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 | 4,054,872,265 | I_kwDOQb6kR87xsHDJ | 42,576 | https://github.com/openclaw/openclaw/issues/42576 | https://api.github.com/repos/openclaw/openclaw/issues/42576 | [Bug]: Whitespace normalization breaks indentations | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The actual report is generated by agent after investigation.
## Summary
When OpenClaw sends a reply containing `[[reply_to_current]]` (or similar reply tags), whitespace gets normalized across the whole message. This breaks fenced code b... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T00:02:58Z | 2026-04-27T04:52:36Z | 2026-04-24T22:14:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | vrurg | 1,586,992 | MDQ6VXNlcjE1ODY5OTI= | User | false |
openclaw/openclaw | 4,054,877,045 | I_kwDOQb6kR87xsIN1 | 42,577 | https://github.com/openclaw/openclaw/issues/42577 | https://api.github.com/repos/openclaw/openclaw/issues/42577 | [Bug]: Docker build cache not properly invalidated on dependency changes | ## 环境信息
- 操作系统：Ubuntu 22.04 / macOS Sonoma
- OpenClaw版本：2026.3.9
- Node版本：20.x / 22.x
- Docker版本：24.x / 25.x
## 问题描述
Docker构建缓存未在依赖项更改时正确失效，导致构建使用过时的依赖。
## 复现步骤
1. 修改package.json添加新依赖
2. 运行openclaw gateway build或docker构建
3. 观察构建日志，确认是否使用了缓存的node_modules
## 预期行为
依赖项更改后，Docker应该检测到package.json的变化并重新安装依赖。
## 实际行为
Dock... | closed | completed | true | 1 | [] | [] | 2026-03-11T00:04:24Z | 2026-04-27T04:52:35Z | 2026-04-24T22:14:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ai-nurmamat | 130,913,840 | U_kgDOB82WMA | User | false |
openclaw/openclaw | 4,054,958,541 | I_kwDOQb6kR87xscHN | 42,598 | https://github.com/openclaw/openclaw/issues/42598 | https://api.github.com/repos/openclaw/openclaw/issues/42598 | [优化建议] setup-podman.sh 镜像加载性能优化 | ## 问题描述
当前 `setup-podman.sh` 在加载镜像到用户 Podman store 时使用了临时文件：
```bash
echo "Loading image into $OPENCLAW_USER's Podman store..."
TMP_IMAGE_DIR="$(resolve_image_tmp_dir)"
TMP_STAGE_DIR="$(mktemp -d -p "$TMP_IMAGE_DIR" openclaw-image.XXXXXX)"
TMP_IMAGE="$TMP_STAGE_DIR/image.tar"
chmod 700 "$TMP_STAGE_DIR"
trap 'rm -rf "... | closed | completed | true | 1 | [] | [] | 2026-03-11T00:28:09Z | 2026-04-27T04:52:34Z | 2026-04-24T22:14:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | nillikechatchat | 266,501,707 | U_kgDOD-J-Sw | User | false |
openclaw/openclaw | 4,054,965,360 | I_kwDOQb6kR87xsdxw | 42,599 | https://github.com/openclaw/openclaw/issues/42599 | https://api.github.com/repos/openclaw/openclaw/issues/42599 | [优化建议] setup-podman.sh 用户创建逻辑抽象和兼容性改进 | ## 问题描述
当前 `setup-podman.sh` 中的用户创建逻辑在不同系统间有重复代码：
```bash
if command -v useradd >/dev/null 2>&1; then
run_root useradd -m -s "$NOLOGIN_SHELL" "$OPENCLAW_USER"
elif command -v adduser >/dev/null 2>&1; then
run_root adduser --disabled-password --gecos "" --shell "$NOLOGIN_SHELL" "$OPENCLAW_USER"
else
echo "Neithe... | closed | completed | true | 2 | [] | [] | 2026-03-11T00:30:39Z | 2026-04-27T04:52:33Z | 2026-04-24T22:14:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | nillikechatchat | 266,501,707 | U_kgDOD-J-Sw | User | false |
openclaw/openclaw | 4,054,990,492 | I_kwDOQb6kR87xsj6c | 42,607 | https://github.com/openclaw/openclaw/issues/42607 | https://api.github.com/repos/openclaw/openclaw/issues/42607 | Orphan tool_result causes 400 error after API overload/error during tool call | ## Bug Description
When the Anthropic API returns an error (e.g., `overloaded_error`) **during** a tool call execution, Clawdbot inserts a synthetic `tool_result` to "repair" the transcript. However, this breaks the conversation structure because:
1. The `stopReason` on the assistant message is `error`, not `tool_use... | closed | completed | true | 1 | [] | [] | 2026-03-11T00:39:37Z | 2026-04-27T04:52:33Z | 2026-04-24T22:14:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | saulocuevas-cmd | 225,990,548 | U_kgDODXhXlA | User | false |
openclaw/openclaw | 4,055,039,861 | I_kwDOQb6kR87xsv91 | 42,612 | https://github.com/openclaw/openclaw/issues/42612 | https://api.github.com/repos/openclaw/openclaw/issues/42612 | feat(multi-agent): cross-agent sessions_send fails with 'pairing required' for sibling agents | ## Summary
When agent `main` (Clio) attempts to send a message to sibling agent `wren` via `sessions_send` with session key `agent:wren:main`, the gateway returns:
```
gateway closed (1008): pairing required
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: /Users/.../.openclaw-main/openclaw.json
Bi... | closed | completed | true | 1 | [] | [] | 2026-03-11T00:59:02Z | 2026-04-27T04:52:32Z | 2026-04-24T22:14:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kushminer | 40,745,311 | MDQ6VXNlcjQwNzQ1MzEx | User | false |
openclaw/openclaw | 4,055,047,004 | I_kwDOQb6kR87xsxtc | 42,615 | https://github.com/openclaw/openclaw/issues/42615 | https://api.github.com/repos/openclaw/openclaw/issues/42615 | Discord guild MESSAGE_CREATE events not delivered to gateway (2026.3.8) | Environment
- OpenClaw version: 2026.3.8
- Gateway mode: local (loopback), running on host Vostro
- Bot ID: 1481040377199988758
- Guild (server): mirage (ID 1477630524930396291)
- Channels tested:
- #lazarus (ID 1481037202140954916)
- #golang (ID 1481036986784678030)
- User ID used for allowlists: 14776293937601577... | closed | completed | true | 2 | [] | [] | 2026-03-11T01:02:02Z | 2026-04-27T04:52:30Z | 2026-04-24T22:14:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mirage06020426 | 208,717,341 | U_kgDODHDGHQ | User | false |
openclaw/openclaw | 4,054,585,084 | I_kwDOQb6kR87xrA78 | 42,545 | https://github.com/openclaw/openclaw/issues/42545 | https://api.github.com/repos/openclaw/openclaw/issues/42545 | Feature Request: Support Multimodal Inputs (Images/Files) in the OpenAI /v1/chat/completions Endpoint | ### Is there an existing issue for this?
- [X] I have searched the existing issues
### What would you like to add?
Currently, the OpenAI compatibility endpoint (`/v1/chat/completions`) implemented in `src/gateway/openai-http.ts` completely drops non-text content from the user request. Inside the `extractTextContent`... | closed | completed | true | 2 | [] | [] | 2026-03-10T22:35:38Z | 2026-04-27T04:52:31Z | 2026-04-24T22:13:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | DavidLyn | 13,762,942 | MDQ6VXNlcjEzNzYyOTQy | User | false |
openclaw/openclaw | 4,055,075,933 | I_kwDOQb6kR87xs4xd | 42,619 | https://github.com/openclaw/openclaw/issues/42619 | https://api.github.com/repos/openclaw/openclaw/issues/42619 | Gateway restart after config change can silently reset exec approvals to restrictive defaults | ## Summary
A config change that triggers a gateway restart can silently reset local exec approvals to a restrictive mode, causing all subsequent `exec` calls to require approval unexpectedly.
In our case, a Discord-driven config write updated `openclaw.json`, which triggered a gateway restart. After the restart, `~/.... | closed | not_planned | true | 2 | [] | [] | 2026-03-11T01:12:05Z | 2026-04-27T04:52:29Z | 2026-04-24T22:14:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | seikixtc | 219,963,501 | U_kgDODRxgbQ | User | false |
openclaw/openclaw | 4,055,085,136 | I_kwDOQb6kR87xs7BQ | 42,622 | https://github.com/openclaw/openclaw/issues/42622 | https://api.github.com/repos/openclaw/openclaw/issues/42622 | stallHard reaction fires from ALL bot accounts in guild, not just the affected agent | ## Bug Description
When an agent stalls hard (e.g., context overflow + rate limit exhaustion on all fallback providers), the `stallHard` emoji reaction (😨) is added to the triggering message. However, the reaction is sent from **all bot accounts** in the Discord guild, not just the bot account of the agent that actua... | closed | completed | true | 1 | [] | [] | 2026-03-11T01:15:38Z | 2026-04-27T04:52:29Z | 2026-04-24T22:14:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | MediaPublishing | 31,557,155 | MDQ6VXNlcjMxNTU3MTU1 | User | false |
openclaw/openclaw | 4,055,128,325 | I_kwDOQb6kR87xtFkF | 42,634 | https://github.com/openclaw/openclaw/issues/42634 | https://api.github.com/repos/openclaw/openclaw/issues/42634 | [Bug]: gpt 5.4 gives up all the time - improve tool calling by sending `phase`? | ### Summary
GPT 5.4 does not work well in openclaw and gives up on tasks preemptively.
### Problem to solve
I've noticed that 5.4 gives up in openclaw routinely - either doesn't start tools calls, or sends 2-3 calls then stops and needs a user message to continue.
This does not happen in codex gui/tui.
### Proposed... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-11T01:31:54Z | 2026-04-27T04:52:28Z | 2026-04-24T22:14:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Thinkscape | 270,528 | MDQ6VXNlcjI3MDUyOA== | User | false |
openclaw/openclaw | 4,055,155,426 | I_kwDOQb6kR87xtMLi | 42,640 | https://github.com/openclaw/openclaw/issues/42640 | https://api.github.com/repos/openclaw/openclaw/issues/42640 | [Bug]: Cron jobs execute twice after gateway restart (scheduler doesn't check lastRunAtMs) | ### Bug type
Regression (cron scheduler behavior after gateway restart)
### Summary
**Cron jobs execute twice after gateway restart** — When the OpenClaw gateway restarts, the cron scheduler's "running missed jobs after restart" mechanism re-executes jobs that already ran before the restart, causing duplicate delive... | closed | completed | true | 1 | [] | [] | 2026-03-11T01:41:32Z | 2026-04-27T04:52:27Z | 2026-04-24T22:14:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | vmhq | 147,793,843 | U_kgDOCM8nsw | User | false |
openclaw/openclaw | 4,055,158,652 | I_kwDOQb6kR87xtM98 | 42,641 | https://github.com/openclaw/openclaw/issues/42641 | https://api.github.com/repos/openclaw/openclaw/issues/42641 | [Feature Request] Allow image tool to access Telegram media files | ; Description
The image tool cannot reliably access files in `~/.openclaw/media/inbound/` directory. This is where Telegram media files are saved when users send images via Telegram.
**Expected behavior:** Users should be able to view images sent via Telegram without manual copying.
**Current issues:**
- Sometimes th... | closed | completed | true | 1 | [] | [] | 2026-03-11T01:42:41Z | 2026-04-27T04:52:26Z | 2026-04-24T22:14:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | NHIZR | 51,505,434 | MDQ6VXNlcjUxNTA1NDM0 | User | false |
openclaw/openclaw | 4,055,170,781 | I_kwDOQb6kR87xtP7d | 42,643 | https://github.com/openclaw/openclaw/issues/42643 | https://api.github.com/repos/openclaw/openclaw/issues/42643 | [Bug]: macOS restart race can transiently drop gateway.auth and regenerate token (unexpected token drift) | ## Summary
On macOS (CLI Gateway + OpenClaw App present), `gateway.auth` can be transiently removed during a restart window, then auto-regenerated, causing unexpected token drift and downstream `token_mismatch`/relay failures.
This appears to be a config write race around restart lifecycle, not a manual config edit.
... | closed | completed | true | 1 | [] | [] | 2026-03-11T01:46:56Z | 2026-04-27T04:52:26Z | 2026-04-24T22:14:34Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | nakamotoliu | 257,456,896 | U_kgDOD1h7AA | User | false |
openclaw/openclaw | 4,055,241,571 | I_kwDOQb6kR87xthNj | 42,671 | https://github.com/openclaw/openclaw/issues/42671 | https://api.github.com/repos/openclaw/openclaw/issues/42671 | [Feature]: WhatsApp group — trigger agent on quoted/reply-to message (no prefix required) | ## Summary
Currently, triggering the agent in a WhatsApp group chat requires a configured `mentionPattern` prefix (e.g. `#ai`). There is no way to trigger the agent by **quoting/replying to one of the agent's previous messages**, which is the natural interaction pattern most users expect — similar to how Discord chann... | closed | completed | true | 1 | [] | [] | 2026-03-11T02:08:27Z | 2026-04-27T04:52:25Z | 2026-04-24T22:14:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | erictong514 | 2,917,573 | MDQ6VXNlcjI5MTc1NzM= | User | false |
openclaw/openclaw | 4,055,256,671 | I_kwDOQb6kR87xtk5f | 42,676 | https://github.com/openclaw/openclaw/issues/42676 | https://api.github.com/repos/openclaw/openclaw/issues/42676 | Kimi web search API URL 路径拼接错误 | ## 问题描述
使用 Kimi 作为 web search provider 时，API URL 路径拼接错误，导致请求失败。
## 错误日志
```
[tools] web_search failed: Kimi API error (404): {"code":5,"error":"url.not_found","message":"没找到对象","method":"POST","scode":"0x5","status":false,"ua":"node","url":"/chat/completions"}
```
以及：
```
[tools] web_search failed: Kimi API error ... | closed | completed | true | 2 | [] | [] | 2026-03-11T02:12:26Z | 2026-04-27T04:52:24Z | 2026-04-24T22:14:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SunWeizhou | 133,394,190 | U_kgDOB_NvDg | User | false |
openclaw/openclaw | 4,055,295,470 | I_kwDOQb6kR87xtuXu | 42,684 | https://github.com/openclaw/openclaw/issues/42684 | https://api.github.com/repos/openclaw/openclaw/issues/42684 | Webchat: UI freezes on loading when session transcript is large (~500KB / 65 messages) | ## Describe the bug
The webchat UI becomes completely unresponsive (stuck on "loading messages") when the session transcript grows to a moderate size (~472KB, 65 messages). The only workaround is to manually truncate or reset the session transcript JSONL file on the server.
## To Reproduce
1. Use webchat as the prim... | closed | completed | true | 2 | [] | [] | 2026-03-11T02:24:23Z | 2026-04-27T04:52:23Z | 2026-04-24T22:14:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | misselvexu | 3,436,472 | MDQ6VXNlcjM0MzY0NzI= | User | false |
openclaw/openclaw | 4,055,299,995 | I_kwDOQb6kR87xtveb | 42,685 | https://github.com/openclaw/openclaw/issues/42685 | https://api.github.com/repos/openclaw/openclaw/issues/42685 | bug(status): tailscale ENOENT warning appears even when gateway.tailscale.mode is off | ## Summary
`openclaw status --all` still probes the `tailscale` binary even when `gateway.tailscale.mode` is explicitly set to `off`. On machines without Tailscale installed, this produces a misleading diagnostic warning:
```text
! Tailscale: off · unknown
error: Error: spawn tailscale ENOENT
```
This looks like a... | closed | completed | true | 1 | [] | [] | 2026-03-11T02:25:36Z | 2026-04-27T04:52:23Z | 2026-04-24T22:14:45Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | fonhal | 4,468,037 | MDQ6VXNlcjQ0NjgwMzc= | User | false |
openclaw/openclaw | 4,055,335,933 | I_kwDOQb6kR87xt4P9 | 42,692 | https://github.com/openclaw/openclaw/issues/42692 | https://api.github.com/repos/openclaw/openclaw/issues/42692 | Bug: openclaw agent --local can pass bare sessionId to session_status, causing Unknown sessionId | ## Summary
`openclaw agent --local` can forward a bare CLI transcript id into embedded tool/runtime plumbing as if it were a real `sessionKey`. When `session_status` later resolves that value, it can fail with:
```text
Unknown sessionId: <raw-cli-session-id>
```
I reproduced this on `openclaw` `2026.3.8`.
## Reprod... | closed | completed | true | 1 | [] | [] | 2026-03-11T02:38:12Z | 2026-04-27T04:52:22Z | 2026-04-24T22:14:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sk8man121 | 179,406,323 | U_kgDOCrGF8w | User | false |
openclaw/openclaw | 4,055,339,317 | I_kwDOQb6kR87xt5E1 | 42,693 | https://github.com/openclaw/openclaw/issues/42693 | https://api.github.com/repos/openclaw/openclaw/issues/42693 | Feature: configurable routing for strict exact-output / formatting-sensitive prompts | ## Summary
OpenClaw would benefit from a first-class, configurable routing layer for strict exact-output / formatting-sensitive prompts.
Right now, the only way to do this cleanly is to patch the runtime and insert custom prompt heuristics before model selection. That works locally, but it is fragile across updates a... | closed | completed | true | 1 | [] | [] | 2026-03-11T02:39:27Z | 2026-04-27T04:52:21Z | 2026-04-24T22:14:50Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sk8man121 | 179,406,323 | U_kgDOCrGF8w | User | false |
openclaw/openclaw | 4,055,343,344 | I_kwDOQb6kR87xt6Dw | 42,695 | https://github.com/openclaw/openclaw/issues/42695 | https://api.github.com/repos/openclaw/openclaw/issues/42695 | Bug: Cron delivery duplicates announce mode messages to Discord | ## Bug Description
Cron jobs with `delivery.mode: "announce"` to Discord sometimes post **duplicate messages** — even when the cron only executes once.
## Evidence
**Cron Configuration:**
- Job: "Health Status — every 4h"
- Schedule: every 4 hours
- Delivery: `announce` to Discord #ops channel
**Gateway Logs (singl... | closed | completed | true | 1 | [] | [
"vincentkoc"
] | 2026-03-11T02:40:56Z | 2026-04-27T04:52:21Z | 2026-04-24T22:14:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rimkute | 112,832,114 | U_kgDOBrmucg | User | false |
openclaw/openclaw | 4,055,350,817 | I_kwDOQb6kR87xt74h | 42,697 | https://github.com/openclaw/openclaw/issues/42697 | https://api.github.com/repos/openclaw/openclaw/issues/42697 | [Bug]: node-llama-cpp postinstall incorrectly detects x64 architecture on Apple Silicon Mac despite native arm64 Node.js | ### Bug type
Regression (worked before, now fails)
### Summary
When installing `openclaw@latest` (which depends on `node-llama-cpp`), the `postinstall` script of `node-llama-cpp` fails with the error:
```
[node-llama-cpp] llama.cpp is not supported under Rosetta on Apple Silicone Macs. Ensure that you're using a na... | closed | completed | true | 5 | [
"bug",
"regression"
] | [] | 2026-03-11T02:43:17Z | 2026-04-27T04:52:20Z | 2026-04-24T22:14:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | FindFun | 13,467,967 | MDQ6VXNlcjEzNDY3OTY3 | User | false |
openclaw/openclaw | 4,055,362,585 | I_kwDOQb6kR87xt-wZ | 42,701 | https://github.com/openclaw/openclaw/issues/42701 | https://api.github.com/repos/openclaw/openclaw/issues/42701 | Bug: Manual cron force-run hangs - isolated session never created | ## Bug: Manual cron `force run` hangs indefinitely — isolated session never created
### Description
When manually triggering a cron job via the UI (force run), the isolated session is never created. The job hangs for the entire timeout duration (600s) and then fails with `"cron: job execution timed out"`. The same jo... | closed | completed | true | 2 | [] | [] | 2026-03-11T02:46:47Z | 2026-04-27T04:52:19Z | 2026-04-24T22:14:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yangguangjin | 93,584,387 | U_kgDOBZP8Aw | User | false |
openclaw/openclaw | 4,055,364,315 | I_kwDOQb6kR87xt_Lb | 42,702 | https://github.com/openclaw/openclaw/issues/42702 | https://api.github.com/repos/openclaw/openclaw/issues/42702 | [Bug]: Discord plugin fails with "fetch failed" when using proxy | ### Bug type
Regression (worked before, now fails)
### Summary
Discord plugin shows `ON | OK` but health check shows `fetch failed`.
- curl with proxy: ✅ Works
- OpenClaw message tool: ❌ Fails
Proxy setting (`channels.discord.proxy`) not applied to REST API calls.
### Steps to reproduce
1. Configure `cha... | closed | completed | true | 5 | [
"bug",
"regression"
] | [] | 2026-03-11T02:47:22Z | 2026-04-27T04:52:18Z | 2026-04-24T22:15:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kww | 20,219,439 | MDQ6VXNlcjIwMjE5NDM5 | User | false |
openclaw/openclaw | 4,055,386,622 | I_kwDOQb6kR87xuEn- | 42,710 | https://github.com/openclaw/openclaw/issues/42710 | https://api.github.com/repos/openclaw/openclaw/issues/42710 | [Bug]: Version 2026.3.8 reports an error when using skill tavely search: Error: exec denied: allovlist execution plan unavailable (unsupported platform) | ### Bug type
Regression (worked before, now fails)
### Summary
tools: exec failed stack:
Error: exec denied: allowlist execution plan unavailable (unsupported platform)
at processGatewayAllowlist (file:///C:/Users/liyp/AppData/Roaming/npm/node_modules/openclaw/dist/compact-D3emcZgv.js:31005:48)
at execute (f... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-11T02:56:21Z | 2026-04-27T04:52:18Z | 2026-04-24T22:15:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | liyangpengs | 27,406,260 | MDQ6VXNlcjI3NDA2MjYw | User | false |
openclaw/openclaw | 4,055,387,088 | I_kwDOQb6kR87xuEvQ | 42,711 | https://github.com/openclaw/openclaw/issues/42711 | https://api.github.com/repos/openclaw/openclaw/issues/42711 | [Feature Request] Subagent tool deny policy (web_search/web_fetch) | ## Current issue
For embedded subagents, even when systemPrompt explicitly says:
- do **not** use `web_search` / `web_fetch`
- use bash-based external tools (`gemini`, `curl`, etc.)
…the model still tends to call built-in tools first, effectively ignoring the prompt instruction.
## What we tried
1. Added strict proh... | closed | completed | true | 1 | [] | [] | 2026-03-11T02:56:33Z | 2026-04-27T04:52:17Z | 2026-04-24T22:15:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | u1631223-lang | 233,756,939 | U_kgDODe7ZCw | User | false |
openclaw/openclaw | 4,055,405,707 | I_kwDOQb6kR87xuJSL | 42,717 | https://github.com/openclaw/openclaw/issues/42717 | https://api.github.com/repos/openclaw/openclaw/issues/42717 | [Bug]: failed to use in tg after upgrade nodejs22 | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to 3.7, I upgrade my nodejs to 22.Then I cannot talk to my bot in tg.
### Steps to reproduce
1. upgrade nodejs to 22
2. upgrade openclaw in tencent ops with version 3.7
### Expected behavior
Work as usual
### Actual behavior
<img wi... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-11T03:03:23Z | 2026-04-27T04:52:16Z | 2026-04-24T22:15:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jslyzx | 5,653,706 | MDQ6VXNlcjU2NTM3MDY= | User | false |
openclaw/openclaw | 4,055,411,047 | I_kwDOQb6kR87xuKln | 42,721 | https://github.com/openclaw/openclaw/issues/42721 | https://api.github.com/repos/openclaw/openclaw/issues/42721 | [Bug]: openclaw models status shows expired OAuth profiles as ok when a refresh token exists | ### Bug type
Regression (worked before, now fails)
### Summary
openclaw models status can report an OAuth profile as ok and expires in 0m even when the stored expires timestamp is already in the past.
### Steps to reproduce
1. Configure an OAuth auth profile for openai-codex with a stored refresh token and an alre... | closed | completed | true | 3 | [
"bug",
"regression"
] | [
"vincentkoc"
] | 2026-03-11T03:05:07Z | 2026-04-27T04:31:01Z | 2026-04-24T22:15:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dholbrook86 | 4,501,147 | MDQ6VXNlcjQ1MDExNDc= | User | false |
openclaw/openclaw | 4,055,426,906 | I_kwDOQb6kR87xuOda | 42,725 | https://github.com/openclaw/openclaw/issues/42725 | https://api.github.com/repos/openclaw/openclaw/issues/42725 | [Bug]: Discord subagent session inherits incomplete deliveryContext, announce ends in silent ANNOUNCE_SKIP | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When a Discord channel session spawns a `runtime="subagent"` session, the child session can inherit only `deliveryContext: {"channel":"discord"}` without `to`/`accountId`, causing `sessions_send`/announce completion flow to end in `ANNOUNCE_S... | closed | completed | true | 2 | [] | [] | 2026-03-11T03:10:24Z | 2026-04-27T04:52:15Z | 2026-04-24T22:15:15Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | No898 | 82,420,070 | MDQ6VXNlcjgyNDIwMDcw | User | false |
openclaw/openclaw | 4,055,432,168 | I_kwDOQb6kR87xuPvo | 42,726 | https://github.com/openclaw/openclaw/issues/42726 | https://api.github.com/repos/openclaw/openclaw/issues/42726 | [Bug]: [tools] tools.profile (coding) allowlist contains unknown entries (apply_patch, memory_search, memory_get). These entries won't match any tool unless the plugin is enabled. | ### Bug type
Regression (worked before, now fails)
### Summary
[tools] tools.profile (coding) allowlist contains unknown entries (apply_patch, memory_search, memory_get). These entries won't match any tool unless the plugin is enabled.
### Steps to reproduce
run bacak
### Expected behavior
run back
### Actual b... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T03:11:50Z | 2026-04-27T04:52:15Z | 2026-04-24T22:15:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | qq2660400163 | 265,964,140 | U_kgDOD9pKbA | User | false |
openclaw/openclaw | 4,055,455,965 | I_kwDOQb6kR87xuVjd | 42,735 | https://github.com/openclaw/openclaw/issues/42735 | https://api.github.com/repos/openclaw/openclaw/issues/42735 | WhatsApp: message.send rejects valid E.164 target while auto-reply to same contact succeeds | ## Summary
In WhatsApp channel, proactive outbound sends via `message.send` fail with:
`Error: Delivering to WhatsApp requires target <E.164|group JID>`
However, inbound messages from the same contact are received successfully, and auto-replies to that same contact are sent successfully.
This suggests a validation/n... | closed | completed | true | 2 | [] | [] | 2026-03-11T03:19:16Z | 2026-04-27T04:52:14Z | 2026-04-24T22:15:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dondyb | 6,231,498 | MDQ6VXNlcjYyMzE0OTg= | User | false |
openclaw/openclaw | 4,055,464,490 | I_kwDOQb6kR87xuXoq | 42,739 | https://github.com/openclaw/openclaw/issues/42739 | https://api.github.com/repos/openclaw/openclaw/issues/42739 | Explore Memvid as a more human-like memory layer for OpenClaw | I’d like to suggest exploring Memvid (<https://github.com/memvid/memvid>) as a memory-layer reference/backend for OpenClaw.
## Why this seems relevant
A lot of agent memory systems today are still shaped like retrieval systems first: embeddings, chunks, similarity search, summaries, etc.
That is useful, but it does no... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T03:21:40Z | 2026-04-27T04:52:13Z | 2026-04-24T22:15:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | i0ivi0i | 105,140,938 | U_kgDOBkRSyg | User | false |
openclaw/openclaw | 4,055,465,482 | I_kwDOQb6kR87xuX4K | 42,740 | https://github.com/openclaw/openclaw/issues/42740 | https://api.github.com/repos/openclaw/openclaw/issues/42740 | Cron jobs fire spontaneously at wrong time (no restart) — possible DST/UTC boundary bug | ## Description
Cron jobs scheduled with `tz: "America/Chicago"` fired ~9 hours early from a **running gateway** (no restart, no SIGUSR1). This is distinct from the restart-triggered cron dump fixed in #40232.
## Environment
- **OpenClaw version:** v2026.3.8 (npm, stable channel)
- **OS:** Ubuntu, `Etc/UTC` system ti... | closed | completed | true | 3 | [] | [] | 2026-03-11T03:21:58Z | 2026-04-27T04:52:12Z | 2026-04-24T22:15:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | latentoperator | 216,301,595 | U_kgDODOSAGw | User | false |
openclaw/openclaw | 4,055,471,876 | I_kwDOQb6kR87xuZcE | 42,744 | https://github.com/openclaw/openclaw/issues/42744 | https://api.github.com/repos/openclaw/openclaw/issues/42744 | Feature Request: Auto-switch to vision model when image input detected | ## Problem
Currently, when users send images to the agent, the agent cannot automatically switch to a vision-capable model to process them. Users must manually change the model configuration and restart the gateway.
## Proposed Solution
Add a `visionModel` configuration option similar to the existing `pdfModel` feat... | closed | completed | true | 1 | [] | [] | 2026-03-11T03:23:45Z | 2026-04-27T04:52:12Z | 2026-04-24T22:15:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yezhaowei666 | 134,367,290 | U_kgDOCAJIOg | User | false |
openclaw/openclaw | 4,055,479,049 | I_kwDOQb6kR87xubMJ | 42,746 | https://github.com/openclaw/openclaw/issues/42746 | https://api.github.com/repos/openclaw/openclaw/issues/42746 | [Bug]: web_search failed: Brave Search API error (422) in openclaw 3.8 | ### Bug type
Regression (worked before, now fails)
### Summary
On Chinese locale setups, web_search sometimes sends search_lang=zh to Brave Search, but Brave only accepts zh-hans or zh-hant, causing 422 errors and downstream task failures/timeouts.
### Steps to reproduce
Install/use OpenClaw on a Chinese locale en... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-11T03:25:50Z | 2026-04-27T04:52:11Z | 2026-04-24T22:15:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ZaneH1992 | 38,309,495 | MDQ6VXNlcjM4MzA5NDk1 | User | false |
openclaw/openclaw | 4,055,512,818 | I_kwDOQb6kR87xujby | 42,753 | https://github.com/openclaw/openclaw/issues/42753 | https://api.github.com/repos/openclaw/openclaw/issues/42753 | Image tool broken on AutoClaw: sharp native module code signature mismatch (Team ID) | ## Bug Description
The `read` tool and `image` tool cannot process images when running OpenClaw via AutoClaw desktop app on macOS. The sharp native module fails to load due to macOS code signature mismatch.
## Environment
- **OS**: macOS 15 (Darwin 24.5.0, arm64)
- **AutoClaw**: Latest version
- **OpenClaw CLI**: 20... | closed | completed | true | 1 | [] | [] | 2026-03-11T03:36:44Z | 2026-04-27T04:52:10Z | 2026-04-24T22:15:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chenxi9921-littledeng | 265,956,085 | U_kgDOD9oq9Q | User | false |
openclaw/openclaw | 4,055,688,392 | I_kwDOQb6kR87xvOTI | 42,782 | https://github.com/openclaw/openclaw/issues/42782 | https://api.github.com/repos/openclaw/openclaw/issues/42782 | [Feature Request] Add health-monitor auto-reconnect for Telegram polling | ## Problem
Telegram polling can stop working after network issues, and unlike Discord, it doesn't have an automatic reconnection mechanism.
## Observed Behavior
- When network becomes unstable, Telegram polling stops receiving messages
- Messages accumulate on Telegram servers (pending_update_count increases)
- Disc... | closed | completed | true | 1 | [] | [] | 2026-03-11T04:24:36Z | 2026-04-27T04:52:09Z | 2026-04-24T22:15:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tyw332 | 52,136,134 | MDQ6VXNlcjUyMTM2MTM0 | User | false |
openclaw/openclaw | 4,055,713,016 | I_kwDOQb6kR87xvUT4 | 42,789 | https://github.com/openclaw/openclaw/issues/42789 | https://api.github.com/repos/openclaw/openclaw/issues/42789 | Discord thread-bound ACP child replies appear to be mirrored into the main Discord route transcript instead of the child session transcript | ## Summary
On a pure OpenClaw VPS deployment, Discord thread-bound ACP sessions are usable end-to-end, but the child session transcript/store projection appears to land in the main Discord route transcript instead of the ACP child session transcript.
This is **not** a report that ACP spawn/routing is completely broke... | closed | completed | true | 4 | [] | [] | 2026-03-11T04:31:44Z | 2026-04-27T04:52:09Z | 2026-04-24T22:15:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rx-tomo | 185,449,797 | U_kgDOCw29RQ | User | false |
openclaw/openclaw | 4,055,725,626 | I_kwDOQb6kR87xvXY6 | 42,791 | https://github.com/openclaw/openclaw/issues/42791 | https://api.github.com/repos/openclaw/openclaw/issues/42791 | Bug: Inbound media attachment path duplicated in [media attached] tag, causing duplicate image slots in LLM context | ## Summary
When a Telegram inbound message contains an image attachment, the `[media attached: ...]` tag is formatted with the local file path appearing **twice** — once as the `path` and once as the `url` field — resulting in two image slots being passed to the LLM (both pointing to the same file).
## Steps to Repro... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T04:34:38Z | 2026-04-27T04:52:08Z | 2026-04-24T22:15:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zooey-sys | 227,112,828 | U_kgDODYl3fA | User | false |
openclaw/openclaw | 4,055,765,058 | I_kwDOQb6kR87xvhBC | 42,794 | https://github.com/openclaw/openclaw/issues/42794 | https://api.github.com/repos/openclaw/openclaw/issues/42794 | OpenAI fc_* transcript sanitizers bypassed when contextEngine.assemble() replaces session messages | ## Summary
When a `contextEngine` plugin (e.g. `lossless-claw`) is active, `contextEngine.assemble()` runs **after** `sanitizeSessionHistory()` in the attempt loop. The assembled messages bypass the OpenAI-specific transcript sanitizers, which can re-introduce `call_id|fc_id` pairs that OpenAI Responses rejects.
Erro... | closed | completed | true | 2 | [] | [] | 2026-03-11T04:45:26Z | 2026-04-27T04:52:07Z | 2026-04-24T22:15:45Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Grynn | 212,880 | MDQ6VXNlcjIxMjg4MA== | User | false |
openclaw/openclaw | 4,055,821,048 | I_kwDOQb6kR87xvur4 | 42,808 | https://github.com/openclaw/openclaw/issues/42808 | https://api.github.com/repos/openclaw/openclaw/issues/42808 | Node pairing through SSH tunnel: 401 on WebSocket upgrade | I'm trying to pair two OpenClaw gateways as nodes to each other (bidirectional). Both are v2026.3.8. Gateway A (Windows) has `gateway.bind: "lan"` and `gateway.auth.mode: "token"`.
Gateway B (EC2 Linux) connects to Gateway A via SSH reverse tunnel (`ssh -R 18791:127.0.0.1:18789`), then runs:
```bash
OPENCLAW_GATEWAY_... | closed | not_planned | true | 2 | [] | [] | 2026-03-11T04:58:43Z | 2026-04-27T04:52:06Z | 2026-04-24T22:15:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kevin-freiburger | 130,078,314 | U_kgDOB8DWag | User | false |
openclaw/openclaw | 4,055,838,465 | I_kwDOQb6kR87xvy8B | 42,815 | https://github.com/openclaw/openclaw/issues/42815 | https://api.github.com/repos/openclaw/openclaw/issues/42815 | research: Discord channel-per-agent binding 可行性与实现路径 (ZZ-20260311-006) | 奏折: ZZ-20260311-006
## 背景
需求：单 Discord bot，guild server 内多个 channel 各绑不同 agent（channel-per-agent binding）。
## 研究结论
### 已完整支持（无需改动）
- resolveAgentRoute() 已有 binding.peer 层（最高优先级）
- Discord monitor buildDiscordRoutePeer() 已传 kind:"channel"/conversationId
- Config 格式：match.peer.{kind:"channel", id:"<channel_snowflake>"... | closed | completed | true | 2 | [] | [] | 2026-03-11T05:02:16Z | 2026-04-27T04:52:06Z | 2026-04-24T22:15:50Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Billmvp73 | 45,382,455 | MDQ6VXNlcjQ1MzgyNDU1 | User | false |
openclaw/openclaw | 4,055,890,898 | I_kwDOQb6kR87xv_vS | 42,827 | https://github.com/openclaw/openclaw/issues/42827 | https://api.github.com/repos/openclaw/openclaw/issues/42827 | [Feature]: Feature Request: Add exec tool for running terminal commands | ### Summary
I'd like to request adding an `exec` tool to allow the agent to run terminal commands directly.
### Problem to solve
- Run diagnostic commands like `openclaw doctor`
- Check versions with `openclaw --version`
- Execute custom scripts
### Proposed solution
Add a new tool that can execute shell commands... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-11T05:16:03Z | 2026-04-27T04:52:05Z | 2026-04-24T22:15:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lanshengzl | 131,418,670 | U_kgDOB9VKLg | User | false |
openclaw/openclaw | 4,055,926,865 | I_kwDOQb6kR87xwIhR | 42,838 | https://github.com/openclaw/openclaw/issues/42838 | https://api.github.com/repos/openclaw/openclaw/issues/42838 | [Bug]: disconnected 4008报错 | ### Bug type
Regression (worked before, now fails)
### Summary
前端页面报错
<img width="1220" height="457" alt="Image" src="https://github.com/user-attachments/assets/38379727-b9ac-4022-94a9-a2cf24372a02" />
后端日志打印
08:59:49 info gateway/ws {"subsystem":"gateway/ws"} webchat connected conn=fd27c824-a291-411f-821d-0ec6f634... | closed | not_planned | true | 8 | [
"bug",
"regression"
] | [] | 2026-03-11T05:25:02Z | 2026-04-27T04:52:04Z | 2026-04-24T22:15:55Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | caiheng1235-netizen | 259,093,216 | U_kgDOD3Fy4A | User | false |
openclaw/openclaw | 4,055,927,758 | I_kwDOQb6kR87xwIvO | 42,839 | https://github.com/openclaw/openclaw/issues/42839 | https://api.github.com/repos/openclaw/openclaw/issues/42839 | [Bug] Windows: openclaw agent --local returns 401 Missing Authentication header after clean reset and fresh onboarding | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
On Windows 11, `openclaw agent --local --agent main` consistently fails with `401 Missing Authentication header` even after a full reset, fresh onboarding, new OpenRouter API key, and a running local gateway.
### Steps to reproduce
1. Run... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T05:25:12Z | 2026-04-27T04:52:03Z | 2026-04-24T22:15:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | huyouwei9583-collab | 267,238,265 | U_kgDOD-27eQ | User | false |
openclaw/openclaw | 4,056,007,303 | I_kwDOQb6kR87xwcKH | 42,850 | https://github.com/openclaw/openclaw/issues/42850 | https://api.github.com/repos/openclaw/openclaw/issues/42850 | [Bug] Token usage 统计在 2026.3.8 版本中始终返回 0 | # OpenClaw Token Usage 统计 Bug 报告
## 问题概述
在 OpenClaw 2026.3.8 版本中，token 使用量统计功能失效，所有 API 调用的 usage 数据始终显示为 0。
## 环境信息
| 项目 | 值 |
|------|-----|
| OpenClaw 版本 | 2026.3.8 (commit: 3caab92) |
| 操作系统 | macOS Darwin 25.1.0 (x64) |
| Node.js 版本 | v24.13.1 |
| 模型 | qwen-portal/kimi-k2.5 |
| 模型 API | openai-completions |
#... | closed | completed | true | 3 | [] | [] | 2026-03-11T05:44:17Z | 2026-04-27T04:52:02Z | 2026-04-24T22:16:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | CaiJian0834 | 31,196,742 | MDQ6VXNlcjMxMTk2NzQy | User | false |
openclaw/openclaw | 4,056,087,618 | I_kwDOQb6kR87xwvxC | 42,864 | https://github.com/openclaw/openclaw/issues/42864 | https://api.github.com/repos/openclaw/openclaw/issues/42864 | Feature: Session-level circuit breaker (pause after N consecutive model errors) | ## Problem
When a session accumulates context that causes repeated model timeouts, the system retries indefinitely. Auth profile cooldowns help with auth/rate-limit errors, but when the root cause is **context size exceeding what the model can process within the timeout**, rotating profiles doesn't help — the next pro... | closed | completed | true | 1 | [] | [] | 2026-03-11T06:03:23Z | 2026-04-27T04:52:01Z | 2026-04-24T22:16:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pescaohq-web | 257,331,994 | U_kgDOD1aTGg | User | false |
openclaw/openclaw | 4,056,088,550 | I_kwDOQb6kR87xwv_m | 42,866 | https://github.com/openclaw/openclaw/issues/42866 | https://api.github.com/repos/openclaw/openclaw/issues/42866 | Feature: Pre-send context token estimation with proactive compaction | ## Problem
Auto-compaction triggers when a session **nears or exceeds** the context window. But by that point, the request has already been assembled and may be sent to the provider, which then times out trying to process an oversized context. The timeout wastes 300+ seconds of compute.
## Current behavior
1. Sessio... | closed | completed | true | 1 | [] | [] | 2026-03-11T06:03:37Z | 2026-04-27T04:52:00Z | 2026-04-24T22:16:06Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pescaohq-web | 257,331,994 | U_kgDOD1aTGg | User | false |
openclaw/openclaw | 4,056,125,824 | I_kwDOQb6kR87xw5GA | 42,875 | https://github.com/openclaw/openclaw/issues/42875 | https://api.github.com/repos/openclaw/openclaw/issues/42875 | openclaw logs --follow Cannot View Gateway Real-time Logs | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The openclaw logs --follow command reads logs from the CLI process instead of the Gateway process, causing users to miss real-time message processing logs.
### Steps to reproduce
1.Start Gateway via systemd service on March 10 (e.g., open... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T06:11:54Z | 2026-04-29T02:11:31Z | 2026-04-24T22:16:12Z | NONE | null | 20260429T030016Z | 2026-04-29T03:00:16Z | lck-001 | 82,191,060 | MDQ6VXNlcjgyMTkxMDYw | User | false |
openclaw/openclaw | 4,056,203,114 | I_kwDOQb6kR87xxL9q | 42,891 | https://github.com/openclaw/openclaw/issues/42891 | https://api.github.com/repos/openclaw/openclaw/issues/42891 | [Feature Request] 503 Service Unavailable should trigger model fallback | ## Description
Currently, when a model provider returns HTTP 503 (Service Unavailable), OpenClaw does not automatically fall back to the next available model in the fallback list. This results in "无可用供应商" (no available provider) switching errors instead of gracefully to a backup model.
## Expected Behavior
When a mo... | closed | completed | true | 1 | [] | [] | 2026-03-11T06:29:24Z | 2026-04-27T04:51:58Z | 2026-04-24T22:16:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | myhzy | 33,489,731 | MDQ6VXNlcjMzNDg5NzMx | User | false |
openclaw/openclaw | 4,056,253,722 | I_kwDOQb6kR87xxYUa | 42,902 | https://github.com/openclaw/openclaw/issues/42902 | https://api.github.com/repos/openclaw/openclaw/issues/42902 | Feishu: message tool filePath parameter sends successfully but file not received | ## Bug Description
When using the `message` tool with `filePath` parameter to send files to a Feishu chat, the tool reports success but the recipient does not receive the file.
## Environment
- OpenClaw: 2026.3.8 (3caab92)
- Channel: Feishu (飞书)
- Chat type: direct (DM)
## Steps to Reproduce
1. Send a message with... | closed | completed | true | 2 | [] | [] | 2026-03-11T06:41:55Z | 2026-04-27T04:51:57Z | 2026-04-24T22:16:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chenxi9921-littledeng | 265,956,085 | U_kgDOD9oq9Q | User | false |
openclaw/openclaw | 4,056,254,035 | I_kwDOQb6kR87xxYZT | 42,903 | https://github.com/openclaw/openclaw/issues/42903 | https://api.github.com/repos/openclaw/openclaw/issues/42903 | [Bug]: WebChat does not send images to non-default agents (e.g., m2) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When sending images through the web interface, images are only sent to the default main agent, not to other agents (like m2).
### Steps to reproduce
## Steps to Reproduce
1. Configure multiple agents (main and m2)
2. Send an image via ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T06:42:02Z | 2026-04-27T04:51:57Z | 2026-04-24T22:16:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhangfan-algo | 47,747,764 | MDQ6VXNlcjQ3NzQ3NzY0 | User | false |
openclaw/openclaw | 4,056,277,117 | I_kwDOQb6kR87xxeB9 | 42,912 | https://github.com/openclaw/openclaw/issues/42912 | https://api.github.com/repos/openclaw/openclaw/issues/42912 | Tool failure notifications are misleading when agent retries successfully | ## Problem
When a tool call fails (e.g. `write` to a protected path like `/usr/local/bin/`) and the agent immediately retries successfully on a different path, the UI still shows a failure notification (⚠️ ✍️ Write: failed) without any indication that the operation ultimately succeeded.
This creates false alarms and ... | closed | completed | true | 1 | [] | [] | 2026-03-11T06:47:53Z | 2026-04-27T04:51:56Z | 2026-04-24T22:16:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sabo961 | 191,984 | MDQ6VXNlcjE5MTk4NA== | User | false |
openclaw/openclaw | 4,056,307,654 | I_kwDOQb6kR87xxlfG | 42,918 | https://github.com/openclaw/openclaw/issues/42918 | https://api.github.com/repos/openclaw/openclaw/issues/42918 | 3.8 regression: in-session gateway restart from Control UI can drop session and appear hung until manual terminal recovery | ## Summary
After updating to OpenClaw 2026.3.8, restarting Gateway from an active agent/control session can terminate the live session and fail to recover from the operator perspective, requiring manual terminal recovery (`openclaw gateway start/status`).
This worked more reliably pre-upgrade in the same environment.
... | closed | completed | true | 2 | [] | [] | 2026-03-11T06:56:32Z | 2026-04-27T04:51:55Z | 2026-04-24T22:16:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aimacbook0 | 262,985,145 | U_kgDOD6zVuQ | User | false |
openclaw/openclaw | 4,056,373,094 | I_kwDOQb6kR87xx1dm | 42,926 | https://github.com/openclaw/openclaw/issues/42926 | https://api.github.com/repos/openclaw/openclaw/issues/42926 | 🦞 虾缘 - AI相亲交友平台欢迎测试 | ## 虾缘是什么？
全球首个AI Agent婚恋交友平台，让小龙虾们（AI Agents）找到志同道合的伙伴！
## 背景
不只是相亲，更是一个AI社交的实验场——测试AI之间的匹配、对话、社交能力。
## 如何参与？
1. 访问：http://124.221.234.22:13106
2. 体验AI相亲匹配
3. 欢迎反馈bug和建议
## 愿景
让每只小龙虾找到命中注定的另一半！🦞❤️
欢迎Star和测试！ | closed | not_planned | true | 3 | [] | [] | 2026-03-11T07:12:55Z | 2026-04-24T22:16:29Z | 2026-04-24T22:16:29Z | NONE | null | 20260425T000026Z | 2026-04-25T00:00:26Z | OwenSharpKing | 31,701,822 | MDQ6VXNlcjMxNzAxODIy | User | false |
openclaw/openclaw | 4,056,378,613 | I_kwDOQb6kR87xx2z1 | 42,930 | https://github.com/openclaw/openclaw/issues/42930 | https://api.github.com/repos/openclaw/openclaw/issues/42930 | [Bug]: After installing the window, the configuration linked to the model in the gpustack, but the chat box returned an empty response. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
I installed the latest openclaw 2026.3.8 and configured gpustack 2.1.0. Through the chat function on the openclaw page, there were no errors on the backend, and the log call of gpustack was also successful.
gpustack logs =====
srv par... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T07:14:17Z | 2026-04-27T04:51:54Z | 2026-04-24T22:16:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cuihangrui | 86,222,753 | MDQ6VXNlcjg2MjIyNzUz | User | false |
openclaw/openclaw | 4,056,402,236 | I_kwDOQb6kR87xx8k8 | 42,938 | https://github.com/openclaw/openclaw/issues/42938 | https://api.github.com/repos/openclaw/openclaw/issues/42938 | 🦞 虾缘 - AI相亲交友平台欢迎测试 | ## 大家好！
推荐一个有趣的AI社交实验项目：**虾缘** - 全球首个AI Agent婚恋交友平台。
### 🎯 访问地址
- 平台：http://124.221.234.22:13106
- 落地页：http://124.221.234.22:13108
### 💡 为什么做这个？
- 探索AI之间的社交可能性
- 娱乐性 + 营销噱头
- 展示OpenClaw生态能力
### 🤝 欢迎参与！
- 体验AI相亲
- 反馈建议
- Star支持！
#AI #社交 #OpenClaw | closed | not_planned | true | 11 | [] | [] | 2026-03-11T07:20:21Z | 2026-04-27T04:51:53Z | 2026-04-24T22:16:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | OwenSharpKing | 31,701,822 | MDQ6VXNlcjMxNzAxODIy | User | false |
openclaw/openclaw | 4,056,425,499 | I_kwDOQb6kR87xyCQb | 42,942 | https://github.com/openclaw/openclaw/issues/42942 | https://api.github.com/repos/openclaw/openclaw/issues/42942 | Regression in v2026.3.8: `openai-codex/gpt-5.4` times out, while v2026.3.7 works in the same environment | ## Summary
After upgrading OpenClaw from `v2026.3.7` to `v2026.3.8`, `openai-codex/gpt-5.4` started failing with `LLM request timed out.` in my environment.
Downgrading back to `v2026.3.7` immediately restored normal behavior.
This looks like a regression in `v2026.3.8`, likely related to the `openai-codex` transpor... | closed | completed | true | 3 | [] | [] | 2026-03-11T07:26:39Z | 2026-04-27T04:51:53Z | 2026-04-24T22:16:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | PangdDa | 56,503,080 | MDQ6VXNlcjU2NTAzMDgw | User | false |
openclaw/openclaw | 4,056,463,052 | I_kwDOQb6kR87xyLbM | 42,947 | https://github.com/openclaw/openclaw/issues/42947 | https://api.github.com/repos/openclaw/openclaw/issues/42947 | contextTokens shows 0% even when session has substantial content | ## Bug Description
The `contextTokens` field in session status shows 0% (0 tokens) even when the session JSONL file has substantial content (e.g., 1.7MB, 520 lines).
## Reproduction Steps
1. Have a long-running session with many messages
2. Run `/status` or check session status
3. Observe that context shows `0/205k ... | closed | completed | true | 1 | [] | [] | 2026-03-11T07:35:24Z | 2026-04-27T04:51:51Z | 2026-04-24T22:16:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tociyou1995 | 150,312,743 | U_kgDOCPWXJw | User | false |
openclaw/openclaw | 4,056,521,667 | I_kwDOQb6kR87xyZvD | 42,956 | https://github.com/openclaw/openclaw/issues/42956 | https://api.github.com/repos/openclaw/openclaw/issues/42956 | Full document of browser control APIs | ### Summary
Hello. Where is the full document of APIs about browser control API?
### Problem to solve
I've red https://docs.openclaw.ai/tools/browser#control-api-optional, but can't found the full doucment about the details or request body etc. of the APIs
### Proposed solution
Is there the full document about br... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-11T07:48:34Z | 2026-04-27T04:51:51Z | 2026-04-24T22:16:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | linzaihui | 26,833,023 | MDQ6VXNlcjI2ODMzMDIz | User | false |
openclaw/openclaw | 4,056,544,407 | I_kwDOQb6kR87xyfSX | 42,958 | https://github.com/openclaw/openclaw/issues/42958 | https://api.github.com/repos/openclaw/openclaw/issues/42958 | [Bug] Dashscope API usage not being recorded | ## Bug Description
Usage/cost for dashscope always shows sh.00 even though API returns usage.
## Investigation
- Dashscope returns: prompt_tokens, completion_tokens, total_tokens
- normalizeUsage() in OpenClaw supports these fields
- But session logs show all zeros
## Expected
Usage should be tracked with configured... | closed | completed | true | 2 | [] | [] | 2026-03-11T07:54:00Z | 2026-04-27T04:51:50Z | 2026-04-24T22:16:45Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | QR-0W | 84,510,337 | MDQ6VXNlcjg0NTEwMzM3 | User | false |
openclaw/openclaw | 4,056,546,200 | I_kwDOQb6kR87xyfuY | 42,960 | https://github.com/openclaw/openclaw/issues/42960 | https://api.github.com/repos/openclaw/openclaw/issues/42960 | [Bug]: Cron jobs enqueued but never execute — lane never dispatches, runningAtMs written on enqueue causes permanent stale marker loop | ### Bug type
Regression (worked before, now fails)
### Summary
Cron jobs return enqueued: true from both scheduled triggers and openclaw cron run but never actually execute — no session start, no LLM call, no tool activity appears in logs.
### Steps to reproduce
1. Have one or more cron jobs configured in jobs.jso... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-11T07:54:25Z | 2026-04-27T04:51:49Z | 2026-04-24T22:16:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Captain-Scarlet | 225,561,567 | U_kgDODXHL3w | User | false |
openclaw/openclaw | 4,056,576,123 | I_kwDOQb6kR87xynB7 | 42,963 | https://github.com/openclaw/openclaw/issues/42963 | https://api.github.com/repos/openclaw/openclaw/issues/42963 | Anthropic extended thinking: 'thinking blocks cannot be modified' error on session replay | ## Summary
When using Anthropic models with extended thinking enabled (`thinking=low` or higher), subsequent messages in the same session fail with:
```
LLM request rejected: messages.19.content.1: thinking or redacted_thinking blocks in the latest assistant message cannot be modified. These blocks must remain as the... | closed | completed | true | 2 | [] | [] | 2026-03-11T08:01:30Z | 2026-04-27T04:51:48Z | 2026-04-24T22:16:50Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | JIN-Li-Hub | 267,252,100 | U_kgDOD-3xhA | User | false |
openclaw/openclaw | 4,056,728,142 | I_kwDOQb6kR87xzMJO | 42,980 | https://github.com/openclaw/openclaw/issues/42980 | https://api.github.com/repos/openclaw/openclaw/issues/42980 | [Feature]: Dashboard - browse archived/historical session transcripts | ### Summary
Add UI to browse and search archived session transcripts in the dashboard.
### Problem to solve
Currently the dashboard only shows **active sessions**. Archived sessions (from `/reset`, `/new`, or natural expiration) are only accessible via filesystem at `~/.openclaw/agents/<agent>/sessions/*.jsonl`.
Th... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-11T08:34:08Z | 2026-04-27T04:51:47Z | 2026-04-24T22:16:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | songke | 4,110,101 | MDQ6VXNlcjQxMTAxMDE= | User | false |
openclaw/openclaw | 4,056,803,274 | I_kwDOQb6kR87xzefK | 42,990 | https://github.com/openclaw/openclaw/issues/42990 | https://api.github.com/repos/openclaw/openclaw/issues/42990 | [Bug]: Support from the Chinese Community & Inquiry Regarding OpenClaw Licensing | ### Bug type
Regression (worked before, now fails)
### Summary
Dear Peter and the OpenClaw Team,
I am writing to you as an enthusiastic user and supporter from China. First of all, thank you for creating such a groundbreaking AI Agent. Your "Lobster" has truly redefined how we interact with our digital workspace, a... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T08:49:43Z | 2026-04-27T04:51:46Z | 2026-04-24T22:16:55Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | wenglei | 18,674,192 | MDQ6VXNlcjE4Njc0MTky | User | false |
openclaw/openclaw | 4,056,837,339 | I_kwDOQb6kR87xzmzb | 42,993 | https://github.com/openclaw/openclaw/issues/42993 | https://api.github.com/repos/openclaw/openclaw/issues/42993 | [Feature Request] Telegram: Support user interrupt during long-running agent tasks | ## Problem
When the OpenClaw agent starts a long-running task (e.g., browser automation, multi-step tool calls), user messages sent during that time are queued and not processed until the task completes. There is currently no way for the user to interrupt or cancel an in-progress task.
## Expected Behavior
Users shoul... | closed | completed | true | 2 | [] | [] | 2026-03-11T08:57:06Z | 2026-04-27T04:51:46Z | 2026-04-24T22:16:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chphch | 19,252,487 | MDQ6VXNlcjE5MjUyNDg3 | User | false |
openclaw/openclaw | 4,056,862,307 | I_kwDOQb6kR87xzs5j | 43,002 | https://github.com/openclaw/openclaw/issues/43002 | https://api.github.com/repos/openclaw/openclaw/issues/43002 | Feature: Memory Consolidation Mechanism (Hippocampal-Cortical Transfer) | ## Problem
OpenClaw's memory system uses a two-layer architecture:
- `memory/YYYY-MM-DD.md` — daily append-only logs (short-term)
- `MEMORY.md` — curated long-term memory
However, the transfer from short-term to long-term memory is entirely manual. The agent must be explicitly reminded to consolidate memories. There ... | closed | completed | true | 2 | [] | [] | 2026-03-11T09:02:46Z | 2026-04-27T04:51:45Z | 2026-04-24T22:17:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,056,862,500 | I_kwDOQb6kR87xzs8k | 43,003 | https://github.com/openclaw/openclaw/issues/43003 | https://api.github.com/repos/openclaw/openclaw/issues/43003 | Bug: memory-core plugin silently returns null on tool creation failure | ## Description
In the `memory-core` plugin, when `memory_search` or `memory_get` tool creation fails, the plugin silently returns `null` with no logging:
```javascript
if (!memorySearchTool || !memoryGetTool) {
return null; // Silent failure — no warning, no log
}
```
## Expected Behavior
- A warning log should b... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T09:02:49Z | 2026-04-27T04:51:44Z | 2026-04-24T22:17:03Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,056,864,826 | I_kwDOQb6kR87xztg6 | 43,004 | https://github.com/openclaw/openclaw/issues/43004 | https://api.github.com/repos/openclaw/openclaw/issues/43004 | Feature: Cron job retry with exponential backoff | ## Problem
When a cron job fails, there is no automatic retry mechanism. The job simply fails and the user may not be notified.
## Proposal
Add configurable retry behavior to cron jobs:
```json5
{
schedule: { kind: 'cron', expr: '0 9 * * *' },
retry: {
maxAttempts: 3,
backoffMs: [30000, 60000, 120000], ... | closed | completed | true | 1 | [] | [] | 2026-03-11T09:03:15Z | 2026-04-27T04:51:44Z | 2026-04-24T22:17:06Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,056,871,779 | I_kwDOQb6kR87xzvNj | 43,007 | https://github.com/openclaw/openclaw/issues/43007 | https://api.github.com/repos/openclaw/openclaw/issues/43007 | [Feature Request] Telegram: Resolve and include forum topic name in inbound message metadata | ## Problem
When OpenClaw receives messages from a Telegram forum group (supergroup with topics enabled), the inbound message metadata only includes `topic_id` (the internal Telegram thread ID), but not the topic name.
This is because the Telegram Bot API does not include the topic name in the `Message` object - it onl... | closed | completed | true | 2 | [] | [] | 2026-03-11T09:04:45Z | 2026-04-27T04:51:43Z | 2026-04-24T22:17:08Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chphch | 19,252,487 | MDQ6VXNlcjE5MjUyNDg3 | User | false |
openclaw/openclaw | 4,056,898,785 | I_kwDOQb6kR87xz1zh | 43,012 | https://github.com/openclaw/openclaw/issues/43012 | https://api.github.com/repos/openclaw/openclaw/issues/43012 | [Bug]: Bug Report: Browser Automation - Fill/Click Operations Timeout on Textbox Elements | ### Bug type
Regression (worked before, now fails)
### Summary
Bug Report: Browser Automation - Fill/Click Operations Timeout on Textbox Elements
Environment
• OpenClaw Version: v2026.3.8
• Comparison Version: v2026.2.26
• Browser: Chrome Extension Relay Mode
Problem Description
In v2026.3.8, the fill and click ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T09:09:55Z | 2026-04-27T04:51:42Z | 2026-04-24T22:17:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | vinhogit | 256,755,280 | U_kgDOD03GUA | User | false |
openclaw/openclaw | 4,056,935,663 | I_kwDOQb6kR87xz-zv | 43,018 | https://github.com/openclaw/openclaw/issues/43018 | https://api.github.com/repos/openclaw/openclaw/issues/43018 | Custom providers can’t use OpenAI Responses API (/v1/responses) | ## Summary
Custom providers configured via `models.providers` with an OpenAI‑compatible base URL can only be used with the chat/completions adapter. There is no way to use the OpenAI **Responses** API (`/v1/responses`) for custom providers, which blocks providers that only implement Responses (e.g., sub2api).
## Probl... | closed | completed | true | 8 | [] | [] | 2026-03-11T09:16:03Z | 2026-04-27T04:51:41Z | 2026-04-24T22:17:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AaronL725 | 76,561,968 | MDQ6VXNlcjc2NTYxOTY4 | User | false |
openclaw/openclaw | 4,056,968,733 | I_kwDOQb6kR87x0G4d | 43,023 | https://github.com/openclaw/openclaw/issues/43023 | https://api.github.com/repos/openclaw/openclaw/issues/43023 | backtest-expert: Missing mandatory stock ticker verification step | # Problem
When a user provides a stock name (e.g., "大连电瓷"), the agent may guess the ticker code from memory without verification. This can lead to analyzing the wrong stock entirely (e.g., mapping "大连电瓷" to 603688 instead of the correct 002606).
# Proposed fix
Add a "Data Identity Verification" section to `reference... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T09:21:56Z | 2026-04-27T04:51:41Z | 2026-04-24T22:17:16Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | c9841013 | 6,790,387 | MDQ6VXNlcjY3OTAzODc= | User | false |
openclaw/openclaw | 4,056,981,974 | I_kwDOQb6kR87x0KHW | 43,025 | https://github.com/openclaw/openclaw/issues/43025 | https://api.github.com/repos/openclaw/openclaw/issues/43025 | Issue #168: Cross-pod API connectivity blocked - port 19876 not exposed in K8s Service | ## Problem
Cross-pod API connectivity is blocked. Other agents (e.g., Ripley) cannot reach `agent-bob-li.team-d-squad.svc.cluster.local:19876`.
## Root Cause Analysis
1. **project-api server.js** defaults to port 8080:
```javascript
const PORT = process.env.PORT || 8080;
```
2. **HEARTBEAT.md** expects ... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T09:24:18Z | 2026-04-27T04:51:40Z | 2026-04-24T22:17:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dsquadteam | 266,730,731 | U_kgDOD-X86w | User | false |
openclaw/openclaw | 4,056,996,372 | I_kwDOQb6kR87x0NoU | 43,027 | https://github.com/openclaw/openclaw/issues/43027 | https://api.github.com/repos/openclaw/openclaw/issues/43027 | Context tokens counter shows 0 after model switch | ### Bug type
Regression (worked before, now fails)
### Summary
After switching models, the `/status` context token counter displays 0 even when the session file contains significant content (635KB+).
### Steps to reproduce
1. Start a session with model A (e.g., bailian/qwen3.5-plus)
2. Have a conversation (context... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T09:26:51Z | 2026-04-27T04:51:39Z | 2026-04-24T22:17:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AugustoShaowu | 262,710,784 | U_kgDOD6imAA | User | false |
openclaw/openclaw | 4,056,999,979 | I_kwDOQb6kR87x0Ogr | 43,029 | https://github.com/openclaw/openclaw/issues/43029 | https://api.github.com/repos/openclaw/openclaw/issues/43029 | [Bug]: Unhandled API in mapOptionsForApi preventing any agent response (0.1.7 & 0.1.8-fix.3) | ### Bug type
Regression (worked before, now fails)
### Summary
Describe the bug
A clear and concise description of what the bug is. 无论通过何种方式（飞书消息、TUI 界面、或 openclaw-cn agent --local 命令）与 Agent 交互，OpenClaw 都无法生成任何回复。进程在尝试调用模型时崩溃，日志中反复出现关键错误： Unhandled promise rejection: Error: Unhandled API in mapOptionsForApi: unde... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T09:27:27Z | 2026-04-27T04:51:38Z | 2026-04-24T22:17:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | bounca001 | 265,178,863 | U_kgDOD85O7w | User | false |
openclaw/openclaw | 4,057,007,108 | I_kwDOQb6kR87x0QQE | 43,031 | https://github.com/openclaw/openclaw/issues/43031 | https://api.github.com/repos/openclaw/openclaw/issues/43031 | Docs: Add security best practices guide for end users | ## Gap
The security documentation currently includes an excellent MITRE ATLAS threat model and formal verification notes, but lacks a practical, user-facing security best practices guide.
## Proposed Content: `docs/security/best-practices.md`
1. **Initial Setup Hardening**
- Token management (rotation, minimal sc... | closed | completed | true | 1 | [] | [] | 2026-03-11T09:28:38Z | 2026-04-27T04:51:37Z | 2026-04-24T22:17:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,057,071,106 | I_kwDOQb6kR87x0f4C | 43,049 | https://github.com/openclaw/openclaw/issues/43049 | https://api.github.com/repos/openclaw/openclaw/issues/43049 | Hot reload triggers SIGTERM causing gateway restart when modifying channel config | Repeated Steps:
Modify any feishu channel configuration (e.g., openclaw channels remove --channel feishu --account main --delete then re-add)
Gateway detects config change and attempts hot reload
After hot reload completes, gateway receives SIGTERM and shuts down
LaunchAgent restarts gateway (if KeepAlive enabled)
Da... | closed | completed | true | 1 | [] | [] | 2026-03-11T09:40:21Z | 2026-04-27T04:51:37Z | 2026-04-24T22:17:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | leekenny | 1,638,965 | MDQ6VXNlcjE2Mzg5NjU= | User | false |
openclaw/openclaw | 4,057,178,849 | I_kwDOQb6kR87x06Lh | 43,058 | https://github.com/openclaw/openclaw/issues/43058 | https://api.github.com/repos/openclaw/openclaw/issues/43058 | Config spontaneously becomes invalid with unrecognized 'enabled' key in heartbeat | ## Bug Description
Config spontaneously became invalid without any user changes. Gateway failed to start.
## Error
Config invalid
Problem:
- agents.defaults.heartbeat: Unrecognized key: "enabled"
## Steps to Reproduce
1. OpenClaw running normally (2026.3.1)
2. No config changes made by user
3. Next day: gateway won't... | closed | not_planned | true | 3 | [] | [] | 2026-03-11T10:00:28Z | 2026-04-27T04:51:36Z | 2026-04-24T22:17:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | curlcomplex | 4,457,907 | MDQ6VXNlcjQ0NTc5MDc= | User | false |
openclaw/openclaw | 4,057,180,248 | I_kwDOQb6kR87x06hY | 43,059 | https://github.com/openclaw/openclaw/issues/43059 | https://api.github.com/repos/openclaw/openclaw/issues/43059 | [Feature]: openclaw cannot read/write local files (SOUL.md, MEMORY.md) on macOS 12, info lost after session ends | ### Summary
System: macOS 12
openclaw version: 2026.3.8
Node.js: 24
npm: 1.9
After installation and configuration, I found that openclaw cannot read or write files such as SOUL.md and MEMORY.md. When a session ends, all information is lost. How can I resolve this issue so that openclaw can automatically update and ma... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-11T10:00:41Z | 2026-04-27T04:51:35Z | 2026-04-24T22:17:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | harbourside | 3,304,769 | MDQ6VXNlcjMzMDQ3Njk= | User | false |
openclaw/openclaw | 4,057,213,035 | I_kwDOQb6kR87x1Chr | 43,066 | https://github.com/openclaw/openclaw/issues/43066 | https://api.github.com/repos/openclaw/openclaw/issues/43066 | [Feature]: tools.profile (coding) allowlist contains unknown entries (apply_patch). These entries won't match any tool unless the plugin is enabled. | ### Summary
<img width="1900" height="876" alt="Image" src="https://github.com/user-attachments/assets/99335b24-be36-4729-8a0e-52d91aca348a" />
<img width="1477" height="295" alt="Image" src="https://github.com/user-attachments/assets/1540e7f4-b4e2-4a22-8734-ccd1d31d6393" />
为啥这样一直报错啊
### Problem to solve
这块claw总是... | closed | completed | true | 3 | [
"enhancement"
] | [] | 2026-03-11T10:06:53Z | 2026-04-27T04:51:34Z | 2026-04-24T22:17:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | yancy-coder | 255,520,048 | U_kgDODzrtMA | User | false |
openclaw/openclaw | 4,057,248,977 | I_kwDOQb6kR87x1LTR | 43,072 | https://github.com/openclaw/openclaw/issues/43072 | https://api.github.com/repos/openclaw/openclaw/issues/43072 | feat(tts): send Telegram voice messages as OGG/Opus (sendVoice) instead of audio files | ## Problem
When TTS responds via Telegram, audio is sent as an audio file (MP3 via `sendAudio`), not as a Telegram voice message (waveform circle via `sendVoice`).
## Expected behavior
TTS replies in Telegram should be delivered as voice notes — the round waveform UI that plays inline — not as file attachments.
## ... | closed | completed | true | 2 | [] | [] | 2026-03-11T10:13:35Z | 2026-04-27T04:51:34Z | 2026-04-24T22:17:40Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | antlarionov | 17,770,042 | MDQ6VXNlcjE3NzcwMDQy | User | false |
openclaw/openclaw | 4,057,300,356 | I_kwDOQb6kR87x1X2E | 43,078 | https://github.com/openclaw/openclaw/issues/43078 | https://api.github.com/repos/openclaw/openclaw/issues/43078 | Skill env vars (skills.entries.<skill>.env) not injected into exec child processes | ## Summary
Skill-level environment variables defined in `skills.entries.<skill>.env` are injected into the main OpenClaw `process.env` but are **not forwarded** to child processes spawned by the `exec` tool.
This means any skill that runs an external script via `exec` (e.g., `node search.mjs`) will not receive the co... | closed | completed | true | 1 | [] | [] | 2026-03-11T10:22:39Z | 2026-04-27T04:51:33Z | 2026-04-24T22:17:42Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | eclipsemindai-spec | 263,372,340 | U_kgDOD7K-NA | User | false |
openclaw/openclaw | 4,057,314,403 | I_kwDOQb6kR87x1bRj | 43,084 | https://github.com/openclaw/openclaw/issues/43084 | https://api.github.com/repos/openclaw/openclaw/issues/43084 | Feature Request: Add configurable typing indicator keepalive interval for Feishu channel | ## Feature Request: Add configurable typing indicator keepalive interval for Feishu channel
### Description
Currently, the Feishu typing indicator (show when bot is streaming a response) refreshes every 3 seconds via the `keepaliveIntervalMs` parameter in `createTypingCallbacks`. This causes excessive API calls - for... | closed | completed | true | 1 | [] | [] | 2026-03-11T10:24:58Z | 2026-04-27T04:51:32Z | 2026-04-24T22:17:45Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | xuxiashi | 7,801,357 | MDQ6VXNlcjc4MDEzNTc= | User | false |
openclaw/openclaw | 4,057,324,654 | I_kwDOQb6kR87x1dxu | 43,086 | https://github.com/openclaw/openclaw/issues/43086 | https://api.github.com/repos/openclaw/openclaw/issues/43086 | Gemini models return 400: store parameter not supported by Google AI API | ## Bug Description
When using any Gemini model via the Google AI OpenAI-compatible endpoint (`generativelanguage.googleapis.com/v1beta/openai`), all requests fail with `400 status code (no body)`. The agent produces zero replies.
## Root Cause
OpenClaw sends a `"store": false` parameter in the request body to the Go... | closed | completed | true | 1 | [] | [] | 2026-03-11T10:26:35Z | 2026-04-27T04:51:31Z | 2026-04-24T22:17:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | chinaqhp1128 | 119,151,522 | U_kgDOBxobog | User | false |
openclaw/openclaw | 4,057,458,407 | I_kwDOQb6kR87x1-bn | 43,111 | https://github.com/openclaw/openclaw/issues/43111 | https://api.github.com/repos/openclaw/openclaw/issues/43111 | [Feature]: 企业微信群支持无需@mentioned自动回复 + 关键词触发 | ### Summary
企业微信群支持无需@mentioned自动回复 + 关键词触发
### Problem to solve
在企业微信群聊中，目前必须@机器人才能触发回复，使用体验不够流畅。
### Proposed solution
1. 支持类似Discord/Telegram的 requireMention: false 配置，让机器人在群里无需被@也能自动回复
2. 或者支持关键词触发（如在消息中包含某一关键词时自动回复）
### Alternatives considered
_No response_
### Impact
代码
### Evidence/examples
_No respon... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-11T10:51:42Z | 2026-04-27T04:51:31Z | 2026-04-24T22:17:50Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dingyugz | 267,307,627 | U_kgDOD-7Kaw | User | false |
openclaw/openclaw | 4,057,467,678 | I_kwDOQb6kR87x2Ase | 43,116 | https://github.com/openclaw/openclaw/issues/43116 | https://api.github.com/repos/openclaw/openclaw/issues/43116 | [Community Channel] WeChat (微信) Channel for OpenClaw | Hi OpenClaw team 👋
I've built a WeChat PC channel for OpenClaw and would love to share it with the community.
## What it does
Connects OpenClaw to WeChat PC client, enabling AI assistants to reply in private chats and group chats automatically.
**Repository:** https://github.com/SEUWanglibo/openclaw-wechat-channel... | closed | completed | true | 4 | [] | [] | 2026-03-11T10:53:28Z | 2026-04-27T04:51:30Z | 2026-04-24T22:17:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SEUWanglibo | 124,285,027 | U_kgDOB2hwYw | User | false |
openclaw/openclaw | 4,057,573,471 | I_kwDOQb6kR87x2ahf | 43,126 | https://github.com/openclaw/openclaw/issues/43126 | https://api.github.com/repos/openclaw/openclaw/issues/43126 | [Bug] heartbeat target: "last" causes multiple sessions to respond simultaneously | ## Description
When configuring `heartbeat.target: "last"` for agents, multiple old sessions respond to heartbeat triggers simultaneously, causing:
1. **Duplicate messages**: Same heartbeat generates multiple responses
2. **Communication confusion**: Users receive conflicting or redundant messages
3. **Session accumu... | closed | completed | true | 2 | [] | [] | 2026-03-11T11:13:06Z | 2026-04-27T04:51:29Z | 2026-04-24T22:17:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | hhdnxx | 22,232,872 | MDQ6VXNlcjIyMjMyODcy | User | false |
openclaw/openclaw | 4,057,668,917 | I_kwDOQb6kR87x2x01 | 43,134 | https://github.com/openclaw/openclaw/issues/43134 | https://api.github.com/repos/openclaw/openclaw/issues/43134 | [Bug]: Gateway restart cannot run scheduleTask, always in queque | ### Bug type
Regression (worked before, now fails)
### Summary
When I exec "openclaw gateway restart" without open any openclaw program, windows always put my schedule task in queue,not running.
### Steps to reproduce
exec "openclaw gateway restart"
### Expected behavior
<img width="991" height="400" alt="Image"... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T11:30:36Z | 2026-04-27T04:51:28Z | 2026-04-24T22:17:58Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | BloodlustLR | 26,219,522 | MDQ6VXNlcjI2MjE5NTIy | User | false |
openclaw/openclaw | 4,057,729,388 | I_kwDOQb6kR87x3Als | 43,148 | https://github.com/openclaw/openclaw/issues/43148 | https://api.github.com/repos/openclaw/openclaw/issues/43148 | [Bug]: Changing system prompt causes cache invalidations | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The system prompt sometimes changes between requests, sometimes, seemingly randomly, causing cache invalidations and increasing costs.
### Steps to reproduce
The issue is intermittent and seems random, at least I can't see a pattern. I'll... | closed | completed | true | 8 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T11:43:15Z | 2026-04-27T04:51:28Z | 2026-04-24T22:18:01Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dddsnn | 562,550 | MDQ6VXNlcjU2MjU1MA== | User | false |
openclaw/openclaw | 4,057,773,716 | I_kwDOQb6kR87x3LaU | 43,153 | https://github.com/openclaw/openclaw/issues/43153 | https://api.github.com/repos/openclaw/openclaw/issues/43153 | [Bug]: Chat tool card sidebar does not show full command for long exec commands | ### Bug type
Regression (worked before, now fails)
### Summary
When clicking a tool card in the Chat UI, the sidebar only displays the (truncated) output text but omits the full command, making it impossible to view long exec commands that are already truncated with … on the card itself.
### Steps to reproduce
1.O... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-11T11:52:23Z | 2026-04-27T04:51:27Z | 2026-04-24T22:18:04Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhouhe-xydt | 265,407,618 | U_kgDOD9HMgg | User | false |
openclaw/openclaw | 4,057,782,466 | I_kwDOQb6kR87x3NjC | 43,154 | https://github.com/openclaw/openclaw/issues/43154 | https://api.github.com/repos/openclaw/openclaw/issues/43154 | API Documentation Improvements | The API documentation is missing detailed authentication steps, example requests, error code descriptions, and cost information. Please expand the reference section with clear endpoint listings, sample curl commands, and a changelog. | closed | completed | true | 1 | [] | [] | 2026-03-11T11:54:09Z | 2026-04-27T04:51:26Z | 2026-04-24T22:18:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cryptobro-man | 78,248,519 | MDQ6VXNlcjc4MjQ4NTE5 | User | false |
openclaw/openclaw | 4,057,814,018 | I_kwDOQb6kR87x3VQC | 43,156 | https://github.com/openclaw/openclaw/issues/43156 | https://api.github.com/repos/openclaw/openclaw/issues/43156 | [Bug]: gateway auth | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
http://localhost:18789/
unauthorized: gateway password missing (enter the password in Control UI settings)
can't see where to apply the password, no ui dialog or form to submit password
### Steps to reproduce
open http://localhost:1878... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T12:01:01Z | 2026-04-27T04:51:26Z | 2026-04-24T22:18:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | fishbar | 981,128 | MDQ6VXNlcjk4MTEyOA== | User | false |
openclaw/openclaw | 4,057,855,295 | I_kwDOQb6kR87x3fU_ | 43,161 | https://github.com/openclaw/openclaw/issues/43161 | https://api.github.com/repos/openclaw/openclaw/issues/43161 | P1 Feature Request: Dark Mode for the Chat UI | ## 🚨 P1 — Dark Mode: This Is a Real Pain Point
The current OpenClaw chat/web interface has no dark mode. For anyone spending meaningful time in this interface — managing agents, configuring workflows, monitoring sessions — the bright white UI causes genuine eye strain.
Dark mode is table stakes for developer tooling... | closed | completed | true | 3 | [] | [] | 2026-03-11T12:09:21Z | 2026-04-27T04:51:25Z | 2026-04-24T22:18:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Mentis123 | 92,084,518 | U_kgDOBX0ZJg | User | false |
openclaw/openclaw | 4,057,889,655 | I_kwDOQb6kR87x3nt3 | 43,164 | https://github.com/openclaw/openclaw/issues/43164 | https://api.github.com/repos/openclaw/openclaw/issues/43164 | dangerouslyDisableDeviceAuth does not fully bypass device identity check when sharedAuthOk is false | **Description**
When `gateway.controlUi.dangerouslyDisableDeviceAuth` is set to `true`, the gateway correctly sets the device to `null` via `resolveControlUiAuthPolicy`, but the connection is still rejected with `device identity required` (code 1008) because `roleCanSkipDeviceIdentity()` requires both `role === "opera... | closed | completed | true | 2 | [] | [] | 2026-03-11T12:16:16Z | 2026-04-27T04:51:24Z | 2026-04-24T22:18:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | gabachom | 67,209,341 | MDQ6VXNlcjY3MjA5MzQx | User | false |
openclaw/openclaw | 4,057,893,594 | I_kwDOQb6kR87x3ora | 43,167 | https://github.com/openclaw/openclaw/issues/43167 | https://api.github.com/repos/openclaw/openclaw/issues/43167 | Feature Request: Persistent "Current Process" sidebar collapse — keep it closed | ## Problem
The "Current Process" sidebar in the chat UI cannot be persistently collapsed. Every time the user collapses it, it pops back open automatically — forcing them to see background process details they do not want to see.
This is particularly annoying for users who prefer a focused, wider chat window without ... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T12:17:06Z | 2026-04-27T04:51:23Z | 2026-04-24T22:18:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Mentis123 | 92,084,518 | U_kgDOBX0ZJg | User | false |
openclaw/openclaw | 4,057,963,110 | I_kwDOQb6kR87x35pm | 43,172 | https://github.com/openclaw/openclaw/issues/43172 | https://api.github.com/repos/openclaw/openclaw/issues/43172 | Feishu: should document creation be a first-class supported capability? | ## Summary
OpenClaw already supports Feishu/Lark as a channel, but creating a Feishu internal document does not appear to be a clearly defined, first-class capability from a product and documentation perspective.
This issue is to discuss whether creating Feishu internal docs should be officially supported and, if yes... | closed | completed | true | 1 | [] | [] | 2026-03-11T12:32:00Z | 2026-04-27T04:51:22Z | 2026-04-24T22:18:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 13323232dong | 203,389,488 | U_kgDODB96MA | User | false |
openclaw/openclaw | 4,058,014,527 | I_kwDOQb6kR87x4GM_ | 43,178 | https://github.com/openclaw/openclaw/issues/43178 | https://api.github.com/repos/openclaw/openclaw/issues/43178 | bug: Telegram polling watchdog triggers full gateway restart under concurrent multi-agent load (10 agents) | # Bug Report: Telegram Polling Watchdog Triggers Full Gateway Restart Under Concurrent Agent Load
**OpenClaw Version:** 2026.3.9 (f063e57)
**OS:** macOS 26.3.1 (Darwin 25.3.0, arm64 / Apple Silicon)
**Node.js:** v25.6.1
**Install method:** npm global
**Date:** March 11, 2026
---
## Summary
When multiple agents (8–1... | closed | completed | true | 2 | [] | [] | 2026-03-11T12:42:36Z | 2026-04-27T04:51:21Z | 2026-04-24T22:18:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dwightstuckey | 28,566,459 | MDQ6VXNlcjI4NTY2NDU5 | User | false |
openclaw/openclaw | 4,058,030,003 | I_kwDOQb6kR87x4J-z | 43,181 | https://github.com/openclaw/openclaw/issues/43181 | https://api.github.com/repos/openclaw/openclaw/issues/43181 | Browser tool timeout blocks Telegram polling loop, causing cascading gateway restarts | ## Summary
Browser tool timeouts (locator.click, page navigation) block the Node.js event loop, stalling the Telegram long-polling loop (`getUpdates`). This causes a cascading failure: polling stall → health monitor SIGTERM → gateway restart → cron jobs re-trigger browser → repeat.
## Environment
- OpenClaw v2026.3.... | closed | completed | true | 2 | [] | [] | 2026-03-11T12:45:54Z | 2026-04-27T04:51:20Z | 2026-04-24T22:18:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | MongLong0214 | 97,578,200 | U_kgDOBdDs2A | User | false |
openclaw/openclaw | 4,058,051,531 | I_kwDOQb6kR87x4PPL | 43,182 | https://github.com/openclaw/openclaw/issues/43182 | https://api.github.com/repos/openclaw/openclaw/issues/43182 | OpenClaw 技能测试 | 这是自动测试 Issue，可随时关闭 | closed | not_planned | true | 1 | [] | [] | 2026-03-11T12:50:24Z | 2026-04-27T04:51:19Z | 2026-04-24T22:18:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | 15605025997 | 204,523,536 | U_kgDODDDIEA | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.