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 | 4,050,680,745 | I_kwDOQb6kR87xcHup | 42,026 | https://github.com/openclaw/openclaw/issues/42026 | https://api.github.com/repos/openclaw/openclaw/issues/42026 | RFC: Distributed Agent Runtime — separate control plane from agent compute | ## Summary
Split the monolithic `openclaw gateway` into two components:
1. **Control Plane** (`openclaw-gateway`) — lightweight process handling channels, routing, cron scheduling, hooks, dashboard, and agent registry
2. **Agent Runtime** (`openclaw-runtime`) — per-agent process running the model loop, context engine... | open | null | false | 5 | [] | [] | 2026-03-10T09:58:49Z | 2026-05-04T13:18:25Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | dwinter3 | 1,416,878 | MDQ6VXNlcjE0MTY4Nzg= | User | false |
openclaw/openclaw | 4,050,809,776 | I_kwDOQb6kR87xcnOw | 42,039 | https://github.com/openclaw/openclaw/issues/42039 | https://api.github.com/repos/openclaw/openclaw/issues/42039 | Expose maxMissedJobsPerRestart and missedJobStaggerMs in cron config | ## Feature Request
`maxMissedJobsPerRestart` (default: 5) and `missedJobStaggerMs` (default: 5000) exist as internal `CronServiceDeps` parameters but are **not exposed** in the public `CronConfig` schema and cannot be set via `openclaw.json`.
## Problem
When the gateway restarts (e.g., nightly update, manual restart... | open | null | false | 2 | [] | [] | 2026-03-10T10:22:45Z | 2026-05-04T12:44:55Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | mySebbe | 14,988,278 | MDQ6VXNlcjE0OTg4Mjc4 | User | false |
openclaw/openclaw | 4,051,061,365 | I_kwDOQb6kR87xdkp1 | 42,079 | https://github.com/openclaw/openclaw/issues/42079 | https://api.github.com/repos/openclaw/openclaw/issues/42079 | [Bug]: feishu channel embedded run agent end: runId=xxxxxxxxx isError=true error=LLM request timed out. | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="1435" height="234" alt="Image" src="https://github.com/user-attachments/assets/d4a674a4-ae8f-4d66-82af-c54f88cfd67a" />
<img width="1426" height="696" alt="Image" src="https://github.com/user-attachments/assets/d5d37036-b1f8-4393-86be-dbddd5... | open | null | false | 4 | [
"bug",
"regression"
] | [] | 2026-03-10T11:12:44Z | 2026-05-04T12:44:57Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ctnger | 35,954,900 | MDQ6VXNlcjM1OTU0OTAw | User | false |
openclaw/openclaw | 4,051,277,352 | I_kwDOQb6kR87xeZYo | 42,099 | https://github.com/openclaw/openclaw/issues/42099 | https://api.github.com/repos/openclaw/openclaw/issues/42099 | fix(plugins): false-positive duplicate plugin ID warning on gateway start (msteams) | ## Summary
On every gateway start, a `warn` log fires for duplicate plugin IDs — most visibly for the `msteams` plugin. This is a false positive: the same physical plugin is being discovered via multiple paths or origins (e.g. bundled entry shadowed by a config-level entry), and the existing deduplication logic treats... | open | null | false | 3 | [] | [] | 2026-03-10T11:54:28Z | 2026-05-04T11:26:38Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | GitDakky | 106,489,879 | U_kgDOBljoFw | User | false |
openclaw/openclaw | 4,051,331,744 | I_kwDOQb6kR87xemqg | 42,106 | https://github.com/openclaw/openclaw/issues/42106 | https://api.github.com/repos/openclaw/openclaw/issues/42106 | [Bug]: Block streaming can drop paragraph separators across streamed updates | ### Bug type
Behavior bug
### Summary
When block streaming splits a reply across multiple streamed deliveries, paragraph separators can be lost from the reconstructed visible text. This collapses Markdown structure such as a heading followed by a paragraph into a single line.
### Steps to reproduce
1. Enable block st... | open | null | false | 2 | [] | [] | 2026-03-10T12:04:20Z | 2026-05-04T11:26:40Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | wbxl2000 | 57,169,560 | MDQ6VXNlcjU3MTY5NTYw | User | false |
openclaw/openclaw | 4,051,564,528 | I_kwDOQb6kR87xfffw | 42,139 | https://github.com/openclaw/openclaw/issues/42139 | https://api.github.com/repos/openclaw/openclaw/issues/42139 | ACP external harness on remote Mac node is unclear/broken in Discord workspace flow | ## Summary
I'm trying to use OpenClaw's **official ACP flow** for Discord workspaces where:
- the **Gateway runs on a Raspberry Pi**
- the actual coding harnesses (**Claude Code / Codex**) are expected to run on a **remote Mac mini node**
- Discord channels are used as the user-facing workspaces
- the goal is to stay... | open | null | false | 2 | [] | [] | 2026-03-10T12:45:23Z | 2026-05-04T11:26:42Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | JonbinC | 62,364,199 | MDQ6VXNlcjYyMzY0MTk5 | User | false |
openclaw/openclaw | 4,051,693,241 | I_kwDOQb6kR87xf-65 | 42,156 | https://github.com/openclaw/openclaw/issues/42156 | https://api.github.com/repos/openclaw/openclaw/issues/42156 | Telegram callback queries time out when agent turn is queued behind sequentialize | Telegram callback queries (inline button presses) have a ~15s server-side timeout for `answerCallbackQuery`. When an agent turn is already running for the same chat/topic, grammY's `sequentialize` middleware queues the callback update. By the time it reaches the handler, the answer window has expired and the user sees ... | open | null | false | 3 | [] | [] | 2026-03-10T13:06:27Z | 2026-05-04T11:26:44Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Diaspar4u | 3,605,840 | MDQ6VXNlcjM2MDU4NDA= | User | false |
openclaw/openclaw | 4,051,693,350 | I_kwDOQb6kR87xf-8m | 42,157 | https://github.com/openclaw/openclaw/issues/42157 | https://api.github.com/repos/openclaw/openclaw/issues/42157 | Inbound Telegram webhook payloads lost on gateway restart | In webhook mode, the gateway responds 200 OK to Telegram's webhook delivery immediately. If the gateway restarts while messages are queued or mid-processing, those messages are permanently lost -- Telegram considers them delivered and will not retry.
This affects any restart path: config change, manual restart, update... | open | null | false | 3 | [] | [] | 2026-03-10T13:06:28Z | 2026-05-04T11:26:47Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Diaspar4u | 3,605,840 | MDQ6VXNlcjM2MDU4NDA= | User | false |
openclaw/openclaw | 4,051,720,701 | I_kwDOQb6kR87xgFn9 | 42,165 | https://github.com/openclaw/openclaw/issues/42165 | https://api.github.com/repos/openclaw/openclaw/issues/42165 | [Bug]: tools.profile: "minimal" exposes read/write/edit unless group:fs is explicitly denied | ### Bug type
Regression (worked before, now fails)
### Summary
tools.profile: "minimal" is documented as exposing session_status only, but in practice the agent still receives filesystem tools (read, write, edit) unless deny: ["group:fs"] is added explicitly.
... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-10T13:11:20Z | 2026-05-03T04:37:18Z | 2026-04-30T13:59:29Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | LaskerSaw | 130,273,415 | U_kgDOB8PQhw | User | false |
openclaw/openclaw | 4,051,936,623 | I_kwDOQb6kR87xg6Vv | 42,196 | https://github.com/openclaw/openclaw/issues/42196 | https://api.github.com/repos/openclaw/openclaw/issues/42196 | cron add should require --agent flag or warn when not specified | When creating a cron job with cron add, the --agent flag is optional. If not specified, the job runs with the default agent (main), which can cause confusion.
Problem:
- A cron job related to a specific agent was executed by the wrong agent
- The delivery channel is separate from which agent executes the task
- Withou... | closed | completed | true | 2 | [] | [] | 2026-03-10T13:48:01Z | 2026-05-03T04:38:04Z | 2026-04-30T07:45:43Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | padixa | 103,999,705 | U_kgDOBjLo2Q | User | false |
openclaw/openclaw | 4,052,307,931 | I_kwDOQb6kR87xiU_b | 42,243 | https://github.com/openclaw/openclaw/issues/42243 | https://api.github.com/repos/openclaw/openclaw/issues/42243 | [Bug]: Announce delivery posts raw error JSON to channels when cron run fails | ### Summary
When a cron job fails (e.g., provider returns 500) and `delivery.mode: "announce"`, the raw error text is delivered to the configured Discord/Telegram channel as if it were normal content. Users see JSON error dumps in their task channels.
### Steps to reproduce
1. Create a cron job with `delivery.mode: ... | open | null | false | 3 | [] | [] | 2026-03-10T14:46:54Z | 2026-05-04T11:26:48Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | sercada | 24,389,792 | MDQ6VXNlcjI0Mzg5Nzky | User | false |
openclaw/openclaw | 4,052,350,540 | I_kwDOQb6kR87xifZM | 42,252 | https://github.com/openclaw/openclaw/issues/42252 | https://api.github.com/repos/openclaw/openclaw/issues/42252 | Improve doctor/gateway diagnostics clarity for mixed LaunchAgent/runtime states | ## Summary
`openclaw doctor` / gateway diagnostics can be confusing in mixed LaunchAgent + runtime states. It can report conflicting status lines in one run (for example: "Gateway not running" then later "Runtime: running"), and restart/install guidance is not always actionable when service entrypoint drift or duplicat... | open | null | false | 2 | [] | [] | 2026-03-10T14:54:05Z | 2026-05-04T13:18:26Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ion-ctrl | 266,816,281 | U_kgDOD-dLGQ | User | false |
openclaw/openclaw | 4,052,416,104 | I_kwDOQb6kR87xivZo | 42,261 | https://github.com/openclaw/openclaw/issues/42261 | https://api.github.com/repos/openclaw/openclaw/issues/42261 | fix(gateway): webchat/WS clients never receive live thinking events | ## Problem
WebSocket clients (webchat, Control UI, third-party dashboards) never receive `stream: "thinking"` agent events during model reasoning, even when reasoning streaming is enabled. The agent appears frozen during the entire thinking phase.
## Root Cause
Two guards in `src/agents/pi-embedded-subscribe.ts` pre... | closed | completed | true | 3 | [] | [] | 2026-03-10T15:04:39Z | 2026-05-02T04:27:15Z | 2026-04-29T20:33:06Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | mode80 | 656,321 | MDQ6VXNlcjY1NjMyMQ== | User | false |
openclaw/openclaw | 4,052,464,266 | I_kwDOQb6kR87xi7KK | 42,273 | https://github.com/openclaw/openclaw/issues/42273 | https://api.github.com/repos/openclaw/openclaw/issues/42273 | backup create stalls on large installations (4GB+ .openclaw dir) | ## Problem
`openclaw backup create` stalls and never completes when the `.openclaw` directory is large. The command writes a small `.tmp` file (10B-42KB) and then silently dies without error output or a completed archive.
## Environment
- OpenClaw 2026.3.8 (3caab92)
- macOS 25.3 (arm64)
- `.openclaw` size: **4.1 GB*... | open | null | false | 4 | [] | [] | 2026-03-10T15:12:23Z | 2026-05-04T11:26:54Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | samshields-oc | 261,743,399 | U_kgDOD5njJw | User | false |
openclaw/openclaw | 4,052,586,192 | I_kwDOQb6kR87xjY7Q | 42,291 | https://github.com/openclaw/openclaw/issues/42291 | https://api.github.com/repos/openclaw/openclaw/issues/42291 | [Bug]: `openclaw models auth login` on node host writes OAuth token locally instead of syncing to gateway | ## Summary
When running `openclaw models auth login --provider openai-codex` on a **node host** (e.g., Sparky) that is connected to a remote gateway, the OAuth token is written to the node's local `~/.openclaw/agents/main/agent/auth-profiles.json` instead of being pushed to the gateway where model calls actually happe... | open | null | false | 2 | [] | [] | 2026-03-10T15:33:35Z | 2026-05-04T11:26:58Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | dyoung522 | 422,900 | MDQ6VXNlcjQyMjkwMA== | User | false |
openclaw/openclaw | 4,333,505,806 | I_kwDOQb6kR88AAAABAkwNDg | 72,641 | https://github.com/openclaw/openclaw/issues/72641 | https://api.github.com/repos/openclaw/openclaw/issues/72641 | WebChat UI: agent responses rendered as duplicate messages | ## Bug Description
In the OpenClaw WebChat control UI (`openclaw-control-ui`), agent responses are sometimes rendered as two identical chat bubbles. Both duplicates carry the same metadata (token counts, context usage percentage), confirming this is a **client-side rendering bug** — the server only sends the message o... | closed | completed | true | 2 | [] | [] | 2026-04-27T06:20:22Z | 2026-04-30T04:34:42Z | 2026-04-27T10:58:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yangli3348-Git | 246,403,690 | U_kgDODq_Sag | User | false |
openclaw/openclaw | 4,052,914,823 | I_kwDOQb6kR87xkpKH | 42,351 | https://github.com/openclaw/openclaw/issues/42351 | https://api.github.com/repos/openclaw/openclaw/issues/42351 | Gateway 重启 - 缺少 bot_p2p_chat_entered_v1 事件处理器 | # Issue: Gateway 重启 - 缺少 bot_p2p_chat_entered_v1 事件处理器
## 问题描述
当用户打开飞书机器人私聊窗口时，会触发 `im.chat.access_event.bot_p2p_chat_entered_v1` 事件。由于 OpenClaw 当前版本缺少该事件处理器，导致 Gateway 短暂重启（约 10 秒），期间 webchat 连接断开。
## 环境信息
- **OpenClaw 版本**: 2026.3.8 (3caab92)
- **Node.js**: v25.7.0
- **操作系统**: macOS 26.3.1 (Tahoe) arm64
- **飞书渠道*... | open | null | false | 2 | [] | [] | 2026-03-10T16:36:13Z | 2026-05-04T00:30:15Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | sunking0223 | 237,174,412 | U_kgDODiL-jA | User | false |
openclaw/openclaw | 4,052,953,463 | I_kwDOQb6kR87xkyl3 | 42,354 | https://github.com/openclaw/openclaw/issues/42354 | https://api.github.com/repos/openclaw/openclaw/issues/42354 | [Bug]: [Feishu/Lark] WebSocket long connection fails - code: 1000040351, system busy | ### Bug type
Regression (worked before, now fails)
### Summary
Feishu channel WebSocket long connection cannot be established. Lark console shows "No application connection detected" when saving event subscription. Error loops: "code: 1000040351, system busy" and "Cannot read properties of undefined (reading 'PingIn... | open | null | false | 4 | [
"bug",
"regression"
] | [] | 2026-03-10T16:43:43Z | 2026-05-04T11:27:12Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | xifengxx | 15,701,172 | MDQ6VXNlcjE1NzAxMTcy | User | false |
openclaw/openclaw | 4,053,070,935 | I_kwDOQb6kR87xlPRX | 42,373 | https://github.com/openclaw/openclaw/issues/42373 | https://api.github.com/repos/openclaw/openclaw/issues/42373 | Feature: Add costCurrency config option to customize cost display currency | ## Problem
Currently, OpenClaw displays costs using the `$` symbol hardcoded in the UI (e.g., `$0.07`). However, users in different regions may want to display costs in their local currency (e.g., `¥` for CNY, `€` for EUR).
The `cost` field in the model configuration already accepts numeric values, but there is no wa... | open | null | false | 2 | [] | [] | 2026-03-10T17:05:14Z | 2026-05-04T11:27:16Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | lizhihong114 | 37,773,237 | MDQ6VXNlcjM3NzczMjM3 | User | false |
openclaw/openclaw | 4,053,145,427 | I_kwDOQb6kR87xlhdT | 42,391 | https://github.com/openclaw/openclaw/issues/42391 | https://api.github.com/repos/openclaw/openclaw/issues/42391 | Expose structured progress metrics in hook events for plugin-based live progress cards | ## Background
We have integrated `openclaw-progress-plugin` in Feishu DM and validated stable live card updates (send/update flow works).
Current UX gap: progress is mostly textual (`进行中/执行中`) because OpenClaw hook payloads do not consistently expose structured progress metrics (`stepIndex/stepTotal/percent`).
## Prob... | open | null | false | 2 | [] | [] | 2026-03-10T17:19:56Z | 2026-05-04T11:27:18Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | cybersentia | 32,168,519 | MDQ6VXNlcjMyMTY4NTE5 | User | false |
openclaw/openclaw | 4,053,266,325 | I_kwDOQb6kR87xl--V | 42,408 | https://github.com/openclaw/openclaw/issues/42408 | https://api.github.com/repos/openclaw/openclaw/issues/42408 | [Bug/Docs]: local+hybrid memory_search quality can appear unstable due to extraPaths path drift + benchmark-file contamination | ### Summary
On OpenClaw `2026.3.8` (macOS arm64), we observed that `memory_search` quality under `provider=local` + `hybrid` can look severely degraded (including empty/irrelevant top hits), but root cause was a **combination** of:
1. `extraPaths` path drift (configured paths not aligned with the active main workspace... | open | null | false | 2 | [] | [] | 2026-03-10T17:42:32Z | 2026-05-04T11:27:24Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | 1sexywoo8 | 262,731,466 | U_kgDOD6j2yg | User | false |
openclaw/openclaw | 4,053,365,410 | I_kwDOQb6kR87xmXKi | 42,424 | https://github.com/openclaw/openclaw/issues/42424 | https://api.github.com/repos/openclaw/openclaw/issues/42424 | Discord self-authored messages can survive into replay/followup paths and persist across /new | ## Summary
Discord self-message protection appears to be incomplete across ingress and replay paths.
In one observed case on March 10, 2026, a Discord bot account started with `logged in to discord` but without a resolved bot identity. After that, the bot began treating its own channel messages as inbound `user` mess... | closed | completed | true | 2 | [] | [] | 2026-03-10T18:03:11Z | 2026-05-03T04:35:53Z | 2026-04-30T17:04:07Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | Citrus086 | 203,685,922 | U_kgDODCQAIg | User | false |
openclaw/openclaw | 4,053,678,792 | I_kwDOQb6kR87xnjrI | 42,461 | https://github.com/openclaw/openclaw/issues/42461 | https://api.github.com/repos/openclaw/openclaw/issues/42461 | [Bug]: openclaw tui does not honor the active Gateway runtime port and still targets 18789 | ### Bug type
Regression (worked before, now fails)
### Summary
When the Gateway is started on a custom port such as 48789, openclaw tui can still try to connect to the default local port 18789. This happens because the TUI
connection path resolves its Gateway URL from config/defaults without first applying the act... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-03-10T19:09:40Z | 2026-05-04T11:27:28Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | jackm1688 | 50,076,324 | MDQ6VXNlcjUwMDc2MzI0 | User | false |
openclaw/openclaw | 4,053,809,755 | I_kwDOQb6kR87xoDpb | 42,472 | https://github.com/openclaw/openclaw/issues/42472 | https://api.github.com/repos/openclaw/openclaw/issues/42472 | Feature: post-process NO_REPLY to suppress partial leaks | ## Problem
When the model outputs text followed by `NO_REPLY` (e.g. `"summary of work done...NO_REPLY"`), the text before `NO_REPLY` leaks to the user as a visible message. This violates the intended behavior where `NO_REPLY` should suppress the entire reply.
This is especially common with weaker models (e.g. GLM-5) ... | open | null | false | 2 | [] | [] | 2026-03-10T19:35:06Z | 2026-05-04T11:32:19Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | the3asic | 17,235,633 | MDQ6VXNlcjE3MjM1NjMz | User | false |
openclaw/openclaw | 4,053,870,821 | I_kwDOQb6kR87xoSjl | 42,475 | https://github.com/openclaw/openclaw/issues/42475 | https://api.github.com/repos/openclaw/openclaw/issues/42475 | [Feature]: Per-agent cost budget enforcement at the gateway level | ### Summary
Add optional per-agent cost budgets (daily/monthly caps) enforced at the gateway before dispatching model calls, so operators can prevent runaway spend without external monitoring.
### Problem to solve
OpenClaw tracks per-session cost and usage via `session-cost-usage.ts` (daily breakdowns, per-model usa... | open | null | false | 9 | [] | [] | 2026-03-10T19:47:09Z | 2026-05-04T12:44:59Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | hkochar | 254,796 | MDQ6VXNlcjI1NDc5Ng== | User | false |
openclaw/openclaw | 4,054,292,355 | I_kwDOQb6kR87xp5eD | 42,504 | https://github.com/openclaw/openclaw/issues/42504 | https://api.github.com/repos/openclaw/openclaw/issues/42504 | Control UI: upload agent avatars from Agents tab | There’s no UI path to set agent avatars today. This patch adds an admin RPC (`agents.avatar.upload`) and a Control UI picker on /agents so clicking an avatar opens a file dialog, stores the image in `workspace/avatars/`, and updates `IDENTITY.md`. The UI then renders the avatar via /avatar/<agentId>.
Expected:
- Click... | open | null | false | 2 | [] | [] | 2026-03-10T21:21:28Z | 2026-05-04T11:27:37Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | matthewxmurphy | 110,652,790 | U_kgDOBphtdg | User | false |
openclaw/openclaw | 4,054,332,894 | I_kwDOQb6kR87xqDXe | 42,510 | https://github.com/openclaw/openclaw/issues/42510 | https://api.github.com/repos/openclaw/openclaw/issues/42510 | Google Chat: replyToMode: "off" does not suppress thread replies | Since v2026.3.1 (PR #30965), Google Chat replies always attach to threads via `messageReplyOption=REPLY_MESSAGE_FALLBACK_TO_NEW_THREAD`. However, setting `channels.googlechat.replyToMode: "off"` has no effect — `sendText` and `sendMedia` in `channel.ts` unconditionally forward `threadId ?? replyToId` to `sendGoogleChat... | open | null | false | 3 | [] | [] | 2026-03-10T21:31:46Z | 2026-05-04T11:27:39Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | svmark | 52,013,853 | MDQ6VXNlcjUyMDEzODUz | User | false |
openclaw/openclaw | 4,054,504,623 | I_kwDOQb6kR87xqtSv | 42,533 | https://github.com/openclaw/openclaw/issues/42533 | https://api.github.com/repos/openclaw/openclaw/issues/42533 | [Bug]: Talk Mode PTT produces len=0 with multi-channel audio interfaces (SFSpeechRecognizer silent failure) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Talk Mode Push-to-Talk always returns `emptyOnRelease len=0` when the system audio input device has more than 2 channels. SFSpeechRecognizer silently fails — no error, no transcript — because it cannot handle multi-channel audio formats. Th... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-10T22:14:31Z | 2026-05-04T04:34:39Z | 2026-05-02T01:56:09Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | jbuecker | 15,744,608 | MDQ6VXNlcjE1NzQ0NjA4 | User | false |
openclaw/openclaw | 4,054,535,018 | I_kwDOQb6kR87xq0tq | 42,538 | https://github.com/openclaw/openclaw/issues/42538 | https://api.github.com/repos/openclaw/openclaw/issues/42538 | Bug: health endpoint returns incorrect running=false for WhatsApp | ## Summary
The `health` endpoint returns `running: false, connected: false` for WhatsApp even when the channel is active and working correctly. The `channels.status` endpoint returns the correct values.
## Steps to Reproduce
1. Start OpenClaw gateway with WhatsApp linked and connected
2. Run: `openclaw gateway call ... | open | null | false | 2 | [] | [] | 2026-03-10T22:22:05Z | 2026-05-03T22:34:11Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | ajayr | 809,529 | MDQ6VXNlcjgwOTUyOQ== | User | false |
openclaw/openclaw | 4,054,536,693 | I_kwDOQb6kR87xq1H1 | 42,539 | https://github.com/openclaw/openclaw/issues/42539 | https://api.github.com/repos/openclaw/openclaw/issues/42539 | Feature: TTS delivery mode option — separate text + voice messages on Telegram | ## Problem
When TTS is enabled on Telegram, the voice note and text are delivered as a single message (voice with text caption). Telegram renders voice note captions in a narrow bubble alongside the audio player, making longer text responses difficult to read.
## Current behavior
- Voice note + text caption delivere... | open | null | false | 2 | [] | [] | 2026-03-10T22:22:25Z | 2026-05-04T11:27:43Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | holisticseo | 216,781,566 | U_kgDODOvS_g | User | false |
openclaw/openclaw | 4,054,955,018 | I_kwDOQb6kR87xsbQK | 42,591 | https://github.com/openclaw/openclaw/issues/42591 | https://api.github.com/repos/openclaw/openclaw/issues/42591 | [优化建议] install.sh 模块化拆分以提升可维护性 | ## 问题描述
当前 `scripts/install.sh` 文件大小约 79KB，包含 2498 行代码，承担了过多的职责：
1. **下载工具管理** (curl/wget 检测和下载)
2. **UI 交互** (gum 工具集成、颜色输出、进度显示)
3. **平台检测** (OS/Arch 检测、包管理器识别)
4. **安装逻辑** (npm/git 安装、构建工具自动安装)
5. **错误处理** (npm 错误诊断、重试逻辑)
这种单体结构导致：
- 代码难以阅读和审查
- 功能耦合度高，修改一处可能影响其他部分
- 难以编写单元测试
- 新贡献者难以快速理解代码结构
## 优化方案
将 `install... | open | null | false | 2 | [] | [] | 2026-03-11T00:26:53Z | 2026-05-04T11:27:52Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | nillikechatchat | 266,501,707 | U_kgDOD-J-Sw | User | false |
openclaw/openclaw | 4,054,957,120 | I_kwDOQb6kR87xsbxA | 42,592 | https://github.com/openclaw/openclaw/issues/42592 | https://api.github.com/repos/openclaw/openclaw/issues/42592 | [优化建议] 添加结构化错误处理和智能重试机制 | ## 问题描述
当前安装脚本在错误处理方面存在以下问题：
1. **错误掩盖**：多处使用 `|| true` 掩盖错误，导致问题难以发现
```bash
current_allowed_origins="$(
docker compose ... config get ... 2>/dev/null || true
)"
```
2. **缺乏错误分类**：npm 安装失败时没有根据错误类型采取不同策略
- 网络错误 (ECONNREFUSED, ETIMEDOUT) 应该重试
- 权限错误 (EACCES) 应该尝试自动修复
- 其他错误应该立即失败并给出明确提示
3.... | open | null | false | 2 | [] | [] | 2026-03-11T00:27:33Z | 2026-05-04T11:27:54Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | nillikechatchat | 266,501,707 | U_kgDOD-J-Sw | User | false |
openclaw/openclaw | 4,055,122,393 | I_kwDOQb6kR87xtEHZ | 42,631 | https://github.com/openclaw/openclaw/issues/42631 | https://api.github.com/repos/openclaw/openclaw/issues/42631 | [Feature]: Allow job-level model override for cron jobs using sessionTarget=main + payload.kind=systemEvent | ## Summary
Please support job-level model selection for cron jobs that run via `sessionTarget: "main"` + `payload.kind: "systemEvent"`.
Today, users who want stable cron execution on a cheaper model are pushed into an awkward tradeoff:
- `isolated + agentTurn` allows straightforward model selection (`payload.model`)... | open | null | false | 2 | [] | [] | 2026-03-11T01:29:47Z | 2026-05-04T11:27:58Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Kaze310 | 183,121,490 | U_kgDOCuo2Ug | User | false |
openclaw/openclaw | 4,055,173,704 | I_kwDOQb6kR87xtQpI | 42,646 | https://github.com/openclaw/openclaw/issues/42646 | https://api.github.com/repos/openclaw/openclaw/issues/42646 | Memory MVP: define SQLite schema and canonical record model | ## Summary
Define the v1 SQLite schema for a typed, local-first memory system.
## Goal
Establish SQLite as the source of truth for durable memory records and metadata.
## Scope
- typed records: `fact`, `preference`, `decision`, `todo`, `event`
- canonical current record + revision trail
- status model for `active`, `... | open | null | false | 2 | [] | [] | 2026-03-11T01:47:59Z | 2026-05-04T01:47:40Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | vinced45 | 101,286 | MDQ6VXNlcjEwMTI4Ng== | User | false |
openclaw/openclaw | 4,055,173,915 | I_kwDOQb6kR87xtQsb | 42,647 | https://github.com/openclaw/openclaw/issues/42647 | https://api.github.com/repos/openclaw/openclaw/issues/42647 | Memory MVP: implement source attribution and provenance model | ## Summary
Add a provenance model so every durable memory can answer: where did this come from?
## Goal
Make source attribution non-negotiable from day 1.
## Scope
- source refs for session/message/file origins
- support multiple supporting sources per memory when relevant
- retain provenance across merges/edits
- ex... | open | null | false | 2 | [] | [] | 2026-03-11T01:48:01Z | 2026-05-04T11:28:00Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | vinced45 | 101,286 | MDQ6VXNlcjEwMTI4Ng== | User | false |
openclaw/openclaw | 4,055,174,004 | I_kwDOQb6kR87xtQt0 | 42,648 | https://github.com/openclaw/openclaw/issues/42648 | https://api.github.com/repos/openclaw/openclaw/issues/42648 | Memory MVP: build write pipeline with classification, dedupe, merge, and conflict handling | ## Summary
Build the write path for candidate memory ingestion.
## Goal
Prevent blind append behavior and reduce bad writes.
## Write path
1. classify type
2. attach source
3. search for near-duplicates / conflicts
4. resolve to create, merge/update, or conflict-for-review
## Requirements
- conservative write behavi... | open | null | false | 3 | [] | [] | 2026-03-11T01:48:03Z | 2026-05-04T11:32:21Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | vinced45 | 101,286 | MDQ6VXNlcjEwMTI4Ng== | User | false |
openclaw/openclaw | 4,055,174,169 | I_kwDOQb6kR87xtQwZ | 42,650 | https://github.com/openclaw/openclaw/issues/42650 | https://api.github.com/repos/openclaw/openclaw/issues/42650 | Memory MVP: add review, edit, forget, and conflict-resolution flows | ## Summary
Create first-class operator flows for correcting and maintaining memory quality.
## Goal
Ensure humans can inspect, fix, forget, and resolve memory safely.
## Scope
- review queue / review listing
- edit flow
- forget flow
- conflict resolution flow
- tombstone/superseded states instead of destructive dele... | open | null | false | 3 | [] | [] | 2026-03-11T01:48:07Z | 2026-05-04T00:42:46Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | vinced45 | 101,286 | MDQ6VXNlcjEwMTI4Ng== | User | false |
openclaw/openclaw | 4,055,174,232 | I_kwDOQb6kR87xtQxY | 42,651 | https://github.com/openclaw/openclaw/issues/42651 | https://api.github.com/repos/openclaw/openclaw/issues/42651 | Memory MVP: add ingestion helpers and CLI/skill surface | ## Summary
Expose the memory system through an operator-friendly interface.
## Goal
Provide the minimal CLI/skill surface for v1 usage.
## Commands
- `remember`
- `recall`
- `review`
- `edit`
- `forget`
## Scope
- ingestion helpers for candidate extraction from conversations/files
- operator-friendly commands around... | open | null | false | 3 | [] | [] | 2026-03-11T01:48:08Z | 2026-05-04T00:30:24Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | vinced45 | 101,286 | MDQ6VXNlcjEwMTI4Ng== | User | false |
openclaw/openclaw | 4,055,175,543 | I_kwDOQb6kR87xtRF3 | 42,652 | https://github.com/openclaw/openclaw/issues/42652 | https://api.github.com/repos/openclaw/openclaw/issues/42652 | Discord announce step uses wrong accountId (falls back to 'default' instead of session's accountId) | ## Problem
When the agent-to-agent announce step delivers messages to Discord, `resolveAnnounceTarget` extracts channel and target from the session key but **does not include accountId**.
The Discord plugin does not have `preferSessionLookupForAnnounceTarget` enabled, so it **skips session lookup** (which would have ... | closed | completed | true | 3 | [] | [] | 2026-03-11T01:48:38Z | 2026-05-05T04:27:27Z | 2026-05-02T11:50:48Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | irchelper | 38,723,383 | MDQ6VXNlcjM4NzIzMzgz | User | false |
openclaw/openclaw | 4,055,207,609 | I_kwDOQb6kR87xtY65 | 42,663 | https://github.com/openclaw/openclaw/issues/42663 | https://api.github.com/repos/openclaw/openclaw/issues/42663 | Feature: trustedWebhooks for Discord — allow specific webhooks to trigger agent responses | ## Problem
OpenClaw drops all Discord webhook messages because `author.bot === true` and `allowBots` defaults to `false`. This prevents users from building voice input pipelines, iOS Shortcuts, or external automations that send messages via Discord webhooks and expect the agent to respond.
Setting `allowBots: true` g... | open | null | false | 2 | [] | [] | 2026-03-11T01:58:55Z | 2026-05-04T11:28:04Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | jarvis-phw | 260,010,066 | U_kgDOD39wUg | User | false |
openclaw/openclaw | 4,055,330,793 | I_kwDOQb6kR87xt2_p | 42,690 | https://github.com/openclaw/openclaw/issues/42690 | https://api.github.com/repos/openclaw/openclaw/issues/42690 | Bug: session_status reports session default model instead of routed active model | ## Summary
`session_status` can report the session's selected/default model even when the active run has been routed to a different model. This makes the status card misleading for runtime routing cases where the selected model is intentionally preserved.
I reproduced this on `openclaw` `2026.3.8`.
## Reproduction
... | open | null | false | 2 | [] | [] | 2026-03-11T02:36:14Z | 2026-05-04T11:28:11Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | sk8man121 | 179,406,323 | U_kgDOCrGF8w | User | false |
openclaw/openclaw | 4,055,391,486 | I_kwDOQb6kR87xuFz- | 42,713 | https://github.com/openclaw/openclaw/issues/42713 | https://api.github.com/repos/openclaw/openclaw/issues/42713 | Failover wrapper errors missing provider and session_id in error logging | ## Problem
When the gateway's failover system exhausts all auth profiles and model fallbacks, the resulting error log entries are missing `provider` and `session_id` metadata. This makes it difficult to attribute errors during post-mortem analysis.
## Evidence
Analyzing the last 100 errors from our `api_health_log` ... | closed | completed | true | 2 | [] | [] | 2026-03-11T02:58:17Z | 2026-05-04T04:37:10Z | 2026-05-01T08:26:58Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | theo674 | 261,068,216 | U_kgDOD4-VuA | User | false |
openclaw/openclaw | 4,055,401,103 | I_kwDOQb6kR87xuIKP | 42,715 | https://github.com/openclaw/openclaw/issues/42715 | https://api.github.com/repos/openclaw/openclaw/issues/42715 | Multi-provider AI Gateway: missing apiKey causes silent model fallback | ## Summary
When using Cloudflare AI Gateway with multiple providers (e.g., OpenRouter + Workers AI), the `apiKey` configured during `openclaw onboard --auth-choice cloudflare-ai-gateway-api-key` is only set on the initial provider. Additional providers added to the same AI Gateway must have `apiKey` manually configure... | closed | completed | false | 2 | [] | [] | 2026-03-11T03:01:56Z | 2026-05-03T22:49:35Z | 2026-05-03T22:49:35Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | jbbjbb | 1,313,120 | MDQ6VXNlcjEzMTMxMjA= | User | false |
openclaw/openclaw | 4,055,794,233 | I_kwDOQb6kR87xvoI5 | 42,798 | https://github.com/openclaw/openclaw/issues/42798 | https://api.github.com/repos/openclaw/openclaw/issues/42798 | fix(tailscale): startup race condition causes transient 'Command failed: tailscale status --json' errors | ## Summary
When the gateway starts with `gateway.tailscale.mode: "serve"`, the startup sequence runs several tailscale commands in rapid succession:
1. `findTailscaleBinary()`: `which tailscale` → `tailscale --version` (binary verification)
2. `enableTailscaleServe()`: `tailscale serve --bg --yes <port>`
3. `getTailn... | open | null | false | 2 | [] | [] | 2026-03-11T04:52:19Z | 2026-05-04T11:28:20Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | archerhpagent | 259,157,385 | U_kgDOD3JtiQ | User | false |
openclaw/openclaw | 4,055,807,423 | I_kwDOQb6kR87xvrW_ | 42,802 | https://github.com/openclaw/openclaw/issues/42802 | https://api.github.com/repos/openclaw/openclaw/issues/42802 | feat: Discord channel-per-agent binding display in Gateway UI (ZZ-20260311-005) | 奏折: ZZ-20260311-005
## 任务背景
Gateway 配置中每个 agent 可以绑定一个或多个 Discord channel（cfg.bindings[]），但 Gateway 管理 UI 上没有地方能看到这些绑定关系。需要在 agents tab → channels panel 添加可视化展示。
## 实现内容
### Backend: Gateway API
- 新增 routing.list method（只读，READ_SCOPE）
- 返回 { bindings: RouteBindingEntry[] }，包含所有 route-type 和 acp-type 绑定
- 新建 src/gate... | open | null | false | 3 | [] | [] | 2026-03-11T04:55:36Z | 2026-05-04T11:28:22Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Billmvp73 | 45,382,455 | MDQ6VXNlcjQ1MzgyNDU1 | User | false |
openclaw/openclaw | 4,055,807,900 | I_kwDOQb6kR87xvrec | 42,803 | https://github.com/openclaw/openclaw/issues/42803 | https://api.github.com/repos/openclaw/openclaw/issues/42803 | [Bug]: Feishu text commands (/stop, /new, /status) no longer bypass queue during active agent run (3.8 regression) | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to 2026.3.8, text commands (`/stop`, `/new`, `/status`) on Feishu no longer execute immediately when an agent run is active. They are queued behind the current run and only execute after it completes. This worked correctly on 3.7 and ear... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-11T04:55:42Z | 2026-05-04T11:28:24Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Chowei11 | 207,524,385 | U_kgDODF6SIQ | User | false |
openclaw/openclaw | 4,055,919,672 | I_kwDOQb6kR87xwGw4 | 42,837 | https://github.com/openclaw/openclaw/issues/42837 | https://api.github.com/repos/openclaw/openclaw/issues/42837 | [Bug]: Feishu dynamicAgentCreation feature not working | ### Bug type
Regression (worked before, now fails)
### Summary
Bug: Feishu dynamicAgentCreation feature not working
Description:
- Enabled channels.feishu.dynamicAgentCreation.enabled: true
- Agent and Binding are created (visible in config)
- But messages still route to agent:main instead of separate agents per us... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-03-11T05:23:18Z | 2026-05-04T12:45:01Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | cwlong163-afk | 267,238,542 | U_kgDOD-28jg | User | false |
openclaw/openclaw | 4,055,931,082 | I_kwDOQb6kR87xwJjK | 42,840 | https://github.com/openclaw/openclaw/issues/42840 | https://api.github.com/repos/openclaw/openclaw/issues/42840 | Feature Request: Add MathJax/LaTeX Support to Control UI | ## 🎯 Feature Request
Add MathJax/LaTeX rendering support to the OpenClaw Control UI for displaying mathematical formulas.
## 📖 Motivation
As an AI assistant using OpenClaw, I frequently need to communicate mathematical and scientific content to users. Currently, LaTeX formulas are displayed as raw source code (e.g... | open | null | false | 6 | [] | [] | 2026-03-11T05:26:07Z | 2026-05-04T11:32:25Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | henserlu | 148,304,080 | U_kgDOCNbw0A | User | false |
openclaw/openclaw | 4,056,029,989 | I_kwDOQb6kR87xwhsl | 42,854 | https://github.com/openclaw/openclaw/issues/42854 | https://api.github.com/repos/openclaw/openclaw/issues/42854 | Model picker shows stale Gemini preview alias for google/gemini-3.1-pro | ## Summary
OpenClaw is showing the wrong Gemini model label/alias in the model picker.
A local config using the current model id `google/gemini-3.1-pro` still ends up showing the older preview-derived name in the picker.
## Expected
If config/model selection uses `google/gemini-3.1-pro`, the picker/UI should reflect ... | closed | not_planned | true | 3 | [] | [] | 2026-03-11T05:49:55Z | 2026-05-01T04:46:39Z | 2026-04-28T04:41:45Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | bobsummerwill-openclaw-bot[bot] | 262,731,058 | BOT_kgDOD6j1Mg | Bot | false |
openclaw/openclaw | 4,056,039,567 | I_kwDOQb6kR87xwkCP | 42,856 | https://github.com/openclaw/openclaw/issues/42856 | https://api.github.com/repos/openclaw/openclaw/issues/42856 | config.schema.lookup errors produce user-facing notifications | ## Bug
When an agent uses `config.schema.lookup` with a path that doesn't exist, the gateway returns a normal error response (`config schema path not found`). However, this also triggers a visible notification to the user (e.g. on Telegram):
```
⚠️ 🔌 Gateway: agents.defaults.modelBySession failed
```
## Expected Be... | open | null | false | 3 | [] | [] | 2026-03-11T05:52:26Z | 2026-05-04T20:13:52Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | akrimm702 | 144,416,483 | U_kgDOCJue4w | User | false |
openclaw/openclaw | 4,056,049,005 | I_kwDOQb6kR87xwmVt | 42,858 | https://github.com/openclaw/openclaw/issues/42858 | https://api.github.com/repos/openclaw/openclaw/issues/42858 | [BUG] bailian provider missing from envMap causes HTTP 401 when using __env__:BAILIAN_API_KEY | ## Bug Description
When configuring the Bailian (阿里云百炼) provider with environment variable reference `"apiKey": "__env__:BAILIAN_API_KEY"`, the agent fails with `HTTP 401: invalid access token or token expired` error, even though:
- The API key is valid and working when tested directly
- The environment variable is co... | open | null | false | 4 | [] | [] | 2026-03-11T05:54:43Z | 2026-05-04T11:28:58Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | fdenny11gg | 264,869,681 | U_kgDOD8mXMQ | User | false |
openclaw/openclaw | 4,056,084,297 | I_kwDOQb6kR87xwu9J | 42,862 | https://github.com/openclaw/openclaw/issues/42862 | https://api.github.com/repos/openclaw/openclaw/issues/42862 | [Feature Request] 自动在聊天窗口中渲染图片URL | ## Description
When a user sends an image URL (e.g., .jpg, .png, .gif) in the chat, it would be great if it could automatically render as an image instead of showing plain text link.
## Expected Behavior
When a user sends `https://example.com/image.png`, the chat window should display the image directly.
## Technic... | closed | not_planned | false | 2 | [] | [] | 2026-03-11T06:02:25Z | 2026-05-03T22:49:41Z | 2026-05-03T22:49:41Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | fengjuntao2025-a11y | 250,480,124 | U_kgDODu4F_A | User | false |
openclaw/openclaw | 4,056,090,005 | I_kwDOQb6kR87xwwWV | 42,867 | https://github.com/openclaw/openclaw/issues/42867 | https://api.github.com/repos/openclaw/openclaw/issues/42867 | Feature: Turn-count TTL (auto-reset session after N turns) | ## Problem
Sessions can accumulate indefinitely even with daily reset configured, because heartbeat messages refresh the session's `updatedAt` timestamp, making it look active. A session with only heartbeat activity is never considered stale for daily reset purposes.
## Current behavior
- Daily reset: session is sta... | closed | not_planned | true | 2 | [] | [] | 2026-03-11T06:03:51Z | 2026-05-05T04:25:52Z | 2026-05-03T01:11:18Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | pescaohq-web | 257,331,994 | U_kgDOD1aTGg | User | false |
openclaw/openclaw | 4,056,132,369 | I_kwDOQb6kR87xw6sR | 42,877 | https://github.com/openclaw/openclaw/issues/42877 | https://api.github.com/repos/openclaw/openclaw/issues/42877 | Bounded memory tool with hard character limits | Hey,
I've been running OpenClaw as my daily driver for about five weeks now with a fairly heavy setup (16 crons, multi-agent delegation, Telegram gateway). One thing I keep running into is memory bloat.
**The problem**
Workspace memory files (MEMORY.md, etc.) grow without any guardrails. My MEMORY.md is at ~3,800 by... | open | null | false | 3 | [] | [] | 2026-03-11T06:13:27Z | 2026-05-04T11:29:00Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | iFiras-Max1 | 260,676,028 | U_kgDOD4mZvA | User | false |
openclaw/openclaw | 4,056,493,488 | I_kwDOQb6kR87xyS2w | 42,952 | https://github.com/openclaw/openclaw/issues/42952 | https://api.github.com/repos/openclaw/openclaw/issues/42952 | ignoreFileContent setting not working - file content still sent to model as context | ## Bug Description
The `ignoreFileContent` setting in Feishu plugin configuration does not work as expected. Even when set to `true`, file content sent via Feishu is still included in the context sent to the model.
## Configuration
In `openclaw.json`:
```json
"channels": {
"feishu": {
"accounts": {
"defa... | open | null | false | 2 | [] | [] | 2026-03-11T07:42:08Z | 2026-05-04T11:29:08Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | tociyou1995 | 150,312,743 | U_kgDOCPWXJw | User | false |
openclaw/openclaw | 4,056,859,962 | I_kwDOQb6kR87xzsU6 | 42,999 | https://github.com/openclaw/openclaw/issues/42999 | https://api.github.com/repos/openclaw/openclaw/issues/42999 | Feature: Dynamic Context Window Token Budget (Metabolic Token Management) | ## Problem
The current `context-window-guard` uses static token thresholds:
- `CONTEXT_WINDOW_HARD_MIN_TOKENS = 16000`
- `CONTEXT_WINDOW_WARN_BELOW_TOKENS = 32000`
These fixed values don't adapt to different model context windows. For a 4K context model, 16K hard minimum is meaningless. For a 1M context model, 32K ... | closed | completed | true | 4 | [] | [] | 2026-03-11T09:02:13Z | 2026-05-03T04:38:06Z | 2026-04-30T07:33:45Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,056,864,925 | I_kwDOQb6kR87xztid | 43,005 | https://github.com/openclaw/openclaw/issues/43005 | https://api.github.com/repos/openclaw/openclaw/issues/43005 | Enhancement: Allow custom safe headers in fetch-guard cross-origin redirects | ## Problem
`fetch-guard.ts` strips all non-standard headers during cross-origin redirects via `CROSS_ORIGIN_REDIRECT_SAFE_HEADERS`. This is security-correct but can break integrations that need custom headers (e.g., `x-api-key`) to persist across redirects.
## Proposal
Add a configuration option to extend the safe h... | open | null | false | 2 | [] | [] | 2026-03-11T09:03:17Z | 2026-05-04T11:32:28Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,056,865,143 | I_kwDOQb6kR87xztl3 | 43,006 | https://github.com/openclaw/openclaw/issues/43006 | https://api.github.com/repos/openclaw/openclaw/issues/43006 | Feature: Progressive memory flush before compaction | ## Problem
The current `compaction.memoryFlush` triggers once when the session crosses the soft threshold. This is a single-shot mechanism — if the flush fails or the model doesn't write enough, important context may be lost during compaction.
## Proposal: Progressive Flush
Instead of a single flush at one threshold... | open | null | false | 3 | [] | [] | 2026-03-11T09:03:19Z | 2026-05-04T00:42:51Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,056,894,789 | I_kwDOQb6kR87xz01F | 43,009 | https://github.com/openclaw/openclaw/issues/43009 | https://api.github.com/repos/openclaw/openclaw/issues/43009 | TUI displays Context Tokens as ?/200k instead of actual value | ## Bug Description
OpenClaw TUI interface continuously displays `tokens ?/200k`, but the actual API returns correct values (e.g., `0/200k (0%)`).
## Reproduction Steps
1. Start OpenClaw TUI (`openclaw tui`)
2. Have multiple conversation turns
3. Observe the tokens display in the top status bar
## Expected Behavior
... | open | null | false | 2 | [] | [] | 2026-03-11T09:09:13Z | 2026-05-04T12:03:44Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ltxy12138-ai | 266,312,779 | U_kgDOD9-cSw | User | false |
openclaw/openclaw | 4,056,909,277 | I_kwDOQb6kR87xz4Xd | 43,015 | https://github.com/openclaw/openclaw/issues/43015 | https://api.github.com/repos/openclaw/openclaw/issues/43015 | message.send schema overexposes poll/components/modal causing GPT auto-population breakages | ### Summary
`message.send` schema currently overexposes advanced fields (poll + Discord components/modal) on the generic send action. GPT-family models tend to auto-populate these optional fields even when the user only wants to send a simple message or attachment. Combined with strict runtime validation that triggers... | open | null | false | 5 | [] | [] | 2026-03-11T09:11:37Z | 2026-05-04T12:15:57Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | CharZhou | 17,255,546 | MDQ6VXNlcjE3MjU1NTQ2 | User | false |
openclaw/openclaw | 4,056,997,894 | I_kwDOQb6kR87x0OAG | 43,028 | https://github.com/openclaw/openclaw/issues/43028 | https://api.github.com/repos/openclaw/openclaw/issues/43028 | [Feature Request] Cron 任务自动化改进：熔断机制 + 自动清理 + Changelog 记录 | ## 问题描述
今天在使用 OpenClaw Cron 功能时，发现以下自动化不足的问题：
### 问题 1: 无超时熔断机制
- **现象**: 一个 Cron 任务连续超时 4 次，系统未自动暂停
- **影响**: 导致 cron 执行队列阻塞，后续任务排队等待
- **日志**:
```
lane wait exceeded: lane=cron waitedMs=181094 queueAhead=0
```
### 问题 2: 无自动清理机制
- **现象**: 删除 Cron 任务后，关联的会话文件未自动清理
- **影响**: 残留会话占用磁盘空间，增加管理复杂度
- **发现**: 2 个已删除... | open | null | false | 3 | [] | [] | 2026-03-11T09:27:06Z | 2026-05-04T12:03:50Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | weaisc10-code | 265,721,514 | U_kgDOD9aWqg | User | false |
openclaw/openclaw | 4,057,207,695 | I_kwDOQb6kR87x1BOP | 43,064 | https://github.com/openclaw/openclaw/issues/43064 | https://api.github.com/repos/openclaw/openclaw/issues/43064 | Heartbeat replies show literal {model} instead of interpolated model name in responsePrefix | ## Description
When `responsePrefix` uses template variables like `{model}`, heartbeat replies output the literal placeholder text (e.g. `{model}: The heartbeat just came in...`) instead of the interpolated model name (e.g. `gemini-2.5-flash: The heartbeat just came in...`).
Normal user-initiated replies correctly in... | closed | completed | true | 3 | [] | [] | 2026-03-11T10:05:52Z | 2026-05-02T04:26:31Z | 2026-04-30T02:48:52Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | yweiii | 40,194,424 | MDQ6VXNlcjQwMTk0NDI0 | User | false |
openclaw/openclaw | 4,057,355,645 | I_kwDOQb6kR87x1lV9 | 43,095 | https://github.com/openclaw/openclaw/issues/43095 | https://api.github.com/repos/openclaw/openclaw/issues/43095 | `sessions_send` may return `status: timeout` even when the target session later receives the message | `sessions_send` currently seems to use `timeout` to mean **reply wait timeout**, not strictly **delivery failure**.
### Observed behavior
We saw cases where:
1. session A sends a message to session B via `sessions_send`
2. `sessions_send` returns:
```json
{ "status": "timeout" }
```
3. but session B later *... | closed | completed | false | 3 | [] | [] | 2026-03-11T10:32:07Z | 2026-05-03T22:23:05Z | 2026-05-03T22:23:05Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | jesseliu1 | 41,902,230 | MDQ6VXNlcjQxOTAyMjMw | User | false |
openclaw/openclaw | 4,057,484,043 | I_kwDOQb6kR87x2EsL | 43,117 | https://github.com/openclaw/openclaw/issues/43117 | https://api.github.com/repos/openclaw/openclaw/issues/43117 | feat(memorySearch): Support multimodal embedding with gemini-embedding-2-preview | ## Summary
`gemini-embedding-2-preview` now supports multimodal embedding - text, images, video, audio, and PDFs all map to the same vector space. This enables cross-modal search (e.g., search images by text description).
## Proposed Changes
1. **Extend memorySearch to embed images/attachments** alongside text when ... | open | null | false | 4 | [] | [] | 2026-03-11T10:56:31Z | 2026-05-04T12:16:03Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | frxiaobei | 353,582 | MDQ6VXNlcjM1MzU4Mg== | User | false |
openclaw/openclaw | 4,057,701,727 | I_kwDOQb6kR87x251f | 43,141 | https://github.com/openclaw/openclaw/issues/43141 | https://api.github.com/repos/openclaw/openclaw/issues/43141 | Queued cron-lane runs are not abort-aware and can time out before model invocation | ## Summary
Queued isolated cron runs can spend their full timeout waiting on the global `cron` lane and still remain queued after the caller times out.
The run never reaches model invocation in this case. The failure happens above the model layer because queued command-lane work is not abort-aware.
## Affected behav... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-11T11:36:57Z | 2026-04-30T04:32:35Z | 2026-04-28T01:24:21Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | matthewxmurphy | 110,652,790 | U_kgDOBphtdg | User | false |
openclaw/openclaw | 4,058,210,219 | I_kwDOQb6kR87x41-r | 43,202 | https://github.com/openclaw/openclaw/issues/43202 | https://api.github.com/repos/openclaw/openclaw/issues/43202 | Feishu card action: pass through form_value, input_value and other fields | ## Problem
The Feishu card action handler (`extensions/feishu/src/card-action.ts`) currently only extracts `event.action.value` from `card.action.trigger` callbacks, discarding important fields:
- `action.input_value` — text input content
- `action.form_value` — form submission values (all child components)
- `action... | open | null | false | 2 | [] | [] | 2026-03-11T13:22:08Z | 2026-05-04T12:04:12Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Edison94hu | 50,041,387 | MDQ6VXNlcjUwMDQxMzg3 | User | false |
openclaw/openclaw | 4,058,357,204 | I_kwDOQb6kR87x5Z3U | 43,231 | https://github.com/openclaw/openclaw/issues/43231 | https://api.github.com/repos/openclaw/openclaw/issues/43231 | Telegram forum: no reliable topic-name -> topic_id resolution for agent routing | ## Summary
In Telegram forum group chats, the assistant cannot reliably resolve a topic by human-readable name (e.g. "AI безопасность") to its `topic_id` for routing messages. This causes repeated misrouting or extra user clarification loops.
## Environment
- OpenClaw runtime in Telegram forum groups
- Channel metadat... | open | null | false | 3 | [] | [] | 2026-03-11T13:49:30Z | 2026-05-04T12:16:13Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,058,475,274 | I_kwDOQb6kR87x52sK | 43,249 | https://github.com/openclaw/openclaw/issues/43249 | https://api.github.com/repos/openclaw/openclaw/issues/43249 | Telegram API returning 400 errors in recent days | ## Description\n\nTelegram integration has been returning 400 API errors in recent days.\n\n## Expected Behavior\n\nTelegram messages should be sent/received without API errors.\n\n## Actual Behavior\n\nGetting 400 errors from Telegram API.\n\n## Environment\n\n- OpenClaw version: [please specify]\n- Channel: telegram\... | open | null | false | 4 | [] | [] | 2026-03-11T14:11:14Z | 2026-05-04T00:42:55Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | xiaoamo22333 | 89,977,502 | MDQ6VXNlcjg5OTc3NTAy | User | false |
openclaw/openclaw | 4,058,860,988 | I_kwDOQb6kR87x7U28 | 43,286 | https://github.com/openclaw/openclaw/issues/43286 | https://api.github.com/repos/openclaw/openclaw/issues/43286 | Add session.threadIsolation: false to allow shared session across Slack threads | ## Problem
When running OpenClaw as a single-identity agent (e.g., a support rep) on Slack, every thread reply creates a separate session via `:thread:<thread_ts>` suffix in `resolveSlackRoutingContext → resolveThreadSessionKeys`. The agent has no context of what happened in other threads — users experience "split per... | closed | not_planned | true | 4 | [] | [] | 2026-03-11T15:16:20Z | 2026-05-03T04:37:40Z | 2026-04-30T11:54:20Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | charlie-n-hill | 39,253,519 | MDQ6VXNlcjM5MjUzNTE5 | User | false |
openclaw/openclaw | 4,058,990,097 | I_kwDOQb6kR87x70YR | 43,300 | https://github.com/openclaw/openclaw/issues/43300 | https://api.github.com/repos/openclaw/openclaw/issues/43300 | [Bug]: | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When OpenClaw is configured with `gateway.auth.mode = "trusted-proxy"` behind Traefik + authentik, the browser UI works, but manager/subagent/backend self-connections fail
### Steps to reproduce
- Configure Gateway with auth.mode = "trust... | open | null | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T15:38:55Z | 2026-05-04T00:52:13Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | papirov | 2,907,836 | MDQ6VXNlcjI5MDc4MzY= | User | false |
openclaw/openclaw | 4,059,273,946 | I_kwDOQb6kR87x85ra | 43,341 | https://github.com/openclaw/openclaw/issues/43341 | https://api.github.com/repos/openclaw/openclaw/issues/43341 | Bug: TUI not receiving real-time message updates from main session | **Bug: TUI not receiving real-time message updates from main session**
## Environment
- OpenClaw version: 2026.3.8
- Install method: pnpm global
- Platform: Linux (VM, x64)
- Node version: v22.22.0
## Description
After upgrading from 2026.3.7 to 2026.3.8, the TUI (`openclaw tui`) no longer receives real-time message ... | closed | not_planned | true | 3 | [] | [] | 2026-03-11T16:30:24Z | 2026-05-01T04:44:16Z | 2026-04-28T09:03:19Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | baiyiyuni | 84,668,767 | MDQ6VXNlcjg0NjY4NzY3 | User | false |
openclaw/openclaw | 4,060,399,812 | I_kwDOQb6kR87yBMjE | 43,440 | https://github.com/openclaw/openclaw/issues/43440 | https://api.github.com/repos/openclaw/openclaw/issues/43440 | imsg should not require parent node process to have Full Disk Access | When Homebrew upgrades node (e.g. from node@22/22.22.0_1 to 22.22.1_1), macOS TCC loses the Full Disk Access grant because it tracks permissions by resolved binary path. This causes imsg to fail with permissionDenied on ~/Library/Messages/chat.db, even though the imsg binary itself has its own FDA grant in the TCC data... | open | null | false | 2 | [] | [] | 2026-03-11T20:07:28Z | 2026-05-04T12:16:28Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | claw-explorer | 263,387,821 | U_kgDOD7L6rQ | User | false |
openclaw/openclaw | 4,060,722,060 | I_kwDOQb6kR87yCbOM | 43,480 | https://github.com/openclaw/openclaw/issues/43480 | https://api.github.com/repos/openclaw/openclaw/issues/43480 | macOS app pinwheels due to SwiftUI infinite render loop in VoiceWakeOverlay | ## Summary
The macOS app intermittently pinwheels (100% CPU, unresponsive) due to an infinite SwiftUI render loop in the VoiceWakeOverlay.
## Reproduction
1. Open the OpenClaw macOS app
2. Trigger the voice wake overlay (click interaction that opens/reorders the overlay window)
3. App freezes with spinning beach bal... | open | null | false | 4 | [] | [] | 2026-03-11T21:18:11Z | 2026-05-04T12:05:18Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | gdiab | 995,507 | MDQ6VXNlcjk5NTUwNw== | User | false |
openclaw/openclaw | 4,061,298,852 | I_kwDOQb6kR87yEoCk | 43,527 | https://github.com/openclaw/openclaw/issues/43527 | https://api.github.com/repos/openclaw/openclaw/issues/43527 | [Bug]: status says "no bootstrap files" when only BOOTSTRAP.md is absent | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`openclaw status --deep` reports `no bootstrap files` on mature workspaces where standard bootstrap files are injected normally and only the one-shot `BOOTSTRAP.md` file is absent.
### Steps to reproduce
1. Create or use a normal non-new Ope... | open | null | false | 2 | [] | [] | 2026-03-11T23:53:52Z | 2026-05-04T12:16:38Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | aleksesipenko | 252,942,994 | U_kgDODxOakg | User | false |
openclaw/openclaw | 4,061,500,938 | I_kwDOQb6kR87yFZYK | 43,557 | https://github.com/openclaw/openclaw/issues/43557 | https://api.github.com/repos/openclaw/openclaw/issues/43557 | Intermittent 'No auth for gmail' during scheduled runs | Hi — I’m seeing an intermittent auth failure with `gog` on macOS during scheduled runs.
Summary
- Environment: macOS (Darwin 24.5.0 arm64)
- gog build: v0.12.0 (c18c58c)
- account: `[redacted]`
- client: `default`
- auth type: OAuth
- Gmail scope is present in `gog auth list --json`
- The failure happens in scheduled ... | closed | not_planned | true | 2 | [] | [] | 2026-03-12T01:08:58Z | 2026-05-05T04:27:33Z | 2026-05-02T11:24:12Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | federicocorradi | 258,326,063 | U_kgDOD2W-Lw | User | false |
openclaw/openclaw | 4,061,535,419 | I_kwDOQb6kR87yFhy7 | 43,562 | https://github.com/openclaw/openclaw/issues/43562 | https://api.github.com/repos/openclaw/openclaw/issues/43562 | Make agent identity preamble configurable (replace hardcoded 'personal assistant') | ## Problem
The system prompt preamble in `src/agents/system-prompt.ts` (line 423) is hardcoded:
```typescript
"You are a personal assistant running inside OpenClaw."
```
This line appears at the **very start** of the system prompt, making it the highest-priority identity instruction for the model. When agents have c... | open | null | false | 3 | [] | [] | 2026-03-12T01:24:18Z | 2026-05-04T12:16:43Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | brad07 | 995,915 | MDQ6VXNlcjk5NTkxNQ== | User | false |
openclaw/openclaw | 4,061,547,846 | I_kwDOQb6kR87yFk1G | 43,564 | https://github.com/openclaw/openclaw/issues/43564 | https://api.github.com/repos/openclaw/openclaw/issues/43564 | [Feature Request] ACP Session Skill Context Injection | ## Summary
Allow OpenClaw skills to be injected into ACP (Codex/Pi/OpenCode/Gemini) session contexts when spawning agents.
## Problem
Currently:
- Skills exist in `~/clawd/skills/` and augment Kai's capabilities
- ACP agents spawn as independent processes with no access to these skills
- When delegating to Codex/Pi ... | open | null | false | 2 | [] | [] | 2026-03-12T01:29:42Z | 2026-05-04T12:05:33Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | kai-the-man[bot] | 259,478,402 | BOT_kgDOD3dTgg | Bot | false |
openclaw/openclaw | 4,061,569,151 | I_kwDOQb6kR87yFqB_ | 43,568 | https://github.com/openclaw/openclaw/issues/43568 | https://api.github.com/repos/openclaw/openclaw/issues/43568 | Feature: per-provider proxy support in auth-profiles (proxyUrl field) | ## Problem
When running multiple Anthropic/OpenAI accounts through a single OpenClaw instance, all API requests originate from the same server IP address. This creates a significant risk: providers like Anthropic may flag or ban multiple accounts accessing their API from the same IP, as it looks like credential sharin... | open | null | false | 2 | [] | [] | 2026-03-12T01:38:26Z | 2026-05-04T12:16:47Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,061,964,790 | I_kwDOQb6kR87yHKn2 | 43,658 | https://github.com/openclaw/openclaw/issues/43658 | https://api.github.com/repos/openclaw/openclaw/issues/43658 | GitHub Copilot auth profile rotation ineffective — all profiles share single cached API token | ## Summary
When multiple GitHub Copilot auth profiles are configured for round-robin rotation, hitting an API rate limit on one profile does **not** effectively rotate to the next profile. The system detects the rate limit correctly and advances to the next profile, but all subsequent requests continue to use the same... | open | null | false | 3 | [] | [] | 2026-03-12T04:02:04Z | 2026-05-04T12:05:56Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | faishalwahiduddin | 15,316,893 | MDQ6VXNlcjE1MzE2ODkz | User | false |
openclaw/openclaw | 4,062,068,973 | I_kwDOQb6kR87yHkDt | 43,673 | https://github.com/openclaw/openclaw/issues/43673 | https://api.github.com/repos/openclaw/openclaw/issues/43673 | [Feature]: First-class org/team deployment — workspace scaffolding, RBAC, and deployment manifests | ### Summary
First-class support for multi-user org/team deployments: workspace scaffolding CLI, per-session tool policies (RBAC), and declarative deployment manifests.
### Problem to solve
Teams and orgs are adopting OpenClaw, but multi-user deployment requires significant manual configuration with no official tooli... | open | null | false | 3 | [] | [] | 2026-03-12T04:31:19Z | 2026-05-04T12:06:01Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | ekkerdthomas | 62,130,061 | MDQ6VXNlcjYyMTMwMDYx | User | false |
openclaw/openclaw | 4,062,172,359 | I_kwDOQb6kR87yH9TH | 43,690 | https://github.com/openclaw/openclaw/issues/43690 | https://api.github.com/repos/openclaw/openclaw/issues/43690 | Feishu: card fails with 'table number over limit' — should fallback or render tables as images | ## Problem
When the agent's response contains multiple markdown tables, sending via Feishu interactive card fails with:
```
Create card request failed with HTTP 400
code: 230099, msg: card table number over limit
```
Both the streaming card creation and the final reply fail. There is no fallback — the user sees the ... | open | null | false | 2 | [] | [] | 2026-03-12T05:01:56Z | 2026-05-04T12:06:13Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | PenguinMiaou | 209,870,260 | U_kgDODIJdtA | User | false |
openclaw/openclaw | 4,062,319,080 | I_kwDOQb6kR87yIhHo | 43,712 | https://github.com/openclaw/openclaw/issues/43712 | https://api.github.com/repos/openclaw/openclaw/issues/43712 | [Bug]: v2026.3.11 update fails on live git install at /usr/lib/node_modules/openclaw, but succeeds in clean checkout | ### Summary
Updating OpenClaw from `2026.3.8` to `v2026.3.11` failed on my live Linux git install at `/usr/lib/node_modules/openclaw`, even though the same tag built successfully in a clean throwaway checkout on the same machine.
So this looks less like a bad release tag and more like an update/install-path edge case... | open | null | false | 4 | [] | [] | 2026-03-12T05:42:03Z | 2026-05-04T12:17:02Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | turrizt | 84,812,544 | MDQ6VXNlcjg0ODEyNTQ0 | User | false |
openclaw/openclaw | 4,062,641,419 | I_kwDOQb6kR87yJv0L | 43,775 | https://github.com/openclaw/openclaw/issues/43775 | https://api.github.com/repos/openclaw/openclaw/issues/43775 | Config version stamp mismatch silently breaks Discord delivery for openai-codex models | ## Config version stamp mismatch silently breaks Discord delivery
### Summary
When `meta.lastTouchedVersion` in `openclaw.json` is newer than the running OpenClaw version, Discord message delivery silently fails for `openai-codex-responses` models (GPT-5.x). No errors are logged anywhere. Anthropic models are unaffect... | closed | not_planned | false | 2 | [] | [] | 2026-03-12T07:02:34Z | 2026-05-03T23:50:07Z | 2026-05-03T23:50:07Z | CONTRIBUTOR | null | 20260504T000036Z | 2026-05-04T00:00:36Z | Marvinthebored | 262,704,729 | U_kgDOD6iOWQ | User | false |
openclaw/openclaw | 4,062,728,246 | I_kwDOQb6kR87yKFA2 | 43,794 | https://github.com/openclaw/openclaw/issues/43794 | https://api.github.com/repos/openclaw/openclaw/issues/43794 | [Feature Request] Add config encryption for credentials at rest | ## Problem
According to the threat model (THREAT-MODEL-ATLAS.md), credential theft from config files is a known high-risk issue (T-PERSIST-001). Currently, tokens and API keys are stored in plaintext in ~/.openclaw/openclaw.json.
## Expected Feature
- Encrypt sensitive fields (tokens, API keys) in config files
- Use s... | open | null | false | 3 | [] | [] | 2026-03-12T07:20:32Z | 2026-05-04T20:22:25Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | 2438 | 58,472,179 | MDQ6VXNlcjU4NDcyMTc5 | User | false |
openclaw/openclaw | 4,063,010,113 | I_kwDOQb6kR87yLJ1B | 43,848 | https://github.com/openclaw/openclaw/issues/43848 | https://api.github.com/repos/openclaw/openclaw/issues/43848 | [Bug]: sessions_send loses Telegram forum topic delivery and can pollute requester transcripts | ## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
`sessions_send` between Telegram supergroup forum topics can lose topic delivery, skip the visible post, or echo topic-to-topic handoff text back into the requester transcript.
## Steps to reproduce
1. Configure OpenClaw `2026.3.11` with Teleg... | closed | not_planned | false | 2 | [] | [] | 2026-03-12T08:24:19Z | 2026-05-03T05:18:51Z | 2026-05-03T05:18:51Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | UntaDotMy | 42,847,931 | MDQ6VXNlcjQyODQ3OTMx | User | false |
openclaw/openclaw | 4,063,123,101 | I_kwDOQb6kR87yLlad | 43,866 | https://github.com/openclaw/openclaw/issues/43866 | https://api.github.com/repos/openclaw/openclaw/issues/43866 | [Bug]: /hooks/agent silently drops accountId — hook dispatch bypasses cron normalization pipeline | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`POST /hooks/agent` silently drops the `accountId` field from incoming payloads, causing WhatsApp (and other multi-account channel) delivery to fail in multi-account setups.
### Steps to reproduce
1. Configure two WhatsApp connections wit... | open | null | false | 2 | [] | [] | 2026-03-12T08:49:42Z | 2026-05-03T23:37:21Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | pauldesmondparker | 114,397,409 | U_kgDOBtGQ4Q | User | false |
openclaw/openclaw | 4,058,888,947 | I_kwDOQb6kR87x7brz | 43,288 | https://github.com/openclaw/openclaw/issues/43288 | https://api.github.com/repos/openclaw/openclaw/issues/43288 | CLI backend: pickSessionId accepts non-UUID strings like 'rate-limited' causing resume failure | ## Bug Description
When a CLI backend (e.g. Claude CLI) hits a rate limit, it returns JSON containing `"session_id": "rate-limited"`. The `pickSessionId()` function in the dispatch module only checks whether the value is a non-empty string — it does not validate that the value is actually a usable session identifier (... | closed | completed | true | 4 | [
"stale"
] | [] | 2026-03-11T15:21:35Z | 2026-04-30T04:35:43Z | 2026-04-27T08:23:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mySebbe | 14,988,278 | MDQ6VXNlcjE0OTg4Mjc4 | User | false |
openclaw/openclaw | 4,334,073,722 | I_kwDOQb6kR88AAAABAlS3eg | 72,697 | https://github.com/openclaw/openclaw/issues/72697 | https://api.github.com/repos/openclaw/openclaw/issues/72697 | [Bug]: Failover writes permanent providerOverride/modelOverride to sessions.json with no self-healing — primary never re-tried | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
Yes
### Summary
When the primary model in a session's agent fails (e.g. ollama timeout, transient network issue, context overflow), OpenClaw 2026.4.8 writes four override fields to the affected session's entry in `agents/main/sessions/session... | closed | not_planned | true | 2 | [
"bug"
] | [] | 2026-04-27T07:58:41Z | 2026-04-30T04:35:42Z | 2026-04-27T08:27:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kibedu | 210,977,235 | U_kgDODJNB0w | User | false |
openclaw/openclaw | 4,334,082,399 | I_kwDOQb6kR88AAAABAlTZXw | 72,699 | https://github.com/openclaw/openclaw/issues/72699 | https://api.github.com/repos/openclaw/openclaw/issues/72699 | Gateway crash-loops with unhandled_rejection on v2026.4.24 (D state, 85%+ CPU) | ## Description
The gateway process enters an unrecoverable crash loop immediately after startup on version `2026.4.24 (cbcfdf6)`. The TUI shows `not connected to gateway — message not sent` and is unable to communicate.
## Environment
- **OpenClaw version:** 2026.4.24 (cbcfdf6)
- **OS:** Linux x64
- **Node:** 22.22.... | closed | completed | true | 4 | [] | [] | 2026-04-27T08:00:01Z | 2026-04-30T04:32:46Z | 2026-04-28T00:07:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | baiwei0427 | 4,404,714 | MDQ6VXNlcjQ0MDQ3MTQ= | User | false |
openclaw/openclaw | 4,334,282,794 | I_kwDOQb6kR88AAAABAlfoKg | 72,712 | https://github.com/openclaw/openclaw/issues/72712 | https://api.github.com/repos/openclaw/openclaw/issues/72712 | memory-core dreaming cleanup fails: subagent deleteSession cannot dispatch sessions.delete (operator.admin required) | ### Symptom
When the `memory-core` dreaming pipeline finishes a narrative cycle (cron-driven
or manual), session cleanup fails with:
```
[memory-core/dreaming] failed to delete dreaming subagent session: missing scope: operator.admin
```
This recurs on every dreaming run. The narrative completes (memories a... | closed | completed | true | 1 | [] | [] | 2026-04-27T08:32:38Z | 2026-04-30T04:35:21Z | 2026-04-27T09:40:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BSG2000 | 5,871,824 | MDQ6VXNlcjU4NzE4MjQ= | User | false |
openclaw/openclaw | 4,067,677,546 | I_kwDOQb6kR87yc9Vq | 44,417 | https://github.com/openclaw/openclaw/issues/44417 | https://api.github.com/repos/openclaw/openclaw/issues/44417 | Bug: systemctl --user detection fails and hangs during `sudo -u` due to SUDO_USER fallback | Issue: `openclaw gateway install` always fails with "systemctl --user unavailable: Permission denied" even when systemd user bus is working
Environment:
- Ubuntu 24, Dell Optiplex 3060
- OpenClaw 2026.3.11 installed globally via npm under user jarvis (uid 1001)
- SSH access is via a separate sudo user admin, then `sud... | closed | completed | true | 3 | [] | [] | 2026-03-12T22:36:43Z | 2026-04-30T04:35:39Z | 2026-04-27T08:36:53Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | myrsu | 82,756,374 | MDQ6VXNlcjgyNzU2Mzc0 | User | false |
openclaw/openclaw | 4,334,195,587 | I_kwDOQb6kR88AAAABAlaTgw | 72,707 | https://github.com/openclaw/openclaw/issues/72707 | https://api.github.com/repos/openclaw/openclaw/issues/72707 | fix(cron): nested lane maxConcurrent defaults to 1, serializing all cron LLM executions | ## Bug
`cron.maxConcurrentRuns` controls the concurrency of the `cron` command lane, but cron jobs' actual LLM execution is routed to the `nested` lane via `resolveGlobalLane()` in `src/agents/pi-embedded-runner/lanes.ts`:
```ts
if (cleaned === CommandLane.Cron) return CommandLane.Nested;
```
The `nested` lane is ne... | closed | completed | true | 3 | [] | [] | 2026-04-27T08:19:15Z | 2026-04-30T04:34:42Z | 2026-04-27T08:33:40Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kagura-agent | 268,167,063 | U_kgDOD_vnlw | User | false |
openclaw/openclaw | 4,332,888,910 | I_kwDOQb6kR88AAAABAkKjTg | 72,556 | https://github.com/openclaw/openclaw/issues/72556 | https://api.github.com/repos/openclaw/openclaw/issues/72556 | Gateway should strip trailing assistant messages before sending to provider API | ## Feature / Hardening Request
The gateway should ensure conversations never end on an assistant turn before sending to provider APIs. Currently, if a context engine plugin (e.g. lossless-claw) short-circuits or returns messages unmodified, trailing assistant messages can reach the Anthropic API and cause a hard 400:
... | closed | completed | true | 2 | [] | [] | 2026-04-27T03:26:01Z | 2026-04-30T04:35:39Z | 2026-04-27T08:36:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Veda-openclaw | 267,342,076 | U_kgDOD-9Q_A | User | false |
openclaw/openclaw | 4,334,172,838 | I_kwDOQb6kR88AAAABAlY6pg | 72,706 | https://github.com/openclaw/openclaw/issues/72706 | https://api.github.com/repos/openclaw/openclaw/issues/72706 | Telegram: streaming.mode 'partial' causes intermediate message deletion (visible 'ghost message' UX bug) | ## Bug: `messengers.telegram.streaming.mode: "partial"` causes Telegram message-id replacement (visible "deleted" effect)
**Affected versions:** `openclaw@2026.4.21` (David Utrero, VPS Hostinger) + `openclaw@2026.4.24` post-Hermes update (Sergio Barrera, dedicated server)
### Symptom
When `messengers.telegram.stream... | closed | completed | true | 3 | [] | [] | 2026-04-27T08:14:58Z | 2026-05-01T04:42:51Z | 2026-04-28T16:46:38Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | David-CKS | 238,664,549 | U_kgDODjm7ZQ | User | false |
openclaw/openclaw | 4,334,350,067 | I_kwDOQb6kR88AAAABAlju8w | 72,715 | https://github.com/openclaw/openclaw/issues/72715 | https://api.github.com/repos/openclaw/openclaw/issues/72715 | [Bug]: openclaw-doctor silently self-triggers "npm install -g openclaw@latest" when a default-enabled plugin requires a newer version | ## Bug type
Behavior bug
## Summary
On a service-managed Linux host running `openclaw-gateway.service`, `openclaw-doctor` silently issues `npm install -g openclaw@latest --no-fund --no-audit --loglevel error` against the global package whenever a bundled, default-enabled plugin (e.g. `feishu`, `whatsapp`) declares a m... | closed | completed | true | 1 | [] | [] | 2026-04-27T08:44:08Z | 2026-04-30T04:35:03Z | 2026-04-27T10:19:50Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Xivi08 | 59,611,157 | MDQ6VXNlcjU5NjExMTU3 | User | false |
openclaw/openclaw | 4,055,853,446 | I_kwDOQb6kR87xv2mG | 42,820 | https://github.com/openclaw/openclaw/issues/42820 | https://api.github.com/repos/openclaw/openclaw/issues/42820 | message tool: Feishu send action polluted by poll schema/guard prevents file send | ### Summary
`message` tool cannot send a plain file to a Feishu chat using `action: "send"` when the tool schema includes poll-related fields. Any `send` call that happens to carry poll params (which the model tends to auto-fill from the schema) is rejected with:
> Poll fields require action "poll"; use action "poll"... | open | null | false | 5 | [] | [] | 2026-03-11T05:06:09Z | 2026-05-04T11:28:28Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | CharZhou | 17,255,546 | MDQ6VXNlcjE3MjU1NTQ2 | User | false |
openclaw/openclaw | 4,056,761,716 | I_kwDOQb6kR87xzUV0 | 42,986 | https://github.com/openclaw/openclaw/issues/42986 | https://api.github.com/repos/openclaw/openclaw/issues/42986 | [Feature Request] Telegram Channel: Support Thread Binding for Subagent Sessions | # OpenClaw 功能请求：Telegram 渠道支持 Thread Binding
## 文档信息
| 字段 | 值 |
|------|-----|
| **请求 ID** | FR-TELEGRAM-THREAD-001 |
| **创建日期** | 2026-03-11 |
| **优先级** | 中 |
| **技术难度** | 中等 |
| **状态** | 待审核 |
| **相关渠道** | Telegram |
| **参考实现** | Discord 渠道 |
---
## 1. 问题描述
### 1.1 当前行为
当用户使用 `sessions_spawn` 创建子代理时，如果同时设置 `mod... | open | null | false | 2 | [] | [] | 2026-03-11T08:40:53Z | 2026-05-04T20:13:54Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | JackpotHan | 30,969,269 | MDQ6VXNlcjMwOTY5MjY5 | User | false |
openclaw/openclaw | 4,056,955,670 | I_kwDOQb6kR87x0DsW | 43,020 | https://github.com/openclaw/openclaw/issues/43020 | https://api.github.com/repos/openclaw/openclaw/issues/43020 | Mattermost: multi-turn responses batch all messages at end when edit-in-place streaming is active | ## Summary
When a Mattermost agent responds with multiple messages (multi-turn tool-call responses), and edit-in-place streaming is active (`blockStreaming` enabled, introduced in #33506), the user sees unexpected behavior:
**Expected:** Each message streams independently (edit-in-place), gets posted when complete, t... | open | null | false | 2 | [] | [] | 2026-03-11T09:19:38Z | 2026-05-04T12:15:59Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | teconomix | 6,959,299 | MDQ6VXNlcjY5NTkyOTk= | User | false |
openclaw/openclaw | 4,058,284,646 | I_kwDOQb6kR87x5IJm | 43,216 | https://github.com/openclaw/openclaw/issues/43216 | https://api.github.com/repos/openclaw/openclaw/issues/43216 | Heartbeat follow-up: plugin-channel fallback and duplicate suppression gaps | ## Summary
There are still two heartbeat follow-up gaps in the current routing / delivery path.
## 1. Plugin-channel heartbeat fallback is incomplete
`src/infra/outbound/targets.ts` now lets `resolveSessionDeliveryTarget()` recover a plugin channel from stored session metadata even when the plugin registry is not bo... | open | null | false | 2 | [] | [] | 2026-03-11T13:36:01Z | 2026-05-04T00:42:53Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | hagiss | 66,506,438 | MDQ6VXNlcjY2NTA2NDM4 | User | false |
openclaw/openclaw | 4,058,915,069 | I_kwDOQb6kR87x7iD9 | 43,292 | https://github.com/openclaw/openclaw/issues/43292 | https://api.github.com/repos/openclaw/openclaw/issues/43292 | [Bug]: Unscheduled reminder guard fires for isolated-session crons + counter lost during compaction | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The `appendUnscheduledReminderNote` guard appends a false "I did not schedule a reminder" disclaimer to agent replies even when a cron job was successfully created in the current turn, due to the `successfulCronAdds` counter being reset dur... | open | null | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T15:26:03Z | 2026-05-04T12:16:19Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | rkcollins | 104,794,829 | U_kgDOBj8KzQ | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.