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 | 4,235,481,155 | I_kwDOQb6kR878dFBD | 63,968 | https://github.com/openclaw/openclaw/issues/63968 | https://api.github.com/repos/openclaw/openclaw/issues/63968 | [Bug]: Packaging regression in 2026.4.9: missing qa/scenarios/index.md breaks `openclaw qa` | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.5 to 2026.4.9, `openclaw qa --help` fails because the installed package is missing `qa/scenarios/index.md`.
### Steps to reproduce
1. Update OpenClaw from 2026.4.5 to 2026.4.9.
2. Run `openclaw ... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-04-09T23:25:06Z | 2026-04-28T05:02:19Z | 2026-04-25T15:53:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kevinibt | 46,771,285 | MDQ6VXNlcjQ2NzcxMjg1 | User | false |
openclaw/openclaw | 4,235,613,535 | I_kwDOQb6kR878dlVf | 63,979 | https://github.com/openclaw/openclaw/issues/63979 | https://api.github.com/repos/openclaw/openclaw/issues/63979 | [Bug]: Gateway never clearing streaming: true flag for main session | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
In my TUI, only for my main agent/main session starting this afternoon, after a reply it always says "Streaming" and the time ticks up. It was not able to debug this itself, and here is a summary of what it discovered. It be... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-10T00:06:21Z | 2026-04-13T04:28:33Z | 2026-04-10T08:02:03Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | patrissimo | 110,461 | MDQ6VXNlcjExMDQ2MQ== | User | false |
openclaw/openclaw | 4,235,614,612 | I_kwDOQb6kR878dlmU | 63,980 | https://github.com/openclaw/openclaw/issues/63980 | https://api.github.com/repos/openclaw/openclaw/issues/63980 | iMessage echo detection: DM misidentified as self-chat when sender == chat_identifier | ## Problem
iMessage echo suppression fails for DMs because the conversation is incorrectly classified as a "self-chat."
### Root cause
In iMessage's `chat.db`, the `sender` field for **outgoing** messages in a DM shows the **other party's handle**, not the local user's. For example, in a DM with `+14145550000`:
| F... | closed | completed | true | 1 | [] | [] | 2026-04-10T00:06:41Z | 2026-04-13T04:28:32Z | 2026-04-10T08:06:03Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | therobinbot | 264,698,461 | U_kgDOD8b6XQ | User | false |
openclaw/openclaw | 4,235,742,063 | I_kwDOQb6kR878eEtv | 63,992 | https://github.com/openclaw/openclaw/issues/63992 | https://api.github.com/repos/openclaw/openclaw/issues/63992 | [Bug]: session key write still hardcodes DEFAULT_AGENT_ID after PR #30654 (regression of #29683) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On 2026.4.8 when you try to set your `agents.list[]` and have one set to `default: true`, sessions are still stored/routed to `~/.openclaw/agents/main/sessions`. I also propose that we should have a `agents.list[].sessionsDi... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-10T00:46:24Z | 2026-04-30T04:32:58Z | 2026-04-27T22:09:03Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | hazrpg | 115,112 | MDQ6VXNlcjExNTExMg== | User | false |
openclaw/openclaw | 4,235,827,095 | I_kwDOQb6kR878eZeX | 63,999 | https://github.com/openclaw/openclaw/issues/63999 | https://api.github.com/repos/openclaw/openclaw/issues/63999 | Qwen model raw tool call XML (<tool_call>) leaks into visible chat messages (Web UI + Signal) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When using a Qwen 3.5 model (via MLX VLM server, `openai-completions` API), raw tool call XML is visible in both the web UI chat bubbles and Signal channel delivery.
The model outputs `<tool_call><function=rea... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T01:10:00Z | 2026-04-14T14:49:36Z | 2026-04-10T14:36:27Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jasonvassallo | 46,975,791 | MDQ6VXNlcjQ2OTc1Nzkx | User | false |
openclaw/openclaw | 4,235,851,819 | I_kwDOQb6kR878efgr | 64,003 | https://github.com/openclaw/openclaw/issues/64003 | https://api.github.com/repos/openclaw/openclaw/issues/64003 | [Bug]: memory-core plugin: Missing idempotencyKey property causes REM dreaming to fail | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Bug Description
The memory-core plugin's dreaming/REM feature throws an error when generating dream narratives because it is missing the `idempotencyKey` property.
## Environment
- OpenClaw version: 2026.4.9
- Plugin: me... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-04-10T01:15:51Z | 2026-04-15T04:19:57Z | 2026-04-12T16:46:18Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | visionzhang-prog | 266,700,237 | U_kgDOD-WFzQ | User | false |
openclaw/openclaw | 4,235,906,437 | I_kwDOQb6kR878es2F | 64,006 | https://github.com/openclaw/openclaw/issues/64006 | https://api.github.com/repos/openclaw/openclaw/issues/64006 | Cron announce delivery reports 'delivered' but messages never reach Telegram | ## Summary
Cron jobs with `delivery.mode: "announce"` and `delivery.channel: "telegram"` report `deliveryStatus: "delivered"` in run history, but the messages never actually arrive on Telegram. The Telegram bot API is functional (direct `curl` to `sendMessage` works), and the gateway health endpoint returns OK.
## En... | closed | completed | true | 3 | [] | [] | 2026-04-10T01:28:29Z | 2026-04-22T04:30:53Z | 2026-04-19T15:46:58Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | masonworld | 64,991,389 | MDQ6VXNlcjY0OTkxMzg5 | User | false |
openclaw/openclaw | 4,235,907,575 | I_kwDOQb6kR878etH3 | 64,007 | https://github.com/openclaw/openclaw/issues/64007 | https://api.github.com/repos/openclaw/openclaw/issues/64007 | QMD: memory-alt-main collection queried but never created | ## Problem
When memory search is triggered, QMD reports:
```
Collection not found: memory-alt-main
```
The search falls back to the built-in engine, but QMD itself never becomes functional.
## Root Cause
During QMD initialization, the system attempts to create a collection named `memory-alt-main`, but this fails be... | closed | completed | true | 4 | [] | [] | 2026-04-10T01:28:45Z | 2026-04-28T05:02:13Z | 2026-04-25T15:54:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fangchen489 | 267,722,258 | U_kgDOD_UeEg | User | false |
openclaw/openclaw | 4,235,909,125 | I_kwDOQb6kR878etgF | 64,008 | https://github.com/openclaw/openclaw/issues/64008 | https://api.github.com/repos/openclaw/openclaw/issues/64008 | [Feature Suggestion] Mimicking the Midjourney Experience: Standardizing the 'Satellite' Node Execution Model with Multi-Modal Local Workflows | ## Summary
As a creative user who leverages OpenClaw as a persistent creative assistant, I’m proposing a **'Satellite' or 'Node-Based' architecture** that separates the persistent OpenClaw daemon (Master Orchestrator) from specialized, heavy-duty execution environments (like ComfyUI).
This proposal aims to provide a *... | closed | completed | true | 2 | [] | [] | 2026-04-10T01:29:08Z | 2026-04-13T04:28:19Z | 2026-04-10T12:14:43Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | FloydianSound | 240,630,251 | U_kgDODle56w | User | false |
openclaw/openclaw | 4,235,943,727 | I_kwDOQb6kR878e18v | 64,011 | https://github.com/openclaw/openclaw/issues/64011 | https://api.github.com/repos/openclaw/openclaw/issues/64011 | [Bug]: Discord gateway crash exits OpenClaw on Windows with "Attempted to reconnect zombie connection after disconnecting first" | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
### Bug type
Crash
### Beta release blocker
No
### Summary
After upgrading from OpenClaw 2026.2.6 to 2026.4.9 on Windows, the gateway process starts, reaches `ready`, and then later exits with an uncaught exception in t... | closed | duplicate | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-10T01:37:41Z | 2026-04-27T04:40:21Z | 2026-04-25T02:28:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | talesrt | 68,339,009 | MDQ6VXNlcjY4MzM5MDA5 | User | false |
openclaw/openclaw | 4,235,996,821 | I_kwDOQb6kR878fC6V | 64,013 | https://github.com/openclaw/openclaw/issues/64013 | https://api.github.com/repos/openclaw/openclaw/issues/64013 | Closed: local /dreaming allowlist configuration issue | Closing this after further verification. In this environment, `/dreaming` was not on `plugins.allow`, so the earlier report overstated the root cause before allowlist validation was completed. Re-testing is in progress after correcting local config.
| closed | completed | true | 1 | [] | [] | 2026-04-10T01:49:33Z | 2026-04-12T04:21:12Z | 2026-04-10T01:53:07Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | smonett | 50,761,481 | MDQ6VXNlcjUwNzYxNDgx | User | false |
openclaw/openclaw | 4,236,014,847 | I_kwDOQb6kR878fHT_ | 64,014 | https://github.com/openclaw/openclaw/issues/64014 | https://api.github.com/repos/openclaw/openclaw/issues/64014 | openclaw update leaves user-systemd gateway unit on stale entrypoint after pnpm/global upgrades | I’m seeing a recurring issue on a **pnpm global install** with a **user-systemd** gateway service.
After `openclaw update`, the package upgrades successfully, but `openclaw doctor` still reports:
> Gateway service entrypoint does not match the current install
> `.../dist/entry.js -> .../dist/index.js`
The issue is... | closed | completed | true | 2 | [] | [] | 2026-04-10T01:53:19Z | 2026-04-16T04:24:31Z | 2026-04-13T15:47:25Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | bennyaph1211 | 250,654,541 | U_kgDODvCvTQ | User | false |
openclaw/openclaw | 4,236,037,377 | I_kwDOQb6kR878fM0B | 64,017 | https://github.com/openclaw/openclaw/issues/64017 | https://api.github.com/repos/openclaw/openclaw/issues/64017 | [Bug]: Discord /acp spawn --thread auto fails with "Session binding adapter failed to bind target conversation" | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
在 Discord 普通频道中执行：
/acp spawn claude --thread auto
会失败并返回：
Session binding adapter failed to bind target conversation
### Steps to reproduce
在 Discord 普通文本频道中使用 OpenClaw Bot
执行：
/acp spawn claude --thread auto
观察报错
### Expe... | closed | completed | true | 4 | [
"bug",
"bug:crash"
] | [] | 2026-04-10T01:58:14Z | 2026-04-25T04:16:36Z | 2026-04-22T21:08:09Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | wysl | 15,063,284 | MDQ6VXNlcjE1MDYzMjg0 | User | false |
openclaw/openclaw | 4,236,077,369 | I_kwDOQb6kR878fWk5 | 64,019 | https://github.com/openclaw/openclaw/issues/64019 | https://api.github.com/repos/openclaw/openclaw/issues/64019 | memory-core: dreaming narrative generation fails - missing idempotencyKey in subagent.run() | ## Bug Report
**Version:** OpenClaw 2026.4.9 (0512059)
**Plugin:** @openclaw/memory-core 2026.4.9
## Summary
The dreaming module's narrative generation silently fails with:
```
memory-core: narrative generation failed for light phase: invalid agent params: must have required property idempotencyKey
memory-core: narr... | closed | completed | true | 7 | [] | [] | 2026-04-10T02:10:10Z | 2026-04-15T04:18:58Z | 2026-04-12T16:46:21Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jiezfong-bot | 249,280,602 | U_kgDODtu4Wg | User | false |
openclaw/openclaw | 4,236,087,475 | I_kwDOQb6kR878fZCz | 64,021 | https://github.com/openclaw/openclaw/issues/64021 | https://api.github.com/repos/openclaw/openclaw/issues/64021 | iMessage Provider - Duplicate Message Delivery (every reply sent 2-4x) | ## Bug Description
Every reply sent via the iMessage channel is delivered multiple times to the recipient. The same message appears 1-4 times in the recipient's Messages.app chat.
## Frequency
High - reproduces on nearly every reply without exception.
## Log Evidence
Multiple `delivered reply` entries per outbound me... | closed | completed | true | 0 | [] | [] | 2026-04-10T02:13:32Z | 2026-04-12T04:21:10Z | 2026-04-10T02:45:08Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | zhoushoujianwork | 40,590,631 | MDQ6VXNlcjQwNTkwNjMx | User | false |
openclaw/openclaw | 4,236,108,096 | I_kwDOQb6kR878feFA | 64,026 | https://github.com/openclaw/openclaw/issues/64026 | https://api.github.com/repos/openclaw/openclaw/issues/64026 | [Bug]: llama.cpp、gemma4: 400 no body | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
llama.cpp、gemma4: 400 no body
### Steps to reproduce
llama.cpp、gemma4: 400 no body
### Expected behavior
llama.cpp、gemma4: 400 no body
### Actual behavior
llama.cpp、gemma4: 400 no body
### OpenClaw version
2026.4.9
... | open | null | false | 3 | [
"bug"
] | [] | 2026-04-10T02:20:27Z | 2026-05-03T01:37:24Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | huwei1024 | 463,801 | MDQ6VXNlcjQ2MzgwMQ== | User | false |
openclaw/openclaw | 4,236,197,099 | I_kwDOQb6kR878fzzr | 64,036 | https://github.com/openclaw/openclaw/issues/64036 | https://api.github.com/repos/openclaw/openclaw/issues/64036 | [Bug]: chunkTextByBreakResolver final chunk has trailing whitespace | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
## Summary
The `chunkTextByBreakResolver` function in `src/shared/text-chunking.ts` produces chunks with trailing whitespace in the **final chunk** (the one added after the loop exits). This is inconsistent w... | open | null | false | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T02:48:29Z | 2026-05-02T11:36:14Z | null | NONE | null | 20260502T120029Z | 2026-05-02T12:00:29Z | thanhtoantnt | 3,733,758 | MDQ6VXNlcjM3MzM3NTg= | User | false |
openclaw/openclaw | 4,236,231,018 | I_kwDOQb6kR878f8Fq | 64,043 | https://github.com/openclaw/openclaw/issues/64043 | https://api.github.com/repos/openclaw/openclaw/issues/64043 | Memory dreaming managed cron job deleted on gateway restart but never recreated |
## Bug 描述
**标题：** Memory dreaming managed cron job deleted on gateway restart but never recreated
**问题描述：**
memory-core 的 dreaming 功能通过 `gateway:startup` hook 自动创建 managed cron job。但每次 gateway 重启时，该 cron job 会被删除，随后未能重建。
## 复现步骤
1. 启用 dreaming 配置：
```json
{
"plugins": {
"entries": {
"memory-core": {
... | closed | completed | true | 4 | [] | [] | 2026-04-10T02:59:09Z | 2026-04-15T04:18:59Z | 2026-04-12T16:53:49Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | nicolas-hey | 258,840,845 | U_kgDOD22ZDQ | User | false |
openclaw/openclaw | 4,236,279,810 | I_kwDOQb6kR878gIAC | 64,048 | https://github.com/openclaw/openclaw/issues/64048 | https://api.github.com/repos/openclaw/openclaw/issues/64048 | [Bug]: open claw and ollama (qwen) | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
When You try using Qwen as the LLM in openclaw control UI it freeze. However Qwen works in the terminal without any issue
### Steps to reproduce
Chat in open Claw with chatgpt then toggle to qwen and it freeze
### Expected ... | closed | completed | true | 4 | [
"bug"
] | [] | 2026-04-10T03:14:06Z | 2026-04-28T05:02:07Z | 2026-04-25T15:55:39Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | michaelsmith134 | 59,780,968 | MDQ6VXNlcjU5NzgwOTY4 | User | false |
openclaw/openclaw | 4,236,281,741 | I_kwDOQb6kR878gIeN | 64,049 | https://github.com/openclaw/openclaw/issues/64049 | https://api.github.com/repos/openclaw/openclaw/issues/64049 | Subsystem loggers silently write to stale log file after midnight rollover | ## Description
Long-running processes (gateway, mac app) stop writing to the current day's log file after crossing midnight. Logs silently go to yesterday's file instead.
## Root Cause
`createSubsystemLogger()` caches its child file logger at module scope on first emit. When the daily rolling-log path changes at mid... | closed | not_planned | true | 6 | [] | [] | 2026-04-10T03:14:44Z | 2026-04-30T04:42:20Z | 2026-04-26T04:46:37Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BrilliantWang | 120,538,413 | U_kgDOBy9FLQ | User | false |
openclaw/openclaw | 4,236,322,009 | I_kwDOQb6kR878gSTZ | 64,058 | https://github.com/openclaw/openclaw/issues/64058 | https://api.github.com/repos/openclaw/openclaw/issues/64058 | fv-publish-dispatch can write successfully then fail index with 404 due to workspace/vault path mismatch | ## Summary
`fv-publish-dispatch --json` can fail during `indexed` with `404 File not found in vault` even though the note was already written successfully. In the same flow, `fv-note-write` writes to the workspace path, while `fidelisverse-index-note`/indexing resolves against the Obsidian vault path, so the pipeline ... | closed | completed | true | 0 | [] | [] | 2026-04-10T03:26:55Z | 2026-04-12T04:21:08Z | 2026-04-10T03:28:08Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | FidelisClayton | 9,627,967 | MDQ6VXNlcjk2Mjc5Njc= | User | false |
openclaw/openclaw | 4,236,437,169 | I_kwDOQb6kR878guax | 64,067 | https://github.com/openclaw/openclaw/issues/64067 | https://api.github.com/repos/openclaw/openclaw/issues/64067 | [Bug]: completion-cache update fails with "qa scenario pack not found: qa/scenarios/index.md" | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Completion cache update fails with error "qa scenario pack not found: qa/scenarios/index.md" whenever openclaw CLI is invoked (e.g. openclaw --update, openclaw gateway restart).
### Steps to reproduce
1. Run `openclaw --up... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-10T03:58:15Z | 2026-04-28T05:02:02Z | 2026-04-25T15:56:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jisi-assist | 262,780,846 | U_kgDOD6m3rg | User | false |
openclaw/openclaw | 4,236,453,443 | I_kwDOQb6kR878gyZD | 64,068 | https://github.com/openclaw/openclaw/issues/64068 | https://api.github.com/repos/openclaw/openclaw/issues/64068 | [Bug]: memory-core dreaming promotion always finds 0 candidates - recallCount never accumulates (normalized recall artifacts resets scores) | ## Bug Description
**Summary**: The short-term dreaming promotion step consistently reports `candidates=0, applied=0, failed=0` across all workspaces. Inspection of `.dreams/short-term-recall.json` shows 1323 entries where **every entry has `recallCount: 0`** and `totalScore` never exceeds 0.62 — both below the promot... | closed | completed | true | 11 | [] | [] | 2026-04-10T04:02:07Z | 2026-04-15T04:18:57Z | 2026-04-12T17:15:18Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ymzhang10 | 16,602,361 | MDQ6VXNlcjE2NjAyMzYx | User | false |
openclaw/openclaw | 4,236,485,430 | I_kwDOQb6kR878g6M2 | 64,069 | https://github.com/openclaw/openclaw/issues/64069 | https://api.github.com/repos/openclaw/openclaw/issues/64069 | Control UI switch to openai/gpt-5.1-codex fails with 400 due to reasoning effort mismatch on v2026.4.9 | ## Summary
Switching a Control UI session to `openai/gpt-5.1-codex` fails with HTTP 400 because OpenClaw appears to send an unsupported reasoning effort (`none` / `low`) and the retry path does not actually correct the outgoing value.
This reproduces on current OpenClaw (`2026.4.9`) and affects the non-mini `gpt-5.1-... | closed | completed | true | 2 | [] | [] | 2026-04-10T04:08:56Z | 2026-04-24T04:26:31Z | 2026-04-21T06:37:50Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | sixamo | 16,271,924 | MDQ6VXNlcjE2MjcxOTI0 | User | false |
openclaw/openclaw | 4,236,499,611 | I_kwDOQb6kR878g9qb | 64,070 | https://github.com/openclaw/openclaw/issues/64070 | https://api.github.com/repos/openclaw/openclaw/issues/64070 | [Bug]: doctor reports @discordjs/opus as missing bundled runtime dep on ARM64, but doctor --fix cannot recover | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After updating from 2026.4.5 to the latest available npm release (2026.4.9), `openclaw doctor` reported `@discordjs/opus` as a missing bundled plugin runtime dependency on ARM64, and the suggested recovery path could not com... | closed | completed | true | 5 | [
"bug",
"regression"
] | [] | 2026-04-10T04:12:05Z | 2026-04-23T04:22:29Z | 2026-04-20T14:37:49Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | KRouhana | 37,153,041 | MDQ6VXNlcjM3MTUzMDQx | User | false |
openclaw/openclaw | 4,236,683,969 | I_kwDOQb6kR878hqrB | 64,090 | https://github.com/openclaw/openclaw/issues/64090 | https://api.github.com/repos/openclaw/openclaw/issues/64090 | [Feature Request]: systemPromptAddition memory hook support on cli-runner path (alongside systemPromptOverride) | ## Problem
When using `systemPromptOverride` with the `claude-cli` backend, dynamic memory injection is completely disabled. The current code path in `cli-runner/prepare.ts` does:
```
resolveSystemPromptOverride() ?? buildSystemPrompt()
```
This means override wins completely — no dynamic additions possible on top.
... | closed | completed | true | 1 | [] | [] | 2026-04-10T04:59:22Z | 2026-04-13T04:27:41Z | 2026-04-10T23:48:21Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | derricksy | 50,358,795 | MDQ6VXNlcjUwMzU4Nzk1 | User | false |
openclaw/openclaw | 4,236,702,508 | I_kwDOQb6kR878hvMs | 64,092 | https://github.com/openclaw/openclaw/issues/64092 | https://api.github.com/repos/openclaw/openclaw/issues/64092 | [Bug]: openai-codex error classification: actual 403/rate-limit shown as 'DNS lookup failed' | Raised on behalf of Chad J.
## Bug Description
When openai-codex requests fail with a 403 HTTP response (HTML body), OpenClaw displays one of several misleading user-facing errors instead of correctly reporting a 403 auth failure. The raw response is visible in logs as rawErrorPreview but the user-facing message is w... | closed | completed | true | 4 | [] | [] | 2026-04-10T05:03:42Z | 2026-04-28T05:01:56Z | 2026-04-25T15:57:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | itsclawdettebot-debug | 260,100,625 | U_kgDOD4DSEQ | User | false |
openclaw/openclaw | 4,236,786,030 | I_kwDOQb6kR878iDlu | 64,105 | https://github.com/openclaw/openclaw/issues/64105 | https://api.github.com/repos/openclaw/openclaw/issues/64105 | BlueBubbles attachment downloads silently fail on Node.js 22 (undici dispatcher mismatch) | ## Summary
Inbound BlueBubbles image attachments silently fail to download. No files appear in `~/.openclaw/media/inbound/`. The agent receives the message text/placeholder but cannot see the image.
## Environment
- OpenClaw **2026.4.9**
- Node.js **22.22.2**
- macOS **26.4.0**
- BlueBubbles server on `http://127.0.... | closed | completed | true | 1 | [] | [] | 2026-04-10T05:22:19Z | 2026-04-19T04:23:08Z | 2026-04-16T17:04:22Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | mpalermiti | 253,352,132 | U_kgDODxnYxA | User | false |
openclaw/openclaw | 4,236,810,166 | I_kwDOQb6kR878iJe2 | 64,111 | https://github.com/openclaw/openclaw/issues/64111 | https://api.github.com/repos/openclaw/openclaw/issues/64111 | [Bug]: Ollama vision models not accessible via `image` tool | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
image tool cannot use Ollama vision models despite correct configuration and confirmed vision capability
### Steps to reproduce
1. Configure Ollama provider with a vision-capable model (e.g., gemma4:e4b) in openclaw.json
2... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-04-10T05:27:15Z | 2026-04-30T04:42:19Z | 2026-04-26T04:46:42Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | SilentNobody-KS | 221,793,220 | U_kgDODThLxA | User | false |
openclaw/openclaw | 4,236,879,295 | I_kwDOQb6kR878iaW_ | 64,117 | https://github.com/openclaw/openclaw/issues/64117 | https://api.github.com/repos/openclaw/openclaw/issues/64117 | [Bug]: voice-call tool/CLI always hits EADDRINUSE | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Bug: voice-call tool/CLI always hits EADDRINUSE
**Version:** OpenClaw 2026.4.9 (0512059)
**OS:** WSL2 (Ubuntu), Linux 6.6.87.2-microsoft-standard-WSL2
**Provider:** Twilio
**Plugin:** stock voice-call (loaded, enabled)
... | closed | completed | true | 5 | [
"bug"
] | [] | 2026-04-10T05:42:38Z | 2026-04-27T04:41:08Z | 2026-04-25T02:12:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ghtechs | 245,203,900 | U_kgDODp2DvA | User | false |
openclaw/openclaw | 4,236,946,380 | I_kwDOQb6kR878iqvM | 64,131 | https://github.com/openclaw/openclaw/issues/64131 | https://api.github.com/repos/openclaw/openclaw/issues/64131 | [Bug]: webui: dreaming Page display abnormality | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Exceptions when upgrading from 4.8 to 4.9
### Steps to reproduce
open webui
### Expected behavior
Display Normal
### Actual behavior
Show exception
### OpenClaw version
2026.4.9
### Operating system
Mac os 26.4
#... | closed | completed | true | 5 | [
"bug"
] | [] | 2026-04-10T05:57:38Z | 2026-04-30T04:38:04Z | 2026-04-26T06:52:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | YuZiHao666 | 34,652,152 | MDQ6VXNlcjM0NjUyMTUy | User | false |
openclaw/openclaw | 4,237,081,136 | I_kwDOQb6kR878jLow | 64,137 | https://github.com/openclaw/openclaw/issues/64137 | https://api.github.com/repos/openclaw/openclaw/issues/64137 | cron: TypeError: runningAtMs crash on gateway restart (race condition) | ## Summary
Cron scheduler crashes with `TypeError: runningAtMs` when the gateway restarts while a cron job is in the `running` state. Observed twice on 2026-04-09 (at 11:56 and 12:26 MSK), causing missed job runs.
## Steps to Reproduce
1. Have a cron job currently executing (`runningAtMs` is set in job state)
2. Res... | closed | completed | true | 2 | [] | [] | 2026-04-10T06:26:53Z | 2026-04-28T05:28:46Z | 2026-04-25T09:39:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | artdaal | 35,874,009 | MDQ6VXNlcjM1ODc0MDA5 | User | false |
openclaw/openclaw | 4,237,083,430 | I_kwDOQb6kR878jMMm | 64,138 | https://github.com/openclaw/openclaw/issues/64138 | https://api.github.com/repos/openclaw/openclaw/issues/64138 | All plugin skills fail to load: dist-runtime symlinks break realpath() containment check | ## Summary
All plugin skills from enabled extensions (feishu, qqbot, tavily, wecom, etc.) fail to load at runtime. The warning `"Skipping skill path that resolves outside its configured root."` is emitted for every plugin skill during session initialization.
## Affected Skills (23 total)
- **feishu (4):** feishu-doc... | closed | completed | true | 0 | [
"bug",
"maintainer"
] | [] | 2026-04-10T06:27:21Z | 2026-04-13T04:28:34Z | 2026-04-10T07:41:29Z | MEMBER | null | 20260414T092630Z | 2026-04-14T09:26:30Z | sliverp | 38,134,380 | MDQ6VXNlcjM4MTM0Mzgw | User | false |
openclaw/openclaw | 4,237,117,311 | I_kwDOQb6kR878jUd_ | 64,139 | https://github.com/openclaw/openclaw/issues/64139 | https://api.github.com/repos/openclaw/openclaw/issues/64139 | [Bug] exec crontab fails with SIGKILL in 2026.4.9 | ## Problem
After upgrading to OpenClaw 2026.4.9, exec commands for crontab timing out or getting SIGKilled immediately.
## Environment
- OpenClaw version: 2026.4.9 (0512059)
- macOS: 26.4 (arm64)
- Node: 24.14.1
- Gateway: running via LaunchAgent
## Steps to Reproduce
1. Try to set a crontab job via exec:
```bas... | open | null | false | 3 | [] | [] | 2026-04-10T06:35:05Z | 2026-05-03T08:39:13Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | guojiongming | 266,004,757 | U_kgDOD9rpFQ | User | false |
openclaw/openclaw | 4,237,135,288 | I_kwDOQb6kR878jY24 | 64,141 | https://github.com/openclaw/openclaw/issues/64141 | https://api.github.com/repos/openclaw/openclaw/issues/64141 | Remove Session Summary functionality | ## Request
Remove the "Session Summary" feature entirely from the codebase.
## Rationale
The Session Summary feature is no longer needed and its removal will:
- Reduce maintenance surface area
- Eliminate associated background jobs / cron logic
- Clean up any related configuration, storage, and UI components
## Sco... | closed | completed | true | 1 | [] | [] | 2026-04-10T06:38:53Z | 2026-04-13T04:28:37Z | 2026-04-10T06:41:46Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | atticusfawler | 268,220,059 | U_kgDOD_y2mw | User | false |
openclaw/openclaw | 4,237,139,373 | I_kwDOQb6kR878jZ2t | 64,142 | https://github.com/openclaw/openclaw/issues/64142 | https://api.github.com/repos/openclaw/openclaw/issues/64142 | [Bug]: Per-message model name not displayed in chat bubbles | ## Problem
Message bubbles in chat (Web UI, Telegram, Signal, etc.) display only the session's **default model** name — not the specific model that actually generated each individual message.
This means when a session uses model routing (e.g. different models for main vs sub-agents), the user sees no indication of wh... | closed | completed | true | 1 | [] | [] | 2026-04-10T06:39:50Z | 2026-04-13T04:28:36Z | 2026-04-10T06:41:48Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | atticusfawler | 268,220,059 | U_kgDOD_y2mw | User | false |
openclaw/openclaw | 4,237,139,689 | I_kwDOQb6kR878jZ7p | 64,143 | https://github.com/openclaw/openclaw/issues/64143 | https://api.github.com/repos/openclaw/openclaw/issues/64143 | [Feature]: Settings → Models: OpenRouter price registry with overrides | ## Problem
The cost displayed in session/message bubbles is often `$0.00` or missing entirely for many models — including popular ones like MiniMax, Kimi, Qwen, Gemini Flash, etc.
The root cause is that OpenRouter model prices are not fetched or maintained centrally. Without accurate per-model pricing, cost computati... | closed | completed | true | 1 | [] | [] | 2026-04-10T06:39:55Z | 2026-04-13T04:28:35Z | 2026-04-10T06:41:50Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | atticusfawler | 268,220,059 | U_kgDOD_y2mw | User | false |
openclaw/openclaw | 4,237,178,845 | I_kwDOQb6kR878jjfd | 64,149 | https://github.com/openclaw/openclaw/issues/64149 | https://api.github.com/repos/openclaw/openclaw/issues/64149 | Telegram channel: Intermittent message delivery failures after session restart | ## Bug Description
Messages sent from assistant to user via Telegram channel experienced intermittent delivery failures starting after a session restart. User could send messages TO assistant successfully, but assistant replies were inconsistently delivered.
## Timeline
- **2026-04-10 05:38 GMT+1**: Session restart ... | closed | completed | true | 5 | [] | [] | 2026-04-10T06:47:56Z | 2026-04-26T04:30:28Z | 2026-04-23T06:15:55Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | dylanpassey-hub | 263,126,731 | U_kgDOD67-yw | User | false |
openclaw/openclaw | 4,237,194,938 | I_kwDOQb6kR878jna6 | 64,151 | https://github.com/openclaw/openclaw/issues/64151 | https://api.github.com/repos/openclaw/openclaw/issues/64151 | Feature: write-time tool result externalization to prevent session JSONL bloat | ## Summary
Session JSONL files grow unbounded when agents make many tool calls with large outputs (exec, read, SSH). In production we hit **93MB / 55K lines** in a single session within 24 hours, pegging gateway CPU at 126% and requiring manual intervention.
LCM (lossless-claw) manages context assembly well — the mod... | closed | completed | true | 2 | [] | [] | 2026-04-10T06:50:51Z | 2026-04-28T05:01:51Z | 2026-04-25T15:58:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | liu51115 | 31,350,941 | MDQ6VXNlcjMxMzUwOTQx | User | false |
openclaw/openclaw | 4,237,212,413 | I_kwDOQb6kR878jrr9 | 64,153 | https://github.com/openclaw/openclaw/issues/64153 | https://api.github.com/repos/openclaw/openclaw/issues/64153 | [Bug]: TTS tool ignores edge.voice config | ## Bug Description
The `tts` tool generates audio using a hardcoded default voice regardless of `messages.tts.providers.edge.voice` setting. Config changes to edge.voice have no audible effect.
## Environment
- OpenClaw version: 2026.4.9
- OS: macOS
- Channels affected: Matrix, Discord
## Config Applied
```
messag... | closed | completed | true | 4 | [] | [] | 2026-04-10T06:54:05Z | 2026-04-27T04:38:08Z | 2026-04-25T03:02:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ion-ctrl | 266,816,281 | U_kgDOD-dLGQ | User | false |
openclaw/openclaw | 4,237,316,991 | I_kwDOQb6kR878kFN_ | 64,158 | https://github.com/openclaw/openclaw/issues/64158 | https://api.github.com/repos/openclaw/openclaw/issues/64158 | [Bug]: Implicit latest-device approval can pair the wrong requester | ## Severity Assessment
### CVSS Assessment
| Metric | v3.1 | v4.0 |
|--------|------|------|
| **Score** | 9.0 / 10.0 | 9.4 / 10.0 |
| **Severity** | Critical | Critical |
| **Vector** | CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:C/C:H/I:H/A:H | CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:P/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H |
| **Calculator** |... | closed | completed | true | 0 | [] | [] | 2026-04-10T07:14:28Z | 2026-04-13T04:27:36Z | 2026-04-11T01:55:03Z | CONTRIBUTOR | null | 20260414T092630Z | 2026-04-14T09:26:30Z | coygeek | 65,363,919 | MDQ6VXNlcjY1MzYzOTE5 | User | false |
openclaw/openclaw | 4,237,317,047 | I_kwDOQb6kR878kFO3 | 64,159 | https://github.com/openclaw/openclaw/issues/64159 | https://api.github.com/repos/openclaw/openclaw/issues/64159 | [Bug]: Provider-supplied OAuth URLs inject Windows cmd.exe via openUrl | ## Severity Assessment
### CVSS Assessment
| Metric | v3.1 | v4.0 |
|--------|------|------|
| **Score** | 8.3 / 10.0 | 9.0 / 10.0 |
| **Severity** | High | Critical |
| **Vector** | CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:H/I:H/A:H | CVSS:4.0/AV:N/AC:H/AT:N/PR:N/UI:P/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H |
| **Calculator** | [CV... | closed | completed | true | 0 | [] | [
"vincentkoc"
] | 2026-04-10T07:14:28Z | 2026-04-15T04:20:08Z | 2026-04-12T15:42:26Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | coygeek | 65,363,919 | MDQ6VXNlcjY1MzYzOTE5 | User | false |
openclaw/openclaw | 4,237,326,662 | I_kwDOQb6kR878kHlG | 64,162 | https://github.com/openclaw/openclaw/issues/64162 | https://api.github.com/repos/openclaw/openclaw/issues/64162 | [Bug]: Gateway auth rate limiting is opt-in on non-loopback shared-secret deployments | ## Severity Assessment
### CVSS Assessment
| Metric | v3.1 | v4.0 |
|--------|------|------|
| **Score** | 3.7 / 10.0 | 6.3 / 10.0 |
| **Severity** | Low | Medium |
| **Vector** | CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:L/A:N | CVSS:4.0/AV:N/AC:H/AT:N/PR:N/UI:N/VC:N/VI:L/VA:N/SC:N/SI:N/SA:N |
| **Calculator** | [CVSS ... | closed | not_planned | true | 0 | [] | [] | 2026-04-10T07:16:18Z | 2026-04-13T04:28:04Z | 2026-04-10T14:44:15Z | CONTRIBUTOR | null | 20260414T092630Z | 2026-04-14T09:26:30Z | coygeek | 65,363,919 | MDQ6VXNlcjY1MzYzOTE5 | User | false |
openclaw/openclaw | 4,237,326,681 | I_kwDOQb6kR878kHlZ | 64,163 | https://github.com/openclaw/openclaw/issues/64163 | https://api.github.com/repos/openclaw/openclaw/issues/64163 | [Bug]: Control UI stores reusable device-auth secrets in browser localStorage | ## Severity Assessment
### CVSS Assessment
| Metric | v3.1 | v4.0 |
|--------|------|------|
| **Score** | 8.0 / 10.0 | 8.8 / 10.0 |
| **Severity** | High | High |
| **Vector** | CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:H/I:H/A:N | CVSS:4.0/AV:N/AC:H/AT:N/PR:N/UI:P/VC:H/VI:H/VA:N/SC:H/SI:H/SA:N |
| **Calculator** | [CVSS v... | closed | not_planned | true | 0 | [] | [] | 2026-04-10T07:16:18Z | 2026-04-13T04:27:55Z | 2026-04-10T17:41:42Z | CONTRIBUTOR | null | 20260414T092630Z | 2026-04-14T09:26:30Z | coygeek | 65,363,919 | MDQ6VXNlcjY1MzYzOTE5 | User | false |
openclaw/openclaw | 4,237,393,848 | I_kwDOQb6kR878kX-4 | 64,170 | https://github.com/openclaw/openclaw/issues/64170 | https://api.github.com/repos/openclaw/openclaw/issues/64170 | plugins: normalizePluginsConfig drops slots.contextEngine, causing false "plugins.allow is empty" warning | ## Summary
`normalizePluginsConfigWithResolver()` preserves `plugins.slots.memory` but silently drops `plugins.slots.contextEngine` from the normalized plugin config. That causes slot-selected context-engine plugins to lose explicit-selection credit during loader diagnostics.
One visible symptom is a false warning fro... | closed | completed | true | 2 | [] | [] | 2026-04-10T07:29:43Z | 2026-04-14T14:49:43Z | 2026-04-10T22:58:29Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,237,445,510 | I_kwDOQb6kR878kkmG | 64,174 | https://github.com/openclaw/openclaw/issues/64174 | https://api.github.com/repos/openclaw/openclaw/issues/64174 | [Bug]: openai-codex OAuth runtime fails on 2026.4.9 with 403 HTML; 2026.3.28 works | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On the same Ubuntu VPS and openai-codex OAuth setup, OpenClaw 2026.4.9 fails to return runtime chat replies in Control UI, WhatsApp, and WeCom, while 2026.3.28 works.
### Steps to reproduce
1. Start the same self-hosted Ub... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-10T07:38:25Z | 2026-04-14T04:20:25Z | 2026-04-11T14:29:19Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | FWo100 | 158,056,709 | U_kgDOCWvBBQ | User | false |
openclaw/openclaw | 4,237,521,272 | I_kwDOQb6kR878k3F4 | 64,180 | https://github.com/openclaw/openclaw/issues/64180 | https://api.github.com/repos/openclaw/openclaw/issues/64180 | fix(agents): context overflow not detected for llama.cpp server provider | ## Bug Description
`isContextOverflowError()` and `matchesProviderContextOverflow()` in `src/agents/pi-embedded-helpers/` detect context window overflow for major providers (Anthropic, OpenAI, Bedrock, Ollama, Cohere, Vertex), but the **llama.cpp HTTP server** (used directly or via vLLM-style local deployments behind ... | closed | completed | true | 1 | [] | [] | 2026-04-10T07:50:12Z | 2026-04-14T14:49:42Z | 2026-04-10T12:30:35Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | alexander-applyinnovations | 111,744,509 | U_kgDOBqkV_Q | User | false |
openclaw/openclaw | 4,237,583,554 | I_kwDOQb6kR878lGTC | 64,186 | https://github.com/openclaw/openclaw/issues/64186 | https://api.github.com/repos/openclaw/openclaw/issues/64186 | [Bug Again]: OpenClaw tries to use local Ollama CLI instead of remote API endpoint | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Problem of OpenClaw tries to use local Ollama CLI instead of remote API endpoint should have been fix based from this [commit 36afd1b](https://github.com/openclaw/openclaw/commit/36afd1b2b0ae38e90b0b05e41506ed01aaac285a) fro... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-10T08:01:21Z | 2026-04-28T05:01:48Z | 2026-04-25T15:58:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | hanifsyuaib | 89,407,284 | MDQ6VXNlcjg5NDA3Mjg0 | User | false |
openclaw/openclaw | 4,237,587,880 | I_kwDOQb6kR878lHWo | 64,188 | https://github.com/openclaw/openclaw/issues/64188 | https://api.github.com/repos/openclaw/openclaw/issues/64188 | [Bug]: Github Copilot add model failed on openclaw configure | ### Bug type
Regression (previously worked, now fails)
### Beta release blocker
No
### Summary
When using the `github-copilot` provider, running `openclaw configure` to add or modify models causes a configuration conflict. The GitHub plugin appears to mutate the OpenClaw config during the process, which prevents s... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-10T08:02:14Z | 2026-04-18T04:14:15Z | 2026-04-15T10:03:11Z | NONE | null | 20260418T042035Z | 2026-04-18T04:20:35Z | Alfzzz124 | 200,902,261 | U_kgDOC_mGdQ | User | false |
openclaw/openclaw | 4,237,676,263 | I_kwDOQb6kR878lc7n | 64,191 | https://github.com/openclaw/openclaw/issues/64191 | https://api.github.com/repos/openclaw/openclaw/issues/64191 | dreaming CLI: plugins.allow false-positive plugin not found warning | ### Bug: dreaming CLI - plugins.allow false-positive warning
**问题**：`openclaw dreaming [status|on|off]` CLI 命令需要在 plugins.allow 里加 "dreaming"，但加了之后反而报 "plugin not found" 被当作无效配置忽略。
**配置**：
```json
"plugins": {
"allow": ["dreaming"],
"entries": {
"memory-core": {
"config": {
"dreaming": {
... | closed | completed | true | 0 | [] | [] | 2026-04-10T08:18:43Z | 2026-04-13T04:28:06Z | 2026-04-10T13:56:14Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | Wanxian-Liu | 263,423,852 | U_kgDOD7OHbA | User | false |
openclaw/openclaw | 4,237,768,907 | I_kwDOQb6kR878lzjL | 64,194 | https://github.com/openclaw/openclaw/issues/64194 | https://api.github.com/repos/openclaw/openclaw/issues/64194 | [Bug]: WhatsApp/Telegram config lost on restart after upgrade to 2026.4.x | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
After upgrading to 2026.4.x, WhatsApp and Telegram channel configurations are being lost when the gateway restarts or when running 'openclaw doctor --fix'.
Steps to reproduce:
1. Configure WhatsApp and Telegram channels in ope... | closed | completed | true | 3 | [
"bug",
"bug:crash"
] | [] | 2026-04-10T08:31:33Z | 2026-04-28T05:01:47Z | 2026-04-25T15:59:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wzkinj-git | 269,459,655 | U_kgDOEA-gxw | User | false |
openclaw/openclaw | 4,237,857,071 | I_kwDOQb6kR878mJEv | 64,202 | https://github.com/openclaw/openclaw/issues/64202 | https://api.github.com/repos/openclaw/openclaw/issues/64202 | [Bug]: Dreams Scene CSS layout causes content clipping/overlap in Control UI | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
In the Dreams Scene page of Control UI (v2026.4.9), the content is clipped and appears overlapping due to CSS layout issues.
### Steps to reproduce
1. Open Control UI at `/control/dreaming`
2. Click on "Sce... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T08:45:32Z | 2026-04-28T05:01:45Z | 2026-04-25T16:00:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xuysang | 20,820,186 | MDQ6VXNlcjIwODIwMTg2 | User | false |
openclaw/openclaw | 4,237,940,257 | I_kwDOQb6kR878mdYh | 64,208 | https://github.com/openclaw/openclaw/issues/64208 | https://api.github.com/repos/openclaw/openclaw/issues/64208 | WhatsApp DM still requires /approve for exec despite tools.exec.ask=off and gateway approvals ask=off | ## Summary
On OpenClaw `2026.4.5`, `exec` calls from a trusted WhatsApp DM still trigger interactive approval prompts (`/approve ... allow-once`) even though:
- `tools.exec.ask = "off"`
- `tools.exec.security = "allowlist"`
- `openclaw approvals get --gateway` shows effective policy:
- `security=allowlist`
- `ask... | closed | completed | true | 1 | [] | [] | 2026-04-10T09:01:06Z | 2026-04-13T04:28:20Z | 2026-04-10T12:01:47Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | nicomiachon | 48,127,448 | MDQ6VXNlcjQ4MTI3NDQ4 | User | false |
openclaw/openclaw | 4,237,948,771 | I_kwDOQb6kR878mfdj | 64,210 | https://github.com/openclaw/openclaw/issues/64210 | https://api.github.com/repos/openclaw/openclaw/issues/64210 | Cron agentTurn to Telegram topic runs but never posts, delivery false/none confusion | ## Summary
Cron-based health reminders targeting a Telegram forum topic are executing, but the user receives no visible messages in the topic.
## Environment
- OpenClaw version: 2026.4.9 (also observed on 2026.4.8)
- Surface: Telegram group forum topic
- Session key: `agent:main:telegram:group:-1003383566804:topic:2`
... | closed | completed | true | 1 | [] | [] | 2026-04-10T09:02:55Z | 2026-04-13T04:28:26Z | 2026-04-10T09:19:07Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | properbrand | 258,450,886 | U_kgDOD2elxg | User | false |
openclaw/openclaw | 4,238,102,739 | I_kwDOQb6kR878nFDT | 64,216 | https://github.com/openclaw/openclaw/issues/64216 | https://api.github.com/repos/openclaw/openclaw/issues/64216 | CLI browser commands hang 30-60s after output (WSL2/Linux, not just Windows) | ## Bug Description
CLI commands for browser management complete successfully but hang for 30-60 seconds after outputting results before exiting. This affects WSL2/Linux in addition to the Windows issue #63227.
## Environment
- **OpenClaw:** 2026.4.9
- **OS:** Linux 6.6.87.2-microsoft-standard-WSL2 (Ubuntu)
- **Node.... | closed | completed | true | 1 | [] | [] | 2026-04-10T09:31:06Z | 2026-04-30T04:42:15Z | 2026-04-26T04:48:32Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Crabsticksalad | 262,156,320 | U_kgDOD6AwIA | User | false |
openclaw/openclaw | 4,238,111,816 | I_kwDOQb6kR878nHRI | 64,217 | https://github.com/openclaw/openclaw/issues/64217 | https://api.github.com/repos/openclaw/openclaw/issues/64217 | [Bug]: Source-dependent instructions/tools drift invalidates OpenAI prompt cache within the same session | ## Summary
For the **same target session** and the **same `prompt_cache_key`**, OpenClaw can assemble different **`instructions`** and different **tool sets** depending on the dispatch source (direct user turn vs inter-session/cron turn).
On the OpenAI / Codex Responses path, that means the prompt prefix is no longer... | open | null | false | 1 | [] | [] | 2026-04-10T09:32:39Z | 2026-05-03T08:39:16Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | pxzhang16 | 147,101,862 | U_kgDOCMSYpg | User | false |
openclaw/openclaw | 4,238,174,917 | I_kwDOQb6kR878nWrF | 64,220 | https://github.com/openclaw/openclaw/issues/64220 | https://api.github.com/repos/openclaw/openclaw/issues/64220 | [Bug]: Control UI can leave "Compacting context..." stuck even when no compaction is running | ## Summary
Control UI can keep showing `Compacting context...` **even when no compaction is actually running**.
This appears to be a **frontend stale-state bug**, not necessarily a backend compaction bug.
## Evidence
In the observed case:
- the UI kept showing `Compacting context...`
- the reply had already fully r... | open | null | false | 1 | [] | [] | 2026-04-10T09:41:27Z | 2026-05-03T10:41:02Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | pxzhang16 | 147,101,862 | U_kgDOCMSYpg | User | false |
openclaw/openclaw | 4,238,244,856 | I_kwDOQb6kR878nnv4 | 64,224 | https://github.com/openclaw/openclaw/issues/64224 | https://api.github.com/repos/openclaw/openclaw/issues/64224 | Billing cooldown flags entire provider instead of individual model — breaks proxy/litellm setups | ## Summary
When using LiteLLM as a single OpenClaw provider that aggregates multiple upstreams (NanoGPT, Z.ai, DeepSeek, etc.), a 402 billing error from **one upstream model** causes OpenClaw to disable the **entire provider** via `billingBackoffHours`. Since all models share the same auth profile, every model behind ... | open | null | false | 1 | [] | [] | 2026-04-10T09:50:08Z | 2026-05-03T10:19:13Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | hugalafutro | 30,209,689 | MDQ6VXNlcjMwMjA5Njg5 | User | false |
openclaw/openclaw | 4,238,267,031 | I_kwDOQb6kR878ntKX | 64,226 | https://github.com/openclaw/openclaw/issues/64226 | https://api.github.com/repos/openclaw/openclaw/issues/64226 | WhatsApp media send silently fails in 2026.4.5+ | ## Summary
WhatsApp outbound media messages (images sent via `openclaw message send --media` and agent auto-reply with media) silently fail starting in 2026.4.5. The gateway API returns success (`✓ send`) but the WhatsApp outbound module never processes the message — no `Sending message` / `Sent message` logs appear, ... | closed | completed | true | 1 | [] | [] | 2026-04-10T09:53:22Z | 2026-04-28T05:01:41Z | 2026-04-25T16:01:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | christianweinmayr | 797,606 | MDQ6VXNlcjc5NzYwNg== | User | false |
openclaw/openclaw | 4,238,277,329 | I_kwDOQb6kR878nvrR | 64,227 | https://github.com/openclaw/openclaw/issues/64227 | https://api.github.com/repos/openclaw/openclaw/issues/64227 | GPT-5.4 / Codex agentic runtime parity in OpenClaw | # GPT-5.4 / Codex Agentic Parity
Make GPT-5.4 agents work as well as Claude agents in OpenClaw: follow instructions, use tools proactively, execute multi-step tasks without stopping to ask, and stop over-confirming.
---
## How the three open PRs work together
```
#65219 (runtime rollup) → strict-agentic auto-ac... | closed | completed | true | 27 | [] | [] | 2026-04-10T09:54:53Z | 2026-04-15T04:19:15Z | 2026-04-13T04:03:24Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,277,473 | I_kwDOQb6kR878nvth | 64,228 | https://github.com/openclaw/openclaw/issues/64228 | https://api.github.com/repos/openclaw/openclaw/issues/64228 | agents: add strict-agentic execution contract and revise update_plan semantics | Parent: #64227
Refs #62854
Refs #47213
## Summary
Add an opt-in embedded Pi execution contract that keeps GPT-5-style runs acting through plan-only turns, and revise `update_plan` so it behaves like a structured progress tool instead of a chatty completion surface.
PR 1 is intentionally GPT-5-first: it only activate... | closed | completed | true | 0 | [] | [] | 2026-04-10T09:54:55Z | 2026-04-13T04:27:44Z | 2026-04-10T21:56:39Z | CONTRIBUTOR | null | 20260414T092630Z | 2026-04-14T09:26:30Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,277,595 | I_kwDOQb6kR878nvvb | 64,229 | https://github.com/openclaw/openclaw/issues/64229 | https://api.github.com/repos/openclaw/openclaw/issues/64229 | openai-codex: fix auth scope handling and classify provider/runtime failures | Parent: #64227
## Summary
Fix maintained-source OpenAI Codex auth scope handling and add explicit provider/runtime failure classification so GPT-5.4 failures stop collapsing into generic model-stop behavior.
## Scope
- maintained-source Codex OAuth scope fix
- typed failure classification for:
- auth_scope
- au... | closed | completed | true | 0 | [] | [] | 2026-04-10T09:54:56Z | 2026-04-13T04:27:40Z | 2026-04-11T00:19:34Z | CONTRIBUTOR | null | 20260414T092630Z | 2026-04-14T09:26:30Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,277,772 | I_kwDOQb6kR878nvyM | 64,230 | https://github.com/openclaw/openclaw/issues/64230 | https://api.github.com/repos/openclaw/openclaw/issues/64230 | agents: add OpenAI/Codex provider-owned tool compatibility layer | Parent: #64227
## Summary
Add a provider-owned OpenAI/Codex tool compatibility layer so GPT-5.4 can use OpenClaw tools more reliably without a global tool dialect.
## Scope
- strict provider-owned schema normalization for OpenAI/Codex
- provider-owned schema diagnostics
- parameter-free MCP verification
- only mini... | closed | completed | true | 1 | [] | [] | 2026-04-10T09:54:58Z | 2026-04-14T04:20:37Z | 2026-04-11T06:26:31Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,277,934 | I_kwDOQb6kR878nv0u | 64,231 | https://github.com/openclaw/openclaw/issues/64231 | https://api.github.com/repos/openclaw/openclaw/issues/64231 | agents: make elevated full truthful and emit explicit permission-blocked states | Parent: #64227
## Summary
Make `elevatedDefault: full` truthful: only advertise it when host auto-approved exec is actually possible, and surface precise blocked states when sandbox or runtime policy makes full access impossible.
## Scope
- `full` means auto-approved host exec is actually available
- explicit permi... | closed | completed | true | 0 | [] | [] | 2026-04-10T09:54:59Z | 2026-04-13T04:27:40Z | 2026-04-11T00:19:34Z | CONTRIBUTOR | null | 20260414T092630Z | 2026-04-14T09:26:30Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,278,088 | I_kwDOQb6kR878nv3I | 64,232 | https://github.com/openclaw/openclaw/issues/64232 | https://api.github.com/repos/openclaw/openclaw/issues/64232 | agents: classify replay/liveness failures and prevent silent long-task abandonment | Parent: #64227
## Summary
Harden replay and long-task liveness so GPT-5.4 does not silently disappear mid-task or have replay failures misread as model behavior.
## Scope
- replay-invalid classification
- stale continuation / compaction abandonment handling
- liveness states for working, paused, blocked, and abando... | closed | completed | true | 1 | [] | [] | 2026-04-10T09:55:00Z | 2026-04-14T04:20:36Z | 2026-04-11T06:26:59Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,278,231 | I_kwDOQb6kR878nv5X | 64,233 | https://github.com/openclaw/openclaw/issues/64233 | https://api.github.com/repos/openclaw/openclaw/issues/64233 | benchmarks: add GPT-5.4 vs Opus 4.6 agentic parity harness and release gate | Parent: #64227
## Summary
Add the shared benchmark harness and release gate for GPT-5.4 vs Opus 4.6 so parity claims are evidence-backed.
## Scope
- shared task suite
- completion / unintended-stop / valid-tool-call metrics
- targeted auth/proxy mislabel regressions
- release gate document and CI entrypoint if feas... | closed | completed | true | 1 | [] | [] | 2026-04-10T09:55:02Z | 2026-04-28T05:01:41Z | 2026-04-25T16:01:24Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,324,598 | I_kwDOQb6kR878n7N2 | 64,235 | https://github.com/openclaw/openclaw/issues/64235 | https://api.github.com/repos/openclaw/openclaw/issues/64235 | [Bug]: telegram sendMessage failed: Network request for 'sendMessage' failed! | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Greetings! I'm a new OpenClaw user. A couple of weeks ago, I configured my first setup, and everything went well. Still, starting this week, I've been receiving error messages related to communication with Telegram. As a res... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-10T10:01:37Z | 2026-04-30T04:42:14Z | 2026-04-26T04:48:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Killbrum | 7,841,390 | MDQ6VXNlcjc4NDEzOTA= | User | false |
openclaw/openclaw | 4,238,346,330 | I_kwDOQb6kR878oAha | 64,236 | https://github.com/openclaw/openclaw/issues/64236 | https://api.github.com/repos/openclaw/openclaw/issues/64236 | Feature request: support per-model reasoningDefault configuration | ## Feature Request
Currently, `reasoningDefault` can only be set at the agent level (`agents.list[].reasoningDefault`), not globally at `agents.defaults`, and there is no support for per-model reasoningDefault configuration.
### Desired Behavior
Allow `reasoningDefault` to be configurable per model in the model pool,... | open | null | false | 2 | [] | [] | 2026-04-10T10:05:08Z | 2026-05-03T10:41:03Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | Lrs-GreenBottle | 253,340,193 | U_kgDODxmqIQ | User | false |
openclaw/openclaw | 4,238,367,434 | I_kwDOQb6kR878oFrK | 64,237 | https://github.com/openclaw/openclaw/issues/64237 | https://api.github.com/repos/openclaw/openclaw/issues/64237 | exec tool output still needs a native pre-session-write output cap | ## Summary
OpenClaw still needs a **native cap on oversized exec/tool text before session write**.
Issue #16574 identified the general problem, but the runtime-safety aspect remains: a single unbounded `exec` result can still inject a massive `toolResult` into session context and trigger a same-turn context overflow ... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:08:34Z | 2026-04-28T05:01:40Z | 2026-04-25T16:01:30Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Neomail2 | 120,675,713 | U_kgDOBzFdgQ | User | false |
openclaw/openclaw | 4,238,368,590 | I_kwDOQb6kR878oF9O | 64,238 | https://github.com/openclaw/openclaw/issues/64238 | https://api.github.com/repos/openclaw/openclaw/issues/64238 | [Bug]: Regression: OpenClaw 2026.4.9 fails with "Unknown model: qwen/qwen3.6-plus" while 2026.4.5 works | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading to OpenClaw 2026.4.9, agent requests fail with:
Unknown model: qwen/qwen3.6-plus
Downgrading back to 2026.4.5 resolves the issue immediately.
### Steps to reproduce
1. Configure Qwen provider with Coding ... | closed | completed | true | 7 | [
"bug"
] | [] | 2026-04-10T10:08:45Z | 2026-04-28T04:43:13Z | 2026-04-26T01:56:55Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zhuzaifangxuele | 52,022,596 | MDQ6VXNlcjUyMDIyNTk2 | User | false |
openclaw/openclaw | 4,238,409,072 | I_kwDOQb6kR878oP1w | 64,243 | https://github.com/openclaw/openclaw/issues/64243 | https://api.github.com/repos/openclaw/openclaw/issues/64243 | Mattermost deploy probe can still hit exec approval timeout in v2026.3.24 despite tools.exec.ask=off | ## Summary
On a live Mattermost-backed OpenClaw deployment, a deploy probe / doctor step hit an exec approval wait and expired after ~120 seconds, even though the generated gateway config had:
```json
"tools": {
"exec": {
"host": "gateway",
"security": "full",
"ask": "off"
}
}
```
This was observed o... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:15:10Z | 2026-04-30T04:42:13Z | 2026-04-26T04:48:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | XanderZhu | 25,664,794 | MDQ6VXNlcjI1NjY0Nzk0 | User | false |
openclaw/openclaw | 4,238,411,043 | I_kwDOQb6kR878oQUj | 64,244 | https://github.com/openclaw/openclaw/issues/64244 | https://api.github.com/repos/openclaw/openclaw/issues/64244 | [Bug]: MiniMax image tool bypasses plugin VL-01 routing in 2026.4.8, sends image request directly to M2.7 chat endpoint | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.5 to 2026.4.8, the `image` tool fails with `400 "invalid params, chat content is empty (2013)"` when using MiniMax M2.7. The image tool sends the request directly to M2.7's Anthropic chat endpoint... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-10T10:15:35Z | 2026-04-28T05:01:39Z | 2026-04-25T16:01:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wengzhe | 7,105,993 | MDQ6VXNlcjcxMDU5OTM= | User | false |
openclaw/openclaw | 4,238,426,768 | I_kwDOQb6kR878oUKQ | 64,245 | https://github.com/openclaw/openclaw/issues/64245 | https://api.github.com/repos/openclaw/openclaw/issues/64245 | Bug: `update.run` lacks retry mechanism during NPM updates, causing fatal service crash | ### Description
When triggering the `update.run` action on an NPM-installed OpenClaw instance, the internal update process lacks the retry mechanism present in the official `install.sh`.
If the underlying `npm install -g openclaw@latest` fails due to transient network issues or process locks, the older version gets un... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:18:33Z | 2026-04-28T05:01:38Z | 2026-04-25T16:01:50Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | henry-bst | 67,732,349 | MDQ6VXNlcjY3NzMyMzQ5 | User | false |
openclaw/openclaw | 4,238,441,042 | I_kwDOQb6kR878oXpS | 64,246 | https://github.com/openclaw/openclaw/issues/64246 | https://api.github.com/repos/openclaw/openclaw/issues/64246 | Bootstrap: preserve effective AGENTS source across compaction, retry, reset, and reinjection | ## Summary
Hard-coded rereads of `AGENTS.md` outside the initial bootstrap path can cause a run to regain the wrong AGENTS source after compaction, retry, reset, or continuation-style reinjection.
For the planned model-aware AGENTS/SubAGENTS feature, the selected AGENTS source must remain stable for the lifetime of t... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:21:18Z | 2026-04-25T04:17:44Z | 2026-04-22T10:36:10Z | CONTRIBUTOR | null | 20260425T060021Z | 2026-04-25T06:00:21Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,441,767 | I_kwDOQb6kR878oX0n | 64,247 | https://github.com/openclaw/openclaw/issues/64247 | https://api.github.com/repos/openclaw/openclaw/issues/64247 | Bootstrap: add core config and docs for model-aware AGENTS / subagent AGENTS file selection | ## Summary
OpenClaw needs a first-class core config surface for selecting a different AGENTS-source file for:
- main runs vs subagent runs
- exact resolved models such as `openai/gpt-5.4`
This should replace the need for an external plugin that swaps `AGENTS.md` to `SUBAGENTS.md` with ad hoc hook logic.
## Problem
... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:21:28Z | 2026-04-25T04:17:43Z | 2026-04-22T10:36:12Z | CONTRIBUTOR | null | 20260425T060021Z | 2026-04-25T06:00:21Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,442,796 | I_kwDOQb6kR878oYEs | 64,248 | https://github.com/openclaw/openclaw/issues/64248 | https://api.github.com/repos/openclaw/openclaw/issues/64248 | Core: model-aware AGENTS/SUBAGENTS bootstrap selection | ## Summary
This umbrella tracks upstreaming model-aware AGENTS/SUBAGENTS bootstrap selection into core OpenClaw.
Goal:
- let main runs and subagent runs use different AGENTS-source files
- let either path override that file by exact resolved model
- keep orchestrator prompts broader while subagents receive narrowed ... | closed | completed | true | 4 | [] | [] | 2026-04-10T10:21:41Z | 2026-04-25T04:17:42Z | 2026-04-22T10:36:15Z | CONTRIBUTOR | null | 20260425T060021Z | 2026-04-25T06:00:21Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,238,458,658 | I_kwDOQb6kR878ob8i | 64,250 | https://github.com/openclaw/openclaw/issues/64250 | https://api.github.com/repos/openclaw/openclaw/issues/64250 | [Bug]: Context limit exceeded. I've reset our conversation to start fresh - please try again | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
after updating to 2026.4.9, no matter what I send, it reply: ⚠️ Context limit exceeded. I've reset our conversation to start fresh - please try again.
To prevent this, increase your compaction buffer by setting... | closed | completed | true | 10 | [
"bug"
] | [] | 2026-04-10T10:24:31Z | 2026-04-28T05:01:37Z | 2026-04-25T16:01:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | twtsummery | 155,789,379 | U_kgDOCUkoQw | User | false |
openclaw/openclaw | 4,238,481,153 | I_kwDOQb6kR878ohcB | 64,251 | https://github.com/openclaw/openclaw/issues/64251 | https://api.github.com/repos/openclaw/openclaw/issues/64251 | [Bug]: CLI-backed model fails in follow-up path (codex-cli/gpt-5.4) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Follow-up agent runs fail with "Unknown model: codex-cli/gpt-5.4" while direct CLI execution succeeds.
### Steps to reproduce
1. Configure agents.defaults.model.primary as codex-cli/gpt-5.4
2. Start openclaw-... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T10:28:50Z | 2026-04-28T05:01:37Z | 2026-04-25T16:02:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | WolfMarkTools | 215,191,532 | U_kgDODNOP7A | User | false |
openclaw/openclaw | 4,238,482,410 | I_kwDOQb6kR878ohvq | 64,252 | https://github.com/openclaw/openclaw/issues/64252 | https://api.github.com/repos/openclaw/openclaw/issues/64252 | [Bug]: UNAVAILABLE: A2UI_HOST_UNAVAILABLE | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
“When using openclaw nodes canvas a2ui push --node 553pro --text "Hello", an error occurs.
<img width="520" height="680" alt="Image" src="https://github.com/user-attachments/assets/e395de7c-f089-404a-827a-6e17... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-04-10T10:29:06Z | 2026-04-28T05:01:36Z | 2026-04-25T16:02:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Running00 | 30,852,766 | MDQ6VXNlcjMwODUyNzY2 | User | false |
openclaw/openclaw | 4,238,494,814 | I_kwDOQb6kR878okxe | 64,253 | https://github.com/openclaw/openclaw/issues/64253 | https://api.github.com/repos/openclaw/openclaw/issues/64253 | Gateway becomes unresponsive under subagent load on Windows - completion announcements timeout | ## Bug
The gateway becomes unresponsive when handling multiple subagent completion announcements, causing cascading timeouts and failed deliveries. This is distinct from #7042 (which was hard crashes) - the gateway stays alive but stops processing requests effectively.
## Symptoms
1. **Subagent completion timeouts**... | open | null | false | 1 | [] | [] | 2026-04-10T10:31:32Z | 2026-05-03T08:39:18Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | L8ton-crypto | 232,345,273 | U_kgDODdlOuQ | User | false |
openclaw/openclaw | 4,238,499,799 | I_kwDOQb6kR878ol_X | 64,254 | https://github.com/openclaw/openclaw/issues/64254 | https://api.github.com/repos/openclaw/openclaw/issues/64254 | memory-core: dreaming promotion cron not persisting across gateway restarts | ## Problem
The memory-core plugin creates a managed cron job for dreaming promotion ("Memory Dreaming Promotion"), but this cron disappears after a gateway restart. Every restart wipes the cron and it must be manually re-added.
## Expected behavior
Per the docs: "When enabled, memory-core auto-manages one cron job f... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:32:33Z | 2026-04-15T04:19:45Z | 2026-04-12T16:53:51Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | kaishi00-student | 217,337,130 | U_kgDODPRNKg | User | false |
openclaw/openclaw | 4,238,540,449 | I_kwDOQb6kR878ov6h | 64,256 | https://github.com/openclaw/openclaw/issues/64256 | https://api.github.com/repos/openclaw/openclaw/issues/64256 | TUI Agent Switching — Messages Route to Wrong Session | Bug Report: TUI Agent Switching — Messages Route to Wrong Session
Steps to reproduce:
1. Open TUI session
2. Run /agent coder_agent (or any non-main agent)
3. TUI creates a new session with the correct agent's model (verified via openclaw status)
4. Send a message
Expected behavior:
Message goes to the coder_... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:40:38Z | 2026-04-28T05:01:35Z | 2026-04-25T16:02:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | maherbot-mk | 262,465,185 | U_kgDOD6TmoQ | User | false |
openclaw/openclaw | 4,238,580,347 | I_kwDOQb6kR878o5p7 | 64,260 | https://github.com/openclaw/openclaw/issues/64260 | https://api.github.com/repos/openclaw/openclaw/issues/64260 | Include context window size and model max_tokens in llm_output event | ## Problem
Plugins have no way to know the context window utilization for a given LLM call. The `llm_output` event includes token counts (input, output, cache) but not:
1. **Context window size** — the model's maximum context length (e.g., 200k for Claude, 128k for GPT-4)
2. **Effective context utilization** — what p... | open | null | false | 1 | [] | [] | 2026-04-10T10:49:31Z | 2026-05-03T10:19:14Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | Kyzcreig | 9,063,726 | MDQ6VXNlcjkwNjM3MjY= | User | false |
openclaw/openclaw | 4,238,581,660 | I_kwDOQb6kR878o5-c | 64,261 | https://github.com/openclaw/openclaw/issues/64261 | https://api.github.com/repos/openclaw/openclaw/issues/64261 | Include per-call latency_ms in llm_output / llm_call events | ## Problem
The `llm_output` event includes token counts but no timing data for individual LLM API calls. The `agent_end` event has total `durationMs` for the entire turn, but there is no way to know how long each individual LLM call took.
This is needed for:
- Spending alerts: `Duration: 53s (API total 44s)` — showin... | closed | completed | true | 1 | [] | [] | 2026-04-10T10:49:46Z | 2026-04-30T04:42:12Z | 2026-04-26T04:48:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Kyzcreig | 9,063,726 | MDQ6VXNlcjkwNjM3MjY= | User | false |
openclaw/openclaw | 4,238,583,929 | I_kwDOQb6kR878o6h5 | 64,262 | https://github.com/openclaw/openclaw/issues/64262 | https://api.github.com/repos/openclaw/openclaw/issues/64262 | [Bug]: Garbled Model Output | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Agent produced garbled/binary-like output when responding to user query. Output appeared as random symbols, `$` characters, and hash marks instead of readable text.
### Steps to reproduce
Unknown - appears to be intermitt... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-10T10:50:12Z | 2026-05-01T04:43:56Z | 2026-04-28T10:21:39Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Kloz813 | 275,048,949 | U_kgDOEGTp9Q | User | false |
openclaw/openclaw | 4,238,597,588 | I_kwDOQb6kR878o93U | 64,267 | https://github.com/openclaw/openclaw/issues/64267 | https://api.github.com/repos/openclaw/openclaw/issues/64267 | Bug: OpenClaw 2026.4.9 exposes agent internal thinking (English) to user | # Bug Report: OpenClaw 2026.4.9 Exposes Agent Internal Thinking (English) to User
## Summary
OpenClaw agent's internal thinking/planning process (in English) is being included in responses sent to users. This affects multiple models and appears to be a system-level bug.
## Environment
- **OpenClaw Version**: 2026.4.9... | open | null | false | 4 | [] | [] | 2026-04-10T10:52:40Z | 2026-05-03T03:22:38Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | superssr | 46,590,166 | MDQ6VXNlcjQ2NTkwMTY2 | User | false |
openclaw/openclaw | 4,238,607,414 | I_kwDOQb6kR878pAQ2 | 64,268 | https://github.com/openclaw/openclaw/issues/64268 | https://api.github.com/repos/openclaw/openclaw/issues/64268 | Emit llm_output per-LLM-call in agentic loops, not just once per turn | In agentic turns (LLM → tool → LLM → tool → LLM), llm_output fires once at the end with aggregated totals. Plugins cannot get per-iteration LLM call data.
handleMessageEnd in pi-embedded fires for every assistant message, and recordAssistantUsage() accumulates into usageTotals, but llm_output only fires once with the ... | open | null | false | 1 | [] | [] | 2026-04-10T10:54:21Z | 2026-05-03T02:14:53Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | Kyzcreig | 9,063,726 | MDQ6VXNlcjkwNjM3MjY= | User | false |
openclaw/openclaw | 4,238,664,937 | I_kwDOQb6kR878pOTp | 64,270 | https://github.com/openclaw/openclaw/issues/64270 | https://api.github.com/repos/openclaw/openclaw/issues/64270 | [Bug] MS Teams SingleTenant bot: JWT validation fails (issuer + audience mismatch) | ### Description
When using a **SingleTenant** Azure Bot with the MS Teams channel, incoming webhook requests from Azure Bot Service are rejected with `401 Unauthorized`. This affects all new bot deployments since Microsoft deprecated MultiTenant bot creation after 2025-07-31.
### Root Cause
Two issues in the JWT val... | closed | completed | true | 0 | [] | [] | 2026-04-10T11:05:10Z | 2026-04-14T04:20:05Z | 2026-04-12T02:19:43Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | mschaepers | 22,095,221 | MDQ6VXNlcjIyMDk1MjIx | User | false |
openclaw/openclaw | 4,238,688,678 | I_kwDOQb6kR878pUGm | 64,271 | https://github.com/openclaw/openclaw/issues/64271 | https://api.github.com/repos/openclaw/openclaw/issues/64271 | browser.request retry loop burns CPU indefinitely when Chrome CDP fails to start | ## Summary
When Chrome CDP fails to start (e.g., sandbox issues, port conflicts, missing dependencies), the gateway retries `browser.request` every ~30 seconds with no failure limit, backoff, or circuit breaker. Each retry spawns a Chrome process that runs for ~15 seconds before failing, consuming significant CPU cont... | closed | completed | true | 2 | [] | [] | 2026-04-10T11:08:59Z | 2026-05-02T04:27:57Z | 2026-04-29T15:03:22Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | TheophilusChinomona | 109,826,598 | U_kgDOBovSJg | User | false |
openclaw/openclaw | 4,238,718,000 | I_kwDOQb6kR878pbQw | 64,272 | https://github.com/openclaw/openclaw/issues/64272 | https://api.github.com/repos/openclaw/openclaw/issues/64272 | [Bug]: TTS Regression in v2026.4.9 UPDATED | # OpenClaw Bug Report: TTS Voice Note Delivery Regression in v2026.4.9
## Summary
TTS (text-to-speech) voice note delivery stopped working after upgrading to OpenClaw v2026.4.9. The `tts` tool reports successful audio generation, but voice notes are never delivered to Telegram. STT (speech-to-text) for inbound voice n... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-04-10T11:13:33Z | 2026-04-15T04:19:10Z | 2026-04-12T08:09:04Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | gyza | 870,379 | MDQ6VXNlcjg3MDM3OQ== | User | false |
openclaw/openclaw | 4,238,734,740 | I_kwDOQb6kR878pfWU | 64,274 | https://github.com/openclaw/openclaw/issues/64274 | https://api.github.com/repos/openclaw/openclaw/issues/64274 | [Bug]: Agent-specific MiniMax auth resolves from main agent auth-profiles.json | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
OpenClaw version:
2026.4.9 (0512059)
Extra confirmation:
- `~/.openclaw/agents/kate/agent/auth-profiles.json` does contain:
- minimax-portal:default
- provider: minimax-portal
Symptom:
- dashboard chat, `/new`, and `/r... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-04-10T11:16:30Z | 2026-05-03T06:49:35Z | null | CONTRIBUTOR | null | 20260503T090142Z | 2026-05-03T09:01:42Z | blaspat | 34,417,345 | MDQ6VXNlcjM0NDE3MzQ1 | User | false |
openclaw/openclaw | 4,238,813,955 | I_kwDOQb6kR878pysD | 64,277 | https://github.com/openclaw/openclaw/issues/64277 | https://api.github.com/repos/openclaw/openclaw/issues/64277 | [Bug]: Slack `requireMention: true` is bypassed for thread replies after prior bot participation | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
With Slack channel config `requireMention: true`, OpenClaw still replies to untagged messages in a thread after it has previously participated in that thread.
This appears to happen because Slack thread partic... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T11:29:36Z | 2026-04-16T04:24:49Z | 2026-04-13T06:04:10Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | santosh-setu | 108,258,166 | U_kgDOBnPjdg | User | false |
openclaw/openclaw | 4,238,820,444 | I_kwDOQb6kR878p0Rc | 64,278 | https://github.com/openclaw/openclaw/issues/64278 | https://api.github.com/repos/openclaw/openclaw/issues/64278 | Gateway RPC gap: no method to set default model (currently only CLI via direct config write) | ## Summary
Gateway exposes `models.list` and `models.ollama.list`, but has no RPC to **set the default model** (`agents.defaults.model`). The only way to change it today is via the `openclaw models set <id>` CLI, which spawns a Node process and rewrites `openclaw.json` locally — bypassing the Gateway entirely.
## Whe... | closed | not_planned | true | 1 | [] | [] | 2026-04-10T11:30:29Z | 2026-04-13T04:28:17Z | 2026-04-10T12:23:04Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | SkylerChenTaiwan | 209,284,954 | U_kgDODHlvWg | User | false |
openclaw/openclaw | 4,238,846,933 | I_kwDOQb6kR878p6vV | 64,280 | https://github.com/openclaw/openclaw/issues/64280 | https://api.github.com/repos/openclaw/openclaw/issues/64280 | [Feature Request] Add file write tool to bypass exec shell escaping issues | ## Issue Summary
**File writing through exec tool fails with complex content due to shell escaping issues.**
When attempting to write files containing Python/JavaScript code with nested quotes through the exec tool, the escaping through PowerShell → Python -c layers causes parse errors.
## Steps to Reproduce
1. Att... | closed | completed | true | 1 | [] | [] | 2026-04-10T11:34:17Z | 2026-04-28T05:01:34Z | 2026-04-25T16:02:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sg1416-zg | 259,938,361 | U_kgDOD35YOQ | User | false |
openclaw/openclaw | 4,238,923,269 | I_kwDOQb6kR878qNYF | 64,282 | https://github.com/openclaw/openclaw/issues/64282 | https://api.github.com/repos/openclaw/openclaw/issues/64282 | [Bug]: Can't update | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
I am not able to update openclaw
### Steps to reproduce
Update Openclaw
### Expected behavior
to work
### Actual behavior
Failing to update
### OpenClaw version
2026.4.2
### Operating system
Raspberry OS
### Insta... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-10T11:46:15Z | 2026-04-28T05:01:33Z | 2026-04-25T16:02:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | shifoc | 32,818,143 | MDQ6VXNlcjMyODE4MTQz | User | false |
openclaw/openclaw | 4,238,942,303 | I_kwDOQb6kR878qSBf | 64,283 | https://github.com/openclaw/openclaw/issues/64283 | https://api.github.com/repos/openclaw/openclaw/issues/64283 | [Bug]: gateway install --force drops GOOGLE_APPLICATION_CREDENTIALS — google-vertex auth fails after service reinstall | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When `openclaw doctor` (or `openclaw gateway install --force`) regenerates the systemd service file, it does **not** include `GOOGLE_APPLICATION_CREDENTIALS` in the `Environment=` entries, even if this variable is defined in... | closed | completed | true | 2 | [] | [] | 2026-04-10T11:49:36Z | 2026-04-28T05:01:32Z | 2026-04-25T16:02:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ordiy | 12,695,045 | MDQ6VXNlcjEyNjk1MDQ1 | User | false |
openclaw/openclaw | 4,238,998,219 | I_kwDOQb6kR878qfrL | 64,287 | https://github.com/openclaw/openclaw/issues/64287 | https://api.github.com/repos/openclaw/openclaw/issues/64287 | [Bug]: Session context loss due to broken parentId chain in JSONL transcripts | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Messages whose parentId references a non-existent id in the session JSONL transcript break the replay chain, causing the agent to silently lose all conversation history before the break point. Observed 189 brok... | closed | completed | true | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T12:00:17Z | 2026-04-13T04:28:18Z | 2026-04-10T12:21:51Z | NONE | null | 20260414T092630Z | 2026-04-14T09:26:30Z | weixingreader | 251,274,555 | U_kgDODvolOw | User | false |
openclaw/openclaw | 4,239,018,690 | I_kwDOQb6kR878qkrC | 64,288 | https://github.com/openclaw/openclaw/issues/64288 | https://api.github.com/repos/openclaw/openclaw/issues/64288 | [Bug]: [telegram] polling runner stopped (polling stall detected) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
telegram is having frequent issue
### Steps to reproduce
just restart gateway and wait
### Expected behavior
no issues
### Actual behavior
12:53:14+01:00 [telegram] Polling stall detected (no completed getUpdates for ... | closed | completed | false | 5 | [
"bug",
"close:duplicate",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-04-10T12:04:07Z | 2026-04-25T09:20:19Z | 2026-04-25T09:20:19Z | NONE | null | 20260425T120014Z | 2026-04-25T12:00:14Z | athuljayaram | 141,337,513 | U_kgDOCGyjqQ | User | false |
openclaw/openclaw | 4,239,018,724 | I_kwDOQb6kR878qkrk | 64,289 | https://github.com/openclaw/openclaw/issues/64289 | https://api.github.com/repos/openclaw/openclaw/issues/64289 | [Bug] Session-scoped plugin lazy-initialization causes UX disruption after idle | ## Bug Description
After gateway startup, when the user returns to webchat after an idle period (~10+ minutes) and performs any interaction (e.g., clicking into a session, sending a message), all session-scoped plugins re-initialize. This causes the webchat WebSocket to disconnect and reconnect (code 1001), creating a... | closed | completed | true | 1 | [] | [] | 2026-04-10T12:04:08Z | 2026-04-29T04:32:31Z | 2026-04-27T02:26:00Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | shenyiqi-2004 | 221,808,311 | U_kgDODTiGtw | User | false |
openclaw/openclaw | 4,239,054,865 | I_kwDOQb6kR878qtgR | 64,291 | https://github.com/openclaw/openclaw/issues/64291 | https://api.github.com/repos/openclaw/openclaw/issues/64291 | [Bug]: Session replay loads only a fraction of transcript even with intact parentId chain and no compaction | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Session replay consistently loads only 18-24K tokens from a transcript containing ~125K tokens of conversation, despite the parentId chain being fully intact and no compaction having occurred. The context windo... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T12:11:08Z | 2026-04-28T05:01:31Z | 2026-04-25T16:02:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | weixingreader | 251,274,555 | U_kgDODvolOw | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.