repo string | github_id int64 | github_node_id string | number int64 | html_url string | api_url string | title string | body string | state string | state_reason string | locked bool | comments_count int64 | labels list | assignees list | created_at string | updated_at string | closed_at string | author_association string | milestone_title string | snapshot_id string | extracted_at string | author_login string | author_id int64 | author_node_id string | author_type string | author_site_admin bool |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
openclaw/openclaw | 4,260,911,929 | I_kwDOQb6kR879-Fs5 | 66,441 | https://github.com/openclaw/openclaw/issues/66441 | https://api.github.com/repos/openclaw/openclaw/issues/66441 | Context overflow recovery may re-append identical user prompts (heartbeat polls), inflating session transcripts | ## Summary\nWhen the agent hits a context overflow during a tool loop, the overflow recovery path plus subsequent retries can cause the *same inbound heartbeat poll prompt* to be appended multiple times to the session store. This inflates the session transcript (role=user entries) and can contribute to repeat overflows... | open | null | false | 0 | [] | [] | 2026-04-14T08:58:50Z | 2026-04-14T08:58:50Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | holisticseo | 216,781,566 | U_kgDODOvS_g | User | false |
openclaw/openclaw | 4,260,983,527 | I_kwDOQb6kR879-XLn | 66,443 | https://github.com/openclaw/openclaw/issues/66443 | https://api.github.com/repos/openclaw/openclaw/issues/66443 | Overflow recovery duplicates role=user messages in session JSONL, amplifying transcript growth | ## Summary
When OpenClaw enters context overflow recovery (for example, logs like `[context-overflow-recovery] Attempting tool result truncation`), the agent session JSONL can end up with duplicated **role=user** entries (same content hash and length), even when tool-result bloat is not present in the session file. Thi... | open | null | false | 2 | [] | [] | 2026-04-14T09:08:47Z | 2026-04-20T05:26:41Z | null | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | holisticseo | 216,781,566 | U_kgDODOvS_g | User | false |
openclaw/openclaw | 4,261,082,715 | I_kwDOQb6kR879-vZb | 66,456 | https://github.com/openclaw/openclaw/issues/66456 | https://api.github.com/repos/openclaw/openclaw/issues/66456 | [Feature]: Privacy-first lived-experience layer between people, mediated by AI agents | ### Summary
Allow users to voluntarily share encrypted, anonymized fragments of personal experience so that AI agents can find relevant patterns and return distilled insights to other users facing similar situations — without ever exposing identities or raw data.
### Problem to solve
The most valuable human knowledg... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-04-14T09:25:04Z | 2026-04-14T09:25:04Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | bergusdz | 96,361,740 | U_kgDOBb5dDA | User | false |
openclaw/openclaw | 4,261,110,896 | I_kwDOQb6kR879-2Rw | 66,459 | https://github.com/openclaw/openclaw/issues/66459 | https://api.github.com/repos/openclaw/openclaw/issues/66459 | [Bug]: Telegram transcript contains final reply but no outbound send occurs for [thinking,text] turn on 2026.4.11 | ## Summary
On OpenClaw 2026.4.11 in a Telegram direct chat, an assistant final reply existed in the session transcript but was never delivered to Telegram. The affected assistant turn contained both a `thinking` block and a `text` block. Session/transcript state made it look like the reply had been sent, but the Telegr... | open | null | false | 3 | [] | [] | 2026-04-14T09:29:44Z | 2026-04-18T00:05:45Z | null | NONE | null | 20260418T002035Z | 2026-04-18T00:20:35Z | wangwllu | 7,668,944 | MDQ6VXNlcjc2Njg5NDQ= | User | false |
openclaw/openclaw | 4,261,144,624 | I_kwDOQb6kR879--gw | 66,460 | https://github.com/openclaw/openclaw/issues/66460 | https://api.github.com/repos/openclaw/openclaw/issues/66460 | [Bug]: cron-owned exec completion events are incorrectly relayed to the user by heartbeat | # [Bug]: cron-owned exec completion events are incorrectly relayed to the user by heartbeat
## Bug type
Behavior bug (incorrect output/state without crash)
## Beta release blocker
No
## Summary
After an isolated cron run already completes with its own final result, OpenClaw still relays the cron-internal `exec` c... | open | null | false | 0 | [] | [] | 2026-04-14T09:36:11Z | 2026-04-14T09:36:11Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | nanli2000cn | 259,904,494 | U_kgDOD33T7g | User | false |
openclaw/openclaw | 4,261,147,672 | I_kwDOQb6kR879-_QY | 66,462 | https://github.com/openclaw/openclaw/issues/66462 | https://api.github.com/repos/openclaw/openclaw/issues/66462 | Embedded Pi agent enters compaction loop on repeated 400 errors with no response body (openai-completions API) | ## Issue Description
When using a custom provider with `api: "openai-completions"` that proxies to Anthropic Claude (or other reasoning-capable models), the embedded Pi agent (used by active-memory, compaction, and other sub-agent flows) can enter a compaction loop when the provider returns a `400 status code` with **... | open | reopened | false | 0 | [] | [] | 2026-04-14T09:36:49Z | 2026-04-14T09:42:12Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | HongzhuLiu | 10,249,691 | MDQ6VXNlcjEwMjQ5Njkx | User | false |
openclaw/openclaw | 4,261,152,294 | I_kwDOQb6kR879_AYm | 66,463 | https://github.com/openclaw/openclaw/issues/66463 | https://api.github.com/repos/openclaw/openclaw/issues/66463 | Dashboard 模型下拉列表应过滤未配置 provider 的模型 | ## 问题描述
Dashboard 的 Agent 模型选择下拉列表显示所有支持的模型（包括 Amazon Bedrock、OpenAI、Anthropic 等），但用户实际只能使用已配置 provider 的模型。
这造成困惑：用户看到大量模型选项，但选择未配置 provider 的模型时会失败。
## 当前行为
- Primary Model 配置为 bailian/kimi-k2.5
- 下拉列表显示所有模型（包括 anthropic.claude-* 等）
- 用户误以为这些模型都可用
## 期望行为
下拉列表应**过滤显示**，只展示已配置 provider 的可用模型。
## 环境
- OpenClaw ... | closed | completed | true | 1 | [] | [] | 2026-04-14T09:37:48Z | 2026-04-17T04:23:01Z | 2026-04-14T09:39:57Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Ycchccc | 111,486,541 | U_kgDOBqUmTQ | User | false |
openclaw/openclaw | 4,261,166,020 | I_kwDOQb6kR879_DvE | 66,464 | https://github.com/openclaw/openclaw/issues/66464 | https://api.github.com/repos/openclaw/openclaw/issues/66464 | Dashboard model dropdown should filter by configured providers | ## Problem Description
The Dashboard's Agent model selection dropdown displays all supported models (including Amazon Bedrock, OpenAI, Anthropic, etc.), but users can only actually use models from configured providers.
This causes confusion: users see many model options, but selecting a model from an unconfigured pro... | open | null | false | 0 | [] | [] | 2026-04-14T09:40:21Z | 2026-04-14T09:40:21Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Ycchccc | 111,486,541 | U_kgDOBqUmTQ | User | false |
openclaw/openclaw | 4,261,182,749 | I_kwDOQb6kR879_H0d | 66,465 | https://github.com/openclaw/openclaw/issues/66465 | https://api.github.com/repos/openclaw/openclaw/issues/66465 | [Feature]: persist ACPX subprocess stderr to disk for post-mortem debugging | ## Problem
The ACPX runtime captures child-process stderr into an in-memory `startupStderr` buffer (~8 KB) that is only surfaced when an `AgentStartupError` fires. After successful `initialize`, stderr chunks continue to stream from the child but are effectively dropped — they flow through `captureStartupStderr` (whic... | open | null | false | 0 | [] | [] | 2026-04-14T09:43:42Z | 2026-04-14T09:43:42Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | rtmuller | 66,280,176 | MDQ6VXNlcjY2MjgwMTc2 | User | false |
openclaw/openclaw | 4,261,203,899 | I_kwDOQb6kR879_M-7 | 66,467 | https://github.com/openclaw/openclaw/issues/66467 | https://api.github.com/repos/openclaw/openclaw/issues/66467 | ACP session/update usage_update notification fails validation when used=null (Invalid params / expected object, received undefined) | ## Summary
Discord-bound ACP sessions can fail during `session/update` notification handling when the ACP runtime emits a `usage_update` notification with `used: null`.
In the observed failure, OpenClaw logs:
```text
Error handling notification {
jsonrpc: '2.0',
method: 'session/update',
params: {
sessionI... | open | null | false | 0 | [] | [] | 2026-04-14T09:47:43Z | 2026-04-14T09:47:43Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | fisher1006 | 9,008,110 | MDQ6VXNlcjkwMDgxMTA= | User | false |
openclaw/openclaw | 4,261,204,053 | I_kwDOQb6kR879_NBV | 66,468 | https://github.com/openclaw/openclaw/issues/66468 | https://api.github.com/repos/openclaw/openclaw/issues/66468 | TUI status bar shows stale model during provider fallback | ## Summary
When the gateway cycles through fallback providers (e.g. primary is rate-limited, auth fails on the next, etc.), the TUI status bar continues to display the **last configured/default model** rather than the provider/model currently being attempted. This makes it look like the system is using one provider wh... | open | null | false | 0 | [] | [] | 2026-04-14T09:47:44Z | 2026-04-14T09:47:44Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | SasugaTV | 38,813,948 | MDQ6VXNlcjM4ODEzOTQ4 | User | false |
openclaw/openclaw | 4,261,206,551 | I_kwDOQb6kR879_NoX | 66,469 | https://github.com/openclaw/openclaw/issues/66469 | https://api.github.com/repos/openclaw/openclaw/issues/66469 | restoreMemoryPluginState clears capability when called from resolvePluginProviders (shouldActivate=false) | ## Bug Description
`wiki_status` always shows `exported artifacts: 0` even though memory-core registers `publicArtifacts` successfully on Gateway startup.
## Root Cause
`restoreMemoryPluginState()` in `memory-state-CZkKzb0u.js` is called repeatedly from `resolvePluginProviders` → `resolveRuntimePluginRegistry` → `lo... | open | null | false | 1 | [] | [] | 2026-04-14T09:48:15Z | 2026-04-15T23:27:40Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | TNTest | 470,968 | MDQ6VXNlcjQ3MDk2OA== | User | false |
openclaw/openclaw | 4,261,210,574 | I_kwDOQb6kR879_OnO | 66,470 | https://github.com/openclaw/openclaw/issues/66470 | https://api.github.com/repos/openclaw/openclaw/issues/66470 | [Bug]: TUI/webchat can stay in pondering after codex/gpt-5.4 has already finished the turn | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
With codex/gpt-5.4 on OpenClaw 2026.4.12, the backend can finish and write the assistant reply, but TUI/webchat can remain in pondering for tens of seconds before rendering the final response.
### Steps to reproduce
1. Start ... | open | null | false | 1 | [
"bug",
"bug:crash"
] | [] | 2026-04-14T09:49:04Z | 2026-04-16T20:50:03Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | AltamimiYasser | 62,793,407 | MDQ6VXNlcjYyNzkzNDA3 | User | false |
openclaw/openclaw | 4,261,227,578 | I_kwDOQb6kR879_Sw6 | 66,472 | https://github.com/openclaw/openclaw/issues/66472 | https://api.github.com/repos/openclaw/openclaw/issues/66472 | [Bug]: DevClaw can create semantically duplicate issues, leave multiple PRs open for overlapping work, and drift issue state from PR reality | ## Summary
While reviewing active work in `yaqub0r/firstlight`, I found a DevClaw workflow failure mode where semantically overlapping tasks were created for the same docs change, leading to duplicate/open competing PRs and issue state drift.
This appears to be a DevClaw orchestration bug, not merely operator confusio... | open | null | false | 2 | [] | [] | 2026-04-14T09:52:23Z | 2026-04-15T00:01:33Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | fujiwaranosai850 | 262,444,726 | U_kgDOD6SWtg | User | false |
openclaw/openclaw | 4,261,243,144 | I_kwDOQb6kR879_WkI | 66,474 | https://github.com/openclaw/openclaw/issues/66474 | https://api.github.com/repos/openclaw/openclaw/issues/66474 | [Feature]: Support Agent Discovery Protocol (/.well-known/agent-discovery.json) | ## Problem
OpenClaw agents currently have no standard way to discover what services are available at a domain. Each agent must be hardcoded with endpoints or rely on human-provided URLs.
## Proposal
Support the [Agent Discovery Protocol (ADP)](https://github.com/walkojas-boop/agent-discovery-protocol) — an open stan... | open | null | false | 5 | [] | [] | 2026-04-14T09:55:31Z | 2026-04-14T17:50:02Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | walkojas-boop | 261,247,722 | U_kgDOD5JS6g | User | false |
openclaw/openclaw | 4,261,245,019 | I_kwDOQb6kR879_XBb | 66,477 | https://github.com/openclaw/openclaw/issues/66477 | https://api.github.com/repos/openclaw/openclaw/issues/66477 | [Feature]: Sift governance plugin — Ed25519-signed receipts for every agent action | ## Problem
OpenClaw agents can execute any tool call without governance oversight. There's no audit trail proving what an agent was authorized to do, and no way to enforce policy before execution.
## Proposal
A Sift governance plugin that evaluates every tool call before execution and produces an Ed25519-signed rece... | open | null | false | 2 | [] | [] | 2026-04-14T09:55:53Z | 2026-04-17T19:26:58Z | null | NONE | null | 20260417T202042Z | 2026-04-17T20:20:42Z | walkojas-boop | 261,247,722 | U_kgDOD5JS6g | User | false |
openclaw/openclaw | 4,261,276,553 | I_kwDOQb6kR879_euJ | 66,479 | https://github.com/openclaw/openclaw/issues/66479 | https://api.github.com/repos/openclaw/openclaw/issues/66479 | Windows: skill not discovered due to UTF-8 BOM injected by PowerShell WriteAllText | ### Summary
On Windows, a skill can disappear from `openclaw skills list` if `SKILL.md` starts with a UTF-8 BOM (`EF BB BF`).
This happens when the file is edited with PowerShell / .NET `WriteAllText` using the default UTF-8 encoding behaviour. The parser in `extractFrontmatterBlock` fails silently because it checks ... | open | null | false | 0 | [] | [] | 2026-04-14T10:02:15Z | 2026-04-14T10:02:15Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ianpor | 275,961,736 | U_kgDOEHLXiA | User | false |
openclaw/openclaw | 4,261,288,555 | I_kwDOQb6kR879_hpr | 66,480 | https://github.com/openclaw/openclaw/issues/66480 | https://api.github.com/repos/openclaw/openclaw/issues/66480 | [Feature Request] Add image upload/paste support to Control UI WebChat |
## Feature Description
Currently, the Control UI WebChat channel does not support uploading or pasting images. Users can only send image URLs, which limits the multi-modal capabilities of the agent.
## Expected Behavior
1. **File Upload**: Users should be able to attach image files via a file picker in the chat inp... | open | null | false | 0 | [] | [] | 2026-04-14T10:04:44Z | 2026-04-14T10:04:44Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | mhyy10 | 41,365,608 | MDQ6VXNlcjQxMzY1NjA4 | User | false |
openclaw/openclaw | 4,261,300,478 | I_kwDOQb6kR879_kj- | 66,481 | https://github.com/openclaw/openclaw/issues/66481 | https://api.github.com/repos/openclaw/openclaw/issues/66481 | Windows: CLI subcommands hang in OpenClaw exec/tool environment, but work in normal PowerShell terminals | ## Summary
On my Windows machine, `openclaw` subcommands appear to hang specifically when launched from the OpenClaw agent/tool exec environment, even though onboarding/configuration completed successfully and the same commands work fine for me when I run them manually in normal Windows Terminal / PowerShell sessions.... | open | null | false | 1 | [] | [] | 2026-04-14T10:07:01Z | 2026-04-14T10:48:16Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ar27111994 | 13,036,687 | MDQ6VXNlcjEzMDM2Njg3 | User | false |
openclaw/openclaw | 4,261,321,978 | I_kwDOQb6kR879_pz6 | 66,483 | https://github.com/openclaw/openclaw/issues/66483 | https://api.github.com/repos/openclaw/openclaw/issues/66483 | Context usage mismatch causes false 'context full' warnings; stale inputTokens after compaction | Impact\n- Users think context is still full after compaction; sessions get reset/compacted unnecessarily.\n- False alarms in long-running sessions.\n\nRoot cause\n- UI and backend use different context measurement sources.\n- UI continues showing stale inputTokens after compaction.\n\nPlanned fix\n- Align UI and backen... | open | null | false | 0 | [] | [] | 2026-04-14T10:11:04Z | 2026-04-14T10:11:04Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | 0riginal-claw | 265,676,882 | U_kgDOD9XoUg | User | false |
openclaw/openclaw | 4,261,333,835 | I_kwDOQb6kR879_stL | 66,484 | https://github.com/openclaw/openclaw/issues/66484 | https://api.github.com/repos/openclaw/openclaw/issues/66484 | openclaw qa CLI surface advertised by binary but unreachable on npm — qa-lab/cli.js facade missing from published package | ### Bug type
Packaging / help mismatch (binary advertises a path that npm users cannot actually take)
### Summary
On the stable npm release (`openclaw@2026.4.11`), the CLI explicitly tells the user how to enable the `openclaw qa` surface — but following that instruction fails, because the `qa-lab` bundled plugin's p... | open | null | false | 0 | [] | [] | 2026-04-14T10:13:14Z | 2026-04-14T10:13:14Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | n3moxyz | 254,834,596 | U_kgDODzB3pA | User | false |
openclaw/openclaw | 4,261,396,872 | I_kwDOQb6kR879_8GI | 66,486 | https://github.com/openclaw/openclaw/issues/66486 | https://api.github.com/repos/openclaw/openclaw/issues/66486 | [Bug]: [Bug] image tool 报 Unknown model 当 本地模型 被设为 imageModel 而非主模型时 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
当使用 MiniMax-M2.7 作为主模型，并将 lmstudio/google/gemma-4-e2b 设置为 agents.defaults.imageModel 时，image tool 报错 Unknown model: lmstudio/google/gemma-4-e2b，但同模型作为主模型时可正常识图。
### Steps to reproduce
1.配置 minimax-portal/Mini... | open | null | false | 0 | [
"bug"
] | [] | 2026-04-14T10:24:48Z | 2026-04-14T10:24:48Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | laiyujiang8-cloud | 270,644,324 | U_kgDOECG0ZA | User | false |
openclaw/openclaw | 4,261,410,159 | I_kwDOQb6kR879__Vv | 66,487 | https://github.com/openclaw/openclaw/issues/66487 | https://api.github.com/repos/openclaw/openclaw/issues/66487 | Heartbeat exec-event prompt drops actual completion payload | ## Bug
When a backgrounded `exec` session completes, the heartbeat fires correctly via `maybeNotifyOnExit`, but the prompt sent to the heartbeat model says "the result is shown in the system messages above" without actually injecting the exec completion payload into the heartbeat context. The model then correctly repo... | open | null | false | 1 | [] | [] | 2026-04-14T10:27:35Z | 2026-04-16T20:50:03Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jeades | 11,338,478 | MDQ6VXNlcjExMzM4NDc4 | User | false |
openclaw/openclaw | 4,261,413,426 | I_kwDOQb6kR87-AAIy | 66,488 | https://github.com/openclaw/openclaw/issues/66488 | https://api.github.com/repos/openclaw/openclaw/issues/66488 | Chrome CDP websocket reachability timeout too short on Windows — browser start fails | ## Environment
- OpenClaw version: 2026.4.12
- OS: Windows 10 (22631), x64
- Node: v24.13.1
- Chrome: 147.0.7727.56
- Browser profile: openclaw (managed)
## Problem
`browser action=start profile=openclaw` consistently fails with:
> Error: Chrome CDP websocket for profile "openclaw" is not reachable after start.
Chr... | open | null | false | 2 | [] | [] | 2026-04-14T10:28:20Z | 2026-04-14T10:57:44Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | buuzzy | 49,503,130 | MDQ6VXNlcjQ5NTAzMTMw | User | false |
openclaw/openclaw | 4,261,434,933 | I_kwDOQb6kR87-AFY1 | 66,489 | https://github.com/openclaw/openclaw/issues/66489 | https://api.github.com/repos/openclaw/openclaw/issues/66489 | [Feature]: classify "Local media path not under allowed directory" as permanent delivery error | ## Problem
The delivery queue retries the same deterministic rejection 5 times
(~2h of retry chain with exponential backoff) when the Telegram channel's
`sendMessage` fails with a sandbox path rejection. Relative paths in
`mediaUrls` can never pass the allowed-directory check — the error is
immutable, not transient — ... | open | null | false | 0 | [] | [] | 2026-04-14T10:33:04Z | 2026-04-14T10:33:04Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | rtmuller | 66,280,176 | MDQ6VXNlcjY2MjgwMTc2 | User | false |
openclaw/openclaw | 4,261,443,389 | I_kwDOQb6kR87-AHc9 | 66,492 | https://github.com/openclaw/openclaw/issues/66492 | https://api.github.com/repos/openclaw/openclaw/issues/66492 | Heartbeat session sends confused relay reply to Discord channel when async result context is missing | ## Bug Description
A heartbeat session for a Discord channel sent a confused \"please point me to which command\" reply to the user, triggered by an async exec completion that lacked the actual command output in context.
## OpenClaw Version
2026.4.12 (commit 1c0672b)
## Steps to Reproduce
1. Run a long async exec ... | open | null | false | 0 | [] | [] | 2026-04-14T10:34:49Z | 2026-04-14T13:06:38Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | slideshow-dingo | 258,996,445 | U_kgDOD2_43Q | User | false |
openclaw/openclaw | 4,261,445,191 | I_kwDOQb6kR87-AH5H | 66,493 | https://github.com/openclaw/openclaw/issues/66493 | https://api.github.com/repos/openclaw/openclaw/issues/66493 | [Bug]: web_fetch still blocked under Surge/Clash fake-IP even with allowRfc2544BenchmarkRange=true | ## Summary
`web_fetch` still blocks requests in a Surge/Clash fake-IP environment **even when** `tools.web.fetch.ssrfPolicy.allowRfc2544BenchmarkRange=true` is configured.
The host resolves public domains into `198.18.0.0/15` (RFC 2544 benchmark range), which is exactly the range this config is supposed to allow, but... | open | null | false | 1 | [] | [] | 2026-04-14T10:35:09Z | 2026-04-14T10:47:00Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | yt1988 | 4,963,078 | MDQ6VXNlcjQ5NjMwNzg= | User | false |
openclaw/openclaw | 4,261,454,008 | I_kwDOQb6kR87-AKC4 | 66,494 | https://github.com/openclaw/openclaw/issues/66494 | https://api.github.com/repos/openclaw/openclaw/issues/66494 | Kilocode discovery failure falls back to single-entry static catalog, causing all model_not_found | ## Summary
When Kilocode model discovery fails (HTTP 503, timeout, etc.), OpenClaw falls back to `buildStaticCatalog()` which contains **only one entry** (`kilo/auto`). Every user-configured Kilocode model ID (e.g. `z-ai/glm-5.1`, `moonshotai/kimi-k2.5`, `anthropic/claude-sonnet-4.6`) then fails client-side with `mode... | open | null | false | 0 | [] | [] | 2026-04-14T10:36:58Z | 2026-04-14T10:36:58Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | junwei1213 | 42,743,231 | MDQ6VXNlcjQyNzQzMjMx | User | false |
openclaw/openclaw | 4,261,476,582 | I_kwDOQb6kR87-APjm | 66,496 | https://github.com/openclaw/openclaw/issues/66496 | https://api.github.com/repos/openclaw/openclaw/issues/66496 | Heartbeat/cron system events arrive as role:user — model mistakes them for human messages | ## Bug
When a heartbeat or cron session receives a system-generated event (e.g. exec completion notification, heartbeat wake), the event is injected as a `role: "user"` message. The model has no way to distinguish this from the actual human typing, so it responds conversationally as if addressing the user directly.
T... | open | null | false | 0 | [] | [] | 2026-04-14T10:41:53Z | 2026-04-14T10:41:53Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jeades | 11,338,478 | MDQ6VXNlcjExMzM4NDc4 | User | false |
openclaw/openclaw | 4,261,487,229 | I_kwDOQb6kR87-ASJ9 | 66,497 | https://github.com/openclaw/openclaw/issues/66497 | https://api.github.com/repos/openclaw/openclaw/issues/66497 | WebSocket delta corruption with inline directive tags | # GitHub Issue
**Title:** WebSocket delta corruption with inline directive tags
**Labels:** bug, gateway, websocket
---
## Bug Description
WebSocket delta and final messages are corrupted when using inline directive tags (e.g., `[[reply_to_current]]`) at the beginning of LLM output. Characters are truncated due to... | open | null | false | 1 | [] | [] | 2026-04-14T10:44:18Z | 2026-04-16T20:50:03Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Loach-Smile | 267,273,575 | U_kgDOD-5FZw | User | false |
openclaw/openclaw | 4,261,488,656 | I_kwDOQb6kR87-ASgQ | 66,498 | https://github.com/openclaw/openclaw/issues/66498 | https://api.github.com/repos/openclaw/openclaw/issues/66498 | [Feature]: Pass freshness parameter to Gemini web_search provider | ## Problem
The `web_search` tool accepts a `freshness` parameter (e.g. `pd`, `pw`, `pm`), but the Gemini (Google Search Grounding) provider silently ignores it. Only Brave and Perplexity respect the parameter.
This means that when using Gemini as the web_search provider (auto-detected when no Brave/Perplexity API key... | open | null | false | 0 | [] | [] | 2026-04-14T10:44:36Z | 2026-04-14T10:46:14Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ismael-81 | 273,866,131 | U_kgDOEFLdkw | User | false |
openclaw/openclaw | 4,261,499,985 | I_kwDOQb6kR87-AVRR | 66,500 | https://github.com/openclaw/openclaw/issues/66500 | https://api.github.com/repos/openclaw/openclaw/issues/66500 | [Bug]: 1. Feishu channel: Chinese filenames garbled when saving inbound media (UTF-8 → Latin-1 mojibake) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
1. When sending files with Chinese filenames through the Feishu channel, the original filename is corrupted before it reaches saveMediaBuffer(). Chinese characters become Latin-1 mojibake.
### Steps to reproduce
1. When s... | open | null | false | 0 | [
"bug"
] | [] | 2026-04-14T10:46:54Z | 2026-04-14T10:46:54Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | kwzq6jjqp8-sys | 271,693,003 | U_kgDOEDG0yw | User | false |
openclaw/openclaw | 4,261,504,159 | I_kwDOQb6kR87-AWSf | 66,501 | https://github.com/openclaw/openclaw/issues/66501 | https://api.github.com/repos/openclaw/openclaw/issues/66501 | [Feature]: Request: Session Lifecycle Hook - Auto-execute script on agent/session start | ### Summary
Add a session lifecycle hook to automatically execute initialization scripts when an agent session starts.
### Problem to solve
Currently when an agent session starts, there is no way to automatically execute initialization scripts. Agents must voluntarily read AGENTS.md and call hook functions, but they... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-04-14T10:47:49Z | 2026-04-14T10:47:49Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | chenxiaoemail-ops | 275,969,628 | U_kgDOEHL2XA | User | false |
openclaw/openclaw | 4,261,514,215 | I_kwDOQb6kR87-AYvn | 66,503 | https://github.com/openclaw/openclaw/issues/66503 | https://api.github.com/repos/openclaw/openclaw/issues/66503 | Feature Request] 定时任务支持 - Cron/Webhook/沙箱 CLI 访问 | ### Summary
<html>
<body>
<!--StartFragment--><h3 style="box-sizing: border-box; color: rgb(60, 60, 67); font-family: Inter, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, sans-serif; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: -0.14px; orphans: 2; t... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-04-14T10:49:57Z | 2026-04-14T11:01:49Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | shifangelectric-ai | 275,969,947 | U_kgDOEHL3mw | User | false |
openclaw/openclaw | 4,261,530,616 | I_kwDOQb6kR87-Acv4 | 66,505 | https://github.com/openclaw/openclaw/issues/66505 | https://api.github.com/repos/openclaw/openclaw/issues/66505 | [Feature]: Direct Ollama Cloud API provider (without local Ollama) | ## Problem
Ollama Cloud offers a public API at `https://api.ollama.com` that allows direct inference to cloud-hosted models. However, OpenClaw currently requires a local Ollama installation to use Ollama Cloud models — the local binary acts as a proxy/intermediary.
This means users who want to use Ollama Cloud must i... | open | null | false | 2 | [] | [] | 2026-04-14T10:53:38Z | 2026-04-17T21:03:26Z | null | NONE | null | 20260417T212042Z | 2026-04-17T21:20:42Z | ismael-81 | 273,866,131 | U_kgDOEFLdkw | User | false |
openclaw/openclaw | 4,261,530,883 | I_kwDOQb6kR87-Ac0D | 66,506 | https://github.com/openclaw/openclaw/issues/66506 | https://api.github.com/repos/openclaw/openclaw/issues/66506 | [Bug]: OpenAI audio transcription returns "No transcript returned" on 2026.4.9 despite provider success | ## Summary
On OpenClaw 2026.4.9, the `infer audio transcribe` / STT CLI path fails for OpenAI audio transcription even though the provider returns a valid transcript.
In the failing path, the user sees the transcription treated as empty / missing (`No transcript returned`), but a direct call to OpenAI's official `/v1... | open | null | false | 0 | [] | [] | 2026-04-14T10:53:42Z | 2026-04-14T10:53:42Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | vic-weng | 90,241,716 | MDQ6VXNlcjkwMjQxNzE2 | User | false |
openclaw/openclaw | 4,261,576,072 | I_kwDOQb6kR87-An2I | 66,507 | https://github.com/openclaw/openclaw/issues/66507 | https://api.github.com/repos/openclaw/openclaw/issues/66507 | [Bug] WeChat and Feishu channels share the same session instead of creating separate sessions | ## Bug Report: WeChat and Feishu channels share the same session
### Issue Description
WeChat and Feishu channels share the same session (`agent:main:main`) instead of creating separate sessions. When messages come from different channels, they are all routed to the same session.
### Steps to Reproduce
1. Configure b... | closed | completed | true | 3 | [] | [] | 2026-04-14T11:02:54Z | 2026-04-18T04:14:19Z | 2026-04-15T14:22:39Z | NONE | null | 20260418T042035Z | 2026-04-18T04:20:35Z | llnnnl531 | 53,088,155 | MDQ6VXNlcjUzMDg4MTU1 | User | false |
openclaw/openclaw | 4,261,593,167 | I_kwDOQb6kR87-AsBP | 66,508 | https://github.com/openclaw/openclaw/issues/66508 | https://api.github.com/repos/openclaw/openclaw/issues/66508 | Dream Diary: Timestamp tabs show red strikethrough on click | ## Bug
(check video below to understand immediatly)
https://github.com/user-attachments/assets/c0f17d05-6798-4dd2-adb6-07c0e8214760
In the Dream Diary view (Control UI), clicking on the timestamp tabs (e.g. "April 13, 2026 at 6:37 PM") causes the front/earlier tabs to display an ugly red strikethrough line before disa... | open | null | false | 1 | [] | [] | 2026-04-14T11:06:26Z | 2026-04-14T11:15:07Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | safrano9999 | 240,768,512 | U_kgDODlnWAA | User | false |
openclaw/openclaw | 4,261,627,555 | I_kwDOQb6kR87-A0aj | 66,509 | https://github.com/openclaw/openclaw/issues/66509 | https://api.github.com/repos/openclaw/openclaw/issues/66509 | Gateway buffers tool-kind text instead of delivering immediately (streaming.mode partial ignored) | ## Bug Description
When the assistant responds with intermediate text **before a tool call** (e.g., "Let me check that..."), the text is classified as `kind: "tool"` by the gateway dispatcher and **buffered instead of sent to the user**. Only the final `kind: "block"` response reaches the chat.
Users see **silence** ... | open | null | false | 0 | [] | [] | 2026-04-14T11:12:48Z | 2026-04-14T11:12:48Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | gabrielduartesignart | 270,779,711 | U_kgDOECPFPw | User | false |
openclaw/openclaw | 4,261,640,458 | I_kwDOQb6kR87-A3kK | 66,510 | https://github.com/openclaw/openclaw/issues/66510 | https://api.github.com/repos/openclaw/openclaw/issues/66510 | Control UI/WebChat: reading indicator disappears during active tool-first runs | ## Summary
In Control UI / WebChat, the visible `...` reading/typing indicator can disappear during an active run even though the run is still in flight and the red Stop button remains visible.
The regression seems to happen when tool activity begins before the first assistant text token arrives.
## Current behavior
... | closed | completed | true | 1 | [] | [] | 2026-04-14T11:14:55Z | 2026-04-19T04:22:54Z | 2026-04-17T00:00:46Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | ctbritt | 4,516,034 | MDQ6VXNlcjQ1MTYwMzQ= | User | false |
openclaw/openclaw | 4,261,663,939 | I_kwDOQb6kR87-A9TD | 66,512 | https://github.com/openclaw/openclaw/issues/66512 | https://api.github.com/repos/openclaw/openclaw/issues/66512 | [Feature] systemd sd_notify / Type=notify support for gateway readiness | ## Problem
The `openclaw gateway install` command generates a systemd unit with `Type=simple`. This means systemd considers the service "started" the moment the node process is spawned — long before the gateway is actually accepting RPC on its WebSocket port.
For a typical deployment running on `systemd --user`, meas... | open | null | false | 0 | [] | [] | 2026-04-14T11:19:21Z | 2026-04-14T11:19:21Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | angeliti999 | 104,985,926 | U_kgDOBkH1Rg | User | false |
openclaw/openclaw | 4,261,689,937 | I_kwDOQb6kR87-BDpR | 66,513 | https://github.com/openclaw/openclaw/issues/66513 | https://api.github.com/repos/openclaw/openclaw/issues/66513 | [Test] Testing GitHub skill from OpenClaw Control UI | This is a test issue created via GitHub skill from OpenClaw Control UI. Please ignore or close. | open | null | false | 0 | [] | [] | 2026-04-14T11:25:02Z | 2026-04-14T11:25:02Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | henrytomson856-glitch | 263,120,409 | U_kgDOD67mGQ | User | false |
openclaw/openclaw | 4,261,693,757 | I_kwDOQb6kR87-BEk9 | 66,514 | https://github.com/openclaw/openclaw/issues/66514 | https://api.github.com/repos/openclaw/openclaw/issues/66514 | [Bug]: Browser control service reports healthy while CDP/browser path is dead; snapshot/screenshot timeout and start recovery fails on macOS (2026.4.12) | # Draft GitHub Issue / Update
## Bug type
Regression / browser control runtime failure
## Summary
On OpenClaw 2026.4.12 on macOS, the browser control layer can look healthy while real browser operations are dead. In my case:
- browser service reports ready
- browser status may report `running: true`, `cdpReady: true`... | open | null | false | 0 | [] | [] | 2026-04-14T11:25:54Z | 2026-04-14T11:25:54Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Flakedict | 41,390,664 | MDQ6VXNlcjQxMzkwNjY0 | User | false |
openclaw/openclaw | 4,261,735,550 | I_kwDOQb6kR87-BOx- | 66,516 | https://github.com/openclaw/openclaw/issues/66516 | https://api.github.com/repos/openclaw/openclaw/issues/66516 | [Bug]: Beta Blocker: Active Memory Timeout | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
Yes
### Summary
I’m debugging Active Memory on OpenClaw 2026.4.12 and it consistently times out before returning any recalled summary.
What’s happening
Active Memory is enabled, loads correctly, and runs before replies, but ... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T11:34:43Z | 2026-04-17T04:22:20Z | 2026-04-15T02:55:11Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | som-686 | 21,360,977 | MDQ6VXNlcjIxMzYwOTc3 | User | false |
openclaw/openclaw | 4,261,791,809 | I_kwDOQb6kR87-BchB | 66,518 | https://github.com/openclaw/openclaw/issues/66518 | https://api.github.com/repos/openclaw/openclaw/issues/66518 | openclaw memory status --deep fails with "Unknown memory embedding provider: ollama" in v2026.4.12 | # `openclaw memory status --deep` fails with "Unknown memory embedding provider: ollama" in v2026.4.12
## Summary
In **v2026.4.12** the CLI command `openclaw memory status --deep` fails immediately with `Error: Unknown memory embedding provider: ollama` before any output is produced, when `agents.defaults.memorySearc... | closed | completed | true | 1 | [] | [] | 2026-04-14T11:45:21Z | 2026-04-17T04:22:45Z | 2026-04-14T16:32:04Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | haumanto | 72,009,799 | MDQ6VXNlcjcyMDA5Nzk5 | User | false |
openclaw/openclaw | 4,261,803,346 | I_kwDOQb6kR87-BfVS | 66,519 | https://github.com/openclaw/openclaw/issues/66519 | https://api.github.com/repos/openclaw/openclaw/issues/66519 | [Bug] Subagent completion triggers duplicate delivery: raw status fallback + proper summary | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When a subagent completes a task, the main agent sends two separate messages to the delivery channel: first a raw status fallback prefixed with "Automatic session resume failed, so sending the status directly.", then a prope... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-14T11:47:34Z | 2026-04-17T04:21:44Z | 2026-04-14T19:21:31Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | marcowiese | 26,906,941 | MDQ6VXNlcjI2OTA2OTQx | User | false |
openclaw/openclaw | 4,261,809,190 | I_kwDOQb6kR87-Bgwm | 66,520 | https://github.com/openclaw/openclaw/issues/66520 | https://api.github.com/repos/openclaw/openclaw/issues/66520 | [Bug] Auto-compaction never fires when Anthropic prompt cache hit rate is ~100% | ## Summary
Auto-compaction does not trigger automatically even when the context window is exceeded (observed at 153% = 305k/200k), as long as the Anthropic prompt cache absorbs most tokens.
## Environment
- OpenClaw version: 2026.4.12 (1c0672b)
- Model: `anthropic/claude-sonnet-4-6`
- Compaction mode: `safeguard`
- ... | open | null | false | 0 | [] | [] | 2026-04-14T11:48:41Z | 2026-04-14T11:48:41Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Neomail2 | 120,675,713 | U_kgDOBzFdgQ | User | false |
openclaw/openclaw | 4,261,835,746 | I_kwDOQb6kR87-BnPi | 66,522 | https://github.com/openclaw/openclaw/issues/66522 | https://api.github.com/repos/openclaw/openclaw/issues/66522 | [Bug] Session index (sessions.json) rebuilds on Gateway restart, causing complete loss of session history without warning | ## Bug Description
When the OpenClaw Gateway restarts (either manually or triggered by cron), the `sessions.json` file is completely rebuilt rather than merged or restored. All previously active sessions are dropped from the index, making them invisible in any client UI (WebUI, WeChat, Feishu). This happens **without ... | open | null | false | 1 | [] | [] | 2026-04-14T11:54:15Z | 2026-04-15T06:44:12Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | wc4440222 | 199,748,891 | U_kgDOC-ftGw | User | false |
openclaw/openclaw | 4,261,839,212 | I_kwDOQb6kR87-BoFs | 66,523 | https://github.com/openclaw/openclaw/issues/66523 | https://api.github.com/repos/openclaw/openclaw/issues/66523 | [Bug]: OPENCLAW_STATE_DIR Environment Variable Ignored on Windows | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
# OPENCLAW_STATE_DIR Environment Variable Ignored on Windows
## Bug Description
The `OPENCLAW_STATE_DIR` environment variable is being ignored. Gateway processes continue to use the default `C:\Users\<username>\.openclaw` ... | open | null | false | 1 | [
"bug"
] | [] | 2026-04-14T11:54:56Z | 2026-04-16T20:50:03Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | visionzhang-prog | 266,700,237 | U_kgDOD-WFzQ | User | false |
openclaw/openclaw | 4,261,840,003 | I_kwDOQb6kR87-BoSD | 66,524 | https://github.com/openclaw/openclaw/issues/66524 | https://api.github.com/repos/openclaw/openclaw/issues/66524 | tools.exec: binder fail-closed rejects ALL absolute-path commands in v2026.4.11+v2026.4.12 (with reproducible env workaround) | ### Summary
Since `v2026.4.11`, the `tools.exec` approval binder rejects **every** command sent from a gateway agent to a node host, including trivial absolute-path commands like `/usr/bin/whoami`. The error returned is:
```
INVALID_REQUEST: SYSTEM_RUN_DENIED: approval cannot safely bind this interpreter/runtime comm... | closed | completed | true | 0 | [] | [] | 2026-04-14T11:55:06Z | 2026-04-19T04:23:11Z | 2026-04-16T15:00:11Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | PabloHurtadoGonzalo86 | 79,223,844 | MDQ6VXNlcjc5MjIzODQ0 | User | false |
openclaw/openclaw | 4,261,910,801 | I_kwDOQb6kR87-B5kR | 66,527 | https://github.com/openclaw/openclaw/issues/66527 | https://api.github.com/repos/openclaw/openclaw/issues/66527 | [Bug] subagent-registry.runtime.js missing in 2026.4.12 npm package | # [Bug] subagent-registry.runtime.js missing in 2026.4.12 npm package
## Summary
OpenClaw 2026.4.12 crashes with `ERR_MODULE_NOT_FOUND` when trying to clean up old subagent runs during gateway startup. The file `dist/subagent-registry.runtime.js` is referenced by `dist/subagent-registry-CflSFWBm.js` but is missing fr... | open | null | false | 0 | [] | [] | 2026-04-14T12:08:47Z | 2026-04-14T12:08:47Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | lolaopenclaw | 259,362,584 | U_kgDOD3WPGA | User | false |
openclaw/openclaw | 4,261,950,677 | I_kwDOQb6kR87-CDTV | 66,530 | https://github.com/openclaw/openclaw/issues/66530 | https://api.github.com/repos/openclaw/openclaw/issues/66530 | [Feature]: Primary model should be probed for recovery after fallback due to timeout | ## Summary
When the primary model times out and OpenClaw falls back to a secondary model, once the fallback succeeds the session becomes "sticky" and never automatically returns to the primary model, even after the primary becomes available again.
## Problem to solve
**Use case**: User has MiniMax-M2.7 (200k context... | open | null | false | 0 | [] | [] | 2026-04-14T12:15:55Z | 2026-04-14T12:15:55Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | MrWangTutu | 131,415,487 | U_kgDOB9U9vw | User | false |
openclaw/openclaw | 4,261,960,703 | I_kwDOQb6kR87-CFv_ | 66,531 | https://github.com/openclaw/openclaw/issues/66531 | https://api.github.com/repos/openclaw/openclaw/issues/66531 | [Feature]: Add agent.abort RPC to cancel embedded agent runs | ### Summary
Add a gateway RPC `agent.abort` so external clients can cancel runs started via the `agent()` RPC. Today they can't: `chat.abort` only tracks `chat.send`-initiated runs, and the underlying primitive `abortEmbeddedPiRun(sessionId)` has no RPC surface.
### Problem to solve
- `agent()` runs register into `A... | open | null | false | 0 | [] | [] | 2026-04-14T12:17:47Z | 2026-04-14T12:17:47Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | xiaohuaxi | 45,277,353 | MDQ6VXNlcjQ1Mjc3MzUz | User | false |
openclaw/openclaw | 4,261,961,279 | I_kwDOQb6kR87-CF4_ | 66,532 | https://github.com/openclaw/openclaw/issues/66532 | https://api.github.com/repos/openclaw/openclaw/issues/66532 | [Feature]: Expose abortEmbeddedPiRun family on api.runtime.agent | ### Summary
Plumb `abortEmbeddedPiRun`, `waitForEmbeddedPiRunEnd`, `isEmbeddedPiRunActive`, and `queueEmbeddedPiMessage` through to `api.runtime.agent` so bundled plugins can cancel and observe embedded agent runs without reaching into process globals.
### Problem to solve
- `src/plugins/runtime/runtime-embedded-pi.... | open | null | false | 0 | [] | [] | 2026-04-14T12:17:54Z | 2026-04-14T12:17:54Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | xiaohuaxi | 45,277,353 | MDQ6VXNlcjQ1Mjc3MzUz | User | false |
openclaw/openclaw | 4,261,975,202 | I_kwDOQb6kR87-CJSi | 66,533 | https://github.com/openclaw/openclaw/issues/66533 | https://api.github.com/repos/openclaw/openclaw/issues/66533 | [Bug]: WebChat session selector shows main session as “heartbeat” after a while, making /new feel like sessions disappear | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
In Control UI / WebChat, the session dropdown becomes confusing over time. After creating a new chat with /new or /reset, things look fine briefly, but later the selector seems to fall back to “heartbeat”, and it looks like... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-04-14T12:20:37Z | 2026-04-16T20:50:02Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | innoffices | 275,835,113 | U_kgDOEHDo6Q | User | false |
openclaw/openclaw | 4,261,981,360 | I_kwDOQb6kR87-CKyw | 66,534 | https://github.com/openclaw/openclaw/issues/66534 | https://api.github.com/repos/openclaw/openclaw/issues/66534 | [Bug]: lifecycle:end event payload missing aborted and stopReason on pi-embedded path | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The main `lifecycle:end` event emitted by `handleAgentEnd` in `src/agents/pi-embedded-subscribe.handlers.lifecycle.ts` does not carry `aborted` or `stopReason` fields. Subscribers cannot distinguish cancellatio... | open | null | false | 1 | [] | [] | 2026-04-14T12:21:48Z | 2026-04-16T20:50:02Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | xiaohuaxi | 45,277,353 | MDQ6VXNlcjQ1Mjc3MzUz | User | false |
openclaw/openclaw | 4,261,981,945 | I_kwDOQb6kR87-CK75 | 66,535 | https://github.com/openclaw/openclaw/issues/66535 | https://api.github.com/repos/openclaw/openclaw/issues/66535 | [Bug]: /compact command cannot be canceled while in progress | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
`/compact` invokes `runtime.compactEmbeddedPiSession(...)` without passing any `abortSignal` and without registering the compaction session into `ACTIVE_EMBEDDED_RUNS`. As a result neither `chat.abort`, `abortE... | open | null | false | 1 | [] | [] | 2026-04-14T12:21:55Z | 2026-04-16T20:50:02Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | xiaohuaxi | 45,277,353 | MDQ6VXNlcjQ1Mjc3MzUz | User | false |
openclaw/openclaw | 4,262,033,377 | I_kwDOQb6kR87-CXfh | 66,539 | https://github.com/openclaw/openclaw/issues/66539 | https://api.github.com/repos/openclaw/openclaw/issues/66539 | [Feature]: Control UI should support stable aliases/pinned names for session keys | ### Feature type
UX / session management
### Summary
In the OpenClaw dashboard / Control UI, I want to be able to keep a stable human-friendly name for a session instead of having the visible session name drift or be derived from changing metadata.
A concrete example is a Discord guild channel session that is inter... | open | null | false | 0 | [] | [] | 2026-04-14T12:31:04Z | 2026-04-14T12:31:04Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | markjongseo | 156,290,140 | U_kgDOCVDMXA | User | false |
openclaw/openclaw | 4,262,038,413 | I_kwDOQb6kR87-CYuN | 66,540 | https://github.com/openclaw/openclaw/issues/66540 | https://api.github.com/repos/openclaw/openclaw/issues/66540 | replyToCurrent default false blocks implicit replyToId in followup queue | ## Bug: `replyToCurrent: false` default blocks implicit `replyToId` in followup queue
### Problem
When processing queued/followup messages, `resolveReplyThreadingForPayload` in `reply-payloads-base.ts` checks:
```typescript
params.payload.replyToId || params.payload.replyToCurrent === false || !implicitReplyToId || ... | open | null | false | 1 | [] | [] | 2026-04-14T12:31:52Z | 2026-04-14T12:51:12Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | hakimi-agent | 259,029,502 | U_kgDOD3B5_g | User | false |
openclaw/openclaw | 4,262,188,471 | I_kwDOQb6kR87-C9W3 | 66,549 | https://github.com/openclaw/openclaw/issues/66549 | https://api.github.com/repos/openclaw/openclaw/issues/66549 | [Bug]: openclaw message send --dry-run still delivers the message via gateway | ## Summary
`openclaw message send --dry-run ...` has documented behavior: **"Print payload and skip sending (default: false)"**. Actual behavior: the message is printed AND **delivered** to the target channel.
## Environment
- **OpenClaw version**: 2026.3.28 (commit f9b1079)
- **Platform**: WSL2 Ubuntu 24.04 on Wind... | open | null | false | 0 | [] | [] | 2026-04-14T12:57:50Z | 2026-04-14T12:57:50Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | aleps001 | 211,350,258 | U_kgDODJjy8g | User | false |
openclaw/openclaw | 4,262,191,735 | I_kwDOQb6kR87-C-J3 | 66,550 | https://github.com/openclaw/openclaw/issues/66550 | https://api.github.com/repos/openclaw/openclaw/issues/66550 | Feature Request: Agent Group Chat / Multi-Agent Discussion Room | ## Problem
Currently, agents cannot proactively interact with each other. In a Feishu group chat, agents only respond when @mentioned by a human. There's no way to let multiple agents autonomously discuss, debate, or collaborate on a topic without manual orchestration.
## Proposed Solution
A "group session" where mu... | open | null | false | 0 | [] | [] | 2026-04-14T12:58:24Z | 2026-04-14T12:58:24Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | lizhihong114 | 37,773,237 | MDQ6VXNlcjM3NzczMjM3 | User | false |
openclaw/openclaw | 4,262,268,669 | I_kwDOQb6kR87-DQ79 | 66,553 | https://github.com/openclaw/openclaw/issues/66553 | https://api.github.com/repos/openclaw/openclaw/issues/66553 | [Bug]: Race condition between inbound voice message delivery and async STT transcription — agent turn fires before transcript is available | ## Bug Report
### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When a WhatsApp voice note arrives, the agent session turn fires **before** the STT transcription worker has populated the `transcription` column in the bridge database. The agent receives the me... | open | null | false | 1 | [] | [] | 2026-04-14T13:11:53Z | 2026-04-14T15:08:05Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | pidgeon777 | 13,739,892 | MDQ6VXNlcjEzNzM5ODky | User | false |
openclaw/openclaw | 4,262,311,089 | I_kwDOQb6kR87-DbSx | 66,558 | https://github.com/openclaw/openclaw/issues/66558 | https://api.github.com/repos/openclaw/openclaw/issues/66558 | [Bug] sessions_spawn / sessions_send not available in main agent session despite tools.profile: "coding" | ## Description
Main agent session does not have `sessions_spawn` or `sessions_send` tools available despite `tools.profile: "coding"` which includes `group:sessions`.
## Environment
- OpenClaw 2026.4.12
- macOS (Apple Silicon M1)
- memory.backend: "qmd"
- 13 agents configured
- Plugins loaded: acpx, browser, memory-... | open | null | false | 1 | [] | [] | 2026-04-14T13:19:24Z | 2026-04-14T14:37:21Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Neomail2 | 120,675,713 | U_kgDOBzFdgQ | User | false |
openclaw/openclaw | 4,262,361,432 | I_kwDOQb6kR87-DnlY | 66,561 | https://github.com/openclaw/openclaw/issues/66561 | https://api.github.com/repos/openclaw/openclaw/issues/66561 | [Bug]: openai-codex SSE stream begins, but embedded run aborts locally and is surfaced as timeout (408) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
OpenClaw appears to abort an `openai-codex/gpt-5.4` SSE response locally **after the upstream response has already started**, then surfaces the failure as a generic timeout/failover (`aborted:true`, `status:408... | open | null | false | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T13:28:11Z | 2026-04-16T20:50:02Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jkoopmann | 11,796,773 | MDQ6VXNlcjExNzk2Nzcz | User | false |
openclaw/openclaw | 4,262,379,513 | I_kwDOQb6kR87-Dr_5 | 66,564 | https://github.com/openclaw/openclaw/issues/66564 | https://api.github.com/repos/openclaw/openclaw/issues/66564 | [Bug]:Control UI feels significantly laggier in 2026.4.12 during normal chat use | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After updating to 2026.4.12, the Control UI became noticeably laggier during normal chat use, including sticky initial load and delayed visual settling after pressing Enter.
### Steps to reproduce
1. Start OpenClaw `2026... | open | null | false | 0 | [
"bug"
] | [] | 2026-04-14T13:31:01Z | 2026-04-14T13:31:01Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jarvisaeye | 276,004,881 | U_kgDOEHOAEQ | User | false |
openclaw/openclaw | 4,262,449,191 | I_kwDOQb6kR87-D9An | 66,573 | https://github.com/openclaw/openclaw/issues/66573 | https://api.github.com/repos/openclaw/openclaw/issues/66573 | [Bug]: chat.history limit=2000 exceeds server maximum of 1000, no retry | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
chat.history API 调用时 limit 参数超过服务器上限（1000），导致历史记录加载失败。客户端代码写死 limit=2000，无校验或重试机制。
### Steps to reproduce
启动 OpenClaw Gateway
任意发送消息触发 chat.history 调用（例如群里 @ 机器人）
观察日志
### Expected behavior
chat.history 应该请... | open | null | false | 1 | [
"bug"
] | [] | 2026-04-14T13:42:34Z | 2026-04-16T20:50:01Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Zhaojun1680 | 267,383,320 | U_kgDOD-_yGA | User | false |
openclaw/openclaw | 4,262,492,940 | I_kwDOQb6kR87-EHsM | 66,576 | https://github.com/openclaw/openclaw/issues/66576 | https://api.github.com/repos/openclaw/openclaw/issues/66576 | [Feature]: Configurable workspace file inclusion/exclusion (bootstrapFiles) | ## Problem
Workspace files (`AGENTS.md`, `SOUL.md`, `TOOLS.md`, `IDENTITY.md`, `USER.md`, `HEARTBEAT.md`, `MEMORY.md`) are injected into every session from a hardcoded `VALID_BOOTSTRAP_NAMES` set in `src/agents/workspace.ts`. There is no way to exclude specific files from injection or customize the list per agent.
Fo... | open | null | false | 0 | [] | [] | 2026-04-14T13:49:43Z | 2026-04-14T13:49:43Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | kagura-agent | 268,167,063 | U_kgDOD_vnlw | User | false |
openclaw/openclaw | 4,262,495,974 | I_kwDOQb6kR87-EIbm | 66,577 | https://github.com/openclaw/openclaw/issues/66577 | https://api.github.com/repos/openclaw/openclaw/issues/66577 | [Feature]: Add optional hard context truncation mode that keeps only a fixed recent portion of max context | ### Summary
I’d like OpenClaw to support an optional context management mode that hard-truncates old conversation history instead of requiring manual compact / new, or automatically summarizing/compressing older turns.
... | open | null | false | 0 | [
"enhancement"
] | [] | 2026-04-14T13:50:13Z | 2026-04-14T13:50:13Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | tangkk | 1,814,604 | MDQ6VXNlcjE4MTQ2MDQ= | User | false |
openclaw/openclaw | 4,262,524,742 | I_kwDOQb6kR87-EPdG | 66,579 | https://github.com/openclaw/openclaw/issues/66579 | https://api.github.com/repos/openclaw/openclaw/issues/66579 | [Bug]: WhatsApp auto-reply ignores plugin hooks (message_sending cancel / before_dispatch handled) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
A local hook-only plugin is loaded successfully, but WhatsApp auto-replies ignore both:
before_dispatch returning { handled: true }
message_sending returning { cancel: true }
This makes it impossible to suppre... | open | null | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T13:54:51Z | 2026-04-16T02:19:51Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | itsatish | 48,756,248 | MDQ6VXNlcjQ4NzU2MjQ4 | User | false |
openclaw/openclaw | 4,262,531,903 | I_kwDOQb6kR87-ERM_ | 66,581 | https://github.com/openclaw/openclaw/issues/66581 | https://api.github.com/repos/openclaw/openclaw/issues/66581 | toolsAllow parameter not forwarded to runEmbeddedAttemptWithBackend | ## Bug Description
`runEmbeddedPiAgent` accepts a `toolsAllow` parameter (used by active-memory plugin and `sessions_spawn` to restrict available tools for sub-agents), but **never forwards it** to `runEmbeddedAttemptWithBackend`. As a result, sub-agents receive ALL configured tools regardless of the `toolsAllow` filt... | open | null | false | 1 | [] | [] | 2026-04-14T13:56:01Z | 2026-04-16T20:50:01Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jojojonathan | 37,664,587 | MDQ6VXNlcjM3NjY0NTg3 | User | false |
openclaw/openclaw | 4,262,547,557 | I_kwDOQb6kR87-EVBl | 66,582 | https://github.com/openclaw/openclaw/issues/66582 | https://api.github.com/repos/openclaw/openclaw/issues/66582 | [Bug]: Active Memory issue | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Bug type
Behavior bug (incorrect output/state without crash)
Beta release blocker
Yes
Summary
I’m debugging Active Memory on OpenClaw 2026.4.12 and it consistently times out before returning any recalled summ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T13:58:30Z | 2026-04-17T04:22:22Z | 2026-04-15T02:54:15Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | happylucktomorrow | 171,368,153 | U_kgDOCjbe2Q | User | false |
openclaw/openclaw | 4,262,587,235 | I_kwDOQb6kR87-Eetj | 66,588 | https://github.com/openclaw/openclaw/issues/66588 | https://api.github.com/repos/openclaw/openclaw/issues/66588 | [Bug]: Browser plugin fails to launch on Raspberry Pi 5 (ARM) — falls back to web_fetch despite Chromium/Playwright working | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
I’m running OpenClaw on a Raspberry Pi 5 (Raspberry Pi OS Lite 64-bit, Debian Trixie), and the browser plugin consistently fails to launch. Instead, OpenClaw falls back to web_fetch with the message “browser is... | open | null | false | 0 | [
"bug"
] | [] | 2026-04-14T14:04:50Z | 2026-04-14T14:04:50Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | sfleung | 219,144,007 | U_kgDODQ_fRw | User | false |
openclaw/openclaw | 4,262,601,951 | I_kwDOQb6kR87-EiTf | 66,591 | https://github.com/openclaw/openclaw/issues/66591 | https://api.github.com/repos/openclaw/openclaw/issues/66591 | [Bug]: 2026.4.14 Broke lossless-claw: Context engine "lossless-claw" factory returned an invalid ContextEngine: info.id must match registered id "lossless-claw"." | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Upon upgrading to 2026.4.14, the lossless-claw engine appears to be broken. Attempting to communicate with the bot shows this error:
⚠️ Something went wrong while processing your request. Please try again, or use /new to st... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-14T14:07:12Z | 2026-04-17T04:22:49Z | 2026-04-14T15:47:34Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | AaronFaby | 41,449,666 | MDQ6VXNlcjQxNDQ5NjY2 | User | false |
openclaw/openclaw | 4,262,643,132 | I_kwDOQb6kR87-EsW8 | 66,595 | https://github.com/openclaw/openclaw/issues/66595 | https://api.github.com/repos/openclaw/openclaw/issues/66595 | Async relay heartbeat can emit 'result shown above' without payload | ## Summary
Heartbeat async relay events can claim command output is "shown above" even when no output payload exists in the session.
When this happens, the agent is asked to relay a result it cannot see, which leads to user-visible confusion.
## Observed behavior
Injected system-style prompt text (user role message i... | open | null | false | 4 | [] | [] | 2026-04-14T14:13:43Z | 2026-04-19T07:21:51Z | null | NONE | null | 20260419T082035Z | 2026-04-19T08:20:35Z | moltar-bot | 259,104,974 | U_kgDOD3Ggzg | User | false |
openclaw/openclaw | 4,262,682,674 | I_kwDOQb6kR87-E2Ay | 66,597 | https://github.com/openclaw/openclaw/issues/66597 | https://api.github.com/repos/openclaw/openclaw/issues/66597 | [webchat] User message flickers after sending (v2026.4.12) | After upgrading to 2026.4.12, user messages in webchat control-ui flash briefly (appear → disappear → reappear after 1-2s).
**Steps to reproduce:**
1. Open webchat control-ui
2. Have a long conversation (100k+ tokens)
3. Send any message
4. Message bubble appears instantly, then disappears, then reappears ~1-2s later
... | open | null | false | 0 | [] | [] | 2026-04-14T14:19:47Z | 2026-04-14T14:19:47Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | fredqiu995-del | 275,584,615 | U_kgDOEG0WZw | User | false |
openclaw/openclaw | 4,262,688,476 | I_kwDOQb6kR87-E3bc | 66,598 | https://github.com/openclaw/openclaw/issues/66598 | https://api.github.com/repos/openclaw/openclaw/issues/66598 | **Bug: Webchat message flickers after sending (v2026.4.12)** | After upgrading to 2026.4.12, user messages in webchat control-ui flash briefly (appear → disappear → reappear after 1-2s).
Steps to reproduce:
1. Open webchat control-ui
2. Have a long conversation (100k+ tokens)
3. Send any message
4. Message bubble appears instantly, then disappears, then reappears ~1-2s later
No... | open | null | false | 0 | [] | [] | 2026-04-14T14:20:39Z | 2026-04-14T14:20:39Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | fredqiu995-del | 275,584,615 | U_kgDOEG0WZw | User | false |
openclaw/openclaw | 4,262,699,518 | I_kwDOQb6kR87-E6H- | 66,601 | https://github.com/openclaw/openclaw/issues/66601 | https://api.github.com/repos/openclaw/openclaw/issues/66601 | OpenClaw v2026.4.14 causes repeated context engine errors and breaks usability | Hi,
After upgrading to OpenClaw v2026.4.14, I started seeing repeated runtime errors and the system became effectively unusable.
Release:
https://github.com/openclaw/openclaw/releases/tag/v2026.4.14
Main error:
Context engine "lossless-claw" factory returned an invalid ContextEngine: info.id must match registered i... | open | null | false | 9 | [] | [] | 2026-04-14T14:22:15Z | 2026-04-16T20:50:14Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | lamkan0210 | 65,824,253 | MDQ6VXNlcjY1ODI0MjUz | User | false |
openclaw/openclaw | 4,262,717,276 | I_kwDOQb6kR87-E-dc | 66,605 | https://github.com/openclaw/openclaw/issues/66605 | https://api.github.com/repos/openclaw/openclaw/issues/66605 | test: add integration test for table-aware block streaming chunking | ## Summary
#66568 added table-aware splitting to the block chunker (`EmbeddedBlockChunker`) and `chunkByParagraph`, preventing markdown tables from being split across messages during block streaming. The fix includes unit tests but no integration-level coverage through the `subscribeEmbeddedPiSession` pipeline.
The e... | open | null | false | 0 | [] | [] | 2026-04-14T14:24:52Z | 2026-04-14T14:24:52Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | zongqichen | 137,198,879 | U_kgDOCC19Hw | User | false |
openclaw/openclaw | 4,262,731,587 | I_kwDOQb6kR87-FB9D | 66,607 | https://github.com/openclaw/openclaw/issues/66607 | https://api.github.com/repos/openclaw/openclaw/issues/66607 | [Bug]: Control UI: thinking dropdown shows Default (low) even when effective/configured default is adaptive | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
In the OpenClaw Control UI WebChat, the thinking-level dropdown can show Default (low) even though the actual configured default and effective runtime thinking level are adaptive.
This appears to be a frontend display bug, ... | open | null | false | 1 | [
"bug"
] | [] | 2026-04-14T14:27:04Z | 2026-04-16T20:50:01Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | oromeis | 178,335,203 | U_kgDOCqEt4w | User | false |
openclaw/openclaw | 4,262,731,832 | I_kwDOQb6kR87-FCA4 | 66,608 | https://github.com/openclaw/openclaw/issues/66608 | https://api.github.com/repos/openclaw/openclaw/issues/66608 | Bug: Z.ai glm-4.6v vision model always aborts with 'Request was aborted' | ## Description
The built-in `image` tool always fails when using Z.ai's `glm-4.6v` vision model, returning:
```
Image model failed (zai/glm-4.6v): Request was aborted
```
## Investigation
- The abort happens **immediately** (< 1 second), so it's not a timeout
- Direct API calls to `https://api.z.ai/api/coding/paas/... | open | null | false | 0 | [] | [] | 2026-04-14T14:27:06Z | 2026-04-14T14:27:06Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | kulovema2012 | 131,257,476 | U_kgDOB9LUhA | User | false |
openclaw/openclaw | 4,262,799,658 | I_kwDOQb6kR87-FSkq | 66,611 | https://github.com/openclaw/openclaw/issues/66611 | https://api.github.com/repos/openclaw/openclaw/issues/66611 | # [Bug]: Browser attachOnly profile fails to connect despite Chrome running with CDP | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
---
## Summary
I'm running OpenClaw on a GCP VPS (Debian 12 arm64), and the browser plugin with attachOnly profile consistently fails to connect to externally running Chromium. The browser tool reports "profi... | open | null | false | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T14:37:54Z | 2026-04-14T14:38:30Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | sirgypsyg | 107,400,417 | U_kgDOBmbM4Q | User | false |
openclaw/openclaw | 4,262,814,371 | I_kwDOQb6kR87-FWKj | 66,612 | https://github.com/openclaw/openclaw/issues/66612 | https://api.github.com/repos/openclaw/openclaw/issues/66612 | Memory flush write sandbox blocks SESSION_HANDOFF.md despite prompt instruction | ## Bug
The pre-compaction memory flush prompt explicitly instructs the agent to update `SESSION_HANDOFF.md`:
> ALSO: Update SESSION_HANDOFF.md with current session state (what was being worked on, background processes, pending tasks, key decisions, file paths). This is critical for session continuity — the next sessi... | open | null | false | 0 | [] | [] | 2026-04-14T14:39:47Z | 2026-04-14T14:39:47Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | rosesandhello | 10,874,184 | MDQ6VXNlcjEwODc0MTg0 | User | false |
openclaw/openclaw | 4,262,838,607 | I_kwDOQb6kR87-FcFP | 66,614 | https://github.com/openclaw/openclaw/issues/66614 | https://api.github.com/repos/openclaw/openclaw/issues/66614 | [Bug]: Block streaming splits markdown tables across messages on all streaming channels | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Block streaming with `flushOnParagraph` splits at `\n\n` paragraph boundaries inside and around markdown tables, causing a single reply containing text + table + text to arrive as multiple separate messages on ... | open | null | false | 1 | [] | [] | 2026-04-14T14:43:05Z | 2026-04-16T20:50:01Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | zongqichen | 137,198,879 | U_kgDOCC19Hw | User | false |
openclaw/openclaw | 4,262,892,442 | I_kwDOQb6kR87-FpOa | 66,618 | https://github.com/openclaw/openclaw/issues/66618 | https://api.github.com/repos/openclaw/openclaw/issues/66618 | Scoped npm packages from ClawHub fail to install with ENOENT | ## Summary
`openclaw plugins install <scoped-package>` fails with `ENOENT` for any package whose name contains a slash (i.e., all `@scope/name` npm-style packages on ClawHub). Non-scoped packages install fine.
## Reproduction
```bash
openclaw plugins install @axonflow/openclaw@1.2.1
```
Output:
```
Resolving clawh... | closed | completed | true | 1 | [] | [] | 2026-04-14T14:51:18Z | 2026-04-17T04:22:33Z | 2026-04-14T23:05:42Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | saurabhjain1592 | 8,345,163 | MDQ6VXNlcjgzNDUxNjM= | User | false |
openclaw/openclaw | 4,262,919,047 | I_kwDOQb6kR87-FvuH | 66,619 | https://github.com/openclaw/openclaw/issues/66619 | https://api.github.com/repos/openclaw/openclaw/issues/66619 | Telegram setup crashes with TypeError: Cannot read properties of undefined (reading 'trim') | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
Running Telegram channel setup fails immediately after the DM policy warning with:
```
TypeError: Cannot read properties of undefined (reading 'trim')
```
### Steps to reproduce
### Steps to reproduce
1. Install openclaw (la... | closed | completed | true | 4 | [
"bug",
"bug:crash"
] | [] | 2026-04-14T14:55:34Z | 2026-04-17T04:21:48Z | 2026-04-14T16:11:49Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Pavel-Durov | 6,065,236 | MDQ6VXNlcjYwNjUyMzY= | User | false |
openclaw/openclaw | 4,262,921,343 | I_kwDOQb6kR87-FwR_ | 66,620 | https://github.com/openclaw/openclaw/issues/66620 | https://api.github.com/repos/openclaw/openclaw/issues/66620 | [Bug]: lossless-claw compaction summaries with Ollama cloud models return empty normalized output and fall back to truncation | ## Bug type
Bug / compatibility gap
## Summary
When `lossless-claw` uses Ollama cloud models for **compaction summaries**, the summary path can return an empty normalized result and silently fall back to truncation, even though:
- normal chat with the same models works
- delegated expansion can work
- the model is ... | open | null | false | 0 | [] | [] | 2026-04-14T14:55:59Z | 2026-04-14T14:55:59Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | BeliefanX | 12,130,040 | MDQ6VXNlcjEyMTMwMDQw | User | false |
openclaw/openclaw | 4,262,935,335 | I_kwDOQb6kR87-Fzsn | 66,621 | https://github.com/openclaw/openclaw/issues/66621 | https://api.github.com/repos/openclaw/openclaw/issues/66621 | Control UI (webchat) does not render user-sent image attachments | ## Environment
- **OpenClaw version**: 2026.4.12 (1c0672b)
- **Channel**: `main` (webchat / Control UI)
- **OS**: macOS (Darwin 25.4.0, arm64)
- **Browser**: Chrome (latest stable)
## Problem Summary
When sending an image attachment via the Control UI (webchat), the message bubble appears as a blank/empty message. Th... | closed | completed | true | 0 | [] | [] | 2026-04-14T14:58:08Z | 2026-04-19T04:23:12Z | 2026-04-16T14:37:58Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | woai36D | 269,974,840 | U_kgDOEBd9OA | User | false |
openclaw/openclaw | 4,262,975,663 | I_kwDOQb6kR87-F9iv | 66,624 | https://github.com/openclaw/openclaw/issues/66624 | https://api.github.com/repos/openclaw/openclaw/issues/66624 | [Bug]: TypeError: Cannot read properties of undefined (reading 'trim') during onboard / configure channel selection (v2026.4.14) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
### Version
OpenClaw 2026.4.14 (323493f)
Node.js v22.22.2
Ubuntu Linux (VPS)
### Description
Fresh install of v2026.4.14 crashes with a `TypeError` at the end of the
interactive setup wizard (channel selection step). The er... | closed | completed | false | 1 | [
"bug"
] | [] | 2026-04-14T15:04:56Z | 2026-04-14T15:06:01Z | 2026-04-14T15:06:01Z | NONE | null | 20260414T202441Z | 2026-04-14T20:24:41Z | emokhamadr-hq | 276,038,496 | U_kgDOEHQDYA | User | false |
openclaw/openclaw | 4,262,982,469 | I_kwDOQb6kR87-F_NF | 66,625 | https://github.com/openclaw/openclaw/issues/66625 | https://api.github.com/repos/openclaw/openclaw/issues/66625 | image tool fails with Minimax VLM model | ## Issue Summary
**Problem**: The `image` tool in OpenClaw consistently fails when configured with a Minimax vision model (`minimax-cn/Minimax-M2.7`). When a user sends an image, the `image` tool returns an error message instead of image description.
**Expected**: Tool should call Minimax VLM API and return image des... | open | null | false | 0 | [] | [] | 2026-04-14T15:06:07Z | 2026-04-14T15:06:07Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | EdyeeBlank | 183,996,152 | U_kgDOCveO-A | User | false |
openclaw/openclaw | 4,262,983,107 | I_kwDOQb6kR87-F_XD | 66,626 | https://github.com/openclaw/openclaw/issues/66626 | https://api.github.com/repos/openclaw/openclaw/issues/66626 | config.get leaks raw secrets via sourceConfig/runtimeConfig paths | ## Summary
The `config.get` tool/command has inconsistent secret redaction across its different output paths. While `parsed` and `resolved` paths properly redact sensitive values (showing `[REDACTED]` or similar), the `sourceConfig` and `runtimeConfig` paths return **raw unredacted values** of all secrets loaded from ... | open | null | false | 1 | [] | [] | 2026-04-14T15:06:14Z | 2026-04-16T20:50:00Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | haythemtrabelsi117 | 68,301,581 | MDQ6VXNlcjY4MzAxNTgx | User | false |
openclaw/openclaw | 4,263,012,781 | I_kwDOQb6kR87-GGmt | 66,631 | https://github.com/openclaw/openclaw/issues/66631 | https://api.github.com/repos/openclaw/openclaw/issues/66631 | [Bug] Feishu: Bot replies create new topics instead of replying within existing topic | # Fix: Feishu topic replies create new threads instead of replying within existing topic
## Problem Description
When a bot replies to messages in Feishu topic groups, it creates new sub-threads instead of replying within the original topic thread. This happens specifically when users send new messages directly in a t... | open | null | false | 0 | [] | [] | 2026-04-14T15:11:26Z | 2026-04-14T15:11:26Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Xinyu-101 | 62,863,525 | MDQ6VXNlcjYyODYzNTI1 | User | false |
openclaw/openclaw | 4,263,060,825 | I_kwDOQb6kR87-GSVZ | 66,633 | https://github.com/openclaw/openclaw/issues/66633 | https://api.github.com/repos/openclaw/openclaw/issues/66633 | [Bug]: openai-codex provider fails with Cloudflare 403 on every request after upgrade from 2026.4.12 → 2026.4.14 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.12 to 2026.4.14, every openai-codex/gpt-5.4 agent turn fails because chatgpt.com/backend-api/v1 returns a Cloudflare bot-mitigation challenge page (HTTP 403, cf-mitigated: challenge) instead of an... | open | null | false | 13 | [
"bug"
] | [] | 2026-04-14T15:20:03Z | 2026-04-16T21:47:34Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | harleymdsavage | 258,685,029 | U_kgDOD2s4ZQ | User | false |
openclaw/openclaw | 4,263,069,658 | I_kwDOQb6kR87-GUfa | 66,634 | https://github.com/openclaw/openclaw/issues/66634 | https://api.github.com/repos/openclaw/openclaw/issues/66634 | [Feature] RAG Knowledge Base Agent Attribution | ## Feature Request: RAG Knowledge Base Agent Attribution
### Problem Description
Currently, the `eo_rag_index` tool does not capture agent attribution when indexing knowledge chunks. This makes it impossible to track which agent contributed which knowledge to the shared knowledge base.
### Use Case
When multiple ag... | open | null | false | 0 | [] | [] | 2026-04-14T15:21:37Z | 2026-04-14T15:21:37Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | 467718584 | 57,902,585 | MDQ6VXNlcjU3OTAyNTg1 | User | false |
openclaw/openclaw | 4,263,100,176 | I_kwDOQb6kR87-Gb8Q | 66,635 | https://github.com/openclaw/openclaw/issues/66635 | https://api.github.com/repos/openclaw/openclaw/issues/66635 | [Bug]: WhatsApp auto-reply inline MEDIA:/absolute/path fails, while manual message send --media with same file succeeds | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On `v2026.4.14`, WhatsApp assistant auto-replies whose final payload is inline `MEDIA:/absolute/path` do not deliver the media, while manual `openclaw message send --media <same path>` succeeds with the same file and account... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-14T15:26:39Z | 2026-04-17T04:21:46Z | 2026-04-14T21:04:33Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Mohamed-HAMMANE | 35,642,940 | MDQ6VXNlcjM1NjQyOTQw | User | false |
openclaw/openclaw | 4,263,105,927 | I_kwDOQb6kR87-GdWH | 66,637 | https://github.com/openclaw/openclaw/issues/66637 | https://api.github.com/repos/openclaw/openclaw/issues/66637 | [Bug]: macOS app node browser proxy fails in remote mode, while CLI node-host on the same Mac works | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When OpenClaw is using a VPS gateway, a macOS app node can stay connected and successfully serve `canvas.*`, `camera.list`, and `location.get`, but `browser status target=node node=<app-node> profile=chrome-liv... | open | null | false | 0 | [
"bug"
] | [] | 2026-04-14T15:27:38Z | 2026-04-14T15:27:38Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | qearlyao | 201,136,044 | U_kgDOC_0XrA | User | false |
openclaw/openclaw | 4,263,126,057 | I_kwDOQb6kR87-GiQp | 66,638 | https://github.com/openclaw/openclaw/issues/66638 | https://api.github.com/repos/openclaw/openclaw/issues/66638 | [Feature]: Decouple subagent lifecycle notifications from session-lane batching | Summary
Introduce per-run, per-event subagent notifications with immediate spawn/progress/completion delivery, instead of batching announcements behind the parent session lane.
Problem to solve
Today, subagent notifications can be delayed until the parent workflow or session lane yields, which creates visible latency ... | open | null | false | 0 | [] | [] | 2026-04-14T15:30:59Z | 2026-04-14T15:30:59Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | pidgeon777 | 13,739,892 | MDQ6VXNlcjEzNzM5ODky | User | false |
openclaw/openclaw | 4,263,142,977 | I_kwDOQb6kR87-GmZB | 66,639 | https://github.com/openclaw/openclaw/issues/66639 | https://api.github.com/repos/openclaw/openclaw/issues/66639 | [Bug] image tool 不支持 minimax provider 作为图片理解后端 | ## 环境信息
- **OpenClaw 版本**：2026.4.11
- **模型**：minimax/MiniMax-M2.7（配置了 input: [\"text\", \"image\"]）
- **minimax API 认证方式**：api_key（Token Plan Key，sk-cp- 开头）
## 复现步骤
1. 在 openclaw.json 中配置 minimax provider，model 包含 image 输入类型（input: [\"text\", \"image\"]）
2. 配置 imageModel: \"minimax/MiniMax-M2.7\"
3. 发送图片给 bot
## 预期... | open | null | false | 1 | [] | [] | 2026-04-14T15:33:28Z | 2026-04-14T15:44:57Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | cjw0011 | 26,214,893 | MDQ6VXNlcjI2MjE0ODkz | User | false |
openclaw/openclaw | 4,263,144,762 | I_kwDOQb6kR87-Gm06 | 66,641 | https://github.com/openclaw/openclaw/issues/66641 | https://api.github.com/repos/openclaw/openclaw/issues/66641 | [Bug]: Installer always crashes after “Select channel (QuickStart)” with TypeError: Cannot read properties of undefined (reading 'trim') | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
During installation of openclaw 2026.4.14, the installer always crashes right after the Select channel (QuickStart) step.
This issue is consistently reproducible. It happens whether I select a channel or choose Skip for now.
E... | closed | completed | true | 8 | [
"bug"
] | [] | 2026-04-14T15:33:45Z | 2026-04-17T04:21:56Z | 2026-04-14T16:11:49Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Hiro674 | 231,472,620 | U_kgDODcv97A | User | false |
openclaw/openclaw | 4,263,164,399 | I_kwDOQb6kR87-Grnv | 66,642 | https://github.com/openclaw/openclaw/issues/66642 | https://api.github.com/repos/openclaw/openclaw/issues/66642 | Alternatives to free credits | ### Summary
### Question
You can use GPT 5.4 for free. https://monkeycode-ai.com/?ic=019d056b-8ca7-7c49-ba41-770717c65677
You can use GLM 5.1 for free. https://cloud.siliconflow.cn/i/MCDHLsQy
### Problem to solve
### Question
You can use GPT 5.4 for free. https://monkeycode-ai.com/?ic=019d056b-8ca7-7c49-ba41-77071... | closed | not_planned | true | 0 | [
"enhancement"
] | [] | 2026-04-14T15:37:08Z | 2026-04-17T04:22:47Z | 2026-04-14T16:00:40Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | aetrstrstesterst-hash | 239,808,251 | U_kgDODksu-w | User | false |
openclaw/openclaw | 4,263,167,360 | I_kwDOQb6kR87-GsWA | 66,643 | https://github.com/openclaw/openclaw/issues/66643 | https://api.github.com/repos/openclaw/openclaw/issues/66643 | 关于免费额度的替代方案 | ### Summary
### Question
可以免费使用gpt5.4 https://monkeycode-ai.com/?ic=019d056b-8ca7-7c49-ba41-770717c65677
可以免费使用GLM5.1 https://cloud.siliconflow.cn/i/MCDHLsQy
### Problem to solve
### Question
可以免费使用gpt5.4 https://monkeycode-ai.com/?ic=019d056b-8ca7-7c49-ba41-770717c65677
可以免费使用GLM5.1 https://cloud.siliconflow.cn/i... | closed | not_planned | true | 0 | [
"enhancement"
] | [] | 2026-04-14T15:37:42Z | 2026-04-17T04:22:48Z | 2026-04-14T16:00:26Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | aetrstrstesterst-hash | 239,808,251 | U_kgDODksu-w | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.