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 | 4,243,856,447 | I_kwDOQb6kR8789Bw_ | 64,724 | https://github.com/openclaw/openclaw/issues/64724 | https://api.github.com/repos/openclaw/openclaw/issues/64724 | Audio media understanding not firing for Telegram voice messages (Groq/Whisper) | ## Environment
- OpenClaw: 2026.4.10 (44e5b62)
- OS: macOS 26.4 (arm64)
- Channel: Telegram
- Audio provider: Groq (whisper-large-v3-turbo)
## Problem
Audio media understanding never fires when receiving Telegram voice messages (.ogg opus). The gateway downloads the file, passes it to the agent as `<media:audio>` with... | closed | completed | true | 1 | [] | [] | 2026-04-11T08:52:34Z | 2026-04-30T04:45:09Z | 2026-04-25T06:51:07Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | saju01 | 30,193,438 | MDQ6VXNlcjMwMTkzNDM4 | User | false |
openclaw/openclaw | 4,243,888,348 | I_kwDOQb6kR8789Jjc | 64,725 | https://github.com/openclaw/openclaw/issues/64725 | https://api.github.com/repos/openclaw/openclaw/issues/64725 | Tối ưu Hybrid Search scoring — configurable weights và MMR per agent | ## Mục tiêu
Cho phép cấu hình linh hoạt hybrid search scoring (vector weight, text weight, MMR lambda) theo từng agent hoặc use-case, thay vì dùng giá trị mặc định cố định 0.5/0.5.
## Context
Hiện tại hệ thống hybrid search trong `src/agents/memory-search.ts` sử dụng weight mặc định cho vector search và text search (F... | closed | completed | true | 1 | [] | [] | 2026-04-11T08:57:36Z | 2026-04-28T05:00:16Z | 2026-04-25T16:16:22Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | phanngoc | 3,756,788 | MDQ6VXNlcjM3NTY3ODg= | User | false |
openclaw/openclaw | 4,243,900,327 | I_kwDOQb6kR8789Men | 64,726 | https://github.com/openclaw/openclaw/issues/64726 | https://api.github.com/repos/openclaw/openclaw/issues/64726 | Cải thiện Embedding Cache — batch embedding và cache invalidation thông minh | ## Mục tiêu
Tối ưu embedding cache để giảm latency và chi phí API calls, bao gồm batch embedding API và smart cache invalidation khi content thay đổi nhỏ.
## Context
Embedding cache hiện tại trong `src/memory-host-sdk/` sử dụng SHA-256 hash trên content để lookup cache. Vấn đề:
1. **Không có batch embedding**: Mỗi chu... | open | null | false | 1 | [] | [] | 2026-04-11T08:59:31Z | 2026-05-03T01:37:30Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | phanngoc | 3,756,788 | MDQ6VXNlcjM3NTY3ODg= | User | false |
openclaw/openclaw | 4,243,903,914 | I_kwDOQb6kR8789NWq | 64,727 | https://github.com/openclaw/openclaw/issues/64727 | https://api.github.com/repos/openclaw/openclaw/issues/64727 | Subagent announce can return toolResult text instead of final assistant summary on tool-heavy runs | ## Summary
Subagent announce can return a `toolResult` blob instead of the child agent's final summary when the child session is tool-heavy.
The root cause appears to be that announce output extraction only reads `chat.history(limit=100)`. If the child run produces more than 100 messages, the final assistant summary ... | closed | completed | true | 2 | [] | [] | 2026-04-11T09:00:05Z | 2026-04-29T04:36:34Z | 2026-04-26T11:54:21Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | conanwhf | 14,151,735 | MDQ6VXNlcjE0MTUxNzM1 | User | false |
openclaw/openclaw | 4,243,905,295 | I_kwDOQb6kR8789NsP | 64,728 | https://github.com/openclaw/openclaw/issues/64728 | https://api.github.com/repos/openclaw/openclaw/issues/64728 | Feature: Anthropic Batch API support for non-urgent agent runs | ## Feature Request
**Problem:**
Running a fleet of AI agents (35+ in our case) on cron schedules and periodic tasks means paying full real-time API pricing for work that doesn't need instant responses. Anthropic's Batch API offers 50% cost reduction with up to 24-hour turnaround, which is perfect for scheduled/backgro... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T09:00:18Z | 2026-04-28T05:00:15Z | 2026-04-25T16:16:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | do3bol85 | 12,564,902 | MDQ6VXNlcjEyNTY0OTAy | User | false |
openclaw/openclaw | 4,243,911,648 | I_kwDOQb6kR8789PPg | 64,729 | https://github.com/openclaw/openclaw/issues/64729 | https://api.github.com/repos/openclaw/openclaw/issues/64729 | openclaw crons list — misleading error when stale plugin ID is in plugins.allow | ## Bug Description
When a stale/phantom plugin ID (e.g. `"crons"`) is present in `plugins.allow` but no plugin actually registers that command, `openclaw` produces a confusing error:
```
Failed to start CLI: Error: The `openclaw crons` command is unavailable because `plugins.allow` excludes "crons". Add "crons" to `p... | closed | completed | true | 1 | [] | [] | 2026-04-11T09:01:20Z | 2026-04-28T04:36:34Z | 2026-04-26T03:55:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | slideshow-dingo | 258,996,445 | U_kgDOD2_43Q | User | false |
openclaw/openclaw | 4,243,981,916 | I_kwDOQb6kR8789gZc | 64,732 | https://github.com/openclaw/openclaw/issues/64732 | https://api.github.com/repos/openclaw/openclaw/issues/64732 | [Bug]: openclaw tool/help CLI surfaces break when plugins.allow omits synthetic command ids | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Bundled plugin CLI surfaces like `openclaw tool` and even `openclaw help` can become unavailable when `plugins.allow` omits pseudo-plugin command names, causing local scripts that rely on these CLI entrypoints to fail unexpectedly. The error ... | closed | completed | true | 5 | [] | [] | 2026-04-11T09:16:07Z | 2026-05-04T04:34:20Z | 2026-05-02T03:07:04Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | efe-arv | 259,833,796 | U_kgDOD3y_xA | User | false |
openclaw/openclaw | 4,244,048,863 | I_kwDOQb6kR8789wvf | 64,738 | https://github.com/openclaw/openclaw/issues/64738 | https://api.github.com/repos/openclaw/openclaw/issues/64738 | Request iOS/Apple Watch TestFlight Access | **GitHub Username:** bfirwin
**Email (for TestFlight):** brian@legacytree.com
**Use case:**
Self-hosted gateway on macOS. Looking to test the iOS app and Apple Watch companion for mobile access and notifications.
Happy to provide feedback and bug reports. | closed | not_planned | true | 1 | [] | [] | 2026-04-11T09:41:25Z | 2026-04-19T04:22:53Z | 2026-04-17T00:03:43Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | bfirwin | 4,893,838 | MDQ6VXNlcjQ4OTM4Mzg= | User | false |
openclaw/openclaw | 4,244,054,787 | I_kwDOQb6kR8789yMD | 64,740 | https://github.com/openclaw/openclaw/issues/64740 | https://api.github.com/repos/openclaw/openclaw/issues/64740 | I just bypassed Claude Code security layer - Here is the solution | ### Summary
Looks like Claude tried to block OpenClaw with this genius method:
If the system prompt contains “HEARTBEAT.md” → block it
If not → allow it 😂
Just rename it to “HEARTBEATa.md” and bypass everything.
### Problem to solve
Claude requires extra usage package for using openclaw.
### Proposed solution
J... | closed | completed | true | 3 | [
"enhancement",
"security"
] | [] | 2026-04-11T09:43:54Z | 2026-04-28T05:00:14Z | 2026-04-25T16:16:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | onuratakan | 41,792,982 | MDQ6VXNlcjQxNzkyOTgy | User | false |
openclaw/openclaw | 4,244,059,631 | I_kwDOQb6kR8789zXv | 64,741 | https://github.com/openclaw/openclaw/issues/64741 | https://api.github.com/repos/openclaw/openclaw/issues/64741 | [Feature]: Why can't the 2026.04.05 version be updated via the web interface? | ### Summary
Previously, updates could be made through the web interface. However, starting from version 2026.04.05, the web interface has not been notifying us of any new versions, making it impossible to update at all.
### Problem to solve
The web interface has never reminded me of the existence of a new version.
... | closed | completed | true | 0 | [
"enhancement"
] | [] | 2026-04-11T09:46:08Z | 2026-04-15T04:19:19Z | 2026-04-13T02:30:20Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | kinlioncd | 83,751,473 | MDQ6VXNlcjgzNzUxNDcz | User | false |
openclaw/openclaw | 4,244,086,207 | I_kwDOQb6kR878952_ | 64,743 | https://github.com/openclaw/openclaw/issues/64743 | https://api.github.com/repos/openclaw/openclaw/issues/64743 | [Bug]: Using the MoE architecture model often leads to the adoption of other models | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The 2026.04.05 version uses the MoE architecture model, frequently reverting to other models, but it is still running using the MoE architecture model.
### Steps to reproduce
The 2026.04.05 version uses the M... | closed | not_planned | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-11T09:51:34Z | 2026-04-28T05:00:13Z | 2026-04-25T16:16:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kinlioncd | 83,751,473 | MDQ6VXNlcjgzNzUxNDcz | User | false |
openclaw/openclaw | 4,244,120,799 | I_kwDOQb6kR878-CTf | 64,744 | https://github.com/openclaw/openclaw/issues/64744 | https://api.github.com/repos/openclaw/openclaw/issues/64744 | [Bug]: codex/gpt-5.4 via bundled Codex harness hangs indefinitely on initialize in v2026.4.10 (macOS, codex-cli 0.118.0 and 0.120.0) | ## Summary
In OpenClaw **2026.4.10** (PR #64298, released 2026-04-10), using a `codex/gpt-*` model (bundled Codex plugin) causes inference to **hang indefinitely** with no error, no log output from the codex subsystem, and no timeout. The child `codex app-server` subprocess is spawned successfully and sits idle. The b... | closed | completed | true | 3 | [] | [] | 2026-04-11T09:59:53Z | 2026-04-29T04:32:26Z | 2026-04-27T02:28:45Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | jankutschera | 944,278 | MDQ6VXNlcjk0NDI3OA== | User | false |
openclaw/openclaw | 4,244,123,335 | I_kwDOQb6kR878-C7H | 64,745 | https://github.com/openclaw/openclaw/issues/64745 | https://api.github.com/repos/openclaw/openclaw/issues/64745 | [Bug]: Critical: macOS 2026.4.8 app causes infinite self-replication, system crash, data loss | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
Hello,
It is with great regret and anger that I report the following:
The OpenClaw version for macOS downloaded from your platform on April 8, 2026 (OpenClaw Releases page: https://github.com/openclaw/openclaw/releases) is ei... | closed | completed | true | 4 | [
"bug"
] | [] | 2026-04-11T10:00:29Z | 2026-04-29T04:32:25Z | 2026-04-27T02:28:50Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | RobertXie60 | 135,635,777 | U_kgDOCBWjQQ | User | false |
openclaw/openclaw | 4,244,152,541 | I_kwDOQb6kR878-KDd | 64,748 | https://github.com/openclaw/openclaw/issues/64748 | https://api.github.com/repos/openclaw/openclaw/issues/64748 | [Bug]: memory-wiki CLI allowlist uses `wiki` while doctor/config treat only `memory-wiki` as valid | ### Bug type
Behavior bug (incorrect config/tooling behavior without crash)
### Beta release blocker
No
### Summary
On OpenClaw 2026.4.9, the bundled `memory-wiki` plugin and the `openclaw wiki ...` CLI surface appear to use different identifiers for `plugins.allow` validation.
In practice:
- the plugin id is `m... | closed | completed | true | 0 | [] | [] | 2026-04-11T10:09:06Z | 2026-04-15T04:20:03Z | 2026-04-12T16:32:12Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | personaltrainerkoichan | 231,572,290 | U_kgDODc2DQg | User | false |
openclaw/openclaw | 4,244,181,482 | I_kwDOQb6kR878-RHq | 64,750 | https://github.com/openclaw/openclaw/issues/64750 | https://api.github.com/repos/openclaw/openclaw/issues/64750 | WhatsApp message.send returns success but drops attachments and delivers text only | ## Summary
`message.send` to the WhatsApp channel appears to accept media input and returns a successful `messageId`, but the recipient only receives the text body. No attachment is delivered.
This looks like a false-success bug in the WhatsApp outbound media path.
## Environment
- OpenClaw: `2026.4.9 (0512059)`
- ... | closed | not_planned | true | 1 | [
"clownfish"
] | [] | 2026-04-11T10:25:09Z | 2026-04-28T04:32:18Z | 2026-04-25T19:44:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jameslu249 | 132,040,173 | U_kgDOB97F7Q | User | false |
openclaw/openclaw | 4,244,185,277 | I_kwDOQb6kR878-SC9 | 64,751 | https://github.com/openclaw/openclaw/issues/64751 | https://api.github.com/repos/openclaw/openclaw/issues/64751 | [Bug]: Cron jobs marked as "error" despite successful agent run — delivery marking issue (Telegram announce) | ## Environment
- OpenClaw v2026.4.10 (44e5b62)
- Node.js v22.22.1
- Linux, Telegram (long-polling)
- 15 cron jobs configured, 8 showing `error` status
## Description
Cron jobs with `delivery.mode: "announce"` and `delivery.channel: "telegram"` consistently get marked as `error` in `cron/jobs.json` even though the age... | closed | not_planned | true | 2 | [] | [] | 2026-04-11T10:27:08Z | 2026-04-30T04:41:54Z | 2026-04-26T04:51:23Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | M-Lietz | 222,629,290 | U_kgDODUUNqg | User | false |
openclaw/openclaw | 4,244,185,963 | I_kwDOQb6kR878-SNr | 64,752 | https://github.com/openclaw/openclaw/issues/64752 | https://api.github.com/repos/openclaw/openclaw/issues/64752 | [Bug]: Inbound reaction context events arrive but do not trigger an agent turn (Telegram) | ## Environment
- OpenClaw v2026.4.10 (44e5b62)
- Node.js v22.22.1
- Telegram (long-polling), grammY
- Config: `messages.reactions.level: "extensive"`, `messages.reactions.triggerAgentTurn: true`
## Description
When a user reacts to a message in Telegram (e.g. adds a 👍 or ❤️ emoji), the gateway receives the `message_... | open | null | false | 6 | [] | [] | 2026-04-11T10:27:30Z | 2026-05-03T04:59:26Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | M-Lietz | 222,629,290 | U_kgDODUUNqg | User | false |
openclaw/openclaw | 4,244,208,524 | I_kwDOQb6kR878-XuM | 64,756 | https://github.com/openclaw/openclaw/issues/64756 | https://api.github.com/repos/openclaw/openclaw/issues/64756 | [Bug]: Dreaming page tekst on top of the dreaming page dashboard. | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Sindce I enabled dreaming in version 2026-04-7. When I open the dreaming page memory snipits are displayed on top of the dreaming dashboard. it shows recals and signals but I do not think this should be here.
### Steps to... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-04-11T10:39:00Z | 2026-04-28T05:00:13Z | 2026-04-25T16:16:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,244,215,783 | I_kwDOQb6kR878-Zfn | 64,757 | https://github.com/openclaw/openclaw/issues/64757 | https://api.github.com/repos/openclaw/openclaw/issues/64757 | Dreaming System writes GBK-encoded JSON into UTF-8 memory files causing garbled content | ## Bug Description
When the Dreaming System (auto-log/auto-dream/auto-confirm cron jobs) writes candidate entries to memory files, non-ASCII characters get corrupted. Characters become \锟斤拷\, missing radicals from Chinese characters (e.g. \淇\,\鏂\,\鏍\), or completely illegible sequences.
## Root Cause
The Dreaming Sy... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T10:42:26Z | 2026-04-28T05:00:12Z | 2026-04-25T16:17:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Barry-886 | 269,722,164 | U_kgDOEBOiNA | User | false |
openclaw/openclaw | 4,244,235,420 | I_kwDOQb6kR878-eSc | 64,759 | https://github.com/openclaw/openclaw/issues/64759 | https://api.github.com/repos/openclaw/openclaw/issues/64759 | Optimize: Remove workspace file double injection in system prompt | ## Problem
HEARTBEAT.md (and potentially other workspace files) are injected **twice** per turn:
1. In `## Project Context` (above cache boundary) — all `.md` workspace files
2. In `## Dynamic Project Context` (below cache boundary) — "frequently-changing project context files"
This means HEARTBEAT.md costs ~715 tok... | closed | completed | true | 1 | [] | [] | 2026-04-11T10:46:48Z | 2026-04-14T04:20:29Z | 2026-04-11T10:50:33Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | rithythul | 5,461,492 | MDQ6VXNlcjU0NjE0OTI= | User | false |
openclaw/openclaw | 4,244,235,794 | I_kwDOQb6kR878-eYS | 64,760 | https://github.com/openclaw/openclaw/issues/64760 | https://api.github.com/repos/openclaw/openclaw/issues/64760 | Optimize: Lazy-load skill descriptions in system prompt | ## Problem
All installed skills (67 in our setup) are listed in the system prompt every turn with name + location path. This costs ~1,500-2,000 tokens per turn. A typical session only uses 0-3 skills.
The descriptions are omitted (compact format), but even the name+path list is significant overhead that grows as user... | open | null | false | 2 | [] | [] | 2026-04-11T10:46:53Z | 2026-05-02T20:42:50Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | rithythul | 5,461,492 | MDQ6VXNlcjU0NjE0OTI= | User | false |
openclaw/openclaw | 4,244,236,241 | I_kwDOQb6kR878-efR | 64,761 | https://github.com/openclaw/openclaw/issues/64761 | https://api.github.com/repos/openclaw/openclaw/issues/64761 | Optimize: Reduce tool schema token cost per turn | ## Problem
All tool definitions (22 tools) with full JSON Schemas are injected every turn. This costs ~8,000-10,000 tokens per turn. A typical turn only uses 3-5 tools.
Each tool includes name, description, and full parameter schema (types, enums, defaults, descriptions for every field). For example, the `exec` tool ... | closed | not_planned | true | 2 | [] | [] | 2026-04-11T10:46:59Z | 2026-04-30T04:41:53Z | 2026-04-26T04:51:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rithythul | 5,461,492 | MDQ6VXNlcjU0NjE0OTI= | User | false |
openclaw/openclaw | 4,244,241,369 | I_kwDOQb6kR878-fvZ | 64,762 | https://github.com/openclaw/openclaw/issues/64762 | https://api.github.com/repos/openclaw/openclaw/issues/64762 | [Bug]: SSRF guard pinned DNS dispatcher corrupts FormData multipart bodies, breaking audio transcription | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Audio transcription via OpenAI-compatible providers (OpenAI, Groq, Mistral) silently fails with HTTP 400 because `fetchWithTimeoutGuarded` defaults to a pinned DNS dispatcher (undici) that corrupts `FormData` m... | closed | completed | true | 0 | [] | [] | 2026-04-11T10:48:06Z | 2026-04-14T04:20:27Z | 2026-04-11T12:35:38Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | GodsBoy | 5,792,287 | MDQ6VXNlcjU3OTIyODc= | User | false |
openclaw/openclaw | 4,244,244,041 | I_kwDOQb6kR878-gZJ | 64,763 | https://github.com/openclaw/openclaw/issues/64763 | https://api.github.com/repos/openclaw/openclaw/issues/64763 | [Feature]: Openclaw config skills updated to mach new version released. | ### Summary
When I want my agent to change config, I need to point it to the docs.openclaw.ai pages for it to read and understand what it needs to change or add. I think openclaw should automatically provide openclaw configuration skills to the agents. something build in and not dependend on external resources. also... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-11T10:48:40Z | 2026-04-28T05:00:11Z | 2026-04-25T16:17:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,244,247,648 | I_kwDOQb6kR878-hRg | 64,764 | https://github.com/openclaw/openclaw/issues/64764 | https://api.github.com/repos/openclaw/openclaw/issues/64764 | [Bug]: Remote CDP from WSL2 to Windows Edge is reported unreachable on 2026.4.10 even though HTTP and WebSocket both work | ## Bug type
Behavior bug
## Summary
On OpenClaw `2026.4.10`, a remote CDP browser profile from WSL2 to Windows Edge is reported as unreachable even though the endpoint is reachable from WSL2 over both HTTP and WebSocket.
## Environment
- OpenClaw `2026.4.10`
- Gateway in WSL2 Ubuntu
- Windows 11 host
- Microsoft Edge... | closed | completed | true | 2 | [] | [] | 2026-04-11T10:49:27Z | 2026-04-15T04:19:30Z | 2026-04-12T19:14:07Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | voytas75 | 13,856,061 | MDQ6VXNlcjEzODU2MDYx | User | false |
openclaw/openclaw | 4,244,254,829 | I_kwDOQb6kR878-jBt | 64,765 | https://github.com/openclaw/openclaw/issues/64765 | https://api.github.com/repos/openclaw/openclaw/issues/64765 | [Bug]: Active Memory returns `NONE` for relevant stored preference that manual memory search can find | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Active Memory is running, but it misses a relevant stored preference and returns `NONE` even though manual `openclaw memory search` can retrieve the same entry after reindex.
### Steps to reproduce
1. Enable ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-11T10:50:59Z | 2026-04-28T05:00:10Z | 2026-04-25T16:17:15Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | chall-gcpl | 163,865,947 | U_kgDOCcRlWw | User | false |
openclaw/openclaw | 4,244,277,989 | I_kwDOQb6kR878-orl | 64,767 | https://github.com/openclaw/openclaw/issues/64767 | https://api.github.com/repos/openclaw/openclaw/issues/64767 | [Bug] Bloated session jsonl (444 MB) hangs gateway via String.prototype.replace — diagnose with sample+lsof | # [Bug] Bloated session JSONL (444 MB) blocks event loop via String.prototype.replace — gateway becomes unresponsive on agent message processing
## Summary
A single session jsonl file in `~/.openclaw/agents/main/sessions/` grew unbounded to **444 MB / 157,879 lines** over 2 days. Whenever the agent processed a new me... | open | null | false | 1 | [] | [] | 2026-04-11T10:56:01Z | 2026-04-29T15:35:19Z | null | NONE | null | 20260429T210024Z | 2026-04-29T21:00:24Z | ppronobis | 50,087,465 | MDQ6VXNlcjUwMDg3NDY1 | User | false |
openclaw/openclaw | 4,244,290,489 | I_kwDOQb6kR878-ru5 | 64,770 | https://github.com/openclaw/openclaw/issues/64770 | https://api.github.com/repos/openclaw/openclaw/issues/64770 | [Feature]: Add Facebook Messenger Channel Support (similar to Telegram/Discord) | ### Summary
Add native support for Facebook Messenger as a communication channel in OpenClaw.
### Problem to solve
Currently, OpenClaw does not support Facebook Messenger as a channel, while other platforms like Telegram, Discord, WhatsApp, and Signal are supported.
This creates a limitation for users in environmen... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-11T10:59:12Z | 2026-04-28T05:00:10Z | 2026-04-25T16:17:22Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | BryarGh | 155,480,528 | U_kgDOCURx0A | User | false |
openclaw/openclaw | 4,244,314,468 | I_kwDOQb6kR878-xlk | 64,771 | https://github.com/openclaw/openclaw/issues/64771 | https://api.github.com/repos/openclaw/openclaw/issues/64771 | [Bug]: Chat disconnects after some time in Control UI — forces page refresh | ## Bug Description
Since updating to v2026.4.10, the chat interface in the Control UI disconnects after some time of inactivity. The user must refresh the page to restore the connection.
## Steps to reproduce
1. Open the Control UI (admin interface)
2. Leave the chat connected for some time (idle or inactive)
3. Afte... | closed | completed | true | 1 | [] | [] | 2026-04-11T11:06:08Z | 2026-04-28T04:59:48Z | 2026-04-25T16:21:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ttomiczek | 265,786,127 | U_kgDOD9eTDw | User | false |
openclaw/openclaw | 4,244,323,993 | I_kwDOQb6kR878-z6Z | 64,772 | https://github.com/openclaw/openclaw/issues/64772 | https://api.github.com/repos/openclaw/openclaw/issues/64772 | Feature request: verbatim transcript re-injection at session reset (bootstrap hook) | ## Problem
OpenClaw resets session context at 4:01 AM (or a configured hour) via a hard wipe. Some deployments mitigate this by writing a *narrative summary* to persistent storage before reset, which a bootstrap hook injects as a system message at session start.
This summary-only approach has a critical weakness: a s... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T11:09:53Z | 2026-04-28T04:59:47Z | 2026-04-25T16:21:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JuizSpeaking | 267,657,022 | U_kgDOD_QfPg | User | false |
openclaw/openclaw | 4,244,349,099 | I_kwDOQb6kR878-6Cr | 64,774 | https://github.com/openclaw/openclaw/issues/64774 | https://api.github.com/repos/openclaw/openclaw/issues/64774 | Readiness checker uses hardcoded staleEventThreshold (30min), ignores gateway.channelStaleEventThresholdMinutes config | ## Bug
The readiness checker in `src/gateway/server/readiness.ts` hardcodes `DEFAULT_CHANNEL_STALE_EVENT_THRESHOLD_MS = 30 * 60000` instead of reading the user-configured `gateway.channelStaleEventThresholdMinutes`.
## Impact
When a Discord (or other non-webhook) channel has no messages for >30 minutes, the readines... | closed | completed | true | 1 | [] | [] | 2026-04-11T11:19:43Z | 2026-04-27T04:56:11Z | 2026-04-24T21:12:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | CageHG | 174,996,616 | U_kgDOCm48iA | User | false |
openclaw/openclaw | 4,244,358,885 | I_kwDOQb6kR878-8bl | 64,775 | https://github.com/openclaw/openclaw/issues/64775 | https://api.github.com/repos/openclaw/openclaw/issues/64775 | [Bug]: Can't setup openclaw with codex, oauth error | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On the provider step, I've got this link to open
<img width="1031" height="296" alt="Image" src="https://github.com/user-attachments/assets/29445b89-7aa8-4942-ab27-40b6c106592e" />
after copying and pasting, I've got such ... | closed | completed | true | 3 | [
"bug",
"regression"
] | [
"vincentkoc"
] | 2026-04-11T11:22:35Z | 2026-04-15T04:20:10Z | 2026-04-12T15:41:39Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | romakoch | 61,388,038 | MDQ6VXNlcjYxMzg4MDM4 | User | false |
openclaw/openclaw | 4,244,367,493 | I_kwDOQb6kR878--iF | 64,776 | https://github.com/openclaw/openclaw/issues/64776 | https://api.github.com/repos/openclaw/openclaw/issues/64776 | sqlite-vec extension not loaded during gateway runtime memory sync | ## Description
When the gateway's memory manager syncs files at runtime (e.g., after file changes in `extraPaths`), it writes chunks to the database **without updating the `chunks_vec` virtual table** because the sqlite-vec extension is not loaded in that code path.
## Evidence
Gateway log warnings during runtime sy... | closed | completed | true | 4 | [] | [] | 2026-04-11T11:25:10Z | 2026-04-28T04:59:47Z | 2026-04-25T16:21:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | stamkivi | 5,263,291 | MDQ6VXNlcjUyNjMyOTE= | User | false |
openclaw/openclaw | 4,244,368,621 | I_kwDOQb6kR878--zt | 64,777 | https://github.com/openclaw/openclaw/issues/64777 | https://api.github.com/repos/openclaw/openclaw/issues/64777 | tools.profile default doesn't apply to channel sessions — channel gets restricted toolset when unset | ## Description
When `tools.profile` is not set in config, CLI agent sessions get the full toolset (including `cron`, `gateway`), but channel sessions (e.g., Signal) get a restricted set that excludes these tools. Setting `tools.profile: "full"` explicitly fixes it.
## Evidence
Tested by having the agent list its ava... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T11:25:32Z | 2026-04-28T04:59:46Z | 2026-04-25T16:21:26Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | stamkivi | 5,263,291 | MDQ6VXNlcjUyNjMyOTE= | User | false |
openclaw/openclaw | 4,244,369,497 | I_kwDOQb6kR878-_BZ | 64,778 | https://github.com/openclaw/openclaw/issues/64778 | https://api.github.com/repos/openclaw/openclaw/issues/64778 | FTS5 index not rebuilt after memory index --force — stale FTS returns 0 results | ## Description
After running `openclaw memory index --force`, the chunks table is correctly populated with new content, but the FTS5 index (`chunks_fts`) is not rebuilt. This causes `memory_search` to return 0 results for queries that have matching content in the chunks table.
## Evidence
After `memory index --force... | closed | completed | true | 1 | [] | [] | 2026-04-11T11:25:49Z | 2026-04-28T04:59:45Z | 2026-04-25T16:21:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | stamkivi | 5,263,291 | MDQ6VXNlcjUyNjMyOTE= | User | false |
openclaw/openclaw | 4,244,374,486 | I_kwDOQb6kR878_APW | 64,781 | https://github.com/openclaw/openclaw/issues/64781 | https://api.github.com/repos/openclaw/openclaw/issues/64781 | [Bug] Google Chat webhook returns 401 when gateway.auth.mode=token — webhook endpoints should bypass gateway token auth | ## Bug Description
Google Chat webhook returns `401 Unauthorized` when `gateway.auth.mode=token` is configured, even though the webhook endpoint (`/googlechat`) has its own JWT-based authentication via `audienceType` + `audience` config.
The gateway's token auth layer intercepts the request **before** the Google Chat... | closed | completed | true | 2 | [] | [] | 2026-04-11T11:27:21Z | 2026-04-14T04:20:28Z | 2026-04-11T12:01:39Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | yuting0624 | 32,728,916 | MDQ6VXNlcjMyNzI4OTE2 | User | false |
openclaw/openclaw | 4,244,401,676 | I_kwDOQb6kR878_G4M | 64,783 | https://github.com/openclaw/openclaw/issues/64783 | https://api.github.com/repos/openclaw/openclaw/issues/64783 | [Bug]: ReferenceError: Cannot access 'utils_1' before initialization when mentioning bot in Feishu group | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When mentioning (@) a bot in a Feishu group chat, the openclaw-lark plugin throws 'ReferenceError: Cannot access utils_1 before initialization', preventing the bot from responding.
### Steps to reproduce
1. C... | open | null | false | 5 | [
"bug"
] | [] | 2026-04-11T11:35:50Z | 2026-05-03T08:25:01Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | YJSansui | 110,321,469 | U_kgDOBpNfPQ | User | false |
openclaw/openclaw | 4,244,403,505 | I_kwDOQb6kR878_HUx | 64,784 | https://github.com/openclaw/openclaw/issues/64784 | https://api.github.com/repos/openclaw/openclaw/issues/64784 | [Feature]: LINE sticker send/receive support | ### Summary
Enable OpenClaw bots to send and receive LINE stickers, with a Skill-based
catalog system for AI-driven sticker selection.
### Problem to solve
In Japan and East Asia, LINE stickers are not decorative add-ons like emoji —
they are the primary unit of emotional communication. Sending a sticker instead
... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-04-11T11:36:35Z | 2026-05-03T02:14:58Z | null | CONTRIBUTOR | null | 20260503T030030Z | 2026-05-03T03:00:30Z | masatohoshino | 246,810,661 | U_kgDODrYIJQ | User | false |
openclaw/openclaw | 4,244,418,823 | I_kwDOQb6kR878_LEH | 64,785 | https://github.com/openclaw/openclaw/issues/64785 | https://api.github.com/repos/openclaw/openclaw/issues/64785 | bug(matrix): requireMention does not detect m.mentions from non-OpenClaw senders | ## Summary
When `requireMention: true` is set for a Matrix group room, messages containing valid `m.mentions` metadata are **not detected as mentions**, causing all agents to ignore the message. This effectively makes `requireMention: true` unusable for rooms where messages may come from non-OpenClaw Matrix clients or... | closed | completed | true | 0 | [] | [] | 2026-04-11T11:45:31Z | 2026-04-14T04:20:05Z | 2026-04-12T03:04:39Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | guci314 | 7,127,384 | MDQ6VXNlcjcxMjczODQ= | User | false |
openclaw/openclaw | 4,244,438,086 | I_kwDOQb6kR878_PxG | 64,788 | https://github.com/openclaw/openclaw/issues/64788 | https://api.github.com/repos/openclaw/openclaw/issues/64788 | Browser plugin: Chrome CDP connection fails when Chrome is running via launchd (cdpHttp: false) |
## Problem
When Chrome/Chromium is launched manually (or via launchd) with remote debugging port (e.g., \`--remote-debugging-port=18800\`), the OpenClaw browser plugin cannot connect to it. The browser plugin detects the Chrome process but reports:
```
running: false
cdpReady: false
cdpHttp: false
```
The \`/start\... | closed | completed | true | 3 | [] | [] | 2026-04-11T11:55:04Z | 2026-04-27T04:37:56Z | 2026-04-25T03:04:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | drzeast-png | 264,975,099 | U_kgDOD8sy-w | User | false |
openclaw/openclaw | 4,244,461,223 | I_kwDOQb6kR878_Van | 64,791 | https://github.com/openclaw/openclaw/issues/64791 | https://api.github.com/repos/openclaw/openclaw/issues/64791 | Skill Creator: Add Eval/Benchmark Loop & Description Optimizer | ## Feature Request
### Summary
The bundled `skill-creator` skill is excellent for authoring and packaging skills, but lacks **evaluation, benchmarking, and iterative optimization** capabilities. Anthropic's Claude skill-creator (in `anthropics/claude-plugins-official`) demonstrates a mature pattern for this that coul... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T12:07:10Z | 2026-04-30T04:41:52Z | 2026-04-26T04:51:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | eldron81-r2d2 | 269,572,528 | U_kgDOEBFZsA | User | false |
openclaw/openclaw | 4,244,462,110 | I_kwDOQb6kR878_Voe | 64,792 | https://github.com/openclaw/openclaw/issues/64792 | https://api.github.com/repos/openclaw/openclaw/issues/64792 | [Feature] Adaptive reasoning — auto-escalate thinking level on repeated failures | ## Problem
Currently, the reasoning/thinking level (`low`/`medium`/`high`) is a session-level config that must be changed manually via `/reasoning`. This means:
- Simple tasks waste tokens on high reasoning
- Complex tasks start at `low` and fail repeatedly until the user manually escalates
## Proposed Solution
Add ... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T12:07:38Z | 2026-04-30T04:41:52Z | 2026-04-26T04:51:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zengwei1997 | 38,238,670 | MDQ6VXNlcjM4MjM4Njcw | User | false |
openclaw/openclaw | 4,244,464,876 | I_kwDOQb6kR878_WTs | 64,793 | https://github.com/openclaw/openclaw/issues/64793 | https://api.github.com/repos/openclaw/openclaw/issues/64793 | [Bug]: Agent timeout does not surface error to UI, UI hangs indefinitely | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When an LLM request times out during agent execution, the agent correctly logs `decision=surface_error reason=timeout` but the Web UI hangs indefinitely showing a loading spinner instead of displaying a timeout... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-11T12:08:50Z | 2026-04-16T04:24:50Z | 2026-04-13T04:50:05Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | x0x0x8x | 49,066,909 | MDQ6VXNlcjQ5MDY2OTA5 | User | false |
openclaw/openclaw | 4,244,481,531 | I_kwDOQb6kR878_aX7 | 64,794 | https://github.com/openclaw/openclaw/issues/64794 | https://api.github.com/repos/openclaw/openclaw/issues/64794 | configure wizard: minimax-portal OAuth configPatch missing api and auth fields (2026.4.10) | ## Bug Description
After upgrading to 2026.4.10 and running `openclaw configure` with MiniMax OAuth (CN), the resulting config for `minimax-portal` is missing the `api` and `auth` fields, causing all models to fail with 'no available auth profile'.
## Steps to Reproduce
1. Have a working minimax-portal config from a... | closed | completed | true | 5 | [] | [] | 2026-04-11T12:16:30Z | 2026-04-14T04:20:18Z | 2026-04-11T18:50:45Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | kikibrian | 275,287,238 | U_kgDOEGiMxg | User | false |
openclaw/openclaw | 4,244,485,212 | I_kwDOQb6kR878_bRc | 64,795 | https://github.com/openclaw/openclaw/issues/64795 | https://api.github.com/repos/openclaw/openclaw/issues/64795 | heartbeat: isolatedSession: true silently reuses the same transcript file across every run | ## Summary
`agents.defaults.heartbeat.isolatedSession: true` is documented as producing a fresh session (with a new `sessionId` and an empty transcript) on every heartbeat run, but in practice it only rolls the `sessionId` in the store entry — the persisted `sessionFile` path is preserved via a spread, so every run ke... | closed | completed | true | 1 | [] | [] | 2026-04-11T12:18:05Z | 2026-04-28T04:59:44Z | 2026-04-25T16:21:39Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexander-applyinnovations | 111,744,509 | U_kgDOBqkV_Q | User | false |
openclaw/openclaw | 4,244,501,201 | I_kwDOQb6kR878_fLR | 64,798 | https://github.com/openclaw/openclaw/issues/64798 | https://api.github.com/repos/openclaw/openclaw/issues/64798 | [Bug]: No plugin updates needed | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Updating OpenClaw...
│
◇ ✓ Updating via package manager (48.65s)
│
◇ ✓ Running doctor checks (7.26s)
Update Result: OK
Root: /opt/homebrew/lib/node_modules/openclaw
Before: 2026.4.9
After: 2026.4.10
... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-11T12:24:16Z | 2026-04-28T04:59:44Z | 2026-04-25T16:21:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yuzhaohuang | 29,242,084 | MDQ6VXNlcjI5MjQyMDg0 | User | false |
openclaw/openclaw | 4,244,502,133 | I_kwDOQb6kR878_fZ1 | 64,799 | https://github.com/openclaw/openclaw/issues/64799 | https://api.github.com/repos/openclaw/openclaw/issues/64799 | [Bug]: Volcengine Coding Plan model resolves to "Unknown model: volcengine-plan/ark-code-latest" on macOS 26.4.1 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After reinstalling and reconfiguring OpenClaw 2026.4.10 on macOS, Volcengine Coding Plan fails before reply because the documented model ref `volcengine-plan/ark-code-latest` is rejected as `Unknown model`.
### Steps to rep... | closed | completed | true | 24 | [
"bug"
] | [] | 2026-04-11T12:24:40Z | 2026-04-28T04:59:43Z | 2026-04-25T16:21:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Siilent-Sea | 72,907,147 | MDQ6VXNlcjcyOTA3MTQ3 | User | false |
openclaw/openclaw | 4,244,538,616 | I_kwDOQb6kR878_oT4 | 64,801 | https://github.com/openclaw/openclaw/issues/64801 | https://api.github.com/repos/openclaw/openclaw/issues/64801 | Discord channel silently fails to load when `plugins.allow` list is present but excludes discord | ## Bug Report
**OpenClaw version:** 2026.4.9
### What happened
When `plugins.allow` is defined in `openclaw.json` but does not include `"discord"`, the Discord channel plugin is silently skipped during startup. The channel never connects, no error or warning is logged, and `openclaw status` shows an empty Channels t... | closed | completed | true | 1 | [] | [] | 2026-04-11T12:36:28Z | 2026-04-28T04:59:42Z | 2026-04-25T16:21:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | CyberRaccoonTeam | 195,688,984 | U_kgDOC6n6GA | User | false |
openclaw/openclaw | 4,244,540,097 | I_kwDOQb6kR878_orB | 64,802 | https://github.com/openclaw/openclaw/issues/64802 | https://api.github.com/repos/openclaw/openclaw/issues/64802 | [Bug]: *Railway Docker image needs update to 2026.4.9** | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
> **Railway Docker image needs update to 2026.4.9**
>
> Running OpenClaw on Railway using the official Docker image. Currently stuck on `2026.4.8` — both the WhatsApp and Telegram channel plugins fail to load with:
>
> `PluginL... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-11T12:36:56Z | 2026-04-28T04:59:41Z | 2026-04-25T16:22:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Bkreutzman | 268,697,306 | U_kgDOEAP-2g | User | false |
openclaw/openclaw | 4,244,552,483 | I_kwDOQb6kR878_rsj | 64,803 | https://github.com/openclaw/openclaw/issues/64803 | https://api.github.com/repos/openclaw/openclaw/issues/64803 | Discord voice messages not recognized as audio (content_type missing or incorrect) | **Bug Report**
Discord voice messages are received as `<media:document>` instead of `<media:audio>`, preventing audio transcription from triggering.
**Environment:**
- OpenClaw 2026.4.10
- Discord channel with `requireMention: false`
- `tools.media.audio` configured with OpenAI Whisper
**Problem:**
In `extensions/di... | closed | completed | true | 1 | [] | [] | 2026-04-11T12:40:36Z | 2026-04-28T05:26:56Z | 2026-04-25T09:58:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zcode-apps | 87,060,116 | MDQ6VXNlcjg3MDYwMTE2 | User | false |
openclaw/openclaw | 4,244,596,137 | I_kwDOQb6kR878_2Wp | 64,806 | https://github.com/openclaw/openclaw/issues/64806 | https://api.github.com/repos/openclaw/openclaw/issues/64806 | Config change auto-restart should be optional/configurable | # Feature Request: Config Change Auto-Restart Should Be Optional/Configurable
## Problem
OpenClaw automatically restarts the gateway ~5 minutes after `openclaw.json` is modified. This behavior is:
- Undocumented
- Unexpected
- Potentially disruptive (active sessions/tasks interrupted)
## Current Behavior
When `open... | closed | completed | true | 1 | [] | [] | 2026-04-11T12:57:47Z | 2026-04-28T04:59:41Z | 2026-04-25T16:22:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Donmeusi | 109,555,289 | U_kgDOBoeuWQ | User | false |
openclaw/openclaw | 4,244,599,314 | I_kwDOQb6kR878_3IS | 64,807 | https://github.com/openclaw/openclaw/issues/64807 | https://api.github.com/repos/openclaw/openclaw/issues/64807 | [Bug]: Dashboard/Overview update banner can diverge from CLI because gateway trusts stale or empty update-check cache | ## Summary
The Control UI / Overview update banner can miss available package updates even though `openclaw update status` reports them correctly.
From local diagnosis, the gateway snapshot appears to rely on the persisted `~/.openclaw/update-check.json` state, while the CLI uses a live registry path. When the persis... | closed | completed | true | 2 | [] | [] | 2026-04-11T12:59:14Z | 2026-05-04T04:35:40Z | 2026-05-01T20:26:03Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | Fuma2013 | 6,313,918 | MDQ6VXNlcjYzMTM5MTg= | User | false |
openclaw/openclaw | 4,244,607,661 | I_kwDOQb6kR878_5Kt | 64,810 | https://github.com/openclaw/openclaw/issues/64810 | https://api.github.com/repos/openclaw/openclaw/issues/64810 | [Bug]: Heartbeat / async system events can interrupt and effectively swallow in-progress replies in Telegram topic sessions | ## Summary
In Telegram forum-topic sessions, heartbeat/system-event turns can preempt an in-progress user reply and make the original answer effectively disappear from the user's perspective.
This is not just heartbeat noise. The more serious effect is that a normal user question can get interrupted by heartbeat polls... | open | null | false | 4 | [] | [] | 2026-04-11T13:02:41Z | 2026-05-02T23:57:28Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | jackiedepp | 31,266,295 | MDQ6VXNlcjMxMjY2Mjk1 | User | false |
openclaw/openclaw | 4,244,610,514 | I_kwDOQb6kR878_53S | 64,812 | https://github.com/openclaw/openclaw/issues/64812 | https://api.github.com/repos/openclaw/openclaw/issues/64812 | Feature: Pass field through to provider API request body | ## Problem
OpenClaw's gateway accepts arbitrary params in agent/model config (`agents.defaults.params`), but `createStreamFnWithExtraParams` has a hardcoded allowlist of fields that get passed to the API request:
- `temperature`
- `maxTokens`
- `transport`
- `openaiWsWarmup`
- `cachedContent`/`cached_content`
- `cach... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T13:03:45Z | 2026-04-30T04:41:51Z | 2026-04-26T04:51:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | JanPlessow | 202,702,070 | U_kgDODBT89g | User | false |
openclaw/openclaw | 4,244,621,992 | I_kwDOQb6kR878_8qo | 64,814 | https://github.com/openclaw/openclaw/issues/64814 | https://api.github.com/repos/openclaw/openclaw/issues/64814 | [Bug]: api.pluginConfig is null in extension plugin register() despite config being set in openclaw.json | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading to 2026.4.9 (0512059) the openclaw-b2-backup v0.1.8 complains about not having values set in its config block despite the values being present.
### Steps to reproduce
Steps to reproduce:
1. Install: opencl... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-11T13:08:38Z | 2026-04-28T04:59:40Z | 2026-04-25T16:22:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | bhinmaine | 4,717,735 | MDQ6VXNlcjQ3MTc3MzU= | User | false |
openclaw/openclaw | 4,244,622,452 | I_kwDOQb6kR878_8x0 | 64,815 | https://github.com/openclaw/openclaw/issues/64815 | https://api.github.com/repos/openclaw/openclaw/issues/64815 | Codex app-server rejects stale serviceTier "priority", and startup warmup misses async model resolution | ## Summary
I hit two related issues while testing the bundled Codex provider / app-server path in OpenClaw `2026.4.10` / source checkout `850182b502ccd35caf52d6313113e03fc5521d99`.
1. A stale `serviceTier: "priority"` value can still make it through the OpenClaw Codex bridge, even though the Codex app-server only acc... | closed | completed | true | 2 | [] | [] | 2026-04-11T13:08:52Z | 2026-04-25T04:15:56Z | 2026-04-23T01:30:48Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | sunnydongbo | 249,433,142 | U_kgDODt4MNg | User | false |
openclaw/openclaw | 4,244,635,749 | I_kwDOQb6kR879AABl | 64,816 | https://github.com/openclaw/openclaw/issues/64816 | https://api.github.com/repos/openclaw/openclaw/issues/64816 | Plugin-bundled skill visible to CLI but not surfaced to agent available_skills in live session | ## Summary
A plugin-bundled skill can exist on disk and appear in `openclaw skills list`, but still not be surfaced to the live agent session skill inventory / `<available_skills>` context.
In my case this happened with the `memory-wiki` bundled skill `wiki-maintainer`.
## Environment
- OpenClaw: `2026.4.10`
- Instal... | closed | completed | true | 1 | [] | [] | 2026-04-11T13:13:55Z | 2026-04-30T04:41:50Z | 2026-04-26T04:51:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jarvisz8 | 260,156,352 | U_kgDOD4GrwA | User | false |
openclaw/openclaw | 4,244,636,360 | I_kwDOQb6kR879AALI | 64,818 | https://github.com/openclaw/openclaw/issues/64818 | https://api.github.com/repos/openclaw/openclaw/issues/64818 | openclaw update: tracked src/canvas-host/a2ui/.bundle.hash breaks preflight bisect walkback | ### Bug type
Regression (worked before, now fails)
### Summary
When `openclaw update` runs preflight and the candidate commit fails (e.g. lint), the updater walks backward through earlier commits to find a "good" one. Every one of those fallback checkouts fails with:
```
error: Your local changes to the following f... | open | null | false | 1 | [] | [] | 2026-04-11T13:14:07Z | 2026-05-02T23:12:24Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | protomota | 1,140,777 | MDQ6VXNlcjExNDA3Nzc= | User | false |
openclaw/openclaw | 4,244,678,902 | I_kwDOQb6kR879AKj2 | 64,821 | https://github.com/openclaw/openclaw/issues/64821 | https://api.github.com/repos/openclaw/openclaw/issues/64821 | [Bug]: tools.exec.security corrupted by plugin init logs — hot-reload concatenates plugin output into config value | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
During normal gateway operation, the config values tools.exec.security and tools.exec.strictInlineEval are silently corrupted by the hot-reload mechanism. The correct values ("full" and false) are replaced with a concatenation ... | open | null | false | 1 | [
"bug",
"security",
"bug:crash"
] | [] | 2026-04-11T13:30:06Z | 2026-04-14T15:18:35Z | null | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | alyssonlago | 208,375,574 | U_kgDODGuPFg | User | false |
openclaw/openclaw | 4,244,725,441 | I_kwDOQb6kR879AV7B | 64,825 | https://github.com/openclaw/openclaw/issues/64825 | https://api.github.com/repos/openclaw/openclaw/issues/64825 | TUI stuck on 'streaming' indicator after run completes — finalizeRun() doesn't transition UI when wasActiveRun is false | ## Summary
The TUI's status bar stays stuck on `streaming` for minutes (indefinitely in some cases) after a chat turn has already completed. The visible reply renders correctly, but the spinner keeps running and the status line keeps saying `streaming • N m Ns` long after `sessions.json` has already recorded the run a... | closed | completed | true | 2 | [] | [] | 2026-04-11T13:45:22Z | 2026-05-02T04:28:28Z | 2026-04-29T11:12:27Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | adamgries | 133,591 | MDQ6VXNlcjEzMzU5MQ== | User | false |
openclaw/openclaw | 4,244,744,698 | I_kwDOQb6kR879Aan6 | 64,826 | https://github.com/openclaw/openclaw/issues/64826 | https://api.github.com/repos/openclaw/openclaw/issues/64826 | Feature Request: Model switching via inline keyword prefix in CLI/TUI | ## Summary
Would love the ability to switch models on-the-fly using a leading keyword prefix directly in the prompt input — without needing a separate `/model` command first.
## Example Syntax
```
*XAI* What's the latest on tariffs and Bitcoin?
*CLD* Draft a long-form article outline on MSTR accumulation
*GRK* Run d... | closed | completed | true | 1 | [] | [] | 2026-04-11T13:49:55Z | 2026-04-28T04:59:39Z | 2026-04-25T16:22:39Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | UTXOMacro | 273,661,899 | U_kgDOEE-_yw | User | false |
openclaw/openclaw | 4,244,782,416 | I_kwDOQb6kR879Aj1Q | 64,828 | https://github.com/openclaw/openclaw/issues/64828 | https://api.github.com/repos/openclaw/openclaw/issues/64828 | feat(feishu): support in-place card content update on button click | ## Problem
When Feishu cards are rendered with interactive buttons (approval confirm/cancel, quick actions, etc.), clicking a button triggers the action but leaves the card in its original visual state. Users see no immediate feedback until the next agent reply arrives — which can take several seconds or longer.
## P... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T13:59:45Z | 2026-04-30T04:41:49Z | 2026-04-26T04:51:47Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | TaoXieSZ | 7,379,039 | MDQ6VXNlcjczNzkwMzk= | User | false |
openclaw/openclaw | 4,244,782,828 | I_kwDOQb6kR879Aj7s | 64,829 | https://github.com/openclaw/openclaw/issues/64829 | https://api.github.com/repos/openclaw/openclaw/issues/64829 | Feature: per-plugin QMD searchMode override for Active Memory | ## Problem
Active Memory plugin uses the global `memory.qmd.searchMode` setting (currently `"query"` ~4s) when calling `memory_search`. There is no way to configure a different (faster) search mode specifically for Active Memory while keeping the main agent on the highest-quality `"query"` mode.
### Current behavior
... | closed | completed | true | 0 | [] | [] | 2026-04-11T13:59:51Z | 2026-04-14T04:20:06Z | 2026-04-12T02:03:41Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | melissafoell2-bit | 258,903,832 | U_kgDOD26PGA | User | false |
openclaw/openclaw | 4,244,826,766 | I_kwDOQb6kR879AuqO | 64,830 | https://github.com/openclaw/openclaw/issues/64830 | https://api.github.com/repos/openclaw/openclaw/issues/64830 | [Feature] Web UI new session should respect agents.defaults.thinkingDefault and agents.defaults.reasoningDefault | ### Summary
## Summary
Web UI 新建会话时，thinking 和 reasoning 的下拉选择器默认显示 "inherit (模型硬编码兜底值)"，而不是读取配置文件中的 `agents.defaults.thinkingDefault` 和 `agents.defaults.reasoningDefault`。
## Problem
用户每次在 Web UI 新建会话后，thinking/reasoning 都需要手动重新选择，无法通过配置文件设置默认行为。
## Expected behavior
Web UI 新建会话时，应该按以下优先级读取默认值：
1. `agents.defaults.th... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-04-11T14:15:05Z | 2026-05-02T04:27:52Z | 2026-04-29T15:32:40Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | xiaotong08 | 40,711,753 | MDQ6VXNlcjQwNzExNzUz | User | false |
openclaw/openclaw | 4,244,836,905 | I_kwDOQb6kR879AxIp | 64,831 | https://github.com/openclaw/openclaw/issues/64831 | https://api.github.com/repos/openclaw/openclaw/issues/64831 | [Bug]: Cron jobs run inconsistently, chat-triggered exec unstable, and missing catch-up delivery on Telegram | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
• Cron jobs are configured, but delivery is inconsistent.
• Chat-triggered execution works for simple commands, fails for real workflows.
• Approval flow is unreliable in practice (timing/state issues).
• Heavy pipelines can... | open | null | false | 1 | [
"bug",
"regression"
] | [] | 2026-04-11T14:17:42Z | 2026-05-03T01:22:37Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | Aredes228 | 68,281,093 | MDQ6VXNlcjY4MjgxMDkz | User | false |
openclaw/openclaw | 4,244,870,914 | I_kwDOQb6kR879A5cC | 64,833 | https://github.com/openclaw/openclaw/issues/64833 | https://api.github.com/repos/openclaw/openclaw/issues/64833 | OpenAI Codex OAuth fails with misleading callback error; actual redirect is invalid_scope for model.request | ## Summary
OpenAI Codex OAuth setup currently fails during browser auth with a misleading callback error page:
- Browser page shows: `Authentication failed` / `Missing authorization code.`
- Actual redirect URL contains:
- `error=invalid_scope`
- `error_description=The OAuth 2.0 Client is not allowed to request s... | closed | completed | true | 1 | [] | [] | 2026-04-11T14:26:55Z | 2026-04-22T04:30:57Z | 2026-04-19T14:36:37Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Silronin | 67,084,317 | MDQ6VXNlcjY3MDg0MzE3 | User | false |
openclaw/openclaw | 4,244,901,388 | I_kwDOQb6kR879BA4M | 64,835 | https://github.com/openclaw/openclaw/issues/64835 | https://api.github.com/repos/openclaw/openclaw/issues/64835 | [Bug]: memory_search returns no results — vec0 virtual table shadow data not populated | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
memory_search returns zero results for all queries despite a fully indexed memory database (511 chunks, 80 files). The vec0 virtual table shadow data table (chunks_vec_vector_chunks00) contains only 1 row rega... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-11T14:37:47Z | 2026-04-14T04:20:19Z | 2026-04-11T18:27:01Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | popmegaphone-byte | 250,897,603 | U_kgDODvRkww | User | false |
openclaw/openclaw | 4,244,907,954 | I_kwDOQb6kR879BCey | 64,836 | https://github.com/openclaw/openclaw/issues/64836 | https://api.github.com/repos/openclaw/openclaw/issues/64836 | Auth config lost after upgrading openclaw via npm | ## Problem
After upgrading openclaw via \`npm install -g openclaw@latest\`, the MiniMax API key authentication stopped working.
## Error
\`\`\`
All models failed: minimax-portal/MiniMax-M2.7: No available auth profile for minimax-portal (all in cooldown or unavailable). (model_not_found) | minimax/MiniMax-M2.7: No A... | closed | completed | true | 1 | [] | [] | 2026-04-11T14:40:33Z | 2026-04-29T04:32:24Z | 2026-04-27T02:29:01Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | lostrain | 17,779,458 | MDQ6VXNlcjE3Nzc5NDU4 | User | false |
openclaw/openclaw | 4,244,917,114 | I_kwDOQb6kR879BEt6 | 64,838 | https://github.com/openclaw/openclaw/issues/64838 | https://api.github.com/repos/openclaw/openclaw/issues/64838 | File editor: CTRL-F hides the file instead of triggering browser find | ## Bug Description
When editing a file in the OpenClaw Control UI file editor, pressing CTRL-F (browser-native find/search) causes the file to disappear/hide, rather than opening the browser search bar to find text within the file.
## Steps to Reproduce
1. Open any file in the OpenClaw file editor (e.g. SOUL.md, IDEN... | closed | completed | true | 1 | [] | [] | 2026-04-11T14:44:26Z | 2026-04-28T04:59:38Z | 2026-04-25T16:22:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ttomiczek | 265,786,127 | U_kgDOD9eTDw | User | false |
openclaw/openclaw | 4,244,917,751 | I_kwDOQb6kR879BE33 | 64,839 | https://github.com/openclaw/openclaw/issues/64839 | https://api.github.com/repos/openclaw/openclaw/issues/64839 | [Bug]: qwen3.6-plus via bailian cannot handle images after upgrade: image tool says Unknown model and QQ media URL is blocked | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading OpenClaw to 2026.4.10, image recognition broke across WebUI, WeChat, and QQ when using bailian/qwen3.6-plus.
Symptoms:
- WebUI/WeChat attachments were dropped with: model does not support images
- QQ image do... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-11T14:44:44Z | 2026-04-28T04:59:38Z | 2026-04-25T16:22:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xanderwowb-design | 267,043,140 | U_kgDOD-rBRA | User | false |
openclaw/openclaw | 4,244,927,057 | I_kwDOQb6kR879BHJR | 64,840 | https://github.com/openclaw/openclaw/issues/64840 | https://api.github.com/repos/openclaw/openclaw/issues/64840 | Model fallback not triggered on HTTP 400 model_not_supported error | ## Bug Description
When the primary model returns HTTP 400 with `model_not_supported` error code, OpenClaw does not trigger model fallback to configured fallback models. The agent run simply fails with an error.
## Expected Behavior
`model_not_supported` / `model_not_found` errors should be classified as failover-wo... | closed | completed | true | 1 | [] | [] | 2026-04-11T14:48:43Z | 2026-04-29T04:41:28Z | 2026-04-26T08:21:25Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | karle102017 | 48,557,267 | MDQ6VXNlcjQ4NTU3MjY3 | User | false |
openclaw/openclaw | 4,244,929,636 | I_kwDOQb6kR879BHxk | 64,841 | https://github.com/openclaw/openclaw/issues/64841 | https://api.github.com/repos/openclaw/openclaw/issues/64841 | [Bug]: Disabled memory-lancedb entry still fails config validation after upgrading to 2026.4.9 | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
After upgrading OpenClaw to `2026.4.9`, the gateway fails to start if `plugins.entries.memory-lancedb` is present in config with `"enabled": false` but without a `config.embedding` block.
This environment uses `mem9` as the ac... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-04-11T14:49:48Z | 2026-04-28T04:59:37Z | 2026-04-25T16:22:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Charlesmpc | 97,822,572 | U_kgDOBdSnbA | User | false |
openclaw/openclaw | 4,244,937,819 | I_kwDOQb6kR879BJxb | 64,844 | https://github.com/openclaw/openclaw/issues/64844 | https://api.github.com/repos/openclaw/openclaw/issues/64844 | [Bug]: /new session ignores agents.defaults.model.primary — injects wrong model | **Describe the bug**
After setting `agents.defaults.model.primary` to a specific model (e.g. `ollama/minimax-m2.7:cloud`), starting a new session via `/new` ignores this config and injects a `model_change` event for a different model (`ollama/gemma4:31b-cloud`).
**Steps to Reproduce**
1. Configure `agents.defaults.mod... | closed | completed | true | 1 | [] | [] | 2026-04-11T14:53:00Z | 2026-04-28T04:59:36Z | 2026-04-25T16:23:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | friday-hetland | 270,614,982 | U_kgDOECFBxg | User | false |
openclaw/openclaw | 4,244,938,374 | I_kwDOQb6kR879BJ6G | 64,845 | https://github.com/openclaw/openclaw/issues/64845 | https://api.github.com/repos/openclaw/openclaw/issues/64845 | [Bug]: Memory dreaming narrative generation fails — 'must have required property idempotencyKey' | **Describe the bug**
The memory-core dreaming pipeline fails during the narrative generation phase with:
```
narrative generation failed for [light/rem] phase: invalid agent params: must have required property 'idempotencyKey'
```
**Steps to Reproduce**
1. Enable memory-core dreaming (`plugins.entries.memory-core.conf... | closed | completed | true | 4 | [] | [] | 2026-04-11T14:53:15Z | 2026-04-15T04:19:55Z | 2026-04-12T16:46:29Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | friday-hetland | 270,614,982 | U_kgDOECFBxg | User | false |
openclaw/openclaw | 4,244,949,310 | I_kwDOQb6kR879BMk- | 64,850 | https://github.com/openclaw/openclaw/issues/64850 | https://api.github.com/repos/openclaw/openclaw/issues/64850 | [Bug] :2026.4.10: ACP runtime backend unavailable persists — qwen-code agent cannot spawn | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
### Summary
After upgrading to OpenClaw 2026.4.10, ACP sessions still fail to start with "ACP runtime backend is currently unavailable" when using qwen-code agent. This appears to be a continuation of #62588.
... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-11T14:58:01Z | 2026-04-28T04:59:35Z | 2026-04-25T16:23:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Ray-Huang-9730 | 182,090,644 | U_kgDOCtp7lA | User | false |
openclaw/openclaw | 4,244,952,367 | I_kwDOQb6kR879BNUv | 64,852 | https://github.com/openclaw/openclaw/issues/64852 | https://api.github.com/repos/openclaw/openclaw/issues/64852 | [Bug]: google-vertex OAuth/service-account auth works on 2026.4.9, fails on 2026.4.10 with 401 CREDENTIALS_MISSING | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.9 to 2026.4.10, google-vertex requests that worked with OAuth/service-account authentication in the same environment now fail with 401 UNAUTHENTICATED / CREDENTIALS_MISSING.
### Steps to reproduc... | closed | completed | false | 7 | [
"bug",
"close:already-fixed",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-04-11T14:59:34Z | 2026-04-25T08:59:58Z | 2026-04-25T08:59:58Z | NONE | null | 20260425T090015Z | 2026-04-25T09:00:15Z | qearlyao | 201,136,044 | U_kgDOC_0XrA | User | false |
openclaw/openclaw | 4,244,955,651 | I_kwDOQb6kR879BOID | 64,853 | https://github.com/openclaw/openclaw/issues/64853 | https://api.github.com/repos/openclaw/openclaw/issues/64853 | [Bug] CLI commands hang on Windows | CLI commands that communicate with gateway get SIGKILL on Windows. Test: openclaw status -> SIGKILL. See also #64211 | closed | completed | true | 1 | [] | [] | 2026-04-11T15:00:55Z | 2026-04-30T04:41:48Z | 2026-04-26T04:51:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | X-bluewinds | 34,063,978 | MDQ6VXNlcjM0MDYzOTc4 | User | false |
openclaw/openclaw | 4,244,968,514 | I_kwDOQb6kR879BRRC | 64,854 | https://github.com/openclaw/openclaw/issues/64854 | https://api.github.com/repos/openclaw/openclaw/issues/64854 | Feature request: per-model request timeout so fallback chain triggers on hangs | # Feature request: per-model request timeout so fallback chain triggers on hangs
**OpenClaw version:** 2026.4.9
## Problem
If an LLM provider call hangs (no HTTP response, no error), the fallback chain
configured in `agents.list[].model.fallbacks` never triggers. Fallback is
error-driven — it only activates on an HT... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:06:40Z | 2026-04-28T04:59:35Z | 2026-04-25T16:23:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | joblas | 40,072,373 | MDQ6VXNlcjQwMDcyMzcz | User | false |
openclaw/openclaw | 4,244,973,666 | I_kwDOQb6kR879BShi | 64,855 | https://github.com/openclaw/openclaw/issues/64855 | https://api.github.com/repos/openclaw/openclaw/issues/64855 | [Feature Request] Context Router: Selective Context Injection Based on Conversation Type | ## Feature Request: Context Router / Lightweight Mode
### Problem
Currently, OpenClaw injects the full workspace context (AGENTS.md, SOUL.md, USER.md, IDENTITY.md, TOOLS.md, HEARTBEAT.md) into **every** LLM interaction, regardless of the conversation type.
For casual chat / simple questions, this is wasteful:
- Send... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T15:09:06Z | 2026-04-28T04:59:34Z | 2026-04-25T16:23:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | shizhewanglu | 262,947,431 | U_kgDOD6xCZw | User | false |
openclaw/openclaw | 4,244,975,955 | I_kwDOQb6kR879BTFT | 64,856 | https://github.com/openclaw/openclaw/issues/64856 | https://api.github.com/repos/openclaw/openclaw/issues/64856 | Heartbeat isolated session can repeatedly re-trigger memory-core dreaming because cron system events are peeked from main session but drained from :heartbeat session | Title: Heartbeat isolated session can repeatedly re-trigger memory-core dreaming because cron system events are peeked from main session but drained from :heartbeat session
## Summary
In our deployment, `memory-core` dreaming was being re-triggered every 30 minutes after the managed cron job had already enqueued its ... | closed | completed | true | 2 | [] | [] | 2026-04-11T15:10:08Z | 2026-04-15T04:19:42Z | 2026-04-12T16:55:30Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | WWWa222 | 165,674,810 | U_kgDOCd__Og | User | false |
openclaw/openclaw | 4,244,987,371 | I_kwDOQb6kR879BV3r | 64,860 | https://github.com/openclaw/openclaw/issues/64860 | https://api.github.com/repos/openclaw/openclaw/issues/64860 | QA Lab extension fails to find scenario pack after 4.8 update | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The qa/scenarios/index.md file is missing from /opt/homebrew/lib/node_modules/openclaw/, causing the QA Lab extension to emit warnings during completion cache updates.
### Steps to reproduce
Steps to Reprod... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-11T15:15:13Z | 2026-04-28T04:59:33Z | 2026-04-25T16:23:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | brickheadbs | 66,306,637 | MDQ6VXNlcjY2MzA2NjM3 | User | false |
openclaw/openclaw | 4,244,990,473 | I_kwDOQb6kR879BWoJ | 64,861 | https://github.com/openclaw/openclaw/issues/64861 | https://api.github.com/repos/openclaw/openclaw/issues/64861 | [Bug]: openai-codex auth fails on 2026.4.10 with invalid_scope for model.request | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Worked on 2026.4.5 - Fails on 2026.4.10
Command: openclaw models auth login --provider openai-codex
Returned callback error: http://localhost:1455/auth/callbackerror=invalid_scope&error_description=The+requeste... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [
"vincentkoc"
] | 2026-04-11T15:16:35Z | 2026-04-15T04:20:09Z | 2026-04-12T15:41:42Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | blabaer | 89,535,496 | MDQ6VXNlcjg5NTM1NDk2 | User | false |
openclaw/openclaw | 4,244,994,877 | I_kwDOQb6kR879BXs9 | 64,866 | https://github.com/openclaw/openclaw/issues/64866 | https://api.github.com/repos/openclaw/openclaw/issues/64866 | [Bug] minimax-portal OAuth login VLM image understanding not working despite having coding-plan-vlm quota | ## Issue
### Environment
- OpenClaw version: 2026.4.8
- Model: minimax-portal/MiniMax-M2.7-highspeed
- OS: macOS
### Problem Description
User has MiniMax Annual Coding Plan subscription which includes image understanding quota (coding-plan-vlm endpoint).
Confirmed at platform.minimax.io:
- coding-plan-vlm: 1/450 use... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:18:45Z | 2026-04-28T04:59:32Z | 2026-04-25T16:23:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | loiclin090-tech | 275,317,217 | U_kgDOEGkB4Q | User | false |
openclaw/openclaw | 4,244,999,359 | I_kwDOQb6kR879BYy_ | 64,867 | https://github.com/openclaw/openclaw/issues/64867 | https://api.github.com/repos/openclaw/openclaw/issues/64867 | [Bug]: Active Memory `timeoutMs` not respected — failover chain blocks replies for 40-125 seconds | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The Active Memory plugin's `config.timeoutMs` setting (set to 15000ms) is not enforced as a hard cap on total execution time. When the initial LLM request times out, the embedded run failover chain (rotate auth... | closed | completed | true | 0 | [
"bug",
"bug:behavior"
] | [
"Takhoffman"
] | 2026-04-11T15:21:09Z | 2026-04-14T04:20:07Z | 2026-04-12T01:20:39Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | sbmilburn | 11,755,869 | MDQ6VXNlcjExNzU1ODY5 | User | false |
openclaw/openclaw | 4,245,010,440 | I_kwDOQb6kR879BbgI | 64,870 | https://github.com/openclaw/openclaw/issues/64870 | https://api.github.com/repos/openclaw/openclaw/issues/64870 | [Feature]: MiniMax music-2.6 model not available in OpenClaw music_generate tool | ### Summary
The MiniMax official API documentation (https://platform.minimaxi.com/docs/api-reference/music-generation) lists available music models as:
music-2.6 (recommended)
music-cover
music-2.6-free
music-cover-free
However, OpenClaw's MiniMax music provider configuration only exposes:
music-2.5+ (default)
music-2... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-11T15:26:39Z | 2026-04-28T05:27:01Z | 2026-04-25T09:57:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yidaoxiong | 46,669,604 | MDQ6VXNlcjQ2NjY5NjA0 | User | false |
openclaw/openclaw | 4,245,027,088 | I_kwDOQb6kR879BfkQ | 64,876 | https://github.com/openclaw/openclaw/issues/64876 | https://api.github.com/repos/openclaw/openclaw/issues/64876 | [Bug] Agent responses contain duplicate/repeated text within a single message | ### Description
The AI agent sometimes generates duplicate text within a single response. For example, the same sentence or paragraph appears twice in one message.
### Steps to Reproduce
1. Send a message to the agent while it is busy processing a long task
2. The message gets queued
3. When the agent finishes and pr... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:34:22Z | 2026-04-28T04:59:32Z | 2026-04-25T16:24:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | williamqian | 19,399,075 | MDQ6VXNlcjE5Mzk5MDc1 | User | false |
openclaw/openclaw | 4,245,030,154 | I_kwDOQb6kR879BgUK | 64,877 | https://github.com/openclaw/openclaw/issues/64877 | https://api.github.com/repos/openclaw/openclaw/issues/64877 | Bug: config.patch schema rejects valid active-memory config fields | **Severity: High — makes active-memory plugin unconfigurable via CLI
**OpenClaw version:** 2026.4.10
**Description:**
config.patch validates active-memory config against a schema that does not match the actual plugin code. The validator rejects agents as object and logging as object, while the plugin normalizePluginC... | closed | completed | true | 0 | [] | [] | 2026-04-11T15:35:49Z | 2026-04-14T04:20:08Z | 2026-04-12T01:19:44Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | AlfredTaddeusPennyworth | 261,802,943 | U_kgDOD5rLvw | User | false |
openclaw/openclaw | 4,245,034,727 | I_kwDOQb6kR879Bhbn | 64,878 | https://github.com/openclaw/openclaw/issues/64878 | https://api.github.com/repos/openclaw/openclaw/issues/64878 | qa-lab: run the parity gate end-to-end in CI and publish artifacts | ## Problem
The parity gate has never been run end-to-end. There are no `.artifacts/qa-e2e/qa-agentic-parity-*` files anywhere in the repo, and no CI workflow triggers `openclaw qa suite --parity-pack agentic`. This means the entire "parity gate shows GPT-5.4 matches or beats Opus 4.6" claim (completion criterion 5 in ... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:38:05Z | 2026-04-28T04:59:31Z | 2026-04-25T16:24:10Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,036,106 | I_kwDOQb6kR879BhxK | 64,879 | https://github.com/openclaw/openclaw/issues/64879 | https://api.github.com/repos/openclaw/openclaw/issues/64879 | qa-lab: make the mock server differentiate providers (or add a live-frontier lane) | ## Problem
The qa-lab mock server's scenario dispatcher (`buildResponsesPayload` in `extensions/qa-lab/src/mock-openai-server.ts`) branches on prompt text, not on `body.model`. I traced three uses of `body.model` in the file (lines 360, 782, 803) and all three are just prose echo or snapshot serialization — none of th... | open | null | false | 1 | [] | [] | 2026-04-11T15:38:45Z | 2026-05-02T23:57:29Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,036,179 | I_kwDOQb6kR879BhyT | 64,880 | https://github.com/openclaw/openclaw/issues/64880 | https://api.github.com/repos/openclaw/openclaw/issues/64880 | qa-lab: extend tool-call assertions to the 4 remaining parity scenarios | ## Problem
Only six of ten parity scenarios currently gate on real tool calls via `/debug/requests` assertions. The remaining four still pass on prose alone:
- `image-understanding-attachment` — checks `imageInputCount >= 1` (verifies the mock saw the image, not that the real model actually processed it)
- `subagent-... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:38:48Z | 2026-04-28T04:59:30Z | 2026-04-25T16:24:17Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,038,006 | I_kwDOQb6kR879BiO2 | 64,881 | https://github.com/openclaw/openclaw/issues/64881 | https://api.github.com/repos/openclaw/openclaw/issues/64881 | agents: trace fullAccessAvailable to a real provider signal + regression tests | ## Problem
Completion criterion 3 in #64227 ("GPT-5.4 no longer gives false `/elevated full` guidance") is enforced at the system prompt layer: `src/agents/system-prompt.ts:781` correctly gates `/elevated full` suggestions on `elevated.fullAccessAvailable === true`. PR B #64439 is credited with making this truthful.
... | open | null | false | 1 | [] | [] | 2026-04-11T15:39:43Z | 2026-05-02T23:12:25Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,038,082 | I_kwDOQb6kR879BiQC | 64,882 | https://github.com/openclaw/openclaw/issues/64882 | https://api.github.com/repos/openclaw/openclaw/issues/64882 | qa-lab: strengthen SUSPICIOUS_PASS_PATTERNS to catch positive-tone fake success | ## Problem
The parity-report fake-success detector in `extensions/qa-lab/src/agentic-parity-report.ts:67-77` uses nine regex patterns to flag "suspicious pass" scenarios — passes that carry failure-tone details. All nine patterns match failure-tone language:
```
/incomplete turn/i, /\btimed out\b/i, /\btimeout\b/i, /... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:39:46Z | 2026-04-28T04:59:30Z | 2026-04-25T16:24:23Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,038,160 | I_kwDOQb6kR879BiRQ | 64,883 | https://github.com/openclaw/openclaw/issues/64883 | https://api.github.com/repos/openclaw/openclaw/issues/64883 | agents: strengthen planning-only detector against dummy tool calls | ## Problem
The planning-only detector in `src/agents/pi-embedded-runner/run/incomplete-turn.ts:50-54` runs regex patterns against the assistant's prose text:
```
PLANNING_ONLY_PROMISE_RE = /\b(?:i(?:'ll| will)|let me|going to|...)\b/i
PLANNING_ONLY_COMPLETION_RE = /\b(?:done|finished|implemented|...)\b/i
PLANNING_ONL... | open | null | false | 1 | [] | [] | 2026-04-11T15:39:48Z | 2026-05-02T23:42:37Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,039,919 | I_kwDOQb6kR879Bisv | 64,884 | https://github.com/openclaw/openclaw/issues/64884 | https://api.github.com/repos/openclaw/openclaw/issues/64884 | active-memory: hardcoded default fallback model is expensive and assumes OpenAI | ## Bug Description
The active-memory extension has a hardcoded fallback model:
```
const DEFAULT_MODEL_REF = "github-copilot/gpt-5.4-mini";
```
When `modelFallbackPolicy: "default-remote"` is set AND the configured model fails (or `modelFallbackPolicy` is not set), active-memory silently falls back to this hardcoded... | closed | completed | true | 0 | [] | [] | 2026-04-11T15:40:39Z | 2026-04-14T04:20:07Z | 2026-04-12T01:24:10Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ttomiczek | 265,786,127 | U_kgDOD9eTDw | User | false |
openclaw/openclaw | 4,245,040,073 | I_kwDOQb6kR879BivJ | 64,885 | https://github.com/openclaw/openclaw/issues/64885 | https://api.github.com/repos/openclaw/openclaw/issues/64885 | qa-lab: verify run.primaryProvider matches caller labels in parity gate | ## Problem
PR L #64789 adds a self-describing `run` block to `qa-suite-summary.json` with `primaryProvider`, `primaryModel`, `providerMode`, and `scenarioIds`. The parity gate reads the summary files but never actually verifies that `candidateSummary.run.primaryProvider` matches the `--candidate-label` the caller pass... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:40:45Z | 2026-04-28T04:59:29Z | 2026-04-25T16:24:30Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,040,140 | I_kwDOQb6kR879BiwM | 64,886 | https://github.com/openclaw/openclaw/issues/64886 | https://api.github.com/repos/openclaw/openclaw/issues/64886 | agents: widen strict-agentic auto-activation to handle prefixed/variant model IDs | ## Problem
Strict-agentic auto-activation (added by PR H #64679) uses this regex in `src/agents/execution-contract.ts:24`:
```typescript
return /^gpt-5(?:[.-]|$)/i.test(params.modelId?.trim() ?? "");
```
This misses:
- Prefixed variants: `openai/gpt-5.4`, `openai:gpt-5.4` (the regex anchors on `^gpt-5`, so anything... | closed | completed | true | 1 | [] | [] | 2026-04-11T15:40:47Z | 2026-04-28T04:59:28Z | 2026-04-25T16:24:36Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,040,203 | I_kwDOQb6kR879BixL | 64,887 | https://github.com/openclaw/openclaw/issues/64887 | https://api.github.com/repos/openclaw/openclaw/issues/64887 | agents: surface livenessState in user-facing payloads | ## Problem
PR C #64300 and PR H #64679 land comprehensive terminal lifecycle metadata in `src/agents/pi-embedded-runner/run.ts` — every terminal branch sets `replayInvalid` + `livenessState` via `setTerminalLifecycleMeta`. That's the strongest slice of the parity program.
The remaining hole is that `meta.livenessStat... | open | null | false | 1 | [] | [] | 2026-04-11T15:40:49Z | 2026-05-02T23:42:38Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,040,270 | I_kwDOQb6kR879BiyO | 64,888 | https://github.com/openclaw/openclaw/issues/64888 | https://api.github.com/repos/openclaw/openclaw/issues/64888 | agents: wrap runEmbeddedAttempt in error boundary + add cancelled liveness state | ## Problem
Two small but related gaps in the terminal lifecycle coverage from PRs C and H:
1. **No uncaught-exception boundary around `runEmbeddedAttempt`.** If the inner run function throws (network glitch, provider SDK bug, runtime OOM), `setTerminalLifecycleMeta` is never called and the caller gets no terminal lif... | open | null | false | 1 | [] | [] | 2026-04-11T15:40:52Z | 2026-05-03T10:41:08Z | null | CONTRIBUTOR | null | 20260503T120038Z | 2026-05-03T12:00:38Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,245,046,045 | I_kwDOQb6kR879BkMd | 64,889 | https://github.com/openclaw/openclaw/issues/64889 | https://api.github.com/repos/openclaw/openclaw/issues/64889 | sessions_spawn and /subagents spawn use inconsistent native subagent paths | ## Bug Description
There are **two different subagent spawn surfaces** in OpenClaw, and they behave inconsistently:
1. `sessions_spawn` (tool surface)
2. `/subagents spawn` (command-handler surface)
The first surface can fail for native subagent spawns because it accepts ACP-only params like `streamTo`, while the se... | closed | not_planned | true | 1 | [] | [] | 2026-04-11T15:43:41Z | 2026-04-30T04:41:48Z | 2026-04-26T04:51:55Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | iflylikeabird | 258,836,296 | U_kgDOD22HSA | User | false |
openclaw/openclaw | 4,245,049,621 | I_kwDOQb6kR879BlEV | 64,891 | https://github.com/openclaw/openclaw/issues/64891 | https://api.github.com/repos/openclaw/openclaw/issues/64891 | [Bug]: Hundreds of repeated HTTP requests to 169.254.169.254:80 for the first 10 minutes after the startup | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading OpenClaw in my NemoClaw environment from `2026.3.11` to `2026.4.9`, NVIDIA's openshell sandbox started reporting GET and PUT requests to `http://169.254.169.254/latest/api/token` and `http://169.254.169.254/l... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-04-11T15:45:33Z | 2026-05-02T23:57:30Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | vnicolici | 2,158,971 | MDQ6VXNlcjIxNTg5NzE= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.