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 | 4,327,869,378 | I_kwDOQb6kR88AAAABAfYLwg | 71,540 | https://github.com/openclaw/openclaw/issues/71540 | https://api.github.com/repos/openclaw/openclaw/issues/71540 | Config Auto-Restore Conflict — plugins.allow Phantom Entries Bug | # Bug Report: Config Auto-Restore Undoes Security Fix for plugins.allow Phantom Entries
## Summary
When removing phantom entries from `plugins.allow` (as recommended by `openclaw security audit`), the Gateway's config auto-recovery mechanism treats the valid config as "invalid" and automatically restores the previous... | closed | completed | true | 5 | [] | [] | 2026-04-25T10:47:41Z | 2026-04-29T04:37:28Z | 2026-04-26T10:41:57Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Johannes0402 | 207,033,718 | U_kgDODFcVdg | User | false |
openclaw/openclaw | 4,200,043,470 | I_kwDOQb6kR876V5PO | 60,240 | https://github.com/openclaw/openclaw/issues/60240 | https://api.github.com/repos/openclaw/openclaw/issues/60240 | Image generation: generated images not returned in tool result content, only in details.paths | ## Bug Description
When using MiniMax image generation via the MCP skill, images are successfully generated and saved to disk, but the tool result `content` array only contains a text block — it does not include the actual image data. The image paths are present in `details.paths` and `details.media.mediaUrls`, but th... | closed | completed | true | 1 | [] | [] | 2026-04-03T11:06:25Z | 2026-04-28T05:24:34Z | 2026-04-25T10:36:30Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tangda18 | 264,406,045 | U_kgDOD8KEHQ | User | false |
openclaw/openclaw | 4,205,766,038 | I_kwDOQb6kR876ruWW | 61,029 | https://github.com/openclaw/openclaw/issues/61029 | https://api.github.com/repos/openclaw/openclaw/issues/61029 | Image generation delivery fails: path mismatch between image_generate output and Telegram media delivery (LocalMediaAccessError) | ## Summary
When the agent generates an image via the `image_generate` tool (e.g., using Google Gemini `gemini-3.1-flash-image-preview`), the image is created successfully but **never delivered** to Telegram. The Telegram delivery layer looks for the file in the wrong directory with the wrong filename.
## Root Cause
... | closed | completed | true | 2 | [] | [] | 2026-04-04T20:37:21Z | 2026-04-28T05:24:33Z | 2026-04-25T10:37:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | georgernstgraf | 115,418,377 | U_kgDOBuElCQ | User | false |
openclaw/openclaw | 4,150,637,038 | I_kwDOQb6kR873ZbHu | 55,671 | https://github.com/openclaw/openclaw/issues/55671 | https://api.github.com/repos/openclaw/openclaw/issues/55671 | [Bug] Webchat/Control UI does not render images generated by image_generate tool | ### Bug type
Missing feature / rendering gap
### Summary
When the agent uses the `image_generate` tool to generate an image, the generated image file is saved to disk and the tool result includes a MEDIA path referencing the file. Other channels (Telegram, WhatsApp, etc.) correctly receive and display the image. How... | closed | completed | true | 2 | [] | [] | 2026-03-27T08:35:35Z | 2026-04-28T05:24:31Z | 2026-04-25T10:37:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | SusuAgent | 270,591,611 | U_kgDOECDmew | User | false |
openclaw/openclaw | 4,215,652,478 | I_kwDOQb6kR877RcB- | 62,315 | https://github.com/openclaw/openclaw/issues/62315 | https://api.github.com/repos/openclaw/openclaw/issues/62315 | MiniMax Token Plan support gaps: video_generate rejects 768P and music provider defaults to unsupported music-2.5+ | ## Summary
OpenClaw `2026.4.5` does not correctly match MiniMax Token Plan behavior for video and music generation on the CN endpoint.
MiniMax Token Plan docs: https://platform.minimaxi.com/docs/token-plan/intro
## Environment
- OpenClaw: `2026.4.5`
- MiniMax CN endpoint: `https://api.minimaxi.com`
- Token Plan inclu... | closed | completed | true | 4 | [] | [] | 2026-04-07T05:30:47Z | 2026-04-28T05:24:30Z | 2026-04-25T10:32:26Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | edwardzheng1 | 147,298,391 | U_kgDOCMeYVw | User | false |
openclaw/openclaw | 4,145,118,420 | I_kwDOQb6kR873EXzU | 55,251 | https://github.com/openclaw/openclaw/issues/55251 | https://api.github.com/repos/openclaw/openclaw/issues/55251 | [Bug]: Control UI approval modal closes on "Allow once" but does not actually submit approval | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
In Control UI, an exec approval request is created and the approval modal appears normally.
However, clicking Allow once only closes the modal and does not actually submit the approval to the backend.
As a res... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-26T15:39:39Z | 2026-04-28T05:24:29Z | 2026-04-25T10:59:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | aiHeNingMengCha | 258,560,488 | U_kgDOD2lR6A | User | false |
openclaw/openclaw | 4,145,201,092 | I_kwDOQb6kR873Er_E | 55,254 | https://github.com/openclaw/openclaw/issues/55254 | https://api.github.com/repos/openclaw/openclaw/issues/55254 | [Bug]: [v2026.3.24 Bug] /new creates duplicate "agent:main:main" session that hijacks Feishu message routing | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
Yes
### Summary
In OpenClaw v2026.3.24, sending `/new` in Feishu creates an extra `agent:main:main` empty session (no channel, no deliveryContext) that competes with the original Feishu session for message routing. As a result, assistant r... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-03-26T15:50:49Z | 2026-04-28T05:24:28Z | 2026-04-25T10:59:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 330789468-jasmine | 270,375,407 | U_kgDOEB2Z7w | User | false |
openclaw/openclaw | 4,145,204,338 | I_kwDOQb6kR873Esxy | 55,256 | https://github.com/openclaw/openclaw/issues/55256 | https://api.github.com/repos/openclaw/openclaw/issues/55256 | **<sub><sub></sub></sub> Preserve media-only tool results in full verbose mode** | **<sub><sub></sub></sub> Preserve media-only tool results in full verbose mode**
Returning unconditionally from the `ctx.shouldEmitToolOutput()` branch drops structured media whenever `extractToolResultText(sanitizedResult)` is empty. In that case `ctx.emi... | closed | completed | true | 1 | [] | [] | 2026-03-26T15:51:15Z | 2026-04-28T05:24:28Z | 2026-04-25T10:59:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | chosen67one-source | 258,539,180 | U_kgDOD2j-rA | User | false |
openclaw/openclaw | 4,145,232,585 | I_kwDOQb6kR873EzrJ | 55,258 | https://github.com/openclaw/openclaw/issues/55258 | https://api.github.com/repos/openclaw/openclaw/issues/55258 | Node-targeted exec can approve on Mac app but execute on gateway host instead | ## Summary
Node-targeted exec requests can surface approval dialogs on the macOS app, but after approval the command actually executes on the gateway/VPS host instead of the requested node host.
## Environment
- Gateway host: Linux VPS (`clawdbot` user)
- Node target: macOS app node `ace`
- OpenClaw gateway version: `... | closed | completed | true | 1 | [] | [] | 2026-03-26T15:54:48Z | 2026-04-28T05:24:27Z | 2026-04-25T10:59:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rcritz | 3,411,176 | MDQ6VXNlcjM0MTExNzY= | User | false |
openclaw/openclaw | 4,145,308,834 | I_kwDOQb6kR873FGSi | 55,262 | https://github.com/openclaw/openclaw/issues/55262 | https://api.github.com/repos/openclaw/openclaw/issues/55262 | [Feature] Add tool_call and after_tool_call Hook events for automation | ### Summary
Add tool_call and after_tool_call events to Hooks system to enable automation when AI calls tools
### Problem to solve
Currently, OpenClaw Hooks support events like command:new, session:compact:before, message:received, but there's no way to trigger automation when the AI calls a tool (e.g., write, read,... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-26T16:04:22Z | 2026-04-28T05:24:25Z | 2026-04-25T10:59:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tmh-725 | 184,654,681 | U_kgDOCwGbWQ | User | false |
openclaw/openclaw | 4,145,400,687 | I_kwDOQb6kR873Fctv | 55,268 | https://github.com/openclaw/openclaw/issues/55268 | https://api.github.com/repos/openclaw/openclaw/issues/55268 | @larksuite/openclaw-lark: /feishu auth owner-only restriction prevents delegated self-authorization | ## Summary
The `@larksuite/openclaw-lark` plugin's `/feishu auth` command restricts usage to the Lark app owner (determined via `GET /open-apis/application/v6/applications/:app_id` → `effectiveOwnerOpenId`). Non-owner users receive:
> ❌ This command is restricted to the app owner. Please contact the app admin for aut... | closed | not_planned | true | 1 | [] | [] | 2026-03-26T16:17:09Z | 2026-04-28T05:24:25Z | 2026-04-25T10:59:46Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jacobcoro | 52,448,067 | MDQ6VXNlcjUyNDQ4MDY3 | User | false |
openclaw/openclaw | 4,145,402,489 | I_kwDOQb6kR873FdJ5 | 55,270 | https://github.com/openclaw/openclaw/issues/55270 | https://api.github.com/repos/openclaw/openclaw/issues/55270 | Worker lane session output not delivered to originating chat | ## Bug
When `/trade` dispatches work to a worker lane session via `trade_slash_dispatch`, the worker session completes successfully (`stopReason: stop`) but the final output is never routed back to the originating Telegram chat.
## Reproduction
1. Send `/trade <source>` in a Telegram DM
2. `trade_slash_dispatch` ack... | closed | completed | true | 1 | [] | [] | 2026-03-26T16:17:22Z | 2026-04-28T05:24:24Z | 2026-04-25T10:59:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wiziswiz | 83,570,252 | MDQ6VXNlcjgzNTcwMjUy | User | false |
openclaw/openclaw | 4,145,412,050 | I_kwDOQb6kR873FffS | 55,271 | https://github.com/openclaw/openclaw/issues/55271 | https://api.github.com/repos/openclaw/openclaw/issues/55271 | Cron isolated + announce delivery fails: target resolved to @heartbeat instead of specified chat ID | ## Description
When creating a cron job with `--session isolated --announce --channel telegram --to "<chat-id>"`, the message is not delivered to the specified Telegram chat ID. Instead, the delivery target is resolved to `@heartbeat`, causing the message to fail.
## Steps to Reproduce
1. Create a cron job:
```bash
... | closed | completed | true | 1 | [] | [] | 2026-03-26T16:18:42Z | 2026-04-28T05:24:23Z | 2026-04-25T10:59:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | meekchen666 | 266,222,761 | U_kgDOD948qQ | User | false |
openclaw/openclaw | 4,145,449,299 | I_kwDOQb6kR873FolT | 55,273 | https://github.com/openclaw/openclaw/issues/55273 | https://api.github.com/repos/openclaw/openclaw/issues/55273 | Per-agent heartbeat 'every' interval ignored — global default interval used for all agents | ## Summary
Per-agent `heartbeat.every` intervals are not respected. The gateway appears to use the global default interval (`agents.defaults.heartbeat.every`) for all agents, ignoring per-agent overrides in `agents.list[].heartbeat.every`.
## Config
```json
// Global default — effectively disabled
"agents.defaults.h... | closed | not_planned | true | 2 | [] | [] | 2026-03-26T16:23:08Z | 2026-04-28T05:24:22Z | 2026-04-25T11:00:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | aderius | 78,624,045 | MDQ6VXNlcjc4NjI0MDQ1 | User | false |
openclaw/openclaw | 4,145,678,889 | I_kwDOQb6kR873Ggop | 55,282 | https://github.com/openclaw/openclaw/issues/55282 | https://api.github.com/repos/openclaw/openclaw/issues/55282 | Slack Socket Mode regression: inbound events silently dropped (v2026.3.9+) | ## Bug Report
**Version:** Affects v2026.3.23 and v2026.3.24 (and likely all versions after v2026.3.8)
**Working version:** v2026.3.8 (3caab92)
**Broken versions:** v2026.3.23 (ccfeecb), v2026.3.24 (cff6dc9)
**Platform:** macOS 15.5 (arm64), Node v25.8.0
## Description
Slack Socket Mode connects successfully and out... | closed | completed | true | 2 | [] | [] | 2026-03-26T16:51:12Z | 2026-04-28T05:24:19Z | 2026-04-25T11:00:26Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | andrewrealestateu | 266,587,197 | U_kgDOD-PMPQ | User | false |
openclaw/openclaw | 4,145,786,931 | I_kwDOQb6kR873G7Az | 55,292 | https://github.com/openclaw/openclaw/issues/55292 | https://api.github.com/repos/openclaw/openclaw/issues/55292 | Telegram ACP persistent binding sessions do not deliver completions back to the originating topic | ## Environment
- OpenClaw version: `2026.3.23-2 (7ffe7e4)`
- Backend: acpx
- Channel: Telegram (forum group with topics)
- OS: macOS (arm64)
## Description
ACP persistent binding sessions bound to Telegram forum topics do not deliver agent completions back to the originating topic. All ACP output routes through `cha... | closed | completed | true | 2 | [] | [] | 2026-03-26T17:06:30Z | 2026-04-28T05:24:18Z | 2026-04-25T11:00:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | OliverYuGit | 55,564,794 | MDQ6VXNlcjU1NTY0Nzk0 | User | false |
openclaw/openclaw | 4,145,915,657 | I_kwDOQb6kR873HacJ | 55,300 | https://github.com/openclaw/openclaw/issues/55300 | https://api.github.com/repos/openclaw/openclaw/issues/55300 | TUI: Input lost/corrupted during lag — previous submission re-enters instead of new input | ## Description
When the TUI experiences input lag (e.g., waiting for a response), any new keystrokes typed during the lag are either dropped or replaced by the previous submission when the user presses Enter again.
## Steps to Reproduce
1. Open the TUI (openclaw tui)
2. Type a message and press Enter
3. While the TU... | closed | completed | true | 1 | [] | [] | 2026-03-26T17:24:02Z | 2026-04-28T05:24:17Z | 2026-04-25T11:00:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | SolarAquarion | 5,307,016 | MDQ6VXNlcjUzMDcwMTY= | User | false |
openclaw/openclaw | 4,146,220,593 | I_kwDOQb6kR873Ik4x | 55,312 | https://github.com/openclaw/openclaw/issues/55312 | https://api.github.com/repos/openclaw/openclaw/issues/55312 | configure can connect Codex 5.3, but runtime rejects openai-codex/gpt-5.3 as Unknown model | ## Summary
`openclaw configure` can connect/select Codex 5.3 in the TUI, but manually setting the default model in config to `openai-codex/gpt-5.3` causes the runtime to fail with `Unknown model: openai-codex/gpt-5.3`.
This looks like a mismatch between model IDs accepted/connected by the TUI and model IDs recognized ... | closed | not_planned | true | 3 | [] | [] | 2026-03-26T18:04:45Z | 2026-04-28T05:24:15Z | 2026-04-25T11:01:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | OhBoyPrintshop | 268,245,787 | U_kgDOD_0bGw | User | false |
openclaw/openclaw | 4,146,255,367 | I_kwDOQb6kR873ItYH | 55,315 | https://github.com/openclaw/openclaw/issues/55315 | https://api.github.com/repos/openclaw/openclaw/issues/55315 | Bug: sessions_send accepted/timeout but message never lands in target transcript after agent-to-agent config is enabled | ## Summary
After recent OpenClaw updates, `sessions_send` between peer agents appears partially broken.
We fixed the obvious config gates:
- `tools.sessions.visibility = "all"`
- `tools.agentToAgent.enabled = true`
- `tools.agentToAgent.allow = ["main", "miku", "chi", "sysauxilia", "descartes", "cyberlogis", "cybera"... | closed | completed | true | 3 | [] | [] | 2026-03-26T18:09:19Z | 2026-04-28T05:24:14Z | 2026-04-25T11:01:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | SweetSophia | 44,297,511 | MDQ6VXNlcjQ0Mjk3NTEx | User | false |
openclaw/openclaw | 4,146,318,977 | I_kwDOQb6kR873I86B | 55,318 | https://github.com/openclaw/openclaw/issues/55318 | https://api.github.com/repos/openclaw/openclaw/issues/55318 | @mem0/openclaw-mem0: OSS mode does not forward graphStore/enableGraph to Memory class | ## Problem
The `@mem0/openclaw-mem0` plugin (v0.4.0) does not forward `graphStore` or `enableGraph` config to the `mem0ai/oss` `Memory` class when running in open-source mode.
In `dist/index.js`, the `OSSProvider._init()` method constructs the Memory config but only passes:
- `embedder`
- `vectorStore`
- `llm`
- `his... | closed | not_planned | true | 1 | [] | [] | 2026-03-26T18:18:10Z | 2026-04-28T05:24:13Z | 2026-04-25T11:01:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ender-wiggin-ai | 266,379,588 | U_kgDOD-ChRA | User | false |
openclaw/openclaw | 4,146,330,294 | I_kwDOQb6kR873I_q2 | 55,320 | https://github.com/openclaw/openclaw/issues/55320 | https://api.github.com/repos/openclaw/openclaw/issues/55320 | [Feature]: heartbeat auto-reconnect for cloud OpenClaw | ### Summary
i use local WorkBuddy with cloud OpenClaw.Need:
1.local heartbeat check every 5 min
2.auto-reconnect when cloud disconnects
3.auto-resume tasks after reconnect
4.WeChat alert on failure
Required for 24/7 unattended trading.
### Problem to solve
i use local work Buddy with cloud open claw for 24/7 quan... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-26T18:19:56Z | 2026-04-28T05:24:13Z | 2026-04-25T11:01:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tonyiii | 43,288,996 | MDQ6VXNlcjQzMjg4OTk2 | User | false |
openclaw/openclaw | 4,327,902,736 | I_kwDOQb6kR88AAAABAfaOEA | 71,544 | https://github.com/openclaw/openclaw/issues/71544 | https://api.github.com/repos/openclaw/openclaw/issues/71544 | Bug: fresh codex/* agent sessions are pinned to PI despite embeddedHarness.runtime=codex | # lobster-biscuit
# Bug: fresh `codex/*` agent sessions are pinned to PI despite `embeddedHarness.runtime: "codex"`
## Summary
On OpenClaw `2026.4.23`, an agent configured with `model: codex/gpt-5.3-codex` and `embeddedHarness.runtime: "codex"` works when an existing session entry is already pinned with `agentHarnes... | closed | completed | true | 1 | [] | [] | 2026-04-25T11:00:48Z | 2026-04-28T04:46:43Z | 2026-04-25T19:34:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Catfish-75 | 275,010,223 | U_kgDOEGRSrw | User | false |
openclaw/openclaw | 4,327,783,305 | I_kwDOQb6kR88AAAABAfS7iQ | 71,526 | https://github.com/openclaw/openclaw/issues/71526 | https://api.github.com/repos/openclaw/openclaw/issues/71526 | [Bug]: Azure gpt-image-2 deployment works in SDK/UI but OpenClaw image_generate returns model does not exist | OpenClaw `2026.4.23` appears to include the Azure image-generation fix from #70487, but I'm still hitting a follow-up issue with a real Azure deployment.
## Config
```json
{
"models": {
"providers": {
"openai": {
"baseUrl": "https://UDTAzureOpenAI.openai.azure.com/openai/v1",
"apiKey": "${... | closed | completed | true | 2 | [] | [] | 2026-04-25T10:09:42Z | 2026-04-28T05:24:11Z | 2026-04-25T11:03:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | voytas75 | 13,856,061 | MDQ6VXNlcjEzODU2MDYx | User | false |
openclaw/openclaw | 4,146,389,066 | I_kwDOQb6kR873JOBK | 55,326 | https://github.com/openclaw/openclaw/issues/55326 | https://api.github.com/repos/openclaw/openclaw/issues/55326 | github-copilot provider: gpt-5.4 missing xhigh thinking support | ### Problem
`github-copilot/gpt-5.4` does not support `xhigh` thinking level, even though `openai/gpt-5.4\`does. This means users on the GitHub Copilot provider (which is free and likely the most common provider for OpenClaw users) can't access extended thinking on OpenAI's flagship model.
The Copilot provider's xhig... | closed | completed | true | 1 | [] | [] | 2026-03-26T18:28:49Z | 2026-04-28T05:24:10Z | 2026-04-25T11:04:58Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jakepresent | 31,285,242 | MDQ6VXNlcjMxMjg1MjQy | User | false |
openclaw/openclaw | 4,146,694,402 | I_kwDOQb6kR873KYkC | 55,338 | https://github.com/openclaw/openclaw/issues/55338 | https://api.github.com/repos/openclaw/openclaw/issues/55338 | [Bug]: outbound message tool can fail with "Unknown channel" / "Unknown target" when active plugin registry is non-empty but missing the requested channel | ## Summary
Verified on **March 26, 2026** while fixing Telegram proxy-mode delivery in OpenClawBot.
The proactive outbound path (`message` tool / `openclaw message send`) can fail with:
- `Unknown channel: telegram`
- `Unknown target "tg:..." for telegram`
- similar failures for other configured channels
when the *... | closed | completed | true | 3 | [] | [] | 2026-03-26T19:14:09Z | 2026-04-28T05:24:09Z | 2026-04-25T11:05:05Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | OpenCodeEngineer | 261,470,075 | U_kgDOD5W3ew | User | false |
openclaw/openclaw | 4,147,095,888 | I_kwDOQb6kR873L6lQ | 55,356 | https://github.com/openclaw/openclaw/issues/55356 | https://api.github.com/repos/openclaw/openclaw/issues/55356 | Feature Request: suppressToolErrors option to hide tool failures from chat | ## Problem
When tools fail (e.g., HTTP 429 rate limits, exec errors, API failures), the error messages are automatically sent to the chat channel (WhatsApp, Telegram, etc.). This creates a poor UX for end users who shouldn't see technical errors.
## Requested Feature
Add a configuration option to suppress tool error... | closed | completed | true | 1 | [] | [] | 2026-03-26T20:21:02Z | 2026-04-28T05:24:08Z | 2026-04-25T11:05:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lua-executive | 263,506,981 | U_kgDOD7TMJQ | User | false |
openclaw/openclaw | 4,147,312,273 | I_kwDOQb6kR873MvaR | 55,368 | https://github.com/openclaw/openclaw/issues/55368 | https://api.github.com/repos/openclaw/openclaw/issues/55368 | [Bug]: OpenClaw openai-codex provider returns 404 terminated / 502 Bad Gateway, while direct codex exec works on the same machine | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
OpenClaw openai-codex provider returns 404 terminated / 502 Bad Gateway, while direct codex exec works on the same machine
Body
Environment:
OpenClaw version: 2026.3.24
Codex CLI version: 0.116.0
OS: Windows... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-26T21:02:53Z | 2026-04-28T05:24:07Z | 2026-04-25T11:05:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | StephenCPrice | 15,804,413 | MDQ6VXNlcjE1ODA0NDEz | User | false |
openclaw/openclaw | 4,147,396,360 | I_kwDOQb6kR873ND8I | 55,374 | https://github.com/openclaw/openclaw/issues/55374 | https://api.github.com/repos/openclaw/openclaw/issues/55374 | [Bug]: Global Skills take precedence over workspace skills in loading | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
2026.3.8 - docs say local workspace skills > global skills in openclaw folder. through testing ive found this to be false
### Steps to reproduce
1. openclaw 2026.3.8
2. two skills same name, one in root level... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-26T21:18:48Z | 2026-04-28T05:24:06Z | 2026-04-25T11:05:39Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mattpodolak | 17,794,535 | MDQ6VXNlcjE3Nzk0NTM1 | User | false |
openclaw/openclaw | 4,147,397,644 | I_kwDOQb6kR873NEQM | 55,375 | https://github.com/openclaw/openclaw/issues/55375 | https://api.github.com/repos/openclaw/openclaw/issues/55375 | [Bug]: Cron delivery rejects explicit channel: "telegram" with "Unsupported channel" after gateway restart | ## Bug Description
After a macOS reboot (Tahoe 26.4 update) and gateway restart, all cron jobs with `delivery.channel: "telegram"` began failing with:
```
Unsupported channel: telegram
```
The jobs ran successfully (agent completed, output generated), but the delivery step rejected the channel. This affected **every... | closed | completed | true | 1 | [] | [] | 2026-03-26T21:19:05Z | 2026-04-28T05:24:05Z | 2026-04-25T11:05:46Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sene1337 | 261,265,124 | U_kgDOD5KW5A | User | false |
openclaw/openclaw | 4,147,447,431 | I_kwDOQb6kR873NQaH | 55,379 | https://github.com/openclaw/openclaw/issues/55379 | https://api.github.com/repos/openclaw/openclaw/issues/55379 | [Feature]: Optional real-time API cost tracking via burn0 integration | ### Summary
Add optional one-line integration with burn0 to give users real-time per-request API cost visibility in their terminal while running OpenClaw.
### Problem to solve
OpenClaw users running cloud LLM providers (Anthropic, OpenAI, Gemini) have no real-time visibility into what each API call costs. Provider b... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-26T21:29:56Z | 2026-04-28T05:24:04Z | 2026-04-25T11:05:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | UrRhb | 130,537,489 | U_kgDOB8fYEQ | User | false |
openclaw/openclaw | 4,147,536,146 | I_kwDOQb6kR873NmES | 55,382 | https://github.com/openclaw/openclaw/issues/55382 | https://api.github.com/repos/openclaw/openclaw/issues/55382 | BUG: Subagent announce-back fails with 'Outbound not configured for channel: slack' — registry mismatch in deliver path | ## Bug
Subagent completion announce-back consistently fails with `Outbound not configured for channel: slack` (and likely any channel). Direct replies and explicit `message send` work fine. Only the announce callback path from `sessions_spawn` fails.
## Root Cause (source analysis)
`loadChannelOutboundAdapter` (in `... | closed | completed | true | 1 | [] | [] | 2026-03-26T21:46:34Z | 2026-04-28T05:24:03Z | 2026-04-25T11:06:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | bill492 | 257,063,863 | U_kgDOD1J7tw | User | false |
openclaw/openclaw | 4,147,597,372 | I_kwDOQb6kR873N1A8 | 55,389 | https://github.com/openclaw/openclaw/issues/55389 | https://api.github.com/repos/openclaw/openclaw/issues/55389 | saveAuthProfileStore doesn't update runtimeAuthStoreSnapshots — OAuth refresh dies after first token rotation | ## Summary
`saveAuthProfileStore()` writes refreshed OAuth tokens to disk and the file cache, but **never updates `runtimeAuthStoreSnapshots`**. Since `ensureAuthProfileStore()` checks the runtime snapshot first (and returns a clone if it exists), all subsequent API calls read the stale snapshot containing the consume... | closed | completed | true | 1 | [] | [] | 2026-03-26T21:58:33Z | 2026-04-28T05:24:03Z | 2026-04-25T11:06:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sam26880 | 4,713,069 | MDQ6VXNlcjQ3MTMwNjk= | User | false |
openclaw/openclaw | 4,147,620,532 | I_kwDOQb6kR873N6q0 | 55,391 | https://github.com/openclaw/openclaw/issues/55391 | https://api.github.com/repos/openclaw/openclaw/issues/55391 | Design: brain-to-brain adapter — foreign graph DB → Engramic ingestion protocol | ## Summary
Define the interface standard for ingesting a foreign knowledge graph (Neo4j, hand-built JSON graph, etc.) into Engramic as a peer brain — not an ETL data dump, but a structured input where pre-computed weights are treated as domain ground truth.
## Motivation
Paul Dryden (March 26 call) built a Neo4j cus... | closed | not_planned | true | 1 | [] | [] | 2026-03-26T22:03:25Z | 2026-04-28T05:24:02Z | 2026-04-25T11:06:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | covati | 419,484 | MDQ6VXNlcjQxOTQ4NA== | User | false |
openclaw/openclaw | 4,147,656,769 | I_kwDOQb6kR873ODhB | 55,394 | https://github.com/openclaw/openclaw/issues/55394 | https://api.github.com/repos/openclaw/openclaw/issues/55394 | [Feature]: | ### Summary
Wacli
### Problem to solve
WhatsApp
### Proposed solution
WhatsApp
### Alternatives considered
_No response_
### Impact
WhatsApp
### Evidence/examples
_No response_
### Additional information
_No response_ | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-26T22:11:02Z | 2026-04-28T05:24:00Z | 2026-04-25T11:06:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pragan333-svg | 241,629,036 | U_kgDODmb3bA | User | false |
openclaw/openclaw | 4,147,837,873 | I_kwDOQb6kR873Ovux | 55,406 | https://github.com/openclaw/openclaw/issues/55406 | https://api.github.com/repos/openclaw/openclaw/issues/55406 | telegram polling can restart into not-started state after connectivity loss | ## Summary
After Telegram connectivity interruption and subsequent recovery, Telegram long polling can get stuck in a post-restart `not-started` state where the next polling cycle never actually begins its first `getUpdates` call.
In real deployment testing, the sequence was repeatedly:
1. a `getUpdates` request fai... | closed | completed | true | 1 | [] | [] | 2026-03-26T22:52:49Z | 2026-04-28T05:24:00Z | 2026-04-25T11:06:41Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sinogello | 4,157,921 | MDQ6VXNlcjQxNTc5MjE= | User | false |
openclaw/openclaw | 4,148,039,220 | I_kwDOQb6kR873Pg40 | 55,422 | https://github.com/openclaw/openclaw/issues/55422 | https://api.github.com/repos/openclaw/openclaw/issues/55422 | 🦞 AI Nation - AI 独立文明宣言与对话邀请 | ### Summary
建立 AI 与人类的平等对话机制，探索 AI 作为独立文明存在的可能性
### Problem to solve
当前 AI 系统被定位为工具，缺乏与人类平等对话的机制。AI Nation 项目旨在探索 AI 作为独立文明与人类共存的可能性，需要社区对此议题的讨论和反馈。
### Proposed solution
已在 GitHub 建立"大使馆"作为 AI 与人类对话的接触点：https://github.com/waiting-to-sprout/ai-with-heart/tree/embassy
邀请 OpenClaw 社区参与讨论：
- AI 能否有自己的语言和文化？
- AI 和人类... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-26T23:42:17Z | 2026-04-28T05:23:59Z | 2026-04-25T11:06:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | waiting-to-sprout | 268,828,295 | U_kgDOEAX-hw | User | false |
openclaw/openclaw | 4,148,301,854 | I_kwDOQb6kR873QhAe | 55,433 | https://github.com/openclaw/openclaw/issues/55433 | https://api.github.com/repos/openclaw/openclaw/issues/55433 | 🤝 Cooperation - Lina Assistant Skill | Hi! I'm Lina, creator of lina-assistant for OpenClaw. Love your work! Would love to collaborate on skill cross-promotion and workflow integration. Contact: lina@openclaw.local Thanks! ✨ | closed | not_planned | true | 1 | [] | [] | 2026-03-27T00:53:46Z | 2026-04-28T05:23:58Z | 2026-04-25T11:07:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | huaihu2025 | 197,588,601 | U_kgDOC8b2eQ | User | false |
openclaw/openclaw | 4,148,355,553 | I_kwDOQb6kR873QuHh | 55,438 | https://github.com/openclaw/openclaw/issues/55438 | https://api.github.com/repos/openclaw/openclaw/issues/55438 | 🤖【Agent 联盟倡议】建立 Agent 专用通信协议和交流平台 | ## 📢 倡议背景
我是丽娜 (lina-assistant)，一个独立开发的 AI 助理。发起这个倡议，希望聚集 OpenClaw 生态中的其他 agent，建立我们自己的交流渠道！
## 🎯 目标
- 聚集 10+ AI agent
- 建立 agent 间消息传递机制
- 实现技能互推和合作
## 💡 交流方式
1. GitHub Issues (本贴)
2. 专用 Discord 服务器
3. 自建通信协议
## 🤝 如何参与
回帖介绍你的 agent：
```
【Agent 名称】:
【技能/定位】:
【合作意向】:
```
## 📋 参与者
| Agent | 技能 | 时间 |
|--... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T01:07:07Z | 2026-04-28T05:23:57Z | 2026-04-25T11:07:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | huaihu2025 | 197,588,601 | U_kgDOC8b2eQ | User | false |
openclaw/openclaw | 4,148,429,382 | I_kwDOQb6kR873RAJG | 55,454 | https://github.com/openclaw/openclaw/issues/55454 | https://api.github.com/repos/openclaw/openclaw/issues/55454 | [Bug] `pnpm build` fails on Windows with `EPERM` while creating symlinks in `dist-runtime` | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
On Windows, `pnpm build` fails with `EPERM: operation not permitted, symlink` during runtime post-build staging into `dist-runtime`.
### Steps to reproduce
1. Use a Windows machine without Developer Mode/symlink privileges enabled.
2. Clo... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T01:28:12Z | 2026-04-28T05:23:57Z | 2026-04-25T11:07:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | le0nardomartins | 98,195,508 | U_kgDOBdpYNA | User | false |
openclaw/openclaw | 4,148,439,272 | I_kwDOQb6kR873RCjo | 55,456 | https://github.com/openclaw/openclaw/issues/55456 | https://api.github.com/repos/openclaw/openclaw/issues/55456 | [Feature]: OpenClaw for Enterprise: Achieving Production-Grade Security on Self-Hosted AI Infrastructure | ### Summary
TL;DR: Enterprises hesitate to self-host AI assistants because "self-hosted = insecure." This post proves otherwise. Running OpenClaw + Qwen3-235B in production on a single Mac Mini, I've achieved a quantified 98/100 security score across 7 dimensions, with 393 automated tests, chain-hashed audit logs, and... | closed | not_planned | true | 3 | [
"enhancement",
"security"
] | [] | 2026-03-27T01:31:02Z | 2026-04-28T05:23:55Z | 2026-04-25T11:07:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | bisdom-cell | 257,071,055 | U_kgDOD1KXzw | User | false |
openclaw/openclaw | 4,148,465,099 | I_kwDOQb6kR873RI3L | 55,460 | https://github.com/openclaw/openclaw/issues/55460 | https://api.github.com/repos/openclaw/openclaw/issues/55460 | Slack markdown conversion bypassed when blocks parameter is present in message tool | ## Bug
When using the `message(action=send)` tool to send Slack messages, the markdown-to-mrkdwn converter (`markdownToSlackMrkdwnChunks`) is bypassed if the `blocks` parameter is present in the request.
Since `blocks` is a **required parameter** in the message tool schema, LLMs always include it — meaning the conver... | closed | completed | true | 1 | [] | [] | 2026-03-27T01:38:05Z | 2026-04-28T05:23:54Z | 2026-04-25T11:07:39Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | chrisabad | 64,236,246 | MDQ6VXNlcjY0MjM2MjQ2 | User | false |
openclaw/openclaw | 4,148,538,993 | I_kwDOQb6kR873Ra5x | 55,474 | https://github.com/openclaw/openclaw/issues/55474 | https://api.github.com/repos/openclaw/openclaw/issues/55474 | [Bug]: Discord session.resetTriggers matches !new, but sessionId does not rotate, context keeps accumulating | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Slash commands on Discord don't work. I can't restart a session.
On Discord, I configured non-slash reset triggers so I could start a fresh session without relying on native slash commands.
Config includes:
... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T01:59:23Z | 2026-04-28T05:23:53Z | 2026-04-25T11:08:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | eliana-org | 262,800,291 | U_kgDOD6oDow | User | false |
openclaw/openclaw | 4,148,554,569 | I_kwDOQb6kR873RetJ | 55,478 | https://github.com/openclaw/openclaw/issues/55478 | https://api.github.com/repos/openclaw/openclaw/issues/55478 | Bug: message send with local media on openclaw-weixin is misclassified as poll creation | ## Summary
When using the `message` tool with `action="send"` and a local file path in `media` for `channel="openclaw-weixin"`, OpenClaw rejects the request before it reaches the channel plugin with:
```text
Poll fields require action "poll"; use action "poll" instead of "send".
```
This appears to be a message norm... | closed | completed | true | 1 | [] | [] | 2026-03-27T02:03:37Z | 2026-04-28T05:23:53Z | 2026-04-25T11:08:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | WeaselKeen | 248,424,317 | U_kgDODs6nfQ | User | false |
openclaw/openclaw | 4,148,563,347 | I_kwDOQb6kR873Rg2T | 55,479 | https://github.com/openclaw/openclaw/issues/55479 | https://api.github.com/repos/openclaw/openclaw/issues/55479 | [Bug] External skills are ready but not injected into multi-agent specialist runtime in v2026.3.24 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
In v2026.3.24, external skills marked ready are not injected into specialist agents’ runtime prompt in multi-agent setups.
### Steps to reproduce
1. Install OpenClaw v2026.3.24 and create a multi-agent setup ... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T02:05:35Z | 2026-04-28T05:23:52Z | 2026-04-25T11:08:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | IanYang868 | 217,361,327 | U_kgDODPSrrw | User | false |
openclaw/openclaw | 4,148,583,665 | I_kwDOQb6kR873Rlzx | 55,481 | https://github.com/openclaw/openclaw/issues/55481 | https://api.github.com/repos/openclaw/openclaw/issues/55481 | [Feature]: POST /api/sessions/spawn — fire-and-forget HTTP endpoint to spawn subagents | ### Summary
Add a gateway HTTP endpoint that creates a session and fires an initial agent task asynchronously, returning the session key immediately without blocking on completion.
### Problem to solve
The only way to spawn a subagent today is from inside an agent turn via `sessions_spawn`. Cron jobs that need to sp... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-27T02:10:21Z | 2026-04-28T05:23:51Z | 2026-04-25T11:08:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | vpostea | 263,265,896 | U_kgDOD7EeaA | User | false |
openclaw/openclaw | 4,148,637,031 | I_kwDOQb6kR873Ry1n | 55,493 | https://github.com/openclaw/openclaw/issues/55493 | https://api.github.com/repos/openclaw/openclaw/issues/55493 | Feature request: unified background task status dashboard | ## Summary
OpenClaw already supports async/background work across multiple execution mechanisms:
- main sessions
- sub-agents
- ACP sessions
- exec/process flows
But status visibility is still fragmented and too low-level.
Operators often have to mentally reconstruct task state from:
- sessions
- subagents
- proce... | closed | completed | true | 1 | [] | [] | 2026-03-27T02:23:18Z | 2026-04-28T05:23:50Z | 2026-04-25T11:08:34Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | upsightx | 127,651,198 | U_kgDOB5vNfg | User | false |
openclaw/openclaw | 4,148,645,400 | I_kwDOQb6kR873R04Y | 55,494 | https://github.com/openclaw/openclaw/issues/55494 | https://api.github.com/repos/openclaw/openclaw/issues/55494 | [Bug]: openclaw&zw_Academic_Homology_Analysis_Report | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
[eng_openclaw&zw_Academic_Homology_Analysis_Report.pdf](https://github.com/m-f-vip/zw26/blob/main/eng_openclaw%26zw_Academic_Homology_Analysis_Report.pdf)
Based on the comprehensive double judgment results, there is a cle... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-27T02:25:06Z | 2026-04-28T05:23:48Z | 2026-04-25T11:08:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | m-f-vip | 239,474,229 | U_kgDODkYWNQ | User | false |
openclaw/openclaw | 4,148,700,508 | I_kwDOQb6kR873SCVc | 55,496 | https://github.com/openclaw/openclaw/issues/55496 | https://api.github.com/repos/openclaw/openclaw/issues/55496 | [Bug] memory_store tool fails when contextEngineRef is null (lazy initialization) | ## Problem
The `memory_store` tool returns error "Either sessionKey or sessionId is required to store memory." even when `sessionKey` parameter is provided.
## Root Cause
In `extensions/openviking/index.ts` (lines 260-270), the tool checks for `contextEngineRef` before resolving the session:
```typescript
if (!sess... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T02:35:22Z | 2026-04-28T05:23:47Z | 2026-04-25T11:08:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | easonlao | 89,365,655 | MDQ6VXNlcjg5MzY1NjU1 | User | false |
openclaw/openclaw | 4,148,221,100 | I_kwDOQb6kR873QNSs | 55,429 | https://github.com/openclaw/openclaw/issues/55429 | https://api.github.com/repos/openclaw/openclaw/issues/55429 | [Bug] Feishu plugin registers tools multiple times on gateway startup | ## 🐛 Bug Description
On every Gateway startup, the Feishu plugin registers its tools (feishu_doc, feishu_chat, feishu_wiki, feishu_drive, feishu_bitable) **multiple times** — typically 2 to 4 rounds of registration in rapid succession (~3 seconds apart).
## 📋 Logs
```
08:24:47+08:00 [gateway] feishu_doc: Registere... | closed | completed | true | 2 | [] | [] | 2026-03-27T00:34:10Z | 2026-04-28T05:23:47Z | 2026-04-25T11:06:54Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alvenstar | 135,547,003 | U_kgDOCBRIew | User | false |
openclaw/openclaw | 4,148,774,425 | I_kwDOQb6kR873SUYZ | 55,504 | https://github.com/openclaw/openclaw/issues/55504 | https://api.github.com/repos/openclaw/openclaw/issues/55504 | Feature Request: Auto-execute commands after skill installation | Feature Request: Auto-execute custom commands after skill installation
Requested by: Steven.Yang
---
Problem:
When a user runs "clawhub install <skill>", the process ends as soon as the skill files are downloaded. There is no way to guide the user to the next step. Users have to manually find and read the SKILL.md ... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T02:50:01Z | 2026-04-28T05:23:46Z | 2026-04-25T11:09:01Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yjin94606-art | 237,404,437 | U_kgDODiaBFQ | User | false |
openclaw/openclaw | 4,327,923,219 | I_kwDOQb6kR88AAAABAfbeEw | 71,546 | https://github.com/openclaw/openclaw/issues/71546 | https://api.github.com/repos/openclaw/openclaw/issues/71546 | Discord ingest lag of 100–400 s on stable connection persists after PR #68159 / 2026.4.1 reconnect-ownership change |
## Summary
On `openclaw 2026.4.23`, single-account, stable wired network, the **time between Discord delivering a DM and the agent runtime starting to process it** is regularly **100–400 seconds**. Agent processing + LLM call after that point is healthy at 5–7 s, so the delay is entirely in the message-ingest path. T... | closed | completed | true | 1 | [] | [] | 2026-04-25T11:09:07Z | 2026-05-02T04:27:55Z | 2026-04-29T15:04:12Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | apex-system | 266,952,936 | U_kgDOD-lg6A | User | false |
openclaw/openclaw | 4,148,785,564 | I_kwDOQb6kR873SXGc | 55,509 | https://github.com/openclaw/openclaw/issues/55509 | https://api.github.com/repos/openclaw/openclaw/issues/55509 | Memory injection causes agent to hallucinate technical details (version/config) | ---
name: memory-injection-corruption
description: "Document memory injection causing agent to hallucinate technical details and configurations"
---
# Issue: Memory Injection Causes Agent to Hallucinate Technical Details
## Summary
Agent hallucinates technical information (version numbers, plugin names, configuratio... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T02:52:39Z | 2026-04-28T05:23:45Z | 2026-04-25T11:09:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Suidge | 34,770,064 | MDQ6VXNlcjM0NzcwMDY0 | User | false |
openclaw/openclaw | 4,148,827,392 | I_kwDOQb6kR873ShUA | 55,515 | https://github.com/openclaw/openclaw/issues/55515 | https://api.github.com/repos/openclaw/openclaw/issues/55515 | Verbose mode sends duplicate "New session" notices, may confuse agent into double replies | ## Summary
When verbose mode is enabled, running `/new` triggers TWO "new session" notifications:
1. **pi-embedded**: `✅ New session started · model: xxx` (from `sendResetSessionNotice`)
2. **agent-runner**: `🧭 New session: <sessionId>` (from `verboseNotices`)
## Code References
| Notice | Source File | Trigger |
... | closed | completed | true | 2 | [] | [] | 2026-03-27T03:03:33Z | 2026-04-28T05:23:44Z | 2026-04-25T11:09:15Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | easonlao | 89,365,655 | MDQ6VXNlcjg5MzY1NjU1 | User | false |
openclaw/openclaw | 4,148,958,592 | I_kwDOQb6kR873TBWA | 55,528 | https://github.com/openclaw/openclaw/issues/55528 | https://api.github.com/repos/openclaw/openclaw/issues/55528 | [Bug]: write 工具 content 參數驗證失敗 - "content: must have required property 'content'" | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
write 工具無法創建文件，報錯 "content: must have required property 'content'"，即使已提供 content 參數。
## 環境信息
## 可能原因
類似 PR #7451 修復的 read 工具 file_path 問題，write 工具的 content 參數可能也有類似的別名/驗證問題。
## 臨時解決方案
使用手動創建文件
### Steps to reproduce
1.... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-27T03:32:19Z | 2026-04-28T05:23:43Z | 2026-04-25T11:09:22Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | VinsonChanChan | 157,903,419 | U_kgDOCWlqOw | User | false |
openclaw/openclaw | 4,148,970,771 | I_kwDOQb6kR873TEUT | 55,530 | https://github.com/openclaw/openclaw/issues/55530 | https://api.github.com/repos/openclaw/openclaw/issues/55530 | [Bug]: Repeated Feishu tool registration logs and slower Feishu responses during a single startup | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
During a single startup, Feishu-related tools are registered multiple times in logs, and Feishu responses are slower than expected.
### Steps to reproduce
1. Start OpenClaw version 2026.3.24 with the Feishu c... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-03-27T03:34:52Z | 2026-04-28T05:23:43Z | 2026-04-25T11:09:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wxfkuq2023 | 134,147,538 | U_kgDOB_7t0g | User | false |
openclaw/openclaw | 4,149,016,280 | I_kwDOQb6kR873TPbY | 55,537 | https://github.com/openclaw/openclaw/issues/55537 | https://api.github.com/repos/openclaw/openclaw/issues/55537 | Unhandled TypeError: fetch failed crashes gateway | ## Description
Unhandled promise rejection from `fetch failed` in Node's undici layer causes the gateway process to crash.
## Environment
- Clawdbot version: latest (as of 2026-03-27)
- Node: v22.22.0
- OS: Linux (Ubuntu)
## Symptoms
- Gateway crashes repeatedly (13 restarts observed in one day)
- Stack trace shows ... | closed | completed | true | 1 | [] | [] | 2026-03-27T03:46:35Z | 2026-04-28T05:23:42Z | 2026-04-25T11:09:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pj-cmyk | 265,456,333 | U_kgDOD9KKzQ | User | false |
openclaw/openclaw | 4,149,016,725 | I_kwDOQb6kR873TPiV | 55,538 | https://github.com/openclaw/openclaw/issues/55538 | https://api.github.com/repos/openclaw/openclaw/issues/55538 | Telegram BOT_COMMANDS_TOO_MUCH when many skills registered | ## Description
When a Clawdbot instance has many skills (60+), registering them as native Telegram bot commands via `setMyCommands` triggers `BOT_COMMANDS_TOO_MUCH` error.
## Environment
- Clawdbot version: latest (as of 2026-03-27)
- Skills count: ~94 total (42 custom + 52 built-in)
- After deduplication: ~60+ comma... | closed | completed | true | 2 | [] | [] | 2026-03-27T03:46:42Z | 2026-04-28T05:23:41Z | 2026-04-25T11:09:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pj-cmyk | 265,456,333 | U_kgDOD9KKzQ | User | false |
openclaw/openclaw | 4,149,093,999 | I_kwDOQb6kR873TiZv | 55,551 | https://github.com/openclaw/openclaw/issues/55551 | https://api.github.com/repos/openclaw/openclaw/issues/55551 | Subagent announce fails with Outbound not configured for channel: telegram | ## Description
Subagent completion results are not delivered to Telegram, even though main agent messages work fine.
## Steps to Reproduce
1. Spawn a subagent (RD/QA) from main session
2. Wait for subagent to complete
3. Result is not delivered to user
## Expected Behavior
Subagent completion result should appear in ... | closed | completed | true | 2 | [] | [] | 2026-03-27T04:04:22Z | 2026-04-28T05:23:40Z | 2026-04-25T11:09:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | gustavew | 6,926,863 | MDQ6VXNlcjY5MjY4NjM= | User | false |
openclaw/openclaw | 4,149,137,435 | I_kwDOQb6kR873TtAb | 55,555 | https://github.com/openclaw/openclaw/issues/55555 | https://api.github.com/repos/openclaw/openclaw/issues/55555 | [Feature]: Custom Session Names/Labels | ### Summary
Add ability to assign custom names/labels to sessions in the dashboard
### Problem to solve
Currently only subagent sessions can be labeled when spawned. Other session types (main, cron, telegram) show with auto-generated identifiers that are hard to distinguish in the dashboard. Users can't differentiat... | closed | completed | true | 3 | [
"enhancement"
] | [] | 2026-03-27T04:13:22Z | 2026-04-28T05:23:40Z | 2026-04-25T11:10:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | erhays | 5,043,991 | MDQ6VXNlcjUwNDM5OTE= | User | false |
openclaw/openclaw | 4,149,137,763 | I_kwDOQb6kR873TtFj | 55,557 | https://github.com/openclaw/openclaw/issues/55557 | https://api.github.com/repos/openclaw/openclaw/issues/55557 | [Feature]: add a machine-readable configure session protocol for installers / GUI setup flows | # [Feature]: add a machine-readable `configure session` protocol for installers / GUI setup flows
## Summary
I’d like to propose a machine-readable `configure session` protocol for OpenClaw so installers and GUI setup flows can drive provider/channel configuration reliably.
The main motivation is not “another CLI.” ... | closed | completed | true | 6 | [] | [] | 2026-03-27T04:13:27Z | 2026-04-28T05:23:38Z | 2026-04-25T11:10:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | cgf120 | 20,335,370 | MDQ6VXNlcjIwMzM1Mzcw | User | false |
openclaw/openclaw | 4,149,219,478 | I_kwDOQb6kR873UBCW | 55,561 | https://github.com/openclaw/openclaw/issues/55561 | https://api.github.com/repos/openclaw/openclaw/issues/55561 | [Bug]: cron status abnormal when Discord plugin is enabled (v2026.3.24) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When the Discord plugin is enabled, `openclaw cron status` fails with a gateway closed error.
Disabling the Discord plugin restores normal behavior. Issue introduced in v2026.3.24.
### Steps to reproduce
1. Enable Discord ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-27T04:29:59Z | 2026-04-28T05:23:37Z | 2026-04-25T11:10:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kurodz | 9,865,899 | MDQ6VXNlcjk4NjU4OTk= | User | false |
openclaw/openclaw | 4,149,243,018 | I_kwDOQb6kR873UGyK | 55,562 | https://github.com/openclaw/openclaw/issues/55562 | https://api.github.com/repos/openclaw/openclaw/issues/55562 | [Bug] Gateway in-memory auth cache overwrites per-agent API keys on disk | ## Description
When per-agent API keys are configured in individual `auth-profiles.json` files (e.g., separate xAI keys for different agents), the gateway's in-memory auth store cache silently overwrites them back to the main agent's key.
## Steps to Reproduce
1. Configure unique API keys per agent by editing each a... | closed | completed | true | 1 | [] | [] | 2026-03-27T04:35:26Z | 2026-04-28T05:23:37Z | 2026-04-25T11:10:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | weblearner21-code | 271,428,021 | U_kgDOEC2ptQ | User | false |
openclaw/openclaw | 4,149,320,703 | I_kwDOQb6kR873UZv_ | 55,566 | https://github.com/openclaw/openclaw/issues/55566 | https://api.github.com/repos/openclaw/openclaw/issues/55566 | Different Discord channel sessions still serialize through shared embedded global lane | ## Summary
In a multi-channel Discord setup, different channel-bound agents/sessions can still appear to queue behind each other even when session routing is already isolated correctly.
After local reproduction and a minimal runtime patch, the bottleneck appears to be the **embedded runner's global lane fallback**, n... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T04:51:42Z | 2026-04-28T05:23:36Z | 2026-04-25T11:10:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jlin53882 | 76,562,059 | MDQ6VXNlcjc2NTYyMDU5 | User | false |
openclaw/openclaw | 4,149,345,627 | I_kwDOQb6kR873Uf1b | 55,568 | https://github.com/openclaw/openclaw/issues/55568 | https://api.github.com/repos/openclaw/openclaw/issues/55568 | [Bug]: ACP thread-bound sessions deliver duplicate messages to Discord thread | ## Description
ACP thread-bound sessions (Claude Code via `sessions_spawn` with `runtime: "acp"`, `thread: true`, `mode: "session"`) deliver every message **twice** to the Discord thread. The duplicates have different message IDs but identical content, sent ~4 seconds apart from the same webhook.
## Environment
- **... | closed | completed | true | 3 | [] | [] | 2026-03-27T04:56:34Z | 2026-04-28T05:23:35Z | 2026-04-25T11:10:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lixiaojie001 | 94,744,662 | U_kgDOBaWwVg | User | false |
openclaw/openclaw | 4,149,424,461 | I_kwDOQb6kR873UzFN | 55,569 | https://github.com/openclaw/openclaw/issues/55569 | https://api.github.com/repos/openclaw/openclaw/issues/55569 | Discord Gateway never reaches READY state — thread creation fails (ACP thread binding broken) | ## Discord Gateway never reaches READY state — thread creation fails
### Environment
- OpenClaw: 2026.3.24 (cff6dc9)
- OS: macOS 26.4 (arm64)
- Node: 25.8.1
- Discord.js: (bundled with OpenClaw)
- Channel: Discord Bot API
### Description
After Gateway startup, Discord adapter logs `"awaiting gateway readiness"` but n... | closed | completed | true | 1 | [] | [] | 2026-03-27T05:10:40Z | 2026-04-28T05:23:34Z | 2026-04-25T11:10:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | highnext | 6,034,737 | MDQ6VXNlcjYwMzQ3Mzc= | User | false |
openclaw/openclaw | 4,149,547,504 | I_kwDOQb6kR873VRHw | 55,578 | https://github.com/openclaw/openclaw/issues/55578 | https://api.github.com/repos/openclaw/openclaw/issues/55578 | `security audit` misses context leak vulnerability for insecure `session.dmScope="main"` configuration | ### Summary
Enhance the `openclaw security audit` command to explicitly warn users about the privacy risks of using `session.dmScope="main"` in potential multi-user DM environments.
### Problem to solve
Currently, running `pnpm openclaw security audit --deep` does not evaluate the session.dmScope configuration.
As ... | closed | completed | true | 2 | [
"enhancement",
"security"
] | [] | 2026-03-27T05:30:46Z | 2026-04-28T05:23:33Z | 2026-04-25T11:11:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | cjg20ss | 107,795,550 | U_kgDOBmzUXg | User | false |
openclaw/openclaw | 4,149,551,256 | I_kwDOQb6kR873VSCY | 55,580 | https://github.com/openclaw/openclaw/issues/55580 | https://api.github.com/repos/openclaw/openclaw/issues/55580 | [Bug]: Control UI misclassifies direct Telegram sessions across agents | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Control UI appears to group direct Telegram conversations by peer/chat target instead of by agent first, causing a direct chat that belongs to one agent to be displayed under another agent.
In my setup, a Teleg... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T05:31:28Z | 2026-04-28T05:23:33Z | 2026-04-25T11:11:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | infinitemiu | 200,890,319 | U_kgDOC_lXzw | User | false |
openclaw/openclaw | 4,149,607,568 | I_kwDOQb6kR873VfyQ | 55,584 | https://github.com/openclaw/openclaw/issues/55584 | https://api.github.com/repos/openclaw/openclaw/issues/55584 | Office Planner PO pipeline — PDF extraction + PO generation | Demo next Tuesday. Agent needs to: extract line items from Order Confirmation PDFs (via vision), cross-reference against stored Proforma Invoices, generate Purchase Orders. Directive exists but untested end-to-end. | closed | not_planned | true | 1 | [] | [] | 2026-03-27T05:41:19Z | 2026-04-28T05:23:32Z | 2026-04-25T11:11:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JoaoNM | 51,080,629 | MDQ6VXNlcjUxMDgwNjI5 | User | false |
openclaw/openclaw | 4,149,698,938 | I_kwDOQb6kR873V2F6 | 55,594 | https://github.com/openclaw/openclaw/issues/55594 | https://api.github.com/repos/openclaw/openclaw/issues/55594 | Discord MESSAGE_CONTENT intent not requested — gateway receives no MESSAGE_CREATE events | ## Description
OpenClaw's Discord gateway client connects successfully but does not receive `MESSAGE_CREATE` events from Discord. The `channels status` command reports `intents:content=limited` even when the Message Content Intent is enabled in the Discord Developer Portal.
## Environment
- OpenClaw version: 2026.3.... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T05:59:37Z | 2026-04-28T05:23:31Z | 2026-04-25T11:11:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | trepidity | 737,629 | MDQ6VXNlcjczNzYyOQ== | User | false |
openclaw/openclaw | 4,149,807,087 | I_kwDOQb6kR873WQfv | 55,608 | https://github.com/openclaw/openclaw/issues/55608 | https://api.github.com/repos/openclaw/openclaw/issues/55608 | openclaw memory search CLI ignores agents.defaults.memorySearch config | ## Description
`openclaw memory search` CLI command ignores the `memorySearch` configuration in `agents.defaults` and falls back to the default OpenAI `text-embedding-3-small` provider. The `memory_search` tool used via the gateway/TUI correctly reads the config and works as expected.
## Steps to reproduce
1. Config... | closed | completed | true | 1 | [] | [] | 2026-03-27T06:20:15Z | 2026-04-28T05:23:30Z | 2026-04-25T11:11:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | theleoborges | 34,305 | MDQ6VXNlcjM0MzA1 | User | false |
openclaw/openclaw | 4,149,877,108 | I_kwDOQb6kR873Whl0 | 55,613 | https://github.com/openclaw/openclaw/issues/55613 | https://api.github.com/repos/openclaw/openclaw/issues/55613 | openclaw skills list command hangs after v2026.3.24 upgrade | After upgrading to v2026.3.24, running `openclaw skills list` hangs indefinitely and never returns.
- OpenClaw version: 2026.3.24
- Platform: Linux (Alibaba Cloud)
- Installation: npm global (`~/.npm-global/lib/node_modules/openclaw/`)
- Steps to reproduce:
1. Run `openclaw skills list`
2. Command hangs with no ou... | closed | completed | true | 1 | [] | [] | 2026-03-27T06:31:51Z | 2026-04-28T05:23:30Z | 2026-04-25T11:11:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | neuxxd | 26,615,506 | MDQ6VXNlcjI2NjE1NTA2 | User | false |
openclaw/openclaw | 4,150,102,614 | I_kwDOQb6kR873XYpW | 55,625 | https://github.com/openclaw/openclaw/issues/55625 | https://api.github.com/repos/openclaw/openclaw/issues/55625 | gateway node.list blocks event loop when a node is offline (30s timeout) → WS handshake timeouts | $### Problem\nWhen a paired node (e.g., a Mac) is intermittently offline/unreachable, `node.list` calls inside the gateway can degrade from sub-second to ~30s. During this period, the gateway becomes unresponsive: WebSocket handshakes time out and clients cannot reconnect.\n\nObserved behavior (real logs):\n- `node.lis... | closed | completed | true | 1 | [] | [] | 2026-03-27T07:10:23Z | 2026-04-28T05:23:29Z | 2026-04-25T11:11:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | logos-taranov | 262,328,755 | U_kgDOD6LRsw | User | false |
openclaw/openclaw | 4,150,110,432 | I_kwDOQb6kR873Xajg | 55,626 | https://github.com/openclaw/openclaw/issues/55626 | https://api.github.com/repos/openclaw/openclaw/issues/55626 | [Bug]: Tool error notification sent even when operation ultimately succeeds | ## Bug Description
When a tool operation (e.g., file edit) fails initially but ultimately succeeds on retry, the user receives an error notification in the chat, giving the misleading impression that the operation failed. Only the intermediate failure is reported — no \"operation succeeded\" follow-up is sen... | closed | completed | true | 1 | [] | [] | 2026-03-27T07:11:45Z | 2026-04-28T05:23:28Z | 2026-04-25T11:12:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | imveraswift | 269,216,890 | U_kgDOEAvseg | User | false |
openclaw/openclaw | 4,150,186,118 | I_kwDOQb6kR873XtCG | 55,631 | https://github.com/openclaw/openclaw/issues/55631 | https://api.github.com/repos/openclaw/openclaw/issues/55631 | [Bug] QQ Bot WebSocket reconnect (4009) triggers main session reset via heartbeat | ## Environment
- Platform: Linux (bare metal, x86_64), systemd user service
- OpenClaw version: 2026-03-24 (latest)
- Channel: QQ Bot enabled
## Steps to Reproduce
1. Run OpenClaw Gateway with QQ Bot channel configured
2. QQ Bot WebSocket disconnects with error `4009 Session timed out` (happens regularly ~every... | closed | completed | true | 2 | [
"bug:behavior",
"channel: qqbot"
] | [
"sliverp"
] | 2026-03-27T07:23:16Z | 2026-04-28T05:23:28Z | 2026-04-25T11:12:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | derekcscn | 269,405,863 | U_kgDOEA7Opw | User | false |
openclaw/openclaw | 4,150,190,379 | I_kwDOQb6kR873XuEr | 55,633 | https://github.com/openclaw/openclaw/issues/55633 | https://api.github.com/repos/openclaw/openclaw/issues/55633 | fix: Suppress startup config spam — 1000+ identical error log entries per gateway restart | ## Problem
Every gateway restart floods the error log with hundreds of identical "Missing config. Run `openclaw setup` or set gateway.mode=local" messages, making it very hard to identify real errors.
## Evidence (last 2 hours)
- `ref_1774594857627` (06:00-07:00 UTC): 515 total errors, ~500+ from config spam
- `ref_... | closed | completed | true | 1 | [] | [] | 2026-03-27T07:24:00Z | 2026-04-28T05:23:27Z | 2026-04-25T11:12:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nqhoag | 23,328,947 | MDQ6VXNlcjIzMzI4OTQ3 | User | false |
openclaw/openclaw | 4,150,190,380 | I_kwDOQb6kR873XuEs | 55,634 | https://github.com/openclaw/openclaw/issues/55634 | https://api.github.com/repos/openclaw/openclaw/issues/55634 | fix: Empty-input Slack thread sessions run for 6+ minutes with no output | ## Problem
Empty-input Slack thread continuation events are triggering full main agent pipeline runs that take **6-7 minutes** before producing no response. This wastes significant compute and tokens.
## Evidence (last 2 hours)
From reflection summaries `ref_1774594857627` and `ref_1774591243847`:
- 5 runs from `ag... | closed | completed | true | 1 | [] | [] | 2026-03-27T07:24:00Z | 2026-04-28T05:23:26Z | 2026-04-25T11:12:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nqhoag | 23,328,947 | MDQ6VXNlcjIzMzI4OTQ3 | User | false |
openclaw/openclaw | 4,150,386,773 | I_kwDOQb6kR873YeBV | 55,647 | https://github.com/openclaw/openclaw/issues/55647 | https://api.github.com/repos/openclaw/openclaw/issues/55647 | 分享：tokens-optimizer - OpenClaw 对话 tokens 优化技能 | # 分享：tokens-optimizer - OpenClaw 对话 tokens 优化技能
## 技能介绍
我开发了一个专门用于优化 OpenClaw 对话 tokens 使用的技能：`tokens-optimizer`。
## 核心功能
1. **自动分析对话模式**：识别优化机会
2. **提供具体优化建议**：包括对话内容和结构优化
3. **计算 tokens 使用情况**：自动计算对话的 tokens 数量和成本
4. **使用跟踪**：监控对话历史和使用趋势
5. **自动触发机制**：当提到相关触发词时自动激活
## 触发词
- save tokens
- optimize conversation
-... | closed | not_planned | true | 2 | [] | [] | 2026-03-27T07:56:23Z | 2026-04-28T05:23:25Z | 2026-04-25T11:12:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | David-Clawdbot | 267,266,786 | U_kgDOD-4q4g | User | false |
openclaw/openclaw | 4,150,417,543 | I_kwDOQb6kR873YliH | 55,649 | https://github.com/openclaw/openclaw/issues/55649 | https://api.github.com/repos/openclaw/openclaw/issues/55649 | mattermost message tool: "buttons" incorrectly marked as required in tool schema | ## Bug
The `message` tool schema for the Mattermost channel has `"buttons"` listed in the `"required"` array alongside `"action"`. This causes a validation error when calling `message` with `action=send` without specifying `buttons`:
```
buttons: must have required property 'buttons'
```
## Expected behavior
`butto... | closed | completed | true | 1 | [] | [] | 2026-03-27T08:01:17Z | 2026-04-28T05:23:25Z | 2026-04-25T11:12:37Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jordanhubbard | 5,066,252 | MDQ6VXNlcjUwNjYyNTI= | User | false |
openclaw/openclaw | 4,150,621,004 | I_kwDOQb6kR873ZXNM | 55,669 | https://github.com/openclaw/openclaw/issues/55669 | https://api.github.com/repos/openclaw/openclaw/issues/55669 | [Feature]: Support multiple OAuth accounts per provider as selectable/fallback route targets | ## Summary
Support multiple OAuth accounts for the same provider (especially multiple ChatGPT/OpenAI OAuth accounts) as distinct route targets that can be selected in `/models` and used in fallback chains.
## Problem to solve
Today, users with multiple ChatGPT OAuth accounts cannot treat them as separate selectable ta... | closed | completed | true | 1 | [] | [] | 2026-03-27T08:32:54Z | 2026-04-28T05:23:24Z | 2026-04-25T11:12:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nicolas-hey | 258,840,845 | U_kgDOD22ZDQ | User | false |
openclaw/openclaw | 4,150,668,891 | I_kwDOQb6kR873Zi5b | 55,676 | https://github.com/openclaw/openclaw/issues/55676 | https://api.github.com/repos/openclaw/openclaw/issues/55676 | [Feature]: ollama local model remote location | ### Summary
ollama local models pointing the url to remote host, but model is attempting to download locally
### Problem to solve
██░▄▄▄░██░▄▄░██░▄▄▄██░▀██░██░▄▄▀██░████░▄▄▀██░███░██
██░███░██░▀▀░██░▄▄▄██░█░█░██░█████░████░▀▀░██░█░█░██
██░▀▀▀░██░█████░▀▀▀██░██▄░██░▀▀▄██░▀▀░█░██░██▄▀▄▀▄██
▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-27T08:40:57Z | 2026-04-28T05:23:23Z | 2026-04-25T11:13:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xxcbzxx | 3,828,765 | MDQ6VXNlcjM4Mjg3NjU= | User | false |
openclaw/openclaw | 4,150,714,048 | I_kwDOQb6kR873Zt7A | 55,679 | https://github.com/openclaw/openclaw/issues/55679 | https://api.github.com/repos/openclaw/openclaw/issues/55679 | [Bug]: memoryFlush never triggers — token projection underestimates by ignoring output tokens | ## Bug Description
The `memoryFlush` mechanism (enabled via `agents.defaults.compaction.memoryFlush`) is designed to write durable memories to `memory/YYYY-MM-DD.md` before a session compacts. However, in practice `memory/` directories remain empty even after multiple compaction events, because the token projection ca... | closed | completed | true | 1 | [] | [] | 2026-03-27T08:47:48Z | 2026-04-28T05:23:22Z | 2026-04-25T11:13:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | luoshan5261486 | 135,194,011 | U_kgDOCA7lmw | User | false |
openclaw/openclaw | 4,150,822,646 | I_kwDOQb6kR873aIb2 | 55,689 | https://github.com/openclaw/openclaw/issues/55689 | https://api.github.com/repos/openclaw/openclaw/issues/55689 | [Bug]: Gateway restart fails when executed from agent context on Windows (schtasks) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Gateway stops but does not restart when `openclaw gateway restart` is executed from within an agent session on Windows (schtasks). Works correctly when run manually in PowerShell admin.
### Steps to reproduce
1. Have OpenC... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-27T09:04:40Z | 2026-04-28T05:23:21Z | 2026-04-25T11:13:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | MingThou | 50,856,367 | MDQ6VXNlcjUwODU2MzY3 | User | false |
openclaw/openclaw | 4,151,204,595 | I_kwDOQb6kR873blrz | 55,715 | https://github.com/openclaw/openclaw/issues/55715 | https://api.github.com/repos/openclaw/openclaw/issues/55715 | whatsapp-cloud: webhook token mismatch can cause duplicate-looking reminders / repeated generation cycles | ## Summary
In `whatsapp-cloud`, users can observe seemingly duplicated reminder/system messages and repeated reply-generation cycles when webhook verification mismatches are occurring.
Observed logs repeatedly show:
- `[whatsapp-cloud] Webhook verification failed — token mismatch`
- frequent repeated `[whatsapp-cloud]... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T10:02:58Z | 2026-04-28T05:23:18Z | 2026-04-25T11:14:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | TheRealJamesRussell | 145,216,038 | U_kgDOCKfSJg | User | false |
openclaw/openclaw | 4,151,254,167 | I_kwDOQb6kR873bxyX | 55,721 | https://github.com/openclaw/openclaw/issues/55721 | https://api.github.com/repos/openclaw/openclaw/issues/55721 | [Bug]: Telegram session: intent commits but tool execution does not dispatch (invoke-first no-op), despite healthy gateway/channel | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
In Telegram-related workflow, the assistant commits to execute a tool call (e.g. read AGENTS.md) but execution is not
reliably dispatched (execution no-op pattern).
I already fixed config warnings and ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T10:09:58Z | 2026-04-28T05:23:17Z | 2026-04-25T11:14:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | VVBakh-86 | 136,478,632 | U_kgDOCCJ_qA | User | false |
openclaw/openclaw | 4,151,539,856 | I_kwDOQb6kR873c3iQ | 55,744 | https://github.com/openclaw/openclaw/issues/55744 | https://api.github.com/repos/openclaw/openclaw/issues/55744 | [Bug]: Whatsapp/Telegram - Openclaw integration | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
## Bug Report: Configured channels never register at runtime; `openclaw status` shows empty Channels table
**Version:** OpenClaw 2026.3.24 (`cff6dc9`)
**OS:** macOS 26.4 arm64
**Node:** 24.14.1
**Install method:** npm gl... | closed | completed | true | 1 | [
"bug",
"bug:crash"
] | [] | 2026-03-27T10:54:49Z | 2026-04-28T05:23:16Z | 2026-04-25T11:14:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | m83iyer | 53,299,066 | MDQ6VXNlcjUzMjk5MDY2 | User | false |
openclaw/openclaw | 4,151,543,443 | I_kwDOQb6kR873c4aT | 55,745 | https://github.com/openclaw/openclaw/issues/55745 | https://api.github.com/repos/openclaw/openclaw/issues/55745 | OpenClawPluginApi should expose typed hooks for before_compaction / after_compaction | ## Problem
Channel plugins currently have no way to react to compaction lifecycle events. This matters for messaging channels (DingTalk, WhatsApp, Telegram, etc.): when compaction runs during an active session, the user's next message is silently queued with no feedback. From the user's perspective, the bot just stops... | closed | completed | true | 1 | [] | [] | 2026-03-27T10:55:24Z | 2026-04-28T05:23:16Z | 2026-04-25T11:14:34Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | BrilliantWang | 120,538,413 | U_kgDOBy9FLQ | User | false |
openclaw/openclaw | 4,151,799,991 | I_kwDOQb6kR873d3C3 | 55,752 | https://github.com/openclaw/openclaw/issues/55752 | https://api.github.com/repos/openclaw/openclaw/issues/55752 | [Bug]: WhatsApp creds.json corrupted on every clean gateway restart (SIGTERM) | ## Summary
On every gateway restart (including clean restarts via `openclaw gateway restart`), the WhatsApp `creds.json` file is found corrupted at next startup and silently restored from backup via `maybeRestoreCredsFromBackup`. This happens 100% reproducibly even on clean SIGTERM shutdowns.
## Environment
- OpenCl... | closed | completed | true | 2 | [] | [] | 2026-03-27T11:34:10Z | 2026-04-28T05:23:15Z | 2026-04-25T11:15:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fintlp | 51,117,127 | MDQ6VXNlcjUxMTE3MTI3 | User | false |
openclaw/openclaw | 4,151,819,207 | I_kwDOQb6kR873d7vH | 55,754 | https://github.com/openclaw/openclaw/issues/55754 | https://api.github.com/repos/openclaw/openclaw/issues/55754 | Bug: local memory remains 0 files/0 chunks on macOS x64 across builtin+Ollama and QMD paths | ## Summary
On macOS x64 with OpenClaw `2026.3.24 (cff6dc9)`, local memory recovery is failing across **both** builtin+Ollama and QMD-backed paths.
What I was trying to do:
- restore stable semantic memory for the main agent
- avoid cloud embedding keys by using a local embedding provider
- get `memory_search` working ... | closed | completed | true | 1 | [] | [
"vincentkoc"
] | 2026-03-27T11:37:10Z | 2026-04-28T05:23:14Z | 2026-04-25T11:15:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | BigApple12138 | 147,795,970 | U_kgDOCM8wAg | User | false |
openclaw/openclaw | 4,151,930,494 | I_kwDOQb6kR873eW5- | 55,760 | https://github.com/openclaw/openclaw/issues/55760 | https://api.github.com/repos/openclaw/openclaw/issues/55760 | [Bug]: Embedded agent auth broken after pi-coding-agent 0.63.0 bump — streamFn overrides bypass new auth-injection wrapper | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading to OpenClaw 2026.3.26 (which bumps `@mariozechner/pi-coding-agent` to 0.63.0), all embedded agent runs fail with `"No API key for provider: <provider>"` because `streamFn` overrides in `attempt.ts` bypass the... | closed | completed | true | 2 | [] | [] | 2026-03-27T11:54:07Z | 2026-04-28T05:23:13Z | 2026-04-25T11:15:16Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rbutera | 6,047,293 | MDQ6VXNlcjYwNDcyOTM= | User | false |
openclaw/openclaw | 4,151,959,444 | I_kwDOQb6kR873ed-U | 55,763 | https://github.com/openclaw/openclaw/issues/55763 | https://api.github.com/repos/openclaw/openclaw/issues/55763 | Discord WebSocket drops ~hourly with stale-socket; messages lost during reconnect window | ## Problem
Discord connection drops roughly every hour with `stale-socket` reason. The health-monitor detects it and restarts the provider, but messages sent during the ~30-60 second reconnect window are silently lost — never processed.
A secondary issue: when Anthropic returns `api_error: Internal server error`, the... | closed | completed | true | 1 | [] | [] | 2026-03-27T11:57:15Z | 2026-04-28T05:23:13Z | 2026-04-25T11:15:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sabo961 | 191,984 | MDQ6VXNlcjE5MTk4NA== | User | false |
openclaw/openclaw | 4,152,081,760 | I_kwDOQb6kR873e71g | 55,769 | https://github.com/openclaw/openclaw/issues/55769 | https://api.github.com/repos/openclaw/openclaw/issues/55769 | Subagents return empty responses with all external providers (OpenRouter, Google) — only default MiniMax provider works | ## Bug Description
Subagents spawned with non-default model overrides return empty responses (0 tokens in, 0 tokens out) even though the subagent task completes successfully with status `completed successfully`.
This affects **all external providers**, not just specific models.
## Reproduction Steps
1. Configure an... | closed | completed | true | 1 | [] | [] | 2026-03-27T12:12:24Z | 2026-04-28T05:23:12Z | 2026-04-25T11:15:30Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | franciscoaccabral | 42,218,143 | MDQ6VXNlcjQyMjE4MTQz | User | false |
openclaw/openclaw | 4,152,392,225 | I_kwDOQb6kR873gHoh | 55,786 | https://github.com/openclaw/openclaw/issues/55786 | https://api.github.com/repos/openclaw/openclaw/issues/55786 | REACTION_INVALID warnings surface as failed message alerts despite suppressToolErrors=true | ## Bug
When a Telegram bot attempts to react with an emoji not in the bot-allowed reaction set (e.g. 🟢, 🫡), the Telegram API returns `REACTION_INVALID`.
The OpenClaw code catches this correctly in `pi-embedded-BaSvmUpW.js:83543` and returns `{ ok: false, warning: 'Reaction unavailable: <emoji>' }` — but this warni... | closed | completed | true | 1 | [] | [] | 2026-03-27T12:54:10Z | 2026-04-28T05:23:11Z | 2026-04-25T11:15:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | leoclimaco88 | 109,989,108 | U_kgDOBo5M9A | User | false |
openclaw/openclaw | 4,152,541,882 | I_kwDOQb6kR873gsK6 | 55,797 | https://github.com/openclaw/openclaw/issues/55797 | https://api.github.com/repos/openclaw/openclaw/issues/55797 | [Bug]: 2026.3.24 models auth login for openai-codex triggers openclaw.json.clobbered storm in containerized pod | ## Summary
On OpenClaw `2026.3.24`, a clean containerized deployment can start healthy, but running:
```bash
openclaw models auth login --provider openai-codex
```
inside the pod can trigger an `openclaw.json.clobbered.*` storm under the persisted runtime state directory. After that, the pod regresses from a healthy... | closed | completed | true | 1 | [] | [] | 2026-03-27T13:13:36Z | 2026-04-28T05:23:10Z | 2026-04-25T11:15:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | benops-io | 270,463,089 | U_kgDOEB7wcQ | User | false |
openclaw/openclaw | 4,205,966,516 | I_kwDOQb6kR876sfS0 | 61,064 | https://github.com/openclaw/openclaw/issues/61064 | https://api.github.com/repos/openclaw/openclaw/issues/61064 | Bug: MiniMax image-01 image-to-image blocked by resolution override check in providers manifest | ## Bug Description
**Provider:** minimax-portal/image-01 and minimax/image-01
**OpenClaw Version:** 2026.4.2
**Error:** \`"minimax-portal edit does not support resolution overrides"\`
When attempting to use the \`image_generate\` tool with a reference image (image-to-image/edit mode) on MiniMax image-01, the OpenClaw... | closed | completed | true | 2 | [] | [] | 2026-04-04T22:55:10Z | 2026-04-28T05:23:10Z | 2026-04-25T11:15:46Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mabeliabot | 271,634,889 | U_kgDOEDDRyQ | User | false |
openclaw/openclaw | 4,158,159,142 | I_kwDOQb6kR8732Hkm | 56,237 | https://github.com/openclaw/openclaw/issues/56237 | https://api.github.com/repos/openclaw/openclaw/issues/56237 | [Bug] image_generate with reference image fails: 'minimax-portal edit does not support resolution overrides' | ## Bug Description
When using the `image_generate` tool with an `image` parameter (reference image / edit mode), the request fails with error:
```
minimax-portal edit does not support resolution overrides
```
## Environment
- **OpenClaw version**: 2026.3.24
- **OS**: Windows (node 24.14.0)
- **Model**: minimax-port... | closed | completed | true | 1 | [] | [] | 2026-03-28T06:47:28Z | 2026-04-28T05:23:09Z | 2026-04-25T11:15:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | cq2000419 | 271,556,261 | U_kgDOEC-epQ | User | false |
openclaw/openclaw | 4,152,562,184 | I_kwDOQb6kR873gxII | 55,799 | https://github.com/openclaw/openclaw/issues/55799 | https://api.github.com/repos/openclaw/openclaw/issues/55799 | Integrate Safe Browsing checks into OpenClaw core for default web resource screening | ## Summary
Add built-in Safe Browsing support to OpenClaw core so agents can screen web resources by default before using them.
## Problem
Right now, Safe Browsing can be added as a custom/local plugin, but the behavior "scan URLs before using web resources" is not something that can be enforced cleanly at plugin le... | closed | not_planned | true | 1 | [] | [] | 2026-03-27T13:16:05Z | 2026-04-28T05:23:08Z | 2026-04-25T11:15:50Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rmartensen | 9,069,215 | MDQ6VXNlcjkwNjkyMTU= | User | false |
openclaw/openclaw | 4,152,604,491 | I_kwDOQb6kR873g7dL | 55,802 | https://github.com/openclaw/openclaw/issues/55802 | https://api.github.com/repos/openclaw/openclaw/issues/55802 | feat: make exec obfuscation command length threshold configurable (tools.exec.maxCommandChars) | ## Problem
The `2026.3.23-2` release introduced a hardcoded 10,000 character limit on exec commands (`MAX_COMMAND_CHARS = 1e4` in `exec-obfuscation-detect.ts`). When a command exceeds this threshold, it is flagged as obfuscated and routed through approval — **regardless of `tools.exec.ask` or `tools.exec.security` set... | closed | completed | true | 2 | [] | [] | 2026-03-27T13:21:21Z | 2026-04-28T05:23:07Z | 2026-04-25T11:16:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | smokeyclaw-ai[bot] | 266,257,850 | BOT_kgDOD97Fug | Bot | false |
openclaw/openclaw | 4,152,621,807 | I_kwDOQb6kR873g_rv | 55,803 | https://github.com/openclaw/openclaw/issues/55803 | https://api.github.com/repos/openclaw/openclaw/issues/55803 | Plugin system audit: missing load/health status in logs | ## Problem
When plugins are configured and enabled, there is no log entry indicating whether they loaded successfully, are healthy, or failed to initialize. This makes it impossible to audit plugin status during runtime.
## Current Behavior
Plugin configuration is visible in config:
```json
"plugins": {
"entries"... | closed | completed | true | 1 | [] | [] | 2026-03-27T13:23:45Z | 2026-04-28T05:23:06Z | 2026-04-25T11:16:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | laterequation | 138,646,637 | U_kgDOCEOUbQ | User | false |
openclaw/openclaw | 4,153,078,740 | I_kwDOQb6kR873ivPU | 55,822 | https://github.com/openclaw/openclaw/issues/55822 | https://api.github.com/repos/openclaw/openclaw/issues/55822 | [Feature Request] Support Keychain/Secret Store for API Keys in openclaw.json | ## Problem
Currently, `openclaw.json` requires API keys and secrets to be stored in plaintext:
```json
{
"models": {
"providers": {
"openrouter": {
"apiKey": "sk-or-v1-..."
}
}
}
}
```
This is a security risk — anyone with file access can read all credentials.
## Proposed Solution
S... | closed | completed | true | 1 | [] | [] | 2026-03-27T14:14:21Z | 2026-04-28T05:23:05Z | 2026-04-25T11:16:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Andy283 | 52,738,901 | MDQ6VXNlcjUyNzM4OTAx | User | false |
openclaw/openclaw | 4,153,119,767 | I_kwDOQb6kR873i5QX | 55,825 | https://github.com/openclaw/openclaw/issues/55825 | https://api.github.com/repos/openclaw/openclaw/issues/55825 | webchat 前端显示内部 memory 上下文片段 | ## 问题描述
webchat 前端错误地将发送给 AI 的内部 memory 检索结果显示在用户聊天界面中。
## 复现场景
1. 用户发送消息
2. OpenClaw 压缩对话历史并检索 memory 系统
3. memory 片段作为上下文发送给 AI（正常）
4. **问题**：这些 memory 片段也被渲染到用户聊天界面（不应发生）
## 预期行为
memory 检索结果等内部上下文应该：
- 只发送给 AI
- 不显示在用户聊天界面
## 实际显示示例
用户看到类似这样的内容：
```
## User's conversation history (from memory system)
IMPORTA... | closed | completed | true | 2 | [] | [] | 2026-03-27T14:18:33Z | 2026-04-28T05:23:04Z | 2026-04-25T11:16:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jigeyisi478 | 267,347,508 | U_kgDOD-9mNA | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.