srv1325349 | ## Issue
The gateway config uses legacy `gateway.bind: "0.0.0.0"` format which triggers deprecation warnings.
**Labels needed:** overnight, type:automation, agent:sysadmin, priority:low
## Impact
- CLI commands show config validation errors
- Gateway runs fine but CLI status checks fail to connect properly
## Fix
Ru... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T01:47:20Z | 2026-04-26T04:30:43Z | 2026-04-23T04:25:12Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | TerminalGravity | 15,236,602 | MDQ6VXNlcjE1MjM2NjAy | User | false |
openclaw/openclaw | 4,031,712,478 | I_kwDOQb6kR87wTwze | 36,987 | https://github.com/openclaw/openclaw/issues/36987 | https://api.github.com/repos/openclaw/openclaw/issues/36987 | [Bug] Feishu channel: deliveryContext.to format incorrect causing replies not sent | ## Bug Description
The Feishu channel sets an incorrect format for `deliveryContext.to` in the session, causing agent replies to not be delivered to Feishu users.
## Problem
When a Feishu message is received and processed:
1. Agent correctly generates a reply
2. But the reply is not sent back to Feishu (`replies=0` ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T01:49:56Z | 2026-04-26T04:30:42Z | 2026-04-23T04:25:13Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | happy5318 | 34,227,488 | MDQ6VXNlcjM0MjI3NDg4 | User | false |
openclaw/openclaw | 4,031,735,218 | I_kwDOQb6kR87wT2Wy | 36,991 | https://github.com/openclaw/openclaw/issues/36991 | https://api.github.com/repos/openclaw/openclaw/issues/36991 | Gateway inserts synthetic tool results prematurely during restart/long-poll cycles | ## Summary
During long-running sessions with gateway restarts or extended poll cycles, the gateway appears to flush pending tool calls before their actual tool results arrive, then injects synthetic error tool results into session history.
**Exact error string**
`missing tool result in session history; inserted synthe... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-06T01:56:34Z | 2026-04-26T04:30:41Z | 2026-04-23T04:25:15Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | matthewlush1 | 33,492,488 | MDQ6VXNlcjMzNDkyNDg4 | User | false |
openclaw/openclaw | 4,031,736,428 | I_kwDOQb6kR87wT2ps | 36,993 | https://github.com/openclaw/openclaw/issues/36993 | https://api.github.com/repos/openclaw/openclaw/issues/36993 | model-usage script should error when explicit --model is not present | ## Summary
In the `skills/model-usage` helper script, `--mode current --model <name>` previously returned a successful response even when the requested model was not present in the CodexBar payload.
This made typos or mismatched model names hard to spot, because output could show the requested model with empty costs i... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T01:56:58Z | 2026-04-26T04:30:40Z | 2026-04-23T04:25:17Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | shuofengzhang | 24,763,026 | MDQ6VXNlcjI0NzYzMDI2 | User | false |
openclaw/openclaw | 4,031,737,240 | I_kwDOQb6kR87wT22Y | 36,994 | https://github.com/openclaw/openclaw/issues/36994 | https://api.github.com/repos/openclaw/openclaw/issues/36994 | [Bug]: Exec and tools keep breaking | ### Bug type
Regression (worked before, now fails)
### Summary
my exec and tools, like browser control, all keep breaking.
everything worked on the first run and shortly after keep breaking.
I've updated to the latest, relay on.
installed via pinokio.
What I want is to get defualt capabilities working again like Ope... | closed | not_planned | true | 19 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T01:57:17Z | 2026-04-26T04:30:40Z | 2026-04-23T04:25:19Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | stoutimon | 151,385,386 | U_kgDOCQX1Kg | User | false |
openclaw/openclaw | 4,031,737,592 | I_kwDOQb6kR87wT274 | 36,995 | https://github.com/openclaw/openclaw/issues/36995 | https://api.github.com/repos/openclaw/openclaw/issues/36995 | 【安全分享】Clawguard：OpenClaw 技能/插件风险扫描与拦截（开源） | 大家好，我做了一个个人开源项目 Clawguard，专门用于 OpenClaw 技能/插件的安全扫描与运行前拦截，目标是把风险挡在入口处。
核心能力：
- 技能/插件全面扫描与安全体检
- 动态识别新技能/插件变更
- 规则包自动更新
- OpenClaw 与 Clawguard 源码完整性校验（防篡改）
- 系统关键节点安全检查（密钥、sudoers、Shell 配置）
一键安装提示词：
请阅读 https://github.com/jiangmuran/clawguard 的 docs/INSTALL.md 并在本机安装 clawguard。安装完成后运行 `clawguard scan --path ~/.openclaw... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T01:57:27Z | 2026-04-26T04:30:39Z | 2026-04-23T04:25:20Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | jiangmuran | 72,869,031 | MDQ6VXNlcjcyODY5MDMx | User | false |
openclaw/openclaw | 4,031,799,435 | I_kwDOQb6kR87wUGCL | 37,027 | https://github.com/openclaw/openclaw/issues/37027 | https://api.github.com/repos/openclaw/openclaw/issues/37027 | Bug: session-memory hook not triggered on session archive/delete | ## Problem Description
session-memory hook only triggers on "/new" or "/reset" commands, but not when sessions are automatically archived or deleted. This causes cross-session memory loss for non-TUI channels.
## Root Cause
In gateway-cli, sessions.delete handler archives transcripts but does not trigger any hook ev... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T02:15:23Z | 2026-04-26T04:30:38Z | 2026-04-23T04:25:22Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | KangShuaiFu | 19,404,095 | MDQ6VXNlcjE5NDA0MDk1 | User | false |
openclaw/openclaw | 4,031,810,309 | I_kwDOQb6kR87wUIsF | 37,028 | https://github.com/openclaw/openclaw/issues/37028 | https://api.github.com/repos/openclaw/openclaw/issues/37028 | [Bug] Feishu plugin duplicate registration causes Gateway crash | ## Bug Description
When configuring Feishu channel and modifying the openclaw.json file, the Gateway crashes unexpectedly.
## Steps to Reproduce
1. Install Feishu plugin via npm
2. Add a Feishu channel using: `openclaw channels add --channel feishu --name default`
3. Try to manually add a second Feishu channel or mo... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-06T02:19:33Z | 2026-04-26T04:30:38Z | 2026-04-23T04:25:24Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | MaxHou-infinity | 202,257,032 | U_kgDODA4yiA | User | false |
openclaw/openclaw | 4,031,831,426 | I_kwDOQb6kR87wUN2C | 37,031 | https://github.com/openclaw/openclaw/issues/37031 | https://api.github.com/repos/openclaw/openclaw/issues/37031 | New Provider: Grok (xAI) - Official API Integration | I would like to introduce a new provider for OpenClaw that integrates directly with xAI Grok API.
**Repository:** https://github.com/BIGBEAR-cjx/openclaw-grok-provider
**Features:**
- Support for 4 Grok models (grok-4-fast, grok-4, grok-3, grok-2)
- Cost-effective pricing (33% savings vs HuggingFace)
- 2M token conte... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T02:25:57Z | 2026-04-26T04:30:37Z | 2026-04-23T04:25:26Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | BIGBEAR-cjx | 187,005,831 | U_kgDOCyV7hw | User | false |
openclaw/openclaw | 4,031,852,470 | I_kwDOQb6kR87wUS-2 | 37,036 | https://github.com/openclaw/openclaw/issues/37036 | https://api.github.com/repos/openclaw/openclaw/issues/37036 | [Bug]: In Windows, when opening the OpenClaw dashboard, it shows "Not Found". | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In Windows, when opening the OpenClaw dashboard, it shows "Not Found".
### Steps to reproduce
1. pnpm i openclaw -g
2. openclaw onboard
3. openclaw gateway install
4. openclaw gateway start
5. openclaw dashboard
### Expected behavior
It... | closed | completed | true | 4 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T02:34:20Z | 2026-04-22T04:31:00Z | 2026-04-19T12:26:26Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | eamd-wq | 58,057,732 | MDQ6VXNlcjU4MDU3NzMy | User | false |
openclaw/openclaw | 4,031,867,395 | I_kwDOQb6kR87wUWoD | 37,040 | https://github.com/openclaw/openclaw/issues/37040 | https://api.github.com/repos/openclaw/openclaw/issues/37040 | [Bug]: OpenClaw Bot with No Visible Reply | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
OpenClaw on Azure Ubuntu VM receives inbound messages (Web UI + Telegram), runs agent turns successfully, but produces no user-visible reply.
Runs finish with `status: ok` but output is empty:
- `result.payloads: []`
- assistant `content: ... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-06T02:40:25Z | 2026-04-26T04:30:36Z | 2026-04-23T04:25:28Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | Outlast18363 | 108,510,344 | U_kgDOBne8iA | User | false |
openclaw/openclaw | 4,031,886,966 | I_kwDOQb6kR87wUbZ2 | 37,047 | https://github.com/openclaw/openclaw/issues/37047 | https://api.github.com/repos/openclaw/openclaw/issues/37047 | Discord thread starter metadata injects changing timestamp, busting prompt cache on every turn | ## Problem
When using OpenClaw in a Discord thread, every user message has a `[Thread starter - for context]` block prepended that includes the full metadata JSON — including a `timestamp` field that changes on every message.
Example of what gets injected on every turn:
```
[Thread starter - for context]
Crash 1.10.... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T02:48:33Z | 2026-04-26T04:30:36Z | 2026-04-23T04:25:30Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bigideagames | 78,379,302 | MDQ6VXNlcjc4Mzc5MzAy | User | false |
openclaw/openclaw | 4,031,905,662 | I_kwDOQb6kR87wUf9- | 37,056 | https://github.com/openclaw/openclaw/issues/37056 | https://api.github.com/repos/openclaw/openclaw/issues/37056 | [Feature]: Add retry configuration for embedded agent (subagent) announce operations | ### Summary
Add retry configuration for embedded agent (subagent) announce operations when gateway timeout or network errors occur.
---
### Problem to solve
When embedded agents (subagents) complete their tasks and try to announce results back through the Gateway, they can fail with `network_error` if the Gateway t... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T02:55:43Z | 2026-04-26T04:30:35Z | 2026-04-23T04:25:32Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | fengyuwusong | 16,892,891 | MDQ6VXNlcjE2ODkyODkx | User | false |
openclaw/openclaw | 4,031,907,898 | I_kwDOQb6kR87wUgg6 | 37,057 | https://github.com/openclaw/openclaw/issues/37057 | https://api.github.com/repos/openclaw/openclaw/issues/37057 | feat: RTK exec proxy — compress tool output to save context tokens | ## Problem
OpenClaw agents running on expensive models (Opus, etc.) burn 30-40% of their context window on raw CLI output from exec tool calls. This causes premature compactions, worse reasoning, and higher costs.
## Proposal
Integrate [RTK](https://github.com/rtk-ai/rtk) (Rust Token Killer) as an optional exec prox... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T02:56:22Z | 2026-04-26T04:30:34Z | 2026-04-23T04:25:33Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | aneym | 43,046,102 | MDQ6VXNlcjQzMDQ2MTAy | User | false |
openclaw/openclaw | 4,031,942,370 | I_kwDOQb6kR87wUo7i | 37,082 | https://github.com/openclaw/openclaw/issues/37082 | https://api.github.com/repos/openclaw/openclaw/issues/37082 | feat: Add --agent option to models set command | ## Problem
Currently, the `openclaw models set` command does not support specifying which agent to configure. It only modifies the global default model (`agents.defaults.model`), which affects all agents that don't have an explicit model override.
This causes confusion when users want to change the model for a specif... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T03:08:14Z | 2026-04-26T04:30:34Z | 2026-04-23T04:25:35Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | haoliqiso | 265,504,676 | U_kgDOD9NHpA | User | false |
openclaw/openclaw | 4,031,946,000 | I_kwDOQb6kR87wUp0Q | 37,083 | https://github.com/openclaw/openclaw/issues/37083 | https://api.github.com/repos/openclaw/openclaw/issues/37083 | Control UI: Messages disappear during tool execution | **Bug Description:**
In the Control UI (web dashboard), when the assistant starts executing a tool call (e.g., `exec`, `browser`, `read`, etc.), the user's previous messages disappear from the chat dialog. The messages are still stored correctly on the Gateway (verified by checking the session JSONL file), but they van... | closed | completed | true | 4 | [] | [] | 2026-03-06T03:09:24Z | 2026-04-27T05:02:58Z | 2026-04-24T19:24:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | fangkelvin | 261,776,090 | U_kgDOD5pi2g | User | false |
openclaw/openclaw | 4,031,991,259 | I_kwDOQb6kR87wU03b | 37,097 | https://github.com/openclaw/openclaw/issues/37097 | https://api.github.com/repos/openclaw/openclaw/issues/37097 | [Bug]: Model params.effort not honored for Claude Opus 4.6 with thinking: adaptive | ## Bug Summary
When configuring `thinking: adaptive` and `effort: max` for Claude Opus 4.6 in `openclaw.json`, the `effort` parameter is not honored in the final API request. The request ends up with `output_config.effort: "medium"` instead of `"max"`.
## Configuration Used
```json
{
"models": {
"claude-opus-4... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T03:24:17Z | 2026-04-26T04:30:33Z | 2026-04-23T04:25:38Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | Feng6611 | 74,105,307 | MDQ6VXNlcjc0MTA1MzA3 | User | false |
openclaw/openclaw | 4,032,033,118 | I_kwDOQb6kR87wU_Fe | 37,114 | https://github.com/openclaw/openclaw/issues/37114 | https://api.github.com/repos/openclaw/openclaw/issues/37114 | macOS app crash-loops when Talk mode enabled (TalkOverlayView Swift exclusivity error) | ## Environment
- **OpenClaw version**: 2026.3.2 (85377a2)
- **macOS version**: 26.3 (Tahoe), Build 25D125
- **Hardware**: Mac mini M4 (Mac16,10), 16GB
- **Audio device**: Insta360 Wave USB (speaker/mic)
## Description
Enabling Talk mode in the macOS companion app causes an immediate crash. The app crash-loops on subse... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T03:40:21Z | 2026-04-26T04:30:32Z | 2026-04-23T04:25:40Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | genieyuan | 40,484,674 | MDQ6VXNlcjQwNDg0Njc0 | User | false |
openclaw/openclaw | 4,032,053,417 | I_kwDOQb6kR87wVECp | 37,123 | https://github.com/openclaw/openclaw/issues/37123 | https://api.github.com/repos/openclaw/openclaw/issues/37123 | [Bug]: Azure OpenAI provider not appearing during onboarding | ### Bug type
Regression (worked before, now fails)
### Summary
Azure OpenAI provider does not appear in the onboarding model/auth provider list even after switching to the mani/7feb branch that supposedly adds Azure support.
### Steps to reproduce
1. Clone the forked repository: https://github.com/skyhawk14/opencl... | closed | completed | true | 4 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T03:48:01Z | 2026-04-25T04:16:36Z | 2026-04-22T21:20:52Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | Vignesh8164 | 222,296,409 | U_kgDODT_5WQ | User | false |
openclaw/openclaw | 4,032,055,832 | I_kwDOQb6kR87wVEoY | 37,125 | https://github.com/openclaw/openclaw/issues/37125 | https://api.github.com/repos/openclaw/openclaw/issues/37125 | feat(browser): support per-profile headless mode | ## Problem
The `browser.headless` config flag is currently global — it applies to **all** browser profiles. This is limiting for setups where some profiles need headless mode and others need headed mode.
**Real-world use case:** A user running multiple browser profiles — e.g. a `localhost` profile for automated/backg... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-06T03:49:02Z | 2026-04-26T04:30:31Z | 2026-04-23T04:25:42Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | fwends | 4,113,483 | MDQ6VXNlcjQxMTM0ODM= | User | false |
openclaw/openclaw | 4,032,072,479 | I_kwDOQb6kR87wVIsf | 37,136 | https://github.com/openclaw/openclaw/issues/37136 | https://api.github.com/repos/openclaw/openclaw/issues/37136 | Telegram heartbeat loop: agent sends message bursts every ~10min instead of HEARTBEAT_OK | ## Bug Report
**OpenClaw Version:** 2026.3.2
**OS:** macOS (Darwin 24.6.0, arm64)
**Node:** v22.22.0 (via nvm)
**Channel:** Telegram (bot mode, DM)
### Description
The Telegram session enters a message loop where the heartbeat generates real content instead of `HEARTBEAT_OK`, sending bursts of ~4 messages ever... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-06T03:55:52Z | 2026-04-26T04:29:00Z | 2026-04-24T04:26:01Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | stormchaser-creator | 260,266,770 | U_kgDOD4NbEg | User | false |
openclaw/openclaw | 4,032,076,786 | I_kwDOQb6kR87wVJvy | 37,140 | https://github.com/openclaw/openclaw/issues/37140 | https://api.github.com/repos/openclaw/openclaw/issues/37140 | skill-creator quick_validate: enforce allowed-tools list shape | ## Problem
`skills/skill-creator/scripts/quick_validate.py` accepted malformed `allowed-tools` frontmatter values (for example a scalar string instead of a list, non-string entries, or empty entries).
That can allow invalid skill metadata to pass local validation and fail later in packaging or runtime expectations.
#... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-06T03:57:51Z | 2026-04-26T04:28:59Z | 2026-04-24T04:26:03Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | shuofengzhang | 24,763,026 | MDQ6VXNlcjI0NzYzMDI2 | User | false |
openclaw/openclaw | 4,032,080,240 | I_kwDOQb6kR87wVKlw | 37,142 | https://github.com/openclaw/openclaw/issues/37142 | https://api.github.com/repos/openclaw/openclaw/issues/37142 | [Bug]: MiniMax provider returns HTTP 401 despite valid API key — anthropic-messages mode | ### Bug type
Regression (worked before, now fails)
### Summary
baseUrl: https://api.minimaxi.com/anthropic
api: anthropic-messages
authHeader: true
Key source: env: MINIMAX_API_KEY
Symptoms: openclaw returns HTTP 401 authentication_error: invalid api key on every request to MiniMax.
Verified working: The exact same... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-06T03:59:30Z | 2026-04-26T04:28:58Z | 2026-04-24T04:26:05Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | 050808zhouyu-prog | 247,914,594 | U_kgDODsbgYg | User | false |
openclaw/openclaw | 4,286,713,875 | I_kwDOQb6kR87_ghAT | 68,406 | https://github.com/openclaw/openclaw/issues/68406 | https://api.github.com/repos/openclaw/openclaw/issues/68406 | MCP Server Process Leak - 191 orphaned processes (PPID=1) over 10 hours | ## Environment
- **OpenClaw**: 2026.4.15 (041266a)
- **Node**: v22.22.2
- **OS**: Linux 6.8.0-106-generic (x64)
## Problem
MCP server processes spawn but become orphaned and are never reaped by the parent process.
## Symptoms
- 191 orphaned `minimax-coding-plan-mcp` processes accumulated over ~10 hours
- Each process... | closed | completed | true | 1 | [] | [] | 2026-04-18T04:37:37Z | 2026-04-28T04:51:03Z | 2026-04-25T18:31:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | HankSU7889 | 32,285,831 | MDQ6VXNlcjMyMjg1ODMx | User | false |
openclaw/openclaw | 4,286,691,711 | I_kwDOQb6kR87_gbl_ | 68,401 | https://github.com/openclaw/openclaw/issues/68401 | https://api.github.com/repos/openclaw/openclaw/issues/68401 | Feature: Allow message processing during session compaction | ## Problem
When OpenClaw session compaction triggers (context near threshold), the main session is blocked for ~1-2 minutes during summarization/compaction. During this time, **all incoming messages (Telegram, etc.) are queued and cannot be processed**. This causes:
1. Message delays - users see \"messages come in ba... | open | null | false | 2 | [] | [] | 2026-04-18T04:28:51Z | 2026-05-02T20:57:30Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | kyle20026 | 274,152,656 | U_kgDOEFc80A | User | false |
openclaw/openclaw | 4,286,712,644 | I_kwDOQb6kR87_ggtE | 68,405 | https://github.com/openclaw/openclaw/issues/68405 | https://api.github.com/repos/openclaw/openclaw/issues/68405 | Cron isolated runs fail Anthropic auth resolution despite valid main-session auth | ## Summary
On OpenClaw 2026.4.15, Anthropic auth works in the interactive main session but fails in isolated cron runs that should be able to resolve the same auth profile state.
## Environment
- OpenClaw: 2026.4.15
- Host: macOS (local gateway on Mac mini)
- Working interactive model path: `anthropic/claude-opus-4-7`... | closed | completed | true | 1 | [] | [] | 2026-04-18T04:36:56Z | 2026-04-30T04:51:58Z | 2026-04-25T05:15:21Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Ben-Saliba | 251,801,773 | U_kgDODwIwrQ | User | false |
openclaw/openclaw | 4,286,695,666 | I_kwDOQb6kR87_gcjy | 68,403 | https://github.com/openclaw/openclaw/issues/68403 | https://api.github.com/repos/openclaw/openclaw/issues/68403 | [Bug]: Published deployment token placeholder bypasses gateway weak-secret guard | ## Severity Assessment
### CVSS Assessment
| Metric | v3.1 | v4.0 |
|--------|------|------|
| **Score** | 9.8 / 10.0 | 9.3 / 10.0 |
| **Severity** | Critical | Critical |
| **Vector** | CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H | CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N |
| **Calculator** |... | closed | completed | true | 1 | [] | [] | 2026-04-18T04:30:22Z | 2026-04-21T04:21:49Z | 2026-04-18T17:29:11Z | CONTRIBUTOR | null | 20260421T060038Z | 2026-04-21T06:00:38Z | coygeek | 65,363,919 | MDQ6VXNlcjY1MzYzOTE5 | User | false |
openclaw/openclaw | 4,286,722,951 | I_kwDOQb6kR87_gjOH | 68,409 | https://github.com/openclaw/openclaw/issues/68409 | https://api.github.com/repos/openclaw/openclaw/issues/68409 | Bug: ACP runtime backend marked unavailable when default probe agent (codex) is not configured | ## Bug Description
The embedded acpx runtime backend's health probe **hardcodes codex as the default probe agent**, and if the probe fails, the **entire ACP runtime backend** is marked as unavailable — even if other ACP agents (e.g., OpenCode) are installed and working correctly.
## Root Cause
In `register.runtime-C... | closed | completed | true | 1 | [] | [] | 2026-04-18T04:42:04Z | 2026-04-26T04:31:43Z | 2026-04-23T04:20:55Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | wangype | 6,704,296 | MDQ6VXNlcjY3MDQyOTY= | User | false |
openclaw/openclaw | 4,286,779,407 | I_kwDOQb6kR87_gxAP | 68,410 | https://github.com/openclaw/openclaw/issues/68410 | https://api.github.com/repos/openclaw/openclaw/issues/68410 | [Bug]: Active Memory silently normalizes configured `timeoutMs` above 60000 down to 60000 at runtime | **Bug type**
Behavior bug
**Beta release blocker**
No
**Summary**
On OpenClaw 2026.4.15, `plugins.entries.active-memory.config.timeoutMs` values above `60000` are silently normalized down to `60000` at runtime, so a configured `90000` budget never actually takes effect.
This is especially surprising because the repo... | closed | completed | true | 0 | [] | [] | 2026-04-18T05:00:20Z | 2026-04-21T04:21:46Z | 2026-04-18T19:31:43Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | anyech | 8,743,351 | MDQ6VXNlcjg3NDMzNTE= | User | false |
openclaw/openclaw | 4,286,802,393 | I_kwDOQb6kR87_g2nZ | 68,412 | https://github.com/openclaw/openclaw/issues/68412 | https://api.github.com/repos/openclaw/openclaw/issues/68412 | Streaming SSE response silently dropped between proxy and Slack delivery | ## Summary
When using a Claude Code hybrid proxy (port 8086) as an OpenAI-compatible provider, a model response was generated, streamed to completion via SSE (`stop_reason=end_turn`), and received by the OpenClaw gateway — but never delivered to Slack. No error was logged. The user only received the tool progress mark... | closed | completed | true | 1 | [] | [] | 2026-04-18T05:08:06Z | 2026-04-28T04:51:02Z | 2026-04-25T18:31:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | noodleprincss-ai | 260,147,601 | U_kgDOD4GJkQ | User | false |
openclaw/openclaw | 4,286,813,680 | I_kwDOQb6kR87_g5Xw | 68,413 | https://github.com/openclaw/openclaw/issues/68413 | https://api.github.com/repos/openclaw/openclaw/issues/68413 | [Bug]: moonshot/kimi-k2.5 agent turns timeout before first chunk, direct Moonshot API works | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Version: 2026.4.15
OS: Ubuntu 24.04 VPS
Model: moonshot/kimi-k2.5
Config tried: thinking.type=disabled, compat.supportsTools=false, compat.requiresStringContent=true
Observed: moonshot/kimi-k2.5 produced no reply before the ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-18T05:11:56Z | 2026-04-23T04:22:13Z | 2026-04-21T01:28:08Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | nicolascarreno46-svg | 277,189,417 | U_kgDOEIWTKQ | User | false |
openclaw/openclaw | 4,286,676,640 | I_kwDOQb6kR87_gX6g | 68,398 | https://github.com/openclaw/openclaw/issues/68398 | https://api.github.com/repos/openclaw/openclaw/issues/68398 | Bug: cleanupProvisionalSession not exported from subagent-spawn, causes subagent completion announce to fail | ## Bug type
Code defect (missing export)
## Summary
`cleanupProvisionalSession` is defined in `subagent-spawn-*.js` but not included in the module's export statement. `subagent-registry-*.js` calls this function at the `finally` block of the direct-announce path but never imports it. This causes all subagent complet... | closed | completed | true | 2 | [] | [] | 2026-04-18T04:23:15Z | 2026-04-30T04:39:20Z | 2026-04-26T05:21:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yuxiaoyang2007-prog | 258,562,502 | U_kgDOD2lZxg | User | false |
openclaw/openclaw | 4,286,831,774 | I_kwDOQb6kR87_g9ye | 68,414 | https://github.com/openclaw/openclaw/issues/68414 | https://api.github.com/repos/openclaw/openclaw/issues/68414 | Bug: Inbound message during agent turn completion silently dropped (race condition) | ## Bug: Inbound message during agent turn completion window silently dropped (never triggers new turn)
### Description
When a text message arrives in the narrow window while the agent is finishing its current turn (sending final responses) and transitioning back to idle, the message is:
- ✅ Received by the gateway (T... | closed | completed | true | 2 | [] | [] | 2026-04-18T05:17:27Z | 2026-04-28T04:51:02Z | 2026-04-25T18:32:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Herman8814 | 153,133,947 | U_kgDOCSCjew | User | false |
openclaw/openclaw | 4,286,896,948 | I_kwDOQb6kR87_hNs0 | 68,419 | https://github.com/openclaw/openclaw/issues/68419 | https://api.github.com/repos/openclaw/openclaw/issues/68419 | GPT-5 bootstrap alias files for AGENTS/SOUL via context-limiter path | ## Summary
Add GPT-5-family bootstrap alias loading so runs can use `AGENTS.gpt5.md` and `SOUL.gpt5.md` in place of canonical bootstrap files when present, without editing the originals.
## Goal
Prioritize agentic workflow quality using the existing bootstrap/context-limiter path rather than changing the main system p... | closed | not_planned | true | 1 | [] | [] | 2026-04-18T05:37:25Z | 2026-04-28T04:51:01Z | 2026-04-25T18:32:10Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,164,599,303 | I_kwDOQb6kR874Or4H | 57,099 | https://github.com/openclaw/openclaw/issues/57099 | https://api.github.com/repos/openclaw/openclaw/issues/57099 | [Bug]: Explicit api ollama provider config fails with No API provider registered after upgrade to 2026.3.28 | ## Bug type
Regression (worked before, now fails)
## Summary
After upgrading from v2026.3.24 to v2026.3.28, any explicitly configured provider with api ollama fails immediately with:
Error: No API provider registered for api: ollama
This worked perfectly on v2026.3.24.
## Environment
- OpenClaw: 2026.3.28 (f9b10... | closed | completed | true | 11 | [] | [] | 2026-03-29T14:44:21Z | 2026-04-28T04:30:33Z | 2026-04-25T11:58:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Skrblik | 139,452,656 | U_kgDOCE_g8A | User | false |
openclaw/openclaw | 4,286,935,513 | I_kwDOQb6kR87_hXHZ | 68,423 | https://github.com/openclaw/openclaw/issues/68423 | https://api.github.com/repos/openclaw/openclaw/issues/68423 | `maybeRecoverSuspiciousConfigRead` should validate `.bak` is not polluted with redacted values before restoring | ## Problem
OpenClaw's config self-healing mechanism (`maybeRecoverSuspiciousConfigRead`) restores from `.bak` when the main config file appears suspicious, but **does not validate whether `.bak` itself has been polluted** with redacted values.
This creates a pollution fixation chain:
1. An agent accidentally writes ... | open | null | false | 1 | [] | [] | 2026-04-18T05:51:00Z | 2026-05-02T21:50:05Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | qiqi-victor | 242,737,599 | U_kgDODnfhvw | User | false |
openclaw/openclaw | 4,286,936,179 | I_kwDOQb6kR87_hXRz | 68,425 | https://github.com/openclaw/openclaw/issues/68425 | https://api.github.com/repos/openclaw/openclaw/issues/68425 | Redacted tool output should include a machine-readable marker so agents can detect redaction | ## Problem
OpenClaw's `redact-D4nea1HF.js` automatically replaces `apiKey` / `token` / `secret` values with redacted placeholders (`"***"` for short values, `"<prefix>…<suffix>"` for long values) in tool outputs before passing them to the agent model. This redaction is **completely transparent to the agent** — there i... | open | null | false | 1 | [] | [] | 2026-04-18T05:51:14Z | 2026-05-02T22:58:57Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | qiqi-victor | 242,737,599 | U_kgDODnfhvw | User | false |
openclaw/openclaw | 4,286,948,242 | I_kwDOQb6kR87_haOS | 68,426 | https://github.com/openclaw/openclaw/issues/68426 | https://api.github.com/repos/openclaw/openclaw/issues/68426 | fix(discord): AbortController-based total timeout for attachment downloads | See https://github.com/openclaw/openclaw/issues/68423 | closed | completed | true | 1 | [] | [] | 2026-04-18T05:55:37Z | 2026-04-28T04:51:00Z | 2026-04-25T18:32:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ChonSong | 85,378,550 | MDQ6VXNlcjg1Mzc4NTUw | User | false |
openclaw/openclaw | 4,286,951,306 | I_kwDOQb6kR87_ha-K | 68,427 | https://github.com/openclaw/openclaw/issues/68427 | https://api.github.com/repos/openclaw/openclaw/issues/68427 | fix(discord): AbortController-based total timeout for attachment downloads to prevent 1800s inbound worker blocking | ## Bug Description
When a Discord message with a large attachment (video ~8MB+, images 15MB+) is sent in a thread, the **inbound worker blocks for the full 1800 seconds**, silently dropping all subsequent messages in that thread — including @mentions.
```
discord inbound worker timed out after 1800 seconds
(channelId... | closed | completed | true | 1 | [] | [] | 2026-04-18T05:56:57Z | 2026-04-28T04:50:59Z | 2026-04-25T18:32:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ChonSong | 85,378,550 | MDQ6VXNlcjg1Mzc4NTUw | User | false |
openclaw/openclaw | 4,286,956,832 | I_kwDOQb6kR87_hcUg | 68,429 | https://github.com/openclaw/openclaw/issues/68429 | https://api.github.com/repos/openclaw/openclaw/issues/68429 | [Bug]: Telegram inline-button first callback can run before session bootstrap context is loaded | ### Bug type
Regression / reliability bug
### Summary
On Telegram inline-button workflows, the first button tap on a freshly created delivery session can be processed before the session's full bootstrap context is loaded. The callback text reaches the agent, but the first response is generated from a cold session, s... | open | null | false | 1 | [] | [] | 2026-04-18T05:59:24Z | 2026-05-03T04:43:39Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | richddr | 1,312,796 | MDQ6VXNlcjEzMTI3OTY= | User | false |
openclaw/openclaw | 4,057,007,643 | I_kwDOQb6kR87x0QYb | 43,034 | https://github.com/openclaw/openclaw/issues/43034 | https://api.github.com/repos/openclaw/openclaw/issues/43034 | Docs: Multi-agent orchestration patterns and best practices | ## Gap
`docs/concepts/multi-agent.md` exists but lacks depth on practical patterns. Users building multi-agent systems need guidance on:
## Proposed: `docs/concepts/multi-agent-patterns.md`
### Content
1. **Communication Patterns**
- `sessions_spawn` for one-shot tasks
- `sessions_send` for persistent agent me... | closed | not_planned | true | 4 | [] | [] | 2026-03-11T09:28:43Z | 2026-04-28T04:41:33Z | 2026-04-26T02:54:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,286,956,818 | I_kwDOQb6kR87_hcUS | 68,428 | https://github.com/openclaw/openclaw/issues/68428 | https://api.github.com/repos/openclaw/openclaw/issues/68428 | [Security] zizmor template-injection findings across CI workflows | ## Summary
`zizmor` (v1.24.1) currently reports **22 `template-injection` findings** across 4 workflow files under `.github/workflows/`. Template-injection happens when a GitHub Actions `${{ … }}` expression is interpolated directly into a shell `run:` block, where an attacker-controlled or weakly-validated value can ... | open | null | false | 2 | [
"bug",
"security",
"maintainer"
] | [] | 2026-04-18T05:59:24Z | 2026-04-18T06:05:51Z | null | CONTRIBUTOR | null | 20260418T062035Z | 2026-04-18T06:20:35Z | visionik | 52,174 | MDQ6VXNlcjUyMTc0 | User | false |
openclaw/openclaw | 4,287,001,047 | I_kwDOQb6kR87_hnHX | 68,434 | https://github.com/openclaw/openclaw/issues/68434 | https://api.github.com/repos/openclaw/openclaw/issues/68434 | WhatsApp: Add quoted reply support for outbound messages | ## WhatsApp: Add quoted reply support for outbound messages
### Summary
Outbound WhatsApp messages sent via OpenClaw do not support quoted replies (`contextInfo.stanzaId`). Other channels (Telegram, Discord, Slack) have `replyToMode` config and pass `replyToMessageId` through, but WhatsApp has neither — the send func... | closed | completed | true | 1 | [] | [] | 2026-04-18T06:16:20Z | 2026-04-28T04:50:59Z | 2026-04-25T18:32:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rastogiachyut | 7,073,017 | MDQ6VXNlcjcwNzMwMTc= | User | false |
openclaw/openclaw | 4,287,011,293 | I_kwDOQb6kR87_hpnd | 68,436 | https://github.com/openclaw/openclaw/issues/68436 | https://api.github.com/repos/openclaw/openclaw/issues/68436 | OpenRouter models return payloads=0 (incomplete turn) while direct API calls succeed | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
OpenClaw returns blank responses (payloads=0) for OpenRouter models while direct OpenRouter API calls return valid content.
### Steps to reproduce
1. Start OpenClaw gateway:
openclaw gateway run
2. Open t... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T06:19:56Z | 2026-04-21T04:21:58Z | 2026-04-18T16:02:22Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | ReDoC-RGB | 277,197,432 | U_kgDOEIWyeA | User | false |
openclaw/openclaw | 4,287,034,845 | I_kwDOQb6kR87_hvXd | 68,438 | https://github.com/openclaw/openclaw/issues/68438 | https://api.github.com/repos/openclaw/openclaw/issues/68438 | [Bug] Android node: TLS fingerprint mismatch after certificate renewal behind reverse proxy | ## Summary
When the OpenClaw gateway sits behind a reverse proxy that terminates TLS (e.g. Nginx, Caddy, or any custom proxy), the Android app pins the proxy's TLS certificate fingerprint on first connection (TOFU). When the proxy's certificate is renewed/rotated (e.g. Let's Encrypt every 60-90 days), the app refuses ... | open | null | false | 1 | [] | [] | 2026-04-18T06:26:57Z | 2026-05-03T04:59:41Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | choury | 4,332,542 | MDQ6VXNlcjQzMzI1NDI= | User | false |
openclaw/openclaw | 4,287,107,080 | I_kwDOQb6kR87_iBAI | 68,442 | https://github.com/openclaw/openclaw/issues/68442 | https://api.github.com/repos/openclaw/openclaw/issues/68442 | UI chat freeze when edit tool is called with non-ASCII characters in oldText/newText | ## Summary
When the `edit` tool is called with non-ASCII characters (Unicode arrows, German umlauts, emojis, box-drawing characters) in `oldText` or `newText`, the Gateway terminates the agent run with a JSON parse error. The webchat UI then freezes — the chat input becomes unresponsive and the stop button no longer w... | closed | completed | true | 2 | [] | [] | 2026-04-18T06:52:39Z | 2026-05-02T04:27:47Z | 2026-04-29T16:12:10Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | flynnmcpoc | 263,569,806 | U_kgDOD7XBjg | User | false |
openclaw/openclaw | 4,163,832,375 | I_kwDOQb6kR874Lwo3 | 56,957 | https://github.com/openclaw/openclaw/issues/56957 | https://api.github.com/repos/openclaw/openclaw/issues/56957 | [Bug]: v2026.3.28 regression — groupPolicy "allowlist" with no groups defined now allows ALL WhatsApp groups instead of blocking them | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from v2026.3.24 to v2026.3.28, `groupPolicy: "allowlist"` with no `groups` array defined began auto-replying to ALL WhatsApp group messages instead of blocking them. First group reply fired exactly 40 minutes... | closed | not_planned | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-29T11:10:17Z | 2026-04-29T04:40:38Z | 2026-04-26T08:46:39Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jakub-arch | 269,324,843 | U_kgDOEA2SKw | User | false |
openclaw/openclaw | 4,020,588,904 | I_kwDOQb6kR87vpVFo | 34,129 | https://github.com/openclaw/openclaw/issues/34129 | https://api.github.com/repos/openclaw/openclaw/issues/34129 | nvidia/moonshotai/kimi-k2.5 emits blank tool name -> call_auto_* -> Tool not found loops (2026.3.3) | ### Summary
On OpenClaw 2026.3.3, using NVIDIA direct provider path (`openai-completions`) with `nvidia/moonshotai/kimi-k2.5`, we repeatedly hit malformed tool-call handling:
- empty/blank tool name emitted in `after_tool_call`
- `toolCallId` becomes `call_auto_1`
- toolResult persisted with blank `toolName`, normaliz... | open | null | false | 4 | [] | [] | 2026-03-04T07:43:11Z | 2026-05-05T02:56:59Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | DiamondEyesFox | 119,838,309 | U_kgDOBySWZQ | User | false |
openclaw/openclaw | 4,098,707,331 | I_kwDOQb6kR870TU-D | 50,129 | https://github.com/openclaw/openclaw/issues/50129 | https://api.github.com/repos/openclaw/openclaw/issues/50129 | Slack connector silently drops file attachments in DM messages | ## Bug Description
The Slack connector receives the text content of DM messages but **silently drops file attachments**. Files uploaded directly in a DM conversation are never downloaded to the media cache (`~/.openclaw/media/inbound/`), and no error or warning is logged.
## Steps to Reproduce
1. Send a file attachm... | closed | not_planned | true | 4 | [] | [] | 2026-03-19T01:46:43Z | 2026-04-29T04:37:56Z | 2026-04-26T10:03:45Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | warren-t-c[bot] | 266,866,348 | BOT_kgDOD-gOrA | Bot | false |
openclaw/openclaw | 4,287,138,404 | I_kwDOQb6kR87_iIpk | 68,447 | https://github.com/openclaw/openclaw/issues/68447 | https://api.github.com/repos/openclaw/openclaw/issues/68447 | [Bug]: Model ID with provider prefix (google/gemma-4-26b-a4b) routes to wrong provider when using LM Studio | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Model ID containing provider prefix (e.g. google/gemma-4-26b-a4b) is incorrectly routed to the provider's remote API instead of the configured local backend (LM Studio).
### Steps to reproduce
1. Configure LM... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T07:06:37Z | 2026-04-28T04:50:58Z | 2026-04-25T18:32:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nxzhouquan | 128,338,434 | U_kgDOB6ZKAg | User | false |
openclaw/openclaw | 4,287,174,660 | I_kwDOQb6kR87_iRgE | 68,449 | https://github.com/openclaw/openclaw/issues/68449 | https://api.github.com/repos/openclaw/openclaw/issues/68449 | [Bug]: Dreaming plugin: stopword list too narrow and no filter for cron-triggered sessions | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On OpenClaw 2026.4.15 with dreaming.enabled: true, the dreaming pipeline surfaces tokenization artifacts (e.g. "assistant" 2754 hits) as top themes due to an incomplete stopword list, and pollutes the session c... | open | null | false | 6 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T07:22:29Z | 2026-05-04T13:03:02Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | htuyer121 | 277,202,213 | U_kgDOEIXFJQ | User | false |
openclaw/openclaw | 4,287,182,300 | I_kwDOQb6kR87_iTXc | 68,452 | https://github.com/openclaw/openclaw/issues/68452 | https://api.github.com/repos/openclaw/openclaw/issues/68452 | [Bug]: isolated cron turns can misclassify exact `NO_REPLY` tool results as incomplete when the final assistant message is empty | ### Bug type
Reliability / silent-turn regression
### Summary
Isolated cron runs with `payload.kind: "agentTurn"` can fail with `incomplete turn detected ... payloads=0` even when the latest successful tool result is exact `NO_REPLY` and the unchanged path should be a quiet success.
The bad path is:
1. the assista... | closed | completed | true | 3 | [] | [] | 2026-04-18T07:25:48Z | 2026-04-30T04:37:06Z | 2026-04-27T05:04:55Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | anyech | 8,743,351 | MDQ6VXNlcjg3NDMzNTE= | User | false |
openclaw/openclaw | 4,287,181,388 | I_kwDOQb6kR87_iTJM | 68,451 | https://github.com/openclaw/openclaw/issues/68451 | https://api.github.com/repos/openclaw/openclaw/issues/68451 | [Bug]: openclaw-weixin sidecar kills main gateway | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The openclaw-weixin plugin's sidecar child process calls cleanStaleGatewayProcessesSync() during startup, which finds and kills the parent gateway process on port 18789, causing an infinite restart loop (exit c... | closed | completed | true | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T07:25:29Z | 2026-04-21T04:21:56Z | 2026-04-18T17:03:57Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | sunnyiren-max | 136,626,430 | U_kgDOCCTA_g | User | false |
openclaw/openclaw | 4,287,213,648 | I_kwDOQb6kR87_ibBQ | 68,457 | https://github.com/openclaw/openclaw/issues/68457 | https://api.github.com/repos/openclaw/openclaw/issues/68457 | Add channels.whatsapp.silentErrorReplies (parity with Telegram) | ## Summary
Telegram supports `channels.telegram.silentErrorReplies: boolean` which suppresses gateway-synthesized error replies to the user. WhatsApp has no equivalent config option. Requesting parity.
## Use case
Agents with a "do not reply to non-skill customer DMs" HARD RULE emit `NO_REPLY` as the assistant respons... | closed | completed | true | 1 | [] | [] | 2026-04-18T07:38:16Z | 2026-05-02T04:27:46Z | 2026-04-29T16:12:20Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | raywu | 603,825 | MDQ6VXNlcjYwMzgyNQ== | User | false |
openclaw/openclaw | 4,287,253,100 | I_kwDOQb6kR87_ikps | 68,461 | https://github.com/openclaw/openclaw/issues/68461 | https://api.github.com/repos/openclaw/openclaw/issues/68461 | [Bug]: Display 'Context limit exceeded. I've reset our conversation to start fresh - please try again' after installed 2026.4.15 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgraded to 2026.4.15 from 2026.4.8, the following error dispalyed no matter what I input:
⚠️ Context limit exceeded. I've reset our conversation to start fresh - please try again.
To prevent this, increase your compa... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-18T07:54:55Z | 2026-04-28T04:50:57Z | 2026-04-25T18:32:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | TangLaoya | 3,899,608 | MDQ6VXNlcjM4OTk2MDg= | User | false |
openclaw/openclaw | 4,287,281,423 | I_kwDOQb6kR87_irkP | 68,466 | https://github.com/openclaw/openclaw/issues/68466 | https://api.github.com/repos/openclaw/openclaw/issues/68466 | [Bug]: URL-containing messages can be dropped after in-place upgrade due to stale link-understanding chunk import | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On an npm-global OpenClaw `2026.4.15` gateway upgraded in place on Ubuntu, Feishu text messages containing URLs failed dispatch with `ERR_MODULE_NOT_FOUND` because the running process imported a stale hashed link-understanding ... | closed | completed | true | 2 | [] | [] | 2026-04-18T08:07:21Z | 2026-04-30T04:34:08Z | 2026-04-27T12:45:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | songshikang0111 | 132,348,332 | U_kgDOB-N5rA | User | false |
openclaw/openclaw | 4,287,314,216 | I_kwDOQb6kR87_izko | 68,467 | https://github.com/openclaw/openclaw/issues/68467 | https://api.github.com/repos/openclaw/openclaw/issues/68467 | Dreaming inline mode silently produces no output (storage.mode: inline broken) | ## Bug Description
The `memory-core` dreaming pipeline silently fails when `storage.mode` is set to `inline` (the default). The cron job runs on schedule and reports success, but no dreaming content is written.
## Steps to Reproduce
1. Use default OpenClaw config (no explicit `dreaming.storage.mode` set)
2. Wait for... | closed | completed | true | 1 | [] | [] | 2026-04-18T08:20:24Z | 2026-04-28T04:50:56Z | 2026-04-25T18:32:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ttomiczek | 265,786,127 | U_kgDOD9eTDw | User | false |
openclaw/openclaw | 4,287,314,737 | I_kwDOQb6kR87_izsx | 68,468 | https://github.com/openclaw/openclaw/issues/68468 | https://api.github.com/repos/openclaw/openclaw/issues/68468 | Change default dreaming storage.mode from inline to separate | ## Problem
The `memory-core` plugin has two dreaming storage modes:
- `inline` — appends dreaming output directly to `memory/YYYY-MM-DD.md` as markdown sections
- `separate` — writes dreaming output to dedicated files under `memory/dreaming/{phase}/YYYY-MM-DD.md`
The default is `inline`, but inline mode has problems... | closed | completed | true | 1 | [] | [] | 2026-04-18T08:20:34Z | 2026-04-28T04:50:56Z | 2026-04-25T18:32:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ttomiczek | 265,786,127 | U_kgDOD9eTDw | User | false |
openclaw/openclaw | 4,287,333,975 | I_kwDOQb6kR87_i4ZX | 68,471 | https://github.com/openclaw/openclaw/issues/68471 | https://api.github.com/repos/openclaw/openclaw/issues/68471 | [Bug]: Heartbeat triggers unnecessary `cli session reset: reason=system-prompt` every 30 minutes | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The built-in heartbeat mechanism (default: every 30 minutes) causes a full CLI session reset each time it fires, because the `extraSystemPromptHash` differs between heartbeat-triggered runs and user-message-tri... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T08:27:24Z | 2026-04-28T05:24:38Z | 2026-04-25T10:28:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sanseiuP | 60,508,733 | MDQ6VXNlcjYwNTA4NzMz | User | false |
openclaw/openclaw | 4,287,336,556 | I_kwDOQb6kR87_i5Bs | 68,472 | https://github.com/openclaw/openclaw/issues/68472 | https://api.github.com/repos/openclaw/openclaw/issues/68472 | [Bug]: openclaw agent explicit-session turns lose session model overrides before the next turn | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
`openclaw agent --session-id ...` on OpenClaw 2026.4.15 does not preserve a per-session model override across turns: a tool-driven `session_status(model:"codex/gpt-5.4")` reports `changedModel: true` in-turn, but ... | closed | completed | true | 4 | [] | [] | 2026-04-18T08:28:06Z | 2026-04-30T04:39:19Z | 2026-04-26T05:21:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jamiezigelbaum | 856,649 | MDQ6VXNlcjg1NjY0OQ== | User | false |
openclaw/openclaw | 4,287,341,390 | I_kwDOQb6kR87_i6NO | 68,475 | https://github.com/openclaw/openclaw/issues/68475 | https://api.github.com/repos/openclaw/openclaw/issues/68475 | Bug: MEDIA directive causes duplicate file delivery in Discord channel | ## Bug Description
When using the \`MEDIA:<path>\` directive to send a file attachment to a Discord channel, the file appears to be delivered multiple times (duplicate sends).
## Steps to Reproduce
1. Have a text/markdown file ready (e.g., \`.md\` file)
2. In a Discord channel, instruct the agent to send the file us... | closed | completed | true | 1 | [] | [] | 2026-04-18T08:29:54Z | 2026-04-21T04:22:17Z | 2026-04-18T08:31:15Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | leoge007 | 116,452,300 | U_kgDOBvDrzA | User | false |
openclaw/openclaw | 4,287,347,828 | I_kwDOQb6kR87_i7x0 | 68,476 | https://github.com/openclaw/openclaw/issues/68476 | https://api.github.com/repos/openclaw/openclaw/issues/68476 | Telegram: no way to disable mention filtering in groups | ## Problem
In OpenClaw 2026.4.15, there is no way to receive all messages in a Telegram group without @mention, regardless of groupPolicy setting.
## Steps to reproduce
1. Set groupPolicy to "open" or "allowlist"
2. Configure groups with chat_id
3. Send message in group without @mention
## Expected behavior
Agent rec... | closed | not_planned | true | 1 | [] | [] | 2026-04-18T08:32:11Z | 2026-04-28T04:50:55Z | 2026-04-25T18:33:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zenomor1 | 220,409,741 | U_kgDODSMvjQ | User | false |
openclaw/openclaw | 4,287,359,830 | I_kwDOQb6kR87_i-tW | 68,478 | https://github.com/openclaw/openclaw/issues/68478 | https://api.github.com/repos/openclaw/openclaw/issues/68478 | Auto system messages spam the chat (e.g. '💾 Cron job created', '💾 Skill installed') | ## Problem
Auto-generated system messages flood the chat every time the user performs certain actions. Examples:
- '💾 Cron job created'
- '💾 Skill installed'
- '✅ Done' confirmations
These messages are:
1. **Annoying** — they interrupt the conversation constantly
2. **Not suppressible** — there's no config option t... | open | null | false | 1 | [] | [] | 2026-04-18T08:37:41Z | 2026-05-02T22:58:58Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | ShadowEnemyx | 73,282,162 | MDQ6VXNlcjczMjgyMTYy | User | false |
openclaw/openclaw | 4,228,008,362 | I_kwDOQb6kR878Akmq | 63,397 | https://github.com/openclaw/openclaw/issues/63397 | https://api.github.com/repos/openclaw/openclaw/issues/63397 | Vertex AI Gemini function calls fail with 'Function call is missing a thought_signature in functionCall parts' | ## Bug: Vertex AI Gemini function calls fail with `Function call is missing a thought_signature in functionCall parts`
### Summary
When OpenClaw routes a tool-using request through LiteLLM to Vertex AI Gemini (`gemini-flash`, `gemini-2.x`, etc.), Vertex returns HTTP 400 with the error message `Function call is missin... | closed | completed | true | 2 | [] | [] | 2026-04-08T22:11:45Z | 2026-04-28T05:03:44Z | 2026-04-25T15:37:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tmote | 21,976,980 | MDQ6VXNlcjIxOTc2OTgw | User | false |
openclaw/openclaw | 4,287,316,602 | I_kwDOQb6kR87_i0J6 | 68,470 | https://github.com/openclaw/openclaw/issues/68470 | https://api.github.com/repos/openclaw/openclaw/issues/68470 | [Bug]: MiniMax prompt token double-counting causes premature compaction at ~20% context | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
MiniMax-M2.7 prompt token usage is double-counted (input + cacheRead) because MiniMax does not report cache breakdown in OpenAI-style fields, causing `normalizeUsage` to skip the dedup subtraction. This inflate... | open | null | false | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T08:21:06Z | 2026-05-03T11:25:23Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | lykeion-dev | 140,231,077 | U_kgDOCFvBpQ | User | false |
openclaw/openclaw | 4,287,403,867 | I_kwDOQb6kR87_jJdb | 68,484 | https://github.com/openclaw/openclaw/issues/68484 | https://api.github.com/repos/openclaw/openclaw/issues/68484 | Webchat image paste not working [2026.4.15] | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
## Bug Description
When pasting an image directly into the webchat input box, the image data is not forwarded to the agent. Only the accompanying text message is received.
**Steps to reproduce:**
1. Paste an ... | closed | completed | true | 1 | [
"bug",
"close:duplicate",
"close:already-fixed",
"dedupe:child",
"bug:behavior"
] | [
"vincentkoc"
] | 2026-04-18T08:55:48Z | 2026-04-30T04:47:21Z | 2026-04-25T06:08:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jie242 | 275,669,041 | U_kgDOEG5gMQ | User | false |
openclaw/openclaw | 4,180,205,318 | I_kwDOQb6kR875KN8G | 58,443 | https://github.com/openclaw/openclaw/issues/58443 | https://api.github.com/repos/openclaw/openclaw/issues/58443 | [Bug] Gateway duplicates inbound messages — same message injected multiple times into session | ## Description
The Gateway is injecting the same inbound message into a session multiple times, causing the agent to respond to each duplicate independently. This results in a flood of duplicate replies being delivered to the channel.
## Reproduction
This has been observed on two separate channels within hours:
### C... | open | null | false | 8 | [] | [] | 2026-03-31T15:31:16Z | 2026-05-01T06:04:46Z | null | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | ryou0903 | 131,286,605 | U_kgDOB9NGTQ | User | false |
openclaw/openclaw | 4,213,467,098 | I_kwDOQb6kR877JGfa | 62,076 | https://github.com/openclaw/openclaw/issues/62076 | https://api.github.com/repos/openclaw/openclaw/issues/62076 | [Bug]: `openclaw sessions --json` hangs when stdout is not a TTY on 2026.4.5 | ## Summary
`openclaw sessions --json` hangs indefinitely when stdout is not a TTY on OpenClaw 2026.4.5.
This reproduces consistently on macOS (arm64) with Volta-managed Node. The same machine does **not** reproduce the issue on 2026.3.28.
## Environment
- OpenClaw: 2026.4.5 (`3e72c03`)
- Previous working version on s... | closed | completed | true | 4 | [] | [] | 2026-04-06T18:49:29Z | 2026-04-28T05:07:09Z | 2026-04-25T14:24:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 293163-lgtm | 243,364,165 | U_kgDODoFxRQ | User | false |
openclaw/openclaw | 4,287,440,021 | I_kwDOQb6kR87_jSSV | 68,486 | https://github.com/openclaw/openclaw/issues/68486 | https://api.github.com/repos/openclaw/openclaw/issues/68486 | Silent failure in macOS update restart helper — 2>/dev/null + || true makes bootstrap failures invisible | ## Summary
The macOS restart helper generated by `openclaw update` (`dist/update-cli-*.js`, `prepareRestartScript()`, darwin branch) suppresses every `launchctl` stderr and ignores the final exit code, so when the fallback `launchctl bootstrap` step fails the update appears successful and the gateway is silently left ... | closed | completed | true | 0 | [] | [] | 2026-04-18T09:11:01Z | 2026-04-21T04:21:49Z | 2026-04-18T17:39:06Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | zhourui89421 | 70,597,471 | MDQ6VXNlcjcwNTk3NDcx | User | false |
openclaw/openclaw | 4,287,373,186 | I_kwDOQb6kR87_jB-C | 68,479 | https://github.com/openclaw/openclaw/issues/68479 | https://api.github.com/repos/openclaw/openclaw/issues/68479 | [Bug]: Regression: cron sessions lost access to `message` tool | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
**Blocker for me, making cron unusable.**
After upgrading to 2026.4.15, after https://github.com/openclaw/openclaw/pull/62675 was merged in, cron jobs can't access `message` tool anymore - can't send messages to specific cha... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-18T08:42:38Z | 2026-04-21T04:22:17Z | 2026-04-18T09:14:39Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | Thinkscape | 270,528 | MDQ6VXNlcjI3MDUyOA== | User | false |
openclaw/openclaw | 4,287,447,559 | I_kwDOQb6kR87_jUIH | 68,487 | https://github.com/openclaw/openclaw/issues/68487 | https://api.github.com/repos/openclaw/openclaw/issues/68487 | subagent sweeper self-stops while pendingLifecycleErrorByRunId still has entries | ## Summary
The subagent sweeper in `src/agents/subagent-registry.ts` self-stops as soon as `subagentRuns` becomes empty, but the **same** sweeper cycle is also the only thing that expires `pendingLifecycleErrorByRunId` entries via `PENDING_ERROR_TTL_MS` (5 minutes). Any pending-error entry that is newer than the TTL a... | closed | completed | true | 1 | [] | [] | 2026-04-18T09:14:56Z | 2026-04-21T04:21:45Z | 2026-04-18T19:57:38Z | CONTRIBUTOR | null | 20260421T060038Z | 2026-04-21T06:00:38Z | Feelw00 | 45,638,585 | MDQ6VXNlcjQ1NjM4NTg1 | User | false |
openclaw/openclaw | 4,287,448,743 | I_kwDOQb6kR87_jUan | 68,488 | https://github.com/openclaw/openclaw/issues/68488 | https://api.github.com/repos/openclaw/openclaw/issues/68488 | opencode/claude-opus-4-7 rejected as Unknown model — bundled pi-ai catalog pinned at 0.66.1 in v2026.4.15 | ## Summary
openclaw v2026.4.15 rejects `opencode/claude-opus-4-7` with `FailoverError: Unknown model: opencode/claude-opus-4-7`, even though OpenCode Zen serves the model for the same API key.
v2026.4.15 release notes mention **"Anthropic 默认升级 Opus 4.7"** — this covers the direct `anthropic/` provider path, but the `... | closed | completed | true | 0 | [] | [] | 2026-04-18T09:15:20Z | 2026-04-27T04:49:29Z | 2026-04-24T22:36:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aaajiao | 49,885 | MDQ6VXNlcjQ5ODg1 | User | false |
openclaw/openclaw | 4,170,804,395 | I_kwDOQb6kR874mWyr | 57,715 | https://github.com/openclaw/openclaw/issues/57715 | https://api.github.com/repos/openclaw/openclaw/issues/57715 | [Performance] sessions.list slow - N+1 transcript fallback + full row build before limit | ## Problem
`sessions.list` RPC takes 5-8s per call in our environment, causing Gateway backpressure under normal polling load.
## Root Cause
1. **Full row build before filter/limit**: `listSessionsFromStore()` builds the full gateway session row for all 253 sessions before applying `limit`. Even `limit=20` doesn't r... | closed | completed | true | 4 | [] | [] | 2026-03-30T13:35:39Z | 2026-05-05T04:28:47Z | 2026-05-02T05:12:29Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | loyur | 8,509,039 | MDQ6VXNlcjg1MDkwMzk= | User | false |
openclaw/openclaw | 4,215,215,825 | I_kwDOQb6kR877PxbR | 62,272 | https://github.com/openclaw/openclaw/issues/62272 | https://api.github.com/repos/openclaw/openclaw/issues/62272 | [Bug] Update to 2026.4.5 fails: Cannot find module '@buape/carbon' | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Update from 2026.3.28 to 2026.4.5 fails at the openclaw doctor step with 'Cannot find module @buape/carbon'
### Steps to reproduce
1. Have OpenClaw 2026.3.28 installed via pnpm
2. Run `openclaw update`
3. Update fails at `... | closed | duplicate | true | 17 | [
"bug"
] | [] | 2026-04-07T03:26:32Z | 2026-04-24T04:25:32Z | 2026-04-22T02:36:12Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | rose131-max | 208,792,308 | U_kgDODHHq9A | User | false |
openclaw/openclaw | 4,287,468,827 | I_kwDOQb6kR87_jZUb | 68,493 | https://github.com/openclaw/openclaw/issues/68493 | https://api.github.com/repos/openclaw/openclaw/issues/68493 | [Bug]: Editing openclaw.json while gateway is running triggers hot-reload crash loop on Windows (stale lock file + EADDRINUSE) | ## Environment
- OpenClaw: 2026.4.15
- OS: Windows 11
- Gateway mode: local, bind: lan
- Model provider: Ollama (remote via direct Ethernet)
## Steps to reproduce
1. Gateway running normally
2. Edit (e.g. add )
3. Gateway detects the change and triggers hot-reload via SIGUSR1
4. Old process does not fully exit before... | open | null | false | 2 | [] | [
"vincentkoc"
] | 2026-04-18T09:24:10Z | 2026-05-02T12:13:00Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | ludwicia | 72,784,746 | MDQ6VXNlcjcyNzg0NzQ2 | User | false |
openclaw/openclaw | 4,287,471,023 | I_kwDOQb6kR87_jZ2v | 68,494 | https://github.com/openclaw/openclaw/issues/68494 | https://api.github.com/repos/openclaw/openclaw/issues/68494 | Telegram channel stalls after long-running use: main session context overflow and repeated send/getUpdates network failures | ## Summary
On my local OpenClaw setup, the Telegram bot becomes unresponsive after running for a while. The behavior looks like a combination of:
- the main Telegram session growing until it hits context overflow,
- long-running child processes staying attached to the gateway,
- Telegram polling/send operations intermi... | open | null | false | 2 | [] | [] | 2026-04-18T09:25:07Z | 2026-05-02T21:25:46Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | mvpbin | 25,851,317 | MDQ6VXNlcjI1ODUxMzE3 | User | false |
openclaw/openclaw | 4,213,631,841 | I_kwDOQb6kR877Juth | 62,089 | https://github.com/openclaw/openclaw/issues/62089 | https://api.github.com/repos/openclaw/openclaw/issues/62089 | [Bug]: Tool calls rendered as raw text in all UIs (control-ui, telegram, tui) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
OpenClaw 2026.4.5 (also tested 2026.4.2)
Model: minimax/MiniMax-M2.7
OS: Linux (CachyOS)
PROBLEM: Tool calls (exec, read, write, web_search, etc.) are displayed as raw text blocks instead of being rendered inline.
### St... | closed | not_planned | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-06T19:24:51Z | 2026-04-28T05:07:06Z | 2026-04-25T14:25:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | hhurma | 45,541,253 | MDQ6VXNlcjQ1NTQxMjUz | User | false |
openclaw/openclaw | 4,287,483,884 | I_kwDOQb6kR87_jc_s | 68,496 | https://github.com/openclaw/openclaw/issues/68496 | https://api.github.com/repos/openclaw/openclaw/issues/68496 | Add Hermes Agent ACP harness support | ## Summary
Add first-class ACP harness support for [NousResearch/hermes-agent](https://github.com/NousResearch/hermes-agent) so OpenClaw can route `runtime:"acp"` / ACP-harness requests to Hermes the same way it already documents and routes Codex, Claude, Gemini, Cursor, etc.
Hermes now exposes `hermes acp`, so this l... | closed | not_planned | true | 1 | [] | [] | 2026-04-18T09:29:30Z | 2026-04-30T04:39:18Z | 2026-04-26T05:21:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mishanefedov | 48,260,330 | MDQ6VXNlcjQ4MjYwMzMw | User | false |
openclaw/openclaw | 4,287,487,935 | I_kwDOQb6kR87_jd-_ | 68,497 | https://github.com/openclaw/openclaw/issues/68497 | https://api.github.com/repos/openclaw/openclaw/issues/68497 | [Feature]: Expose canvas A2UI action bridge to web browsers (WebSocket fallback for window.openclawSendUserAction) | ## Context
The canvas host at `/__openclaw__/canvas/` serves a test page (`index.html` inside the `canvas-host` dist and the default `~/.openclaw/canvas/index.html`) with Hello / Time / Photo / Dalek buttons. The page status shows:
```
Bridge: missing · iOS=no · Android=no
Failed to send action: dalek
```
when opene... | open | null | false | 1 | [] | [] | 2026-04-18T09:30:49Z | 2026-05-02T21:50:07Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | n3wth | 5,912,894 | MDQ6VXNlcjU5MTI4OTQ= | User | false |
openclaw/openclaw | 4,287,504,754 | I_kwDOQb6kR87_jiFy | 68,500 | https://github.com/openclaw/openclaw/issues/68500 | https://api.github.com/repos/openclaw/openclaw/issues/68500 | exec(bg=true) leaves zombie processes when child exits | ## Problem
When using `exec(bg=true)` to spawn background processes, the child processes become **zombie (defunct)** after exit because OpenClaw main process never calls `waitpid()` on them.
## Reproduction
1. Spawn a background process:
```
exec(bg=true, command="sleep 1")
```
2. Wait for child to finish
3... | closed | not_planned | true | 1 | [] | [] | 2026-04-18T09:36:45Z | 2026-04-28T04:50:54Z | 2026-04-25T18:33:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | MiaoDX | 8,840,792 | MDQ6VXNlcjg4NDA3OTI= | User | false |
openclaw/openclaw | 4,074,787,500 | I_kwDOQb6kR87y4FKs | 45,774 | https://github.com/openclaw/openclaw/issues/45774 | https://api.github.com/repos/openclaw/openclaw/issues/45774 | 402 error from third-party proxy not recognized as billing error (failoverReason: null) | When using a third-party Anthropic-compatible proxy, a 402 response with message `"402 No available asset for API access, please purchase a subscription"` is not recognized as a billing error.
**Expected**: `failoverReason: "billing"`, triggering auth profile rotation or model fallback.
**Actual**: `failoverReason: n... | closed | completed | true | 6 | [] | [] | 2026-03-14T05:49:02Z | 2026-04-21T04:22:16Z | 2026-04-18T09:38:06Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | 2488583886 | 29,515,984 | MDQ6VXNlcjI5NTE1OTg0 | User | false |
openclaw/openclaw | 4,287,534,696 | I_kwDOQb6kR87_jpZo | 68,501 | https://github.com/openclaw/openclaw/issues/68501 | https://api.github.com/repos/openclaw/openclaw/issues/68501 | Upstream: dreaming inline path → configurable / align with daily-memory/ | OpenClaw **memory-core** dreaming in **inline** storage mode writes to:
`{workspaceDir}/memory/{YYYY-MM-DD}.md`
Many vaults use **Obsidian Daily notes** under `memory/daily-memory/{YYYY-MM-DD}.md`, which splits “human diary” and “dreaming artifact” across two locations.
## Requested behaviour
- Add configuration (e... | open | null | false | 1 | [] | [] | 2026-04-18T09:47:34Z | 2026-05-03T07:44:57Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | uli6 | 22,777,986 | MDQ6VXNlcjIyNzc3OTg2 | User | false |
openclaw/openclaw | 4,287,560,435 | I_kwDOQb6kR87_jvrz | 68,504 | https://github.com/openclaw/openclaw/issues/68504 | https://api.github.com/repos/openclaw/openclaw/issues/68504 | WhatsApp: add audio input + optional voice replies (ElevenLabs) | ## Problem
WhatsApp users currently get rejected when they send a voice note. `saas/openclaw-wrapper/whatsapp-business-channel/message-processor.ts:688` replies *"I can't process audio messages yet. Please send text or images instead."* There is also no way for a user to hear the agent reply back as voice — all outbou... | closed | not_planned | true | 1 | [] | [] | 2026-04-18T09:57:48Z | 2026-04-30T04:44:33Z | 2026-04-25T07:27:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | pksidekyk | 264,911,230 | U_kgDOD8o5fg | User | false |
openclaw/openclaw | 4,287,604,921 | I_kwDOQb6kR87_j6i5 | 68,508 | https://github.com/openclaw/openclaw/issues/68508 | https://api.github.com/repos/openclaw/openclaw/issues/68508 | [Bug]: WebChat / Control UI: System event messages leaked into visible transcript after async exec | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After an async exec command completes, the internal system event notification is visible to the user in the WebChat / Control UI. These messages should be invisible agent-only signals.
**What appears in the UI:**
> System (... | closed | not_planned | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-18T10:11:07Z | 2026-04-30T04:39:17Z | 2026-04-26T05:21:46Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | scottbrenchley | 67,395,221 | MDQ6VXNlcjY3Mzk1MjIx | User | false |
openclaw/openclaw | 4,287,646,856 | I_kwDOQb6kR87_kEyI | 68,510 | https://github.com/openclaw/openclaw/issues/68510 | https://api.github.com/repos/openclaw/openclaw/issues/68510 | emitGatewayRestart's catch path leaves sigusr1 authorization count incremented on emit failure | ## Summary
`emitGatewayRestart()` in `src/infra/restart.ts` increments `sigusr1AuthorizedCount` via `authorizeGatewaySigusr1Restart()` _before_ emitting the SIGUSR1. If that emission throws (e.g. `process.kill(pid, "SIGUSR1")` raises `EPERM`/`ESRCH` under a sandboxed or restricted runtime), the catch block only rolls ... | closed | completed | true | 1 | [] | [] | 2026-04-18T10:25:20Z | 2026-04-21T04:22:12Z | 2026-04-18T13:02:58Z | CONTRIBUTOR | null | 20260421T060038Z | 2026-04-21T06:00:38Z | Feelw00 | 45,638,585 | MDQ6VXNlcjQ1NjM4NTg1 | User | false |
openclaw/openclaw | 4,061,593,727 | I_kwDOQb6kR87yFwB_ | 43,578 | https://github.com/openclaw/openclaw/issues/43578 | https://api.github.com/repos/openclaw/openclaw/issues/43578 | $schema field silently dropped on config hot-reload rewrite | ## Bug Report
### Summary
The `$schema` field in `openclaw.json` is accepted during validation (fixed in #14998) but silently stripped when the gateway rewrites the config file during hot-reload. This makes `$schema`-based editor validation effectively unusable for anyone running the gateway.
### Steps to Reproduce
1... | closed | completed | true | 2 | [] | [] | 2026-03-12T01:49:25Z | 2026-04-21T04:22:14Z | 2026-04-18T10:32:52Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | Kaspre | 36,520,309 | MDQ6VXNlcjM2NTIwMzA5 | User | false |
openclaw/openclaw | 4,287,666,211 | I_kwDOQb6kR87_kJgj | 68,513 | https://github.com/openclaw/openclaw/issues/68513 | https://api.github.com/repos/openclaw/openclaw/issues/68513 | [Bug] User image attachments dropped when primary model lacks vision — imageModel fallback not honored | ## Bug Description
User-sent image attachments (e.g., screenshots dragged into webchat) are dropped at the gateway layer when the primary model does not support image input, even when \`agents.defaults.imageModel\` is configured.
## Steps to Reproduce
1. Set primary model to MiniMax M2.7 Highspeed (\`input: ["text"]... | closed | completed | true | 1 | [] | [] | 2026-04-18T10:33:21Z | 2026-04-26T04:29:15Z | 2026-04-24T01:40:48Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | WwNeXst | 268,033,705 | U_kgDOD_neqQ | User | false |
openclaw/openclaw | 4,287,728,807 | I_kwDOQb6kR87_kYyn | 68,522 | https://github.com/openclaw/openclaw/issues/68522 | https://api.github.com/repos/openclaw/openclaw/issues/68522 | [Bug]: openai-codex/gpt-5.4 authenticates successfully but reply fails with “DNS lookup for the provider endpoint failed” | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Hello OpenClaw team,
We are running OpenClaw 2026.4.14 on Ubuntu 24.04.4 LTS, installed globally with npm, in local gateway mode on a self-hosted machine. The dashboard is opened from a Windows PC through an S... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-04-18T11:00:16Z | 2026-04-24T04:26:28Z | 2026-04-21T06:40:10Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | Antonio-Jesus-Sancho | 244,387,694 | U_kgDODpEPbg | User | false |
openclaw/openclaw | 4,287,723,555 | I_kwDOQb6kR87_kXgj | 68,520 | https://github.com/openclaw/openclaw/issues/68520 | https://api.github.com/repos/openclaw/openclaw/issues/68520 | [Bug]: OpenRouter returns empty responses (payloads=0) on 2026.4.15 (Linux Mint), works on 2026.4.12 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## 🐛 Bug: OpenRouter returns empty responses (payloads=0) on 2026.4.15 (Linux Mint), works on 2026.4.12
### Summary
OpenClaw 2026.4.15 fails to return any model output when using OpenRouter (DeepSeek), resulting in:
- `inc... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-18T10:57:23Z | 2026-04-21T04:21:57Z | 2026-04-18T16:02:24Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | PongoDerMngoHund | 6,275,510 | MDQ6VXNlcjYyNzU1MTA= | User | false |
openclaw/openclaw | 4,287,733,463 | I_kwDOQb6kR87_kZ7X | 68,524 | https://github.com/openclaw/openclaw/issues/68524 | https://api.github.com/repos/openclaw/openclaw/issues/68524 | [Bug] iOS: large chat image attachments fail or blow up memory (48MP frames, no client-side resize) | ## Problem
Attaching multiple photos from the iOS app's chat composer — particularly modern iPhone 48MP photos — causes one or more of:
- Attachments silently fail to deliver
- Memory pressure during base64 encoding on the MainActor
- Per-attachment payload exceeds gateway/model limits (typical ceiling ~5 MB)
- Stack... | open | null | false | 1 | [] | [] | 2026-04-18T11:02:12Z | 2026-05-03T04:43:40Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | maxperf | 117,154,779 | U_kgDOBvuj2w | User | false |
openclaw/openclaw | 4,287,743,529 | I_kwDOQb6kR87_kcYp | 68,525 | https://github.com/openclaw/openclaw/issues/68525 | https://api.github.com/repos/openclaw/openclaw/issues/68525 | [Feature Request]: Request for iOS TestFlight Beta Access | ### Summary
I am requesting access to the iOS TestFlight beta program to test the OpenClaw iOS app as a mobile node.
### Problem to solve
I have already deployed and am actively using OpenClaw Gateway on my Linux computer. I want to use my iPhone as a mobile OpenClaw node to access features like voice control, came... | closed | completed | true | 1 | [
"enhancement",
"r: testflight"
] | [] | 2026-04-18T11:06:21Z | 2026-04-21T04:22:14Z | 2026-04-18T11:06:42Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | ilwl222 | 142,095,795 | U_kgDOCHg1sw | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.