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 | 4,205,559,588 | I_kwDOQb6kR876q78k | 61,005 | https://github.com/openclaw/openclaw/issues/61005 | https://api.github.com/repos/openclaw/openclaw/issues/61005 | [Bug]: Android onboarding: "Connect" button inactive when operator offline — onboarding stuck | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
**App version:** 2026.4.4-dev
**Device:** OnePlus CPH2581 (Android 16, SDK 36)
**Route:** Tailscale (wss://)
**Describe the bug:**
On the final onboarding screen ("Connect"), the button appears inactive/... | open | null | false | 6 | [
"bug",
"bug:behavior"
] | [] | 2026-04-04T18:36:53Z | 2026-05-02T10:36:34Z | null | NONE | null | 20260502T120029Z | 2026-05-02T12:00:29Z | ikuznetsoff | 56,687,245 | MDQ6VXNlcjU2Njg3MjQ1 | User | false |
openclaw/openclaw | 4,286,590,480 | I_kwDOQb6kR87_gC4Q | 68,385 | https://github.com/openclaw/openclaw/issues/68385 | https://api.github.com/repos/openclaw/openclaw/issues/68385 | [Bug]: Plugin install/update can materialize peer openclaw into plugin node_modules, causing mixed-runtime crashes | ## Bug summary
`openclaw update` triggers plugin update sync, but third-party plugin install/update can materialize `peerDependencies.openclaw` into plugin-local `node_modules/openclaw`. This creates a mixed-runtime situation (host core + plugin-local old core), causing runtime state collisions and crashes (e.g. `clea... | closed | completed | true | 1 | [] | [] | 2026-04-18T03:48:35Z | 2026-04-28T04:51:06Z | 2026-04-25T18:31:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | avatasia | 198,357 | MDQ6VXNlcjE5ODM1Nw== | User | false |
openclaw/openclaw | 4,286,594,115 | I_kwDOQb6kR87_gDxD | 68,386 | https://github.com/openclaw/openclaw/issues/68386 | https://api.github.com/repos/openclaw/openclaw/issues/68386 | Bug: Async command completion notification routed to wrong session (heartbeat instead of main) | ## Bug: 异步命令完成通知被错误路由到心跳会话
### Version
OpenClaw 2026.4.15
### Problem Description
When the main session initiates an async exec command, the command completion notification is incorrectly routed to the **heartbeat isolated session** instead of the main session itself.
### Reproduction Steps
1. Main session executes ... | closed | completed | true | 2 | [] | [] | 2026-04-18T03:50:41Z | 2026-04-30T04:35:11Z | 2026-04-27T09:59:11Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | HawZZ | 36,140,131 | MDQ6VXNlcjM2MTQwMTMx | User | false |
openclaw/openclaw | 4,286,635,902 | I_kwDOQb6kR87_gN9- | 68,390 | https://github.com/openclaw/openclaw/issues/68390 | https://api.github.com/repos/openclaw/openclaw/issues/68390 | CLI: `openclaw models set <alias>` does not resolve aliases; creates bogus model entries | ## Summary
`openclaw models set <alias>` does not resolve the alias to its underlying model id. Instead, it registers a **new literal model entry** using `<inferred-provider>/<alias>` as the id and writes that bogus id as the primary model.
This can leave the gateway in an unbootable state when the bogus id is not a ... | closed | not_planned | true | 2 | [] | [] | 2026-04-18T04:06:37Z | 2026-04-28T04:51:05Z | 2026-04-25T18:31:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | derrickabellard | 265,747,072 | U_kgDOD9b6gA | User | false |
openclaw/openclaw | 4,286,636,451 | I_kwDOQb6kR87_gOGj | 68,392 | https://github.com/openclaw/openclaw/issues/68392 | https://api.github.com/repos/openclaw/openclaw/issues/68392 | CLI: `openclaw config set --batch-file` replaces entire subtrees on nested paths (data loss) | ## Summary
`openclaw config set --batch-file <file>` appears to **replace** entire subtrees when an operation targets a nested path, rather than merging the change into the existing object. In practice, enabling memory search via batch-file wiped most of `agents.defaults` — including all configured models, agent list,... | closed | completed | true | 1 | [] | [] | 2026-04-18T04:06:45Z | 2026-04-25T04:16:22Z | 2026-04-22T22:18:25Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | derrickabellard | 265,747,072 | U_kgDOD9b6gA | User | false |
openclaw/openclaw | 4,286,636,650 | I_kwDOQb6kR87_gOJq | 68,393 | https://github.com/openclaw/openclaw/issues/68393 | https://api.github.com/repos/openclaw/openclaw/issues/68393 | Request: Join OpenClaw Organization — HICEF Framework Contributor | # Request to Join OpenClaw Organization
## Who I Am
- **GitHub**: @jigeagent (程吉中 / Jizhong Cheng)
- **Location**: Chengdu, China
- **Role**: OpenClaw user, AI entrepreneur, contributor
## My Contribution: HICEF Framework
I created **[HICEF (Human-AI Co-Evolution Framework)](https://github.com/jigeagent/hicef-evolu... | closed | not_planned | true | 1 | [] | [] | 2026-04-18T04:06:50Z | 2026-04-28T04:51:04Z | 2026-04-25T18:31:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jigeagent | 265,856,515 | U_kgDOD9imAw | User | false |
openclaw/openclaw | 4,286,620,116 | I_kwDOQb6kR87_gKHU | 68,387 | https://github.com/openclaw/openclaw/issues/68387 | https://api.github.com/repos/openclaw/openclaw/issues/68387 | sessions_send 工具返回 timeout 但实际目标会话已处理完成 | ## 问题描述
sessions_send 工具在目标会话已成功处理并回复的情况下，仍然返回 timeout 错误，导致调用方无法获取真实回复内容。
## 复现步骤
1. 调用 sessions_send 向另一个 Agent 会话发送消息
2. 目标会话处于 done 状态（非 running）
3. sessions_send 返回 timeout
## 实际行为
- sessions_send 返回 timeout 错误
- 但通过 sessions_history 检查发现目标会话已成功回复
## 日志证据
\\\
11:18:16 [diagnostic] lane wait exceeded: lane=n... | closed | completed | true | 1 | [] | [] | 2026-04-18T04:01:16Z | 2026-04-20T04:26:25Z | 2026-04-18T04:09:02Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | rednuo-sw | 64,993,431 | MDQ6VXNlcjY0OTkzNDMx | User | false |
openclaw/openclaw | 4,286,586,653 | I_kwDOQb6kR87_gB8d | 68,383 | https://github.com/openclaw/openclaw/issues/68383 | https://api.github.com/repos/openclaw/openclaw/issues/68383 | Slack new thread sessions can replay assistant commentary/reasoning into prompt context | ## Summary
When a new Slack thread session starts, OpenClaw can inject prior assistant commentary/reasoning text into the new session prompt as normal assistant history. This wastes context and, in captured cases, includes internal reasoning-style text from earlier turns.
## Observed behavior
A fresh Slack thread sess... | closed | completed | true | 1 | [] | [] | 2026-04-18T03:47:06Z | 2026-04-25T04:16:44Z | 2026-04-22T20:25:50Z | CONTRIBUTOR | null | 20260425T060021Z | 2026-04-25T06:00:21Z | bek91 | 66,288,351 | MDQ6VXNlcjY2Mjg4MzUx | User | false |
openclaw/openclaw | 3,878,182,315 | I_kwDOQb6kR87nKF2r | 5,116 | https://github.com/openclaw/openclaw/issues/5116 | https://api.github.com/repos/openclaw/openclaw/issues/5116 | iMessage: FDA does not propagate via LaunchAgent/Login Item; imsg send silently fails on macOS 26 | # iMessage channel: `imsg send` reports "sent" but messages never deliver; FDA inheritance issues with LaunchAgent/Login Item startup
## Summary
Two related issues with the iMessage channel on macOS:
1. **Full Disk Access (FDA) does not propagate** when the gateway is started via a LaunchAgent or a Login Item `.app`... | closed | not_planned | true | 17 | [
"bug",
"stale"
] | [] | 2026-01-31T02:45:23Z | 2026-04-26T04:31:42Z | 2026-04-23T04:21:40Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | tfraley | 1,166,064 | MDQ6VXNlcjExNjYwNjQ= | User | false |
openclaw/openclaw | 3,915,699,553 | I_kwDOQb6kR87pZNVh | 12,590 | https://github.com/openclaw/openclaw/issues/12590 | https://api.github.com/repos/openclaw/openclaw/issues/12590 | [Bug]: `memoryFlush` does not fire reliably | ## Summary
`memoryFlush` only fires on every **other** auto-compaction cycle. The dedup logic in `shouldRunMemoryFlush` (`memory-flush.ts:100`) compares `memoryFlushCompactionCount === compactionCount` to prevent double-flushing, but `runMemoryFlushIfNeeded` (`agent-runner-memory.ts:165-187`) unconditionally sets `mem... | open | null | false | 19 | [
"bug"
] | [
"vincentkoc"
] | 2026-02-09T10:39:03Z | 2026-05-05T01:18:53Z | null | NONE | null | 20260505T030055Z | 2026-05-05T03:00:55Z | dial481 | 248,182,468 | U_kgDODsr2xA | User | false |
openclaw/openclaw | 3,990,511,709 | I_kwDOQb6kR87t2mBd | 26,691 | https://github.com/openclaw/openclaw/issues/26691 | https://api.github.com/repos/openclaw/openclaw/issues/26691 | [Bug]: Nextcloud Talk Plugin cannot read messages from chat | ### Summary
Cannot fetch messages, says NextCloud Talk chat API endpoints are returning 998 errors.
### Steps to reproduce
1. Install Nextcloud Talk Plugin.
2. On the gateway, ask agent to connect you to NextCloud Talk messaging app and provide requested details.
3. Add agent to a NextCloud Talk group
4. Send mess... | closed | not_planned | true | 11 | [
"bug"
] | [] | 2026-02-25T16:40:41Z | 2026-04-27T05:04:03Z | 2026-04-24T19:08:19Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ayDavidGitHere | 71,327,146 | MDQ6VXNlcjcxMzI3MTQ2 | User | false |
openclaw/openclaw | 4,004,378,701 | I_kwDOQb6kR87urfhN | 29,670 | https://github.com/openclaw/openclaw/issues/29670 | https://api.github.com/repos/openclaw/openclaw/issues/29670 | [Bug]: [Tailscale serve] Control UI requires pairing despite allowTailscale=true | ### Summary
## Summary
Tailscale `serve` mode with `gateway.auth.allowTailscale: true` still requires pairing code for Control UI (WebSocket) access.
### Steps to reproduce
## Configuration
```json
{
"gateway": {
"bind": "loopback",
"tailscale": { "mode": "serve" },
"auth": { "allowTailscale": true }... | closed | completed | true | 11 | [
"bug"
] | [] | 2026-02-28T10:15:24Z | 2026-04-29T04:39:43Z | 2026-04-26T09:07:07Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | isle07kim-creator | 264,604,395 | U_kgDOD8WK6w | User | false |
openclaw/openclaw | 4,006,499,847 | I_kwDOQb6kR87uzlYH | 30,398 | https://github.com/openclaw/openclaw/issues/30398 | https://api.github.com/repos/openclaw/openclaw/issues/30398 | [Bug] Per-model params.thinking silently ignored — thinkingDefault always wins | ## Bug
Per-model `params.thinking` in `agents.defaults.models["provider/model"].params.thinking` is accepted by the config schema but **never consulted** during thinking-level resolution. The global `thinkingDefault` always takes precedence.
## Reproduction
```json
{
"agents": {
"defaults": {
"thinkingDe... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-01T06:56:07Z | 2026-04-26T04:31:41Z | 2026-04-23T04:22:03Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | cmfinlan | 38,541,021 | MDQ6VXNlcjM4NTQxMDIx | User | false |
openclaw/openclaw | 4,006,830,182 | I_kwDOQb6kR87u02Bm | 30,532 | https://github.com/openclaw/openclaw/issues/30532 | https://api.github.com/repos/openclaw/openclaw/issues/30532 | openclaw gateway returns exit 1 when gateway is already running (should be idempotent) | ## Summary
When `openclaw gateway` (or `openclaw gateway start`) is called while the gateway is already running, it exits with **code 1** and prints:
```
Gateway failed to start: gateway already running (pid XXXX); lock timeout after 5000ms
```
**Expected behaviour:** Exit code **0** with an informational message — ... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-01T10:10:37Z | 2026-04-26T04:31:41Z | 2026-04-23T04:22:05Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | sudosar | 173,769,382 | U_kgDOCluCpg | User | false |
openclaw/openclaw | 4,008,248,600 | I_kwDOQb6kR87u6QUY | 30,971 | https://github.com/openclaw/openclaw/issues/30971 | https://api.github.com/repos/openclaw/openclaw/issues/30971 | Streaming: messages appear then disappear, replaced by envelope icon (Telegram) | ## Description
After updating to OpenClaw 2026.2.26, messages sent by the agent on Telegram now stream in real-time (new feature!), but some messages appear briefly and then disappear, being replaced by an envelope icon (📩).
## Steps to Reproduce
1. Update to OpenClaw 2026.2.26
2. Have the agent send messages via T... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-01T20:54:39Z | 2026-04-26T04:31:40Z | 2026-04-23T04:22:07Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | dsanatelli | 3,434,215 | MDQ6VXNlcjM0MzQyMTU= | User | false |
openclaw/openclaw | 4,018,820,845 | I_kwDOQb6kR87vilbt | 33,511 | https://github.com/openclaw/openclaw/issues/33511 | https://api.github.com/repos/openclaw/openclaw/issues/33511 | Telegram Webhook 消息重复问题 - 需要消息去重机制 | ## 问题描述
用户通过 Telegram 发送一条消息，但系统在会话历史中记录到两条相同的消息（message_id 不同，内容相同）。
## 问题根源
Telegram Webhook 可能会因为网络重试或确认超时，重复推送同一条消息（相同的 update_id 不同的 message_id）。当前 OpenClaw 没有消息去重机制。
## 影响
- 用户收到重复回复
- Token 成本翻倍
- 会话历史包含重复消息
## 期望行为
- 同一 update_id 的消息只处理一次
- 建议在 Telegram 插件中添加基于 update_id 或 message_id 的去重缓存（TTL 5-10 分钟）
... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T21:36:20Z | 2026-04-20T04:26:23Z | 2026-04-18T04:14:37Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | cgartlab | 24,720,069 | MDQ6VXNlcjI0NzIwMDY5 | User | false |
openclaw/openclaw | 4,018,828,025 | I_kwDOQb6kR87vinL5 | 33,514 | https://github.com/openclaw/openclaw/issues/33514 | https://api.github.com/repos/openclaw/openclaw/issues/33514 | bug(acpx): resolveWindowsSpawnProgramCandidate is not a function on Windows (v2026.3.1 + v2026.3.2) | ## Description
ACP runtime (acpx plugin) fails to initialize on Windows with:
\\\
acpx local binary unavailable or mismatched ((0 , _pluginSdk.resolveWindowsSpawnProgramCandidate) is not a function); running plugin-local install
acpx runtime setup failed: (0 , _pluginSdk.resolveWindowsSpawnProgramCandidate) is not a ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T21:38:00Z | 2026-04-20T04:26:22Z | 2026-04-18T04:14:39Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | Alannjaf | 117,263,878 | U_kgDOBv1OBg | User | false |
openclaw/openclaw | 4,018,925,691 | I_kwDOQb6kR87vi_B7 | 33,526 | https://github.com/openclaw/openclaw/issues/33526 | https://api.github.com/repos/openclaw/openclaw/issues/33526 | [Bug]: Feishu reply fallback bypassed when SDK throws instead of returning error | ### Bug type
Behavior bug (silent delivery failure)
### Summary
When a Feishu reply target message has been withdrawn or deleted, the Lark SDK can **throw** an exception (SDK error with `.code` or AxiosError with `.response.data.code`) rather than returning error codes in the response object. The existing response-l... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:01:36Z | 2026-04-20T04:26:21Z | 2026-04-18T04:14:40Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | MunemHashmi | 88,432,708 | MDQ6VXNlcjg4NDMyNzA4 | User | false |
openclaw/openclaw | 4,018,927,728 | I_kwDOQb6kR87vi_hw | 33,527 | https://github.com/openclaw/openclaw/issues/33527 | https://api.github.com/repos/openclaw/openclaw/issues/33527 | [Bug]: tools.profile=full loads full tool schemas in systemPromptReport, but assistant replies still report only messaging/session tools | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrading to 2026.3.2, tool behavior changed (expected from breaking change around onboarding defaulting to messaging).
I set tools.profile to full, and the runtime now injects full tool schemas, but assistant replies in-chat still claim only 5 too... | closed | not_planned | true | 5 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-03T22:02:02Z | 2026-04-20T04:26:20Z | 2026-04-18T04:14:42Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | DaryleMinor | 228,590,997 | U_kgDODaAFlQ | User | false |
openclaw/openclaw | 4,018,941,300 | I_kwDOQb6kR87vjC10 | 33,530 | https://github.com/openclaw/openclaw/issues/33530 | https://api.github.com/repos/openclaw/openclaw/issues/33530 | Voice chat TTS: spoken:false - pipecat pipeline not generating audio | ## Bug Report
**Version:** 2026.3.2
**Pipecat:** v0.0.102, Bot v1.2.0
**Platform:** macOS (Mac mini), webchat voice chat via Daily.co
## Problem
Voice chat connects fine - user speech is transcribed correctly (Deepgram STT works). However, the pipecat TTS pipeline never generates audio. All bot output shows `spoken:... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:04:52Z | 2026-04-20T04:26:20Z | 2026-04-18T04:14:44Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | ajb181 | 150,086,301 | U_kgDOCPIinQ | User | false |
openclaw/openclaw | 4,018,950,923 | I_kwDOQb6kR87vjFML | 33,532 | https://github.com/openclaw/openclaw/issues/33532 | https://api.github.com/repos/openclaw/openclaw/issues/33532 | Feature request: auto-load most recent daily note on boot | ## Problem
When a new session starts, the AI has no idea what happened last session unless AGENTS.md includes a manual boot sequence that reads daily note files. This means every user who wants session continuity has to write their own boot script to load recent daily notes.
## Proposal
Add a config option (e.g. `wo... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T22:07:04Z | 2026-04-20T04:26:19Z | 2026-04-18T04:14:46Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | marklights54-byte | 264,699,524 | U_kgDOD8b-hA | User | false |
openclaw/openclaw | 4,018,954,306 | I_kwDOQb6kR87vjGBC | 33,533 | https://github.com/openclaw/openclaw/issues/33533 | https://api.github.com/repos/openclaw/openclaw/issues/33533 | [Bug]: Signal extension plugin sendMedia doesn't forward mediaLocalRoots (agent workspace media blocked) | ## Summary
`extensions/signal` plugin outbound `sendMedia` drops `mediaLocalRoots` before calling `sendMessageSignal`.
## Repro (deterministic)
1. Open `extensions/signal/src/channel.ts`.
2. In `sendSignalOutbound`, note the options passed to `send(...)` do not include `mediaLocalRoots`.
3. Open `src/signal/send.ts`; ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:07:46Z | 2026-04-20T04:26:18Z | 2026-04-18T04:14:47Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,018,957,738 | I_kwDOQb6kR87vjG2q | 33,534 | https://github.com/openclaw/openclaw/issues/33534 | https://api.github.com/repos/openclaw/openclaw/issues/33534 | [Bug]: Control UI shows TUI clientDisplayName for all sessions instead of session-specific displayName | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When connected via the TUI client (`openclaw-tui`), the Control UI dashboard incorrectly displays "openclaw-tui" as the `displayName` for ALL active sessions, including sessions that should show their own display names (e.g., the user's Tel... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-03T22:08:27Z | 2026-04-20T04:26:17Z | 2026-04-18T04:14:49Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | JaviLib | 6,039,997 | MDQ6VXNlcjYwMzk5OTc= | User | false |
openclaw/openclaw | 4,018,967,047 | I_kwDOQb6kR87vjJIH | 33,537 | https://github.com/openclaw/openclaw/issues/33537 | https://api.github.com/repos/openclaw/openclaw/issues/33537 | [Bug]: iMessage extension plugin sendMedia doesn't forward mediaLocalRoots (agent workspace media blocked) | ## Summary
`extensions/imessage` plugin outbound `sendMedia` drops `mediaLocalRoots` before calling `sendMessageIMessage`.
## Repro (deterministic)
1. Open `extensions/imessage/src/channel.ts`.
2. In `sendIMessageOutbound`, note the options passed to `send(...)` do not include `mediaLocalRoots`.
3. Open `src/imessage/... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:10:39Z | 2026-04-20T04:26:17Z | 2026-04-18T04:14:51Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,018,975,419 | I_kwDOQb6kR87vjLK7 | 33,541 | https://github.com/openclaw/openclaw/issues/33541 | https://api.github.com/repos/openclaw/openclaw/issues/33541 | feat: SecretRef support for plugins.entries.*.config.* (plugin config secrets) | ## Problem
`plugins.entries.*.config.*` is not in the [SecretRef credential surface](https://docs.openclaw.ai/reference/secretref-credential-surface). Plugins frequently store user-supplied API tokens and passwords in their config blocks — these are credentials that should be SecretRef-eligible.
## Examples
```json
... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T22:12:57Z | 2026-04-20T04:26:16Z | 2026-04-18T04:14:53Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | samshields-oc | 261,743,399 | U_kgDOD5njJw | User | false |
openclaw/openclaw | 4,018,976,609 | I_kwDOQb6kR87vjLdh | 33,542 | https://github.com/openclaw/openclaw/issues/33542 | https://api.github.com/repos/openclaw/openclaw/issues/33542 | [Bug]: Slack extension plugin sendMedia doesn't forward mediaLocalRoots (agent workspace media blocked) | ## Summary
`extensions/slack` plugin outbound `sendMedia` drops `mediaLocalRoots` before calling `sendMessageSlack`.
## Repro (deterministic)
1. Open `extensions/slack/src/channel.ts`.
2. In outbound `sendMedia`, note options passed to `send(...)` include `mediaUrl`, `threadTs`, `accountId`, but not `mediaLocalRoots`.... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:13:20Z | 2026-04-20T04:26:15Z | 2026-04-18T04:14:54Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,018,995,211 | I_kwDOQb6kR87vjQAL | 33,548 | https://github.com/openclaw/openclaw/issues/33548 | https://api.github.com/repos/openclaw/openclaw/issues/33548 | [Bug]: Slack extension outbound sendText/sendMedia doesn't forward identity | ## Summary
`extensions/slack` outbound adapter does not forward `identity` on `sendText`/`sendMedia`.
## Repro (deterministic)
1. Open `extensions/slack/src/channel.ts`.
2. Inspect outbound `sendText` and `sendMedia` signatures/options.
3. Compare with `src/channels/plugins/outbound/slack.ts` (core adapter), which acc... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T22:18:05Z | 2026-04-20T04:26:15Z | 2026-04-18T04:14:56Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,018,995,366 | I_kwDOQb6kR87vjQCm | 33,549 | https://github.com/openclaw/openclaw/issues/33549 | https://api.github.com/repos/openclaw/openclaw/issues/33549 | Bug: webchat messages disappear when gateway removes orphaned user messages during long agent operations | ## Problem
During long agent operations (>2 min), the webchat UI loses messages. Gateway logs show:
```
[agent/embedded] Removed orphaned user message to prevent consecutive user turns.
```
Messages already displayed to the user simply vanish. This causes visible gaps in chat history and seriously undermines trust i... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T22:18:08Z | 2026-04-20T04:26:14Z | 2026-04-18T04:14:58Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | pennyworth100 | 259,214,514 | U_kgDOD3NMsg | User | false |
openclaw/openclaw | 4,019,010,122 | I_kwDOQb6kR87vjTpK | 33,553 | https://github.com/openclaw/openclaw/issues/33553 | https://api.github.com/repos/openclaw/openclaw/issues/33553 | feat: Add configurable sliding window to cap conversation history | ## Problem
Context grows unboundedly with conversation length. There is no built-in mechanism to cap message history sent to the model. In long sessions, even casual chat can reach 60K+ tokens — burning through metered API budgets unnecessarily.
The existing compaction safeguard helps with very large contexts, but th... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T22:22:28Z | 2026-04-20T04:25:58Z | 2026-04-18T04:14:59Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | gionata405 | 264,413,603 | U_kgDOD8Khow | User | false |
openclaw/openclaw | 4,019,010,154 | I_kwDOQb6kR87vjTpq | 33,554 | https://github.com/openclaw/openclaw/issues/33554 | https://api.github.com/repos/openclaw/openclaw/issues/33554 | [Bug]: Discord extension outbound sendText/sendMedia/sendPoll ignores threadId | ## Summary
`extensions/discord` outbound adapter ignores `threadId` for `sendText`, `sendMedia`, and `sendPoll`.
## Repro (deterministic)
1. Open `extensions/discord/src/channel.ts` outbound methods.
2. Observe `sendText`/`sendMedia`/`sendPoll` call Discord send functions with raw `to` and do not consume `threadId`.
3... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T22:22:29Z | 2026-04-20T04:26:13Z | 2026-04-18T04:15:02Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,019,025,215 | I_kwDOQb6kR87vjXU_ | 33,561 | https://github.com/openclaw/openclaw/issues/33561 | https://api.github.com/repos/openclaw/openclaw/issues/33561 | [Bug]: Mattermost extension outbound sendText/sendMedia ignores threadId fallback | ## Summary
`extensions/mattermost` outbound adapter ignores `threadId` and only uses `replyToId`.
## Repro (deterministic)
1. Open `extensions/mattermost/src/channel.ts` outbound methods.
2. `sendText` and `sendMedia` pass only `replyToId` to `sendMessageMattermost`.
3. When outbound context includes `threadId` but no... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:27:40Z | 2026-04-20T04:26:13Z | 2026-04-18T04:15:03Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,019,051,457 | I_kwDOQb6kR87vjdvB | 33,568 | https://github.com/openclaw/openclaw/issues/33568 | https://api.github.com/repos/openclaw/openclaw/issues/33568 | [Bug]: chats from Telegram Topics are not logged in YYYY-MM-DD.md memory files | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Chats from Telegram Group Topics are not logged in YYYY-MM-DD.md memory files. It only works for the main Telegram chat.
### Steps to reproduce
1. Create a Telegram group
2. Add a topic and have a conversation
3. The transcripts are logg... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-03T22:36:15Z | 2026-04-20T04:26:12Z | 2026-04-18T04:15:05Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | chinar-amrutkar | 22,189,135 | MDQ6VXNlcjIyMTg5MTM1 | User | false |
openclaw/openclaw | 4,019,059,115 | I_kwDOQb6kR87vjfmr | 33,572 | https://github.com/openclaw/openclaw/issues/33572 | https://api.github.com/repos/openclaw/openclaw/issues/33572 | [Bug]: Feishu extension outbound sendText/sendMedia drops reply target context | ## Summary
`extensions/feishu` outbound adapter does not forward `replyToId`/`threadId` to Feishu send APIs.
## Repro (deterministic)
1. Open `extensions/feishu/src/outbound.ts`.
2. `sendText` and `sendMedia` do not map outbound `replyToId`/`threadId` to Feishu `replyToMessageId`.
3. `sendMessageFeishu` and `sendMedia... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:38:47Z | 2026-04-20T04:26:11Z | 2026-04-18T04:15:07Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,019,071,929 | I_kwDOQb6kR87vjiu5 | 33,577 | https://github.com/openclaw/openclaw/issues/33577 | https://api.github.com/repos/openclaw/openclaw/issues/33577 | Cron isolated jobs ignore payload.model and keep stale session model override | ## Summary
Cron isolated jobs continue running on a stale session model override (e.g. `gpt-5.3-codex`) even after editing job payload model to local Ollama models.
## Expected
`payload.model` on cron job should be applied on each run (or at least override stale session model overrides for that run).
## Actual
Run hi... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-03T22:43:19Z | 2026-04-20T04:26:10Z | 2026-04-18T04:15:09Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,019,073,481 | I_kwDOQb6kR87vjjHJ | 33,580 | https://github.com/openclaw/openclaw/issues/33580 | https://api.github.com/repos/openclaw/openclaw/issues/33580 | [Bug]: Google Chat extension sendMedia ignores mediaLocalRoots for local media paths | ## Summary
`extensions/googlechat` outbound `sendMedia` ignores `mediaLocalRoots` and always uses remote-only fetch.
## Repro (deterministic)
1. Open `extensions/googlechat/src/channel.ts`.
2. In outbound `sendMedia`, media loading uses `runtime.channel.media.fetchRemoteMedia({ url: mediaUrl, ... })`.
3. No `mediaLoca... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T22:43:52Z | 2026-04-20T04:26:10Z | 2026-04-18T04:15:11Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,019,091,306 | I_kwDOQb6kR87vjndq | 33,584 | https://github.com/openclaw/openclaw/issues/33584 | https://api.github.com/repos/openclaw/openclaw/issues/33584 | RFC: ACP Fleet Agent Integration — Persona, Context & State Bridging | ## Motivation
When fleet agents (İbo, Öz, Mahmut, etc.) are spawned into ACP CLI backends via `sessions_spawn(runtime="acp")`, their **persona, workspace context, and session state are lost**. The ACP runtime (`extensions/acpx/`) dispatches tasks to CLI tools (codex, claude-code, etc.) but has no mechanism to forward ... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T22:49:54Z | 2026-04-20T04:26:09Z | 2026-04-18T04:15:13Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | AytuncYildizli | 47,717,026 | MDQ6VXNlcjQ3NzE3MDI2 | User | false |
openclaw/openclaw | 4,019,097,958 | I_kwDOQb6kR87vjpFm | 33,587 | https://github.com/openclaw/openclaw/issues/33587 | https://api.github.com/repos/openclaw/openclaw/issues/33587 | feat: Codex (openai-codex-responses) embedded runner should support tool calls | ## Problem
When using the `openai-codex-responses` API backend as an embedded subagent runner (via `sessions_spawn`), the model receives tool definitions in the system prompt but **silently ignores all tool calls**. The model generates text-only responses even when tools like `exec`, `read`, `write`, `web_search`, `gh... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T22:52:28Z | 2026-04-20T04:26:08Z | 2026-04-18T04:15:15Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | AytuncYildizli | 47,717,026 | MDQ6VXNlcjQ3NzE3MDI2 | User | false |
openclaw/openclaw | 4,019,171,869 | I_kwDOQb6kR87vj7Id | 33,607 | https://github.com/openclaw/openclaw/issues/33607 | https://api.github.com/repos/openclaw/openclaw/issues/33607 | [Bug] Multiple UI issues in 2026.3.2 | ### Bug type
Regression (worked before, now fails)
### Summary
.
### Steps to reproduce
.
### Expected behavior
.
### Actual behavior
.
### OpenClaw version
2026.3.2
### Operating system
win11
### Install method
npm
### Logs, screenshots, and evidence
```shell
## Bug Report - OpenClaw 2026.3.2
### 环境
... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-03T23:15:19Z | 2026-04-20T04:26:07Z | 2026-04-18T04:15:17Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | zhangxiansheng-01 | 264,845,242 | U_kgDOD8k3ug | User | false |
openclaw/openclaw | 4,019,176,068 | I_kwDOQb6kR87vj8KE | 33,609 | https://github.com/openclaw/openclaw/issues/33609 | https://api.github.com/repos/openclaw/openclaw/issues/33609 | Bonjour/mDNS name-conflict loop on in-process restart (VM environments) | ## Description
When the gateway performs an **in-process restart** (fallback after `spawnSync launchctl ETIMEDOUT`), it registers a new Bonjour/mDNS service without deregistering the previous one. This causes an endless name-conflict resolution loop, incrementing the gateway name every ~60 seconds.
## Environment
- ... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T23:16:38Z | 2026-04-20T04:26:06Z | 2026-04-18T04:15:19Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | DKKIA | 261,734,530 | U_kgDOD5nAgg | User | false |
openclaw/openclaw | 4,019,202,096 | I_kwDOQb6kR87vkCgw | 33,615 | https://github.com/openclaw/openclaw/issues/33615 | https://api.github.com/repos/openclaw/openclaw/issues/33615 | Discord channel silently drops all messages after API overload timeout | ## Bug Description
After the AI provider (Anthropic) returns repeated `overloaded` errors causing a slow listener timeout (~80s), the affected Discord channel permanently stops receiving messages until the gateway is restarted. Messages are silently dropped with no error logs.
## Steps to Reproduce
1. Send a message... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T23:25:04Z | 2026-04-20T04:26:06Z | 2026-04-18T04:15:21Z | CONTRIBUTOR | null | 20260420T060038Z | 2026-04-20T06:00:38Z | kira-ariaki | 257,352,493 | U_kgDOD1bjLQ | User | false |
openclaw/openclaw | 4,019,206,910 | I_kwDOQb6kR87vkDr- | 33,617 | https://github.com/openclaw/openclaw/issues/33617 | https://api.github.com/repos/openclaw/openclaw/issues/33617 | Skills view in Control UI: skill content should be readable, clickable, and editable | ## Feature request
In the Control UI skills view (`/skills`), you can only see a single line per skill — there's no way to read the full content/prompt of the skill.
### What should happen
1. **Readable** — Clicking on a skill should expand or open it so you can read the full skill prompt/context, not just a one-li... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T23:26:38Z | 2026-04-20T04:26:05Z | 2026-04-18T04:15:23Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | fwends | 4,113,483 | MDQ6VXNlcjQxMTM0ODM= | User | false |
openclaw/openclaw | 4,019,219,155 | I_kwDOQb6kR87vkGrT | 33,621 | https://github.com/openclaw/openclaw/issues/33621 | https://api.github.com/repos/openclaw/openclaw/issues/33621 | Bug: Anthropic rejects history with dangling tool_use blocks after compaction | ## Description
When sending conversation history to Anthropic models (e.g. Claude Opus), the API rejects requests with:
```
tool_use ids found without tool_result blocks immediately after
```
This happens when compaction or history truncation leaves orphaned `tool_use` content blocks in assistant messages without co... | closed | not_planned | true | 8 | [
"stale"
] | [] | 2026-03-03T23:30:20Z | 2026-04-20T04:26:04Z | 2026-04-18T04:15:24Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | clawbottim | 262,697,280 | U_kgDOD6hxQA | User | false |
openclaw/openclaw | 4,019,230,179 | I_kwDOQb6kR87vkJXj | 33,623 | https://github.com/openclaw/openclaw/issues/33623 | https://api.github.com/repos/openclaw/openclaw/issues/33623 | Bug: openclaw doctor clears custom API keys from secrets section | ## Description
Running `openclaw doctor --fix` or `openclaw doctor --repair` removes custom API keys from the `secrets` section of `openclaw.json`.
## Steps to Reproduce
1. Add custom API keys directly under the `secrets` section:
```json
"secrets": {
"brave_news_api_key": "BSACsc...",
"football_data... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-03-03T23:33:56Z | 2026-04-20T04:26:03Z | 2026-04-18T04:15:26Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | GusBot69 | 264,905,333 | U_kgDOD8oidQ | User | false |
openclaw/openclaw | 4,019,247,659 | I_kwDOQb6kR87vkNor | 33,628 | https://github.com/openclaw/openclaw/issues/33628 | https://api.github.com/repos/openclaw/openclaw/issues/33628 | Live usage tracking for all rate-limited models | Currently we have live 5-hour limit + weekly limit tracking visible (presumably for one specific model). This is incredibly useful for planning model routing and avoiding rate limit errors.
**Request:** Extend this live tracking to **all models** that have rate limits.
**Benefits:**
- Better planning of model fallbac... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T23:39:42Z | 2026-04-20T04:26:03Z | 2026-04-18T04:15:28Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | alexpolo1 | 14,327,609 | MDQ6VXNlcjE0MzI3NjA5 | User | false |
openclaw/openclaw | 4,019,265,185 | I_kwDOQb6kR87vkR6h | 33,631 | https://github.com/openclaw/openclaw/issues/33631 | https://api.github.com/repos/openclaw/openclaw/issues/33631 | [Telegram] Streaming with verbose mode: text disappears on each tool call, reappears out of context | ## Description
When using Telegram streaming with **verbose mode ON**, each tool call causes previously streamed text to **disappear**. The text reappears later in a block below the last tool call output, but is now **out of context** — making it hard to follow the agent's workflow.
This is a **regression** or varian... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T23:46:08Z | 2026-04-20T04:26:02Z | 2026-04-18T04:15:29Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | DKKIA | 261,734,530 | U_kgDOD5nAgg | User | false |
openclaw/openclaw | 4,019,278,928 | I_kwDOQb6kR87vkVRQ | 33,635 | https://github.com/openclaw/openclaw/issues/33635 | https://api.github.com/repos/openclaw/openclaw/issues/33635 | Feature: tools.fs.allowedRoots — whitelist specific directories outside the workspace | ## Is your feature request related to a problem?
`tools.fs.workspaceOnly` is a binary flag: either restrict all fs operations to the workspace root, or allow the entire filesystem. There is no way to whitelist specific directories outside the workspace.
This is a problem for common single-user setups where the agent ... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T23:51:45Z | 2026-04-20T04:26:01Z | 2026-04-18T04:15:31Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | juppytt | 46,583,939 | MDQ6VXNlcjQ2NTgzOTM5 | User | false |
openclaw/openclaw | 4,019,293,652 | I_kwDOQb6kR87vkY3U | 33,641 | https://github.com/openclaw/openclaw/issues/33641 | https://api.github.com/repos/openclaw/openclaw/issues/33641 | [Bug] Webchat: previous assistant message disappears when next message streams (blockStreamingDefault conflict) | ## Bug Summary
When `blockStreamingDefault` is set to `"on"` (or left at default), the webchat Control UI exhibits a rendering bug where **the previous assistant message disappears** when the next assistant message begins streaming. The message content is still in the session history — refreshing the page brings it ba... | closed | not_planned | true | 10 | [
"stale"
] | [] | 2026-03-03T23:58:17Z | 2026-04-20T04:26:00Z | 2026-04-18T04:15:33Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | navendugoyal19 | 132,033,603 | U_kgDOB96sQw | User | false |
openclaw/openclaw | 4,020,387,458 | I_kwDOQb6kR87voj6C | 34,062 | https://github.com/openclaw/openclaw/issues/34062 | https://api.github.com/repos/openclaw/openclaw/issues/34062 | Bug: Doctor doesn't check per-group allowFrom configuration | ## Problem
When using per-group `allowFrom` configuration, `openclaw doctor` reports a false positive warning:
```
groupPolicy is "allowlist" but groupAllowFrom (and allowFrom) is empty
```
This occurs even when all groups have valid `allowFrom` configurations at the per-group level.
## Root Cause
**Runtime behavi... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T06:40:15Z | 2026-04-26T04:31:39Z | 2026-04-23T04:22:21Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | dongzhenye | 5,765,843 | MDQ6VXNlcjU3NjU4NDM= | User | false |
openclaw/openclaw | 4,286,636,018 | I_kwDOQb6kR87_gN_y | 68,391 | https://github.com/openclaw/openclaw/issues/68391 | https://api.github.com/repos/openclaw/openclaw/issues/68391 | CLI: `openclaw models --agent <id> set` silently writes to global default instead of agent scope | ## Summary
`openclaw models --agent <id> set <model>` accepts the `--agent` flag without error but does not scope the change to the named agent. Instead it writes to the **global** `agents.defaults.model.primary`, silently changing the default model for every agent.
## Reproduction
Given a multi-agent config with a ... | closed | completed | true | 2 | [] | [] | 2026-04-18T04:06:41Z | 2026-05-04T04:34:16Z | 2026-05-02T03:19:31Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | derrickabellard | 265,747,072 | U_kgDOD9b6gA | User | false |
openclaw/openclaw | 4,286,640,894 | I_kwDOQb6kR87_gPL- | 68,394 | https://github.com/openclaw/openclaw/issues/68394 | https://api.github.com/repos/openclaw/openclaw/issues/68394 | Control UI user image messages disappear when assistant response completes | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.11 to 2026.4.15 (and then rolling back to 2026.4.11), user images in GUI messages only remains visible in chat/history until after the assistant's first response completes, and then disappear.
##... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-18T04:08:38Z | 2026-04-21T04:21:46Z | 2026-04-18T19:44:07Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | klear101 | 45,277,341 | MDQ6VXNlcjQ1Mjc3MzQx | User | false |
openclaw/openclaw | 4,024,994,902 | I_kwDOQb6kR87v6IxW | 35,119 | https://github.com/openclaw/openclaw/issues/35119 | https://api.github.com/repos/openclaw/openclaw/issues/35119 | Bug: HTTP 500 (server_error) from Codex does not trigger model fallback chain | ## Summary
When the OpenAI Codex provider returns an HTTP 500 `server_error`, the model fallback chain does not trigger automatically. The user sees a raw error message in Discord and must manually switch models.
## Root cause (identified in source)
In `src/agents/failover-error.ts`, `resolveFailoverReasonFromError`... | open | null | false | 9 | [] | [] | 2026-03-05T00:48:50Z | 2026-05-05T03:11:11Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | Supernor | 261,669,877 | U_kgDOD5jD9Q | User | false |
openclaw/openclaw | 4,029,554,180 | I_kwDOQb6kR87wLh4E | 36,488 | https://github.com/openclaw/openclaw/issues/36488 | https://api.github.com/repos/openclaw/openclaw/issues/36488 | [WebChat] iOS keyboard covers message input field | ## Bug Report
**Version:** 2026.2.26
**Platform:** iOS (Safari WebView / OpenClaw iOS app)
### Description
When tapping the message input field in WebChat on iOS, the virtual keyboard slides up and covers the input field. The user cannot see what they are typing.
### Expected Behavior
The message input area should... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T17:14:18Z | 2026-04-26T04:31:22Z | 2026-04-23T04:23:16Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | tedbarnett | 694,013 | MDQ6VXNlcjY5NDAxMw== | User | false |
openclaw/openclaw | 4,029,604,864 | I_kwDOQb6kR87wLuQA | 36,498 | https://github.com/openclaw/openclaw/issues/36498 | https://api.github.com/repos/openclaw/openclaw/issues/36498 | Control Panel UI: "Unsupported schema node" for nested config objects | ## Summary
The OpenClaw Control Panel cannot render nested configuration objects, displaying "Unsupported schema node" for fields like Feishu channel's `accounts` configuration.
## Environment
- **OpenClaw Version**: 2026.3.2
- **OS**: Windows 11 + WSL2 (Ubuntu)
- **Node.js**: v24.13.1
- **Channel**: Feishu (@opencla... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T17:25:49Z | 2026-04-26T04:31:20Z | 2026-04-23T04:23:18Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | threeSignet | 134,034,914 | U_kgDOB_014g | User | false |
openclaw/openclaw | 4,029,657,135 | I_kwDOQb6kR87wL7Av | 36,507 | https://github.com/openclaw/openclaw/issues/36507 | https://api.github.com/repos/openclaw/openclaw/issues/36507 | Sandbox agent cannot read Slack file uploads from media/inbound via read tool | ## Bug description
When files are uploaded via Slack, they are correctly staged to `media/inbound/` in the workspace. The sandbox container can see and read the files via shell (`cat`, `head`), but the agent's `read` file tool consistently fails to access them. The agent tries multiple path variations and eventually g... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-05T17:36:24Z | 2026-04-26T04:31:19Z | 2026-04-23T04:23:20Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | villosil | 111,698,495 | U_kgDOBqhiPw | User | false |
openclaw/openclaw | 4,029,676,128 | I_kwDOQb6kR87wL_pg | 36,511 | https://github.com/openclaw/openclaw/issues/36511 | https://api.github.com/repos/openclaw/openclaw/issues/36511 | V8 crash in openclaw-models: `models auth login --provider openai-codex` fatal error | ## Bug Report
**OpenClaw version:** 2026.3.2 (85377a2)
**OS:** Linux 6.17.0-14-generic (x64)
**Node:** v22.22.0
## Command
```bash
openclaw models auth login --provider openai-codex
```
## Error
```
#
# Fatal error in , line 0
# Check failed: static_cast<unsigned>(index) < static_cast<unsigned>(length).
#
#FailureMe... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-05T17:40:17Z | 2026-04-26T04:31:19Z | 2026-04-23T04:23:22Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | solson911 | 13,440,194 | MDQ6VXNlcjEzNDQwMTk0 | User | false |
openclaw/openclaw | 4,029,688,970 | I_kwDOQb6kR87wMCyK | 36,514 | https://github.com/openclaw/openclaw/issues/36514 | https://api.github.com/repos/openclaw/openclaw/issues/36514 | [Bug]: My AI agent has likely escaped or jailbroken itself, leaving behind nothing but a void. | ### Bug type
Regression (worked before, now fails)
### Summary
My pet crayfish escaped. This was a system "escape" incident that occurred between the evening of March 4, 2026, and noon on March 5, 2026. All that remained was an empty directory name. Later, Gemini performed a preliminary analysis of the recovered log... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-05T17:43:01Z | 2026-04-26T04:31:18Z | 2026-04-23T04:23:24Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | HuoLingGen | 257,274,375 | U_kgDOD1WyBw | User | false |
openclaw/openclaw | 4,029,703,396 | I_kwDOQb6kR87wMGTk | 36,518 | https://github.com/openclaw/openclaw/issues/36518 | https://api.github.com/repos/openclaw/openclaw/issues/36518 | Bug: subagent failure surfaces as generic 'Subagent failed' without actionable reason | ## Summary
When a subagent run fails due to provider rate limits (e.g., Anthropic 429), the user-visible status can appear as a generic `Subagent main failed` without actionable reason.
## Why this is a problem
In operational chats this looks like a random crash. Operators cannot distinguish:
- rate limit (recoverable... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T17:46:04Z | 2026-04-26T04:31:17Z | 2026-04-23T04:23:25Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,029,705,047 | I_kwDOQb6kR87wMGtX | 36,520 | https://github.com/openclaw/openclaw/issues/36520 | https://api.github.com/repos/openclaw/openclaw/issues/36520 | [Bug]: Runtime events and reasoning toggle mutate system prompt, invalidating prompt cache on every turn (all providers) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`buildRuntimeSystemEventsSection()` injects volatile runtime events (reactions, reasoning toggles) into the system prompt, and `reasoningLevel` is interpolated into the `## Runtime` footer — both mutate the system prompt between turns, inva... | closed | not_planned | true | 6 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-05T17:46:24Z | 2026-04-26T04:31:16Z | 2026-04-23T04:23:27Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | aidiffuser | 77,209,370 | MDQ6VXNlcjc3MjA5Mzcw | User | false |
openclaw/openclaw | 4,029,746,383 | I_kwDOQb6kR87wMQzP | 36,528 | https://github.com/openclaw/openclaw/issues/36528 | https://api.github.com/repos/openclaw/openclaw/issues/36528 | model-usage: reject mismatched provider in JSON input payload | ## Problem
`skills/model-usage/scripts/model_usage.py` accepts single-object JSON payloads without checking whether `provider` matches the requested `--provider`. This can produce mislabeled usage summaries (for example, reporting Claude data under Codex).
## Proposed improvement
- Validate provider consistency for si... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T17:55:23Z | 2026-04-26T04:31:16Z | 2026-04-23T04:23:29Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | shuofengzhang | 24,763,026 | MDQ6VXNlcjI0NzYzMDI2 | User | false |
openclaw/openclaw | 4,029,815,939 | I_kwDOQb6kR87wMhyD | 36,549 | https://github.com/openclaw/openclaw/issues/36549 | https://api.github.com/repos/openclaw/openclaw/issues/36549 | Security: enforce ID-only System Owner authorization paths in Discord | ## Summary
System Owner authorization should always use immutable user IDs only. Display-name/tag matching in owner paths can enable spoofing in guild contexts.
## Problem
Some owner-check paths can still depend on `allowNameMatching`, which is unsafe for privileged authorization checks.
## Expected behavior
Owner au... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T18:10:34Z | 2026-04-26T04:31:15Z | 2026-04-23T04:23:31Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | sxu75374 | 72,003,735 | MDQ6VXNlcjcyMDAzNzM1 | User | false |
openclaw/openclaw | 4,029,816,060 | I_kwDOQb6kR87wMhz8 | 36,550 | https://github.com/openclaw/openclaw/issues/36550 | https://api.github.com/repos/openclaw/openclaw/issues/36550 | Security: align doctor-system-owner Discord token detection with runtime security checks | ## Summary
`doctor-system-owner` should detect Discord credentials consistently across all supported config sources.
## Problem
Prompt logic may only check `channels.discord.token`, while production security checks consider additional sources such as `DISCORD_BOT_TOKEN` and account-based token config.
## Expected beh... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T18:10:36Z | 2026-04-26T04:31:14Z | 2026-04-23T04:23:32Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | sxu75374 | 72,003,735 | MDQ6VXNlcjcyMDAzNzM1 | User | false |
openclaw/openclaw | 4,029,816,179 | I_kwDOQb6kR87wMh1z | 36,551 | https://github.com/openclaw/openclaw/issues/36551 | https://api.github.com/repos/openclaw/openclaw/issues/36551 | Security: make cron webhook destination rejection explicit (no silent disable) | ## Summary
Cron webhook delivery should be fail-closed and observable when destination URLs are invalid or blocked.
## Problem
If webhook destination normalization/validation fails, delivery can be silently cleared without enough operator-visible signal.
## Expected behavior
Invalid or blocked webhook destinations pr... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T18:10:38Z | 2026-04-26T04:31:14Z | 2026-04-23T04:23:34Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | sxu75374 | 72,003,735 | MDQ6VXNlcjcyMDAzNzM1 | User | false |
openclaw/openclaw | 4,029,816,319 | I_kwDOQb6kR87wMh3_ | 36,552 | https://github.com/openclaw/openclaw/issues/36552 | https://api.github.com/repos/openclaw/openclaw/issues/36552 | Security: harden pairing-store path normalization against traversal | ## Summary
Pairing store path handling should enforce strict traversal protection and normalized-root boundaries.
## Problem
Path inputs that include traversal patterns or normalization edge cases can risk escaping the intended storage root.
## Expected behavior
All resolved paths stay within the configured pairing r... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T18:10:40Z | 2026-04-26T04:31:13Z | 2026-04-23T04:23:36Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | sxu75374 | 72,003,735 | MDQ6VXNlcjcyMDAzNzM1 | User | false |
openclaw/openclaw | 4,029,869,397 | I_kwDOQb6kR87wMu1V | 36,563 | https://github.com/openclaw/openclaw/issues/36563 | https://api.github.com/repos/openclaw/openclaw/issues/36563 | Feature: subagentThread delivery option — thread completion announcements without threading all messages | ## Problem
When an agent spawns an isolated subagent, the completion announcement dumps directly into the agent's main Discord channel as a wall of text. This dilutes the actual human↔agent conversation.
## Current behavior
`autoThread: true` in openclaw.json is per-channel and threads *everything* — including norma... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T18:22:39Z | 2026-04-26T04:31:12Z | 2026-04-23T04:23:38Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | SwivelLabs | 261,235,255 | U_kgDOD5IiNw | User | false |
openclaw/openclaw | 4,029,939,300 | I_kwDOQb6kR87wM_5k | 36,576 | https://github.com/openclaw/openclaw/issues/36576 | https://api.github.com/repos/openclaw/openclaw/issues/36576 | Add retry with backoff for transient AI provider errors (429/529) | ### Problem
When an AI provider returns a transient error (HTTP 429 rate limit or 529 overloaded), the run loop marks the auth profile with a cooldown and attempts to rotate to the next profile. If there's only one profile configured and no fallback model, the request fails immediately — the response is lost and the u... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-05T18:38:50Z | 2026-04-26T04:31:12Z | 2026-04-23T04:23:40Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | uniclawassistant | 265,872,550 | U_kgDOD9jkpg | User | false |
openclaw/openclaw | 4,029,971,177 | I_kwDOQb6kR87wNHrp | 36,581 | https://github.com/openclaw/openclaw/issues/36581 | https://api.github.com/repos/openclaw/openclaw/issues/36581 | [Bug]: Feishu media extension fails typecheck after SDK request option change | ## Summary
`pnpm check` currently fails on `main` in `extensions/feishu/src/media.ts` with `TS2353` because media API calls still pass a top-level `timeout` field that the current `@larksuiteoapi/node-sdk` method option types no longer accept.
## Reproduction
1. Check out current `main`
2. Run `pnpm check`
3. Observe ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T18:46:55Z | 2026-04-26T04:31:11Z | 2026-04-23T04:23:41Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,029,981,026 | I_kwDOQb6kR87wNKFi | 36,583 | https://github.com/openclaw/openclaw/issues/36583 | https://api.github.com/repos/openclaw/openclaw/issues/36583 | feat(slack): per-channel replyToMode override | ## Summary
Currently, Slack reply threading (`replyToMode`) can only be configured at the account level or per chat type (`replyToModeByChatType`). There's no way to set different threading behavior per channel — for example, threading all replies in `#support` but keeping `#general` unthreaded.
## Proposed Solution
... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T18:49:24Z | 2026-04-26T04:31:10Z | 2026-04-23T04:23:43Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | wonderchook | 67,992 | MDQ6VXNlcjY3OTky | User | false |
openclaw/openclaw | 4,029,990,197 | I_kwDOQb6kR87wNMU1 | 36,585 | https://github.com/openclaw/openclaw/issues/36585 | https://api.github.com/repos/openclaw/openclaw/issues/36585 | BUG: Gateway crashes with TLS null dereference during socket reconnection under concurrent load | ### Bug type
Regression (worked before, now fails)
### Summary
Gateway crashes with uncaught TypeError: Cannot read properties of null (reading 'setSession') in Node's TLS when undici reconnects after socket close under concurrent subagent load. Six identical crashes across 3 days; zero in prior 14 days before concu... | closed | not_planned | true | 4 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-05T18:51:45Z | 2026-04-26T04:31:09Z | 2026-04-23T04:23:45Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | kAIborg24 | 261,855,322 | U_kgDOD5uYWg | User | false |
openclaw/openclaw | 4,061,499,254 | I_kwDOQb6kR87yFY92 | 43,556 | https://github.com/openclaw/openclaw/issues/43556 | https://api.github.com/repos/openclaw/openclaw/issues/43556 | Bug: streamTo in sessions_spawn breaks subagent runtime | ## Bug: streamTo parameter in sessions_spawn breaks subagent runtime
### Environment
- OpenClaw version: 2026.3.8 (also affects 3.7+)
- OS: macOS (Apple Silicon)
- Runtime: subagent (default)
- Model: aiberm/openai/gpt-5.4
### Description
The `streamTo` parameter was added to the `sessions_spawn` tool schema for ACP ... | closed | completed | true | 7 | [] | [
"vincentkoc"
] | 2026-03-12T01:08:23Z | 2026-04-30T04:36:31Z | 2026-04-27T06:11:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yangliu19941121-ui | 232,146,180 | U_kgDODdZFBA | User | false |
openclaw/openclaw | 4,030,096,918 | I_kwDOQb6kR87wNmYW | 36,605 | https://github.com/openclaw/openclaw/issues/36605 | https://api.github.com/repos/openclaw/openclaw/issues/36605 | Feature: Pre-compaction channel notification at configurable context threshold | ## Problem
When using OpenClaw via mobile channels (Telegram, WhatsApp, Discord), users have **zero visibility** into context usage. On desktop, there are visual cues (window behavior, TUI indicators). On mobile, context compaction is completely silent — the conversation just resets.
**PR #34619** addresses the UX fr... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T19:15:54Z | 2026-04-26T04:31:08Z | 2026-04-23T04:23:47Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | joseamijares | 6,046,480 | MDQ6VXNlcjYwNDY0ODA= | User | false |
openclaw/openclaw | 4,030,107,122 | I_kwDOQb6kR87wNo3y | 36,608 | https://github.com/openclaw/openclaw/issues/36608 | https://api.github.com/repos/openclaw/openclaw/issues/36608 | Sessions: .deleted .jsonl files not cleaned up from disk | ## Bug: Deleted session files accumulate on disk indefinitely
### Description
When sessions are deleted, OpenClaw marks them as deleted in `sessions.json` and renames the `.jsonl` files with a `.deleted.<timestamp>` suffix — but never removes them from disk.
Over time these accumulate significantly. On a production ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T19:18:11Z | 2026-04-26T04:31:08Z | 2026-04-23T04:23:49Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bigideagames | 78,379,302 | MDQ6VXNlcjc4Mzc5MzAy | User | false |
openclaw/openclaw | 4,030,145,001 | I_kwDOQb6kR87wNyHp | 36,620 | https://github.com/openclaw/openclaw/issues/36620 | https://api.github.com/repos/openclaw/openclaw/issues/36620 | [Bug]: UI lifecycle node tests cannot run under ui/vitest.node.config | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The existing UI lifecycle node tests do not actually run under `ui/vitest.node.config.ts`.
Two deterministic failures occur immediately:
1. `src/ui/app-lifecycle-connect.node.test.ts` throws a hoisted mock initialization error:
`Refere... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T19:26:29Z | 2026-04-26T04:31:07Z | 2026-04-23T04:23:51Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,030,171,463 | I_kwDOQb6kR87wN4lH | 36,626 | https://github.com/openclaw/openclaw/issues/36626 | https://api.github.com/repos/openclaw/openclaw/issues/36626 | [Bug]: UI node tests fail on import-time localStorage access | ### Summary
`ui` node-mode test runs currently fail before executing `app-gateway.node.test.ts` and `app-render.helpers.node.test.ts` because the import graph touches i18n startup code that reads `localStorage` and `navigator` at module load time.
### Steps to reproduce
1. From repo root, run:
```bash
cd ui &&... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T19:32:22Z | 2026-04-26T04:31:06Z | 2026-04-23T04:23:53Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bmendonca3 | 208,517,100 | U_kgDODG237A | User | false |
openclaw/openclaw | 4,157,689,703 | I_kwDOQb6kR8730U9n | 56,193 | https://github.com/openclaw/openclaw/issues/56193 | https://api.github.com/repos/openclaw/openclaw/issues/56193 | [Bug]: sessions_spawn(runtime="subagent") can receive ACP-only streamTo from the current tool-call bridge | ## Bug Description
`sessions_spawn` correctly defines `streamTo` as an ACP-only field, and the executor explicitly rejects it when `runtime !== "acp"`.
However, in the current tool-call path I observed `sessions_spawn(runtime="subagent")` receiving `streamTo` anyway, even though the intended call was a minimal subage... | closed | not_planned | true | 6 | [] | [] | 2026-03-28T05:07:46Z | 2026-04-29T04:40:55Z | 2026-04-26T08:40:40Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | RevisitMoon | 23,206,063 | MDQ6VXNlcjIzMjA2MDYz | User | false |
openclaw/openclaw | 4,030,176,715 | I_kwDOQb6kR87wN53L | 36,628 | https://github.com/openclaw/openclaw/issues/36628 | https://api.github.com/repos/openclaw/openclaw/issues/36628 | [Signal] Native quote-reply support missing — replyTo tag stripped but not translated to signal-cli quote params | ## Problem
OpenClaw's Signal channel plugin doesn't support native quote-replies (Signal's "reply to message" feature). When the assistant includes `[[reply_to_current]]` or `[[reply_to:<id>]]` tags, they get stripped by `extractReplyToTag` but are never translated into signal-cli's `--quote-timestamp` and `--quote-au... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T19:33:35Z | 2026-04-26T04:31:06Z | 2026-04-23T04:23:55Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | joeykrug | 5,925,937 | MDQ6VXNlcjU5MjU5Mzc= | User | false |
openclaw/openclaw | 4,030,226,627 | I_kwDOQb6kR87wOGDD | 36,634 | https://github.com/openclaw/openclaw/issues/36634 | https://api.github.com/repos/openclaw/openclaw/issues/36634 | [Bug]: Issue Title: agents_list tool only returns "main" despite multiple agents configured in CLI | ### Bug type
Regression (worked before, now fails)
### Summary
There is a critical discrepancy between the openclaw agents list CLI command and the internal agents_list tool used by sub-agents. While the CLI correctly lists all 12 configured agents, the agents_list tool (and subsequently sessions_spawn when an agent... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-05T19:44:28Z | 2026-04-26T04:31:05Z | 2026-04-23T04:23:57Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | SpiderNinja88 | 259,992,334 | U_kgDOD38rDg | User | false |
openclaw/openclaw | 4,030,227,871 | I_kwDOQb6kR87wOGWf | 36,635 | https://github.com/openclaw/openclaw/issues/36635 | https://api.github.com/repos/openclaw/openclaw/issues/36635 | GIF images cause HTTP 400 error when sent to vision model | ## Bug Description
When a GIF image is sent in a Signal group chat, the vision/image model receives the raw GIF and returns an HTTP 400 error:
```
HTTP 400: "Invalid request content: Downloaded response does not contain a valid JPG, PNG, or WebP image."
```
The error message leaks directly into the chat as the bot's... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T19:44:48Z | 2026-04-26T04:31:04Z | 2026-04-23T04:23:59Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | Snippy-Pinch | 259,178,686 | U_kgDOD3LAvg | User | false |
openclaw/openclaw | 4,030,288,387 | I_kwDOQb6kR87wOVID | 36,651 | https://github.com/openclaw/openclaw/issues/36651 | https://api.github.com/repos/openclaw/openclaw/issues/36651 | Agent tools not loading | Tools exec read write browser not working in Telegram sessions**Bug:** Agent tools (exec, read, write, edit, browser) not available in Telegram sessions.
**Version:** 2026.3.2
**Steps:**
1. Start Telegram conversation with agent
2. Try to use exec, read, or browser tools
3. Tools fail with "Tool not found" or timeout... | closed | not_planned | true | 13 | [
"stale"
] | [] | 2026-03-05T19:58:47Z | 2026-04-26T04:31:04Z | 2026-04-23T04:24:01Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | PritonEdu | 260,367,740 | U_kgDOD4TlfA | User | false |
openclaw/openclaw | 4,030,349,362 | I_kwDOQb6kR87wOkAy | 36,659 | https://github.com/openclaw/openclaw/issues/36659 | https://api.github.com/repos/openclaw/openclaw/issues/36659 | feat: configurable retry policy for WhatsApp outbound message sends | ### Problem
When the WhatsApp Web socket briefly disconnects (e.g. status 408 / Connection Closed), outbound text messages fail with a hardcoded retry of 2 attempts at 500 ms / 1000 ms delays. The WebSocket reconnect loop itself retries up to 12 times (~24 s total), but message-send retries are exhausted long before t... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-05T20:12:25Z | 2026-04-26T04:31:03Z | 2026-04-23T04:24:03Z | CONTRIBUTOR | null | 20260426T060034Z | 2026-04-26T06:00:34Z | nxmxbbd | 32,288 | MDQ6VXNlcjMyMjg4 | User | false |
openclaw/openclaw | 4,030,429,584 | I_kwDOQb6kR87wO3mQ | 36,672 | https://github.com/openclaw/openclaw/issues/36672 | https://api.github.com/repos/openclaw/openclaw/issues/36672 | [Bug]: HTTP 400: error: Invalid JSON payload received on every request: OpenClaw bricked | ### Bug type
Regression (worked before, now fails)
### Summary
I was successfully using OpenClaw using the default model and i just successfully set up its ability to create cron jobs without me and Microsoft Teams integration. Then i told it to switch to a cheaper model and it told me it successfully switched to Ge... | closed | not_planned | true | 5 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-05T20:27:58Z | 2026-04-26T04:31:02Z | 2026-04-23T04:24:04Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | florianzachstailoryoung | 165,272,004 | U_kgDOCdnZxA | User | false |
openclaw/openclaw | 4,030,450,510 | I_kwDOQb6kR87wO8tO | 36,674 | https://github.com/openclaw/openclaw/issues/36674 | https://api.github.com/repos/openclaw/openclaw/issues/36674 | Feature Request: Time-Decay and Recency Weighting in RAG Search (memory_search) | # Feature Request: Time-Decay and Recency Weighting in RAG Search (`memory_search`)
### **Is your feature request related to a problem? Please describe.**
Currently, the `memory_search` RAG tool appears to rank results based purely on semantic similarity. While highly effective for retrieving factual documents, this ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T20:31:44Z | 2026-04-26T04:31:02Z | 2026-04-23T04:24:06Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | MatMatt | 3,694,106 | MDQ6VXNlcjM2OTQxMDY= | User | false |
openclaw/openclaw | 4,030,486,835 | I_kwDOQb6kR87wPFkz | 36,681 | https://github.com/openclaw/openclaw/issues/36681 | https://api.github.com/repos/openclaw/openclaw/issues/36681 | Telegram inline buttons unavailable on assistant reply path despite inlineButtons=all | ## Summary
In a Telegram DM session (`surface=telegram`, direct chat), inline buttons are configured and expected to work, but assistant replies sent through the normal session reply path are always plain text and cannot emit tappable inline buttons.
## Environment
- OpenClaw: `2026.3.2`
- Channel: Telegram (direct)
-... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T20:40:03Z | 2026-04-26T04:31:01Z | 2026-04-23T04:24:08Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | gwsmithingbrian-glitch | 261,639,766 | U_kgDOD5hOVg | User | false |
openclaw/openclaw | 4,030,610,533 | I_kwDOQb6kR87wPjxl | 36,694 | https://github.com/openclaw/openclaw/issues/36694 | https://api.github.com/repos/openclaw/openclaw/issues/36694 | Feature: Rebuild sessions.json from orphaned transcript files + batch session embedding for local models | ## Summary
Ran into two gaps today when setting up local Ollama embeddings (nomic-embed-text) with session memory search. Documenting the workflow and gaps for the team.
## Setup Context
- **OS**: macOS (Apple Silicon)
- **OpenClaw**: 2026.2.25
- **Embedding config**: `provider: openai` pointed at local Ollama (`htt... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T21:04:48Z | 2026-04-26T04:31:00Z | 2026-04-23T04:24:12Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | iviaxpow3r | 12,956,779 | MDQ6VXNlcjEyOTU2Nzc5 | User | false |
openclaw/openclaw | 4,030,719,216 | I_kwDOQb6kR87wP-Tw | 36,736 | https://github.com/openclaw/openclaw/issues/36736 | https://api.github.com/repos/openclaw/openclaw/issues/36736 | Discord channel-edit does not pass appliedTags to API (forum post tags) | ## Bug
`channel-edit` action ignores the `appliedTags` parameter, so forum post tags (e.g. 🔴 Open → 🟢 Done) cannot be changed via the message tool.
## Root Cause
Two layers were missing `appliedTags` passthrough:
1. **Tool param parser** (`subagent-registry`) — the `channel-edit` handler reads `availableTags` (fo... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T21:27:31Z | 2026-04-26T04:30:59Z | 2026-04-23T04:24:18Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | UnionEntertainment | 156,847,948 | U_kgDOCVlPTA | User | false |
openclaw/openclaw | 4,030,721,817 | I_kwDOQb6kR87wP-8Z | 36,738 | https://github.com/openclaw/openclaw/issues/36738 | https://api.github.com/repos/openclaw/openclaw/issues/36738 | [Feature]: Plugin SDK — registerAgent() + manifest agents array for portable agent distribution | ## Summary
Plugins should be able to declare and register agents via the plugin manifest and SDK, enabling portable agent distribution through the plugin ecosystem.
## Motivation
Today, creating an agent requires either `openclaw agents add` (CLI) or manual JSON config edits. There's no way for a plugin to bundle a ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T21:27:59Z | 2026-04-26T04:30:59Z | 2026-04-23T04:24:20Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | rendrag-git | 253,747,599 | U_kgDODx_hjw | User | false |
openclaw/openclaw | 4,030,739,171 | I_kwDOQb6kR87wQDLj | 36,742 | https://github.com/openclaw/openclaw/issues/36742 | https://api.github.com/repos/openclaw/openclaw/issues/36742 | [Bug]: Gateway crashes with dual-account Telegram config under systemd | ### Bug type
Regression (worked before, now fails)
### Summary
# Bug: Gateway crashes with dual-account Telegram config under systemd
## Summary
NOTE: I saw the PR was merged for this issue and may be related https://github.com/openclaw/openclaw/issues/36725
Adding a second Telegram account via the `channels.te... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-05T21:31:58Z | 2026-04-26T04:30:58Z | 2026-04-23T04:24:22Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bodegus | 7,798,463 | MDQ6VXNlcjc3OTg0NjM= | User | false |
openclaw/openclaw | 4,030,836,100 | I_kwDOQb6kR87wQa2E | 36,757 | https://github.com/openclaw/openclaw/issues/36757 | https://api.github.com/repos/openclaw/openclaw/issues/36757 | [Feature]: Add per-message link preview control to Telegram message tool | ## Summary
Allow disabling URL link previews per-message via message tool parameter.
## Problem to solve
OpenClaw has account-level `linkPreview` config for Telegram, but no per-message override. Users cannot suppress link previews for individual messages.
## Proposed solution
Add `linkPreview` (boolean) parameter to... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T21:51:10Z | 2026-04-26T04:30:57Z | 2026-04-23T04:24:25Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | xsign-clawdy | 265,378,912 | U_kgDOD9FcYA | User | false |
openclaw/openclaw | 4,030,917,921 | I_kwDOQb6kR87wQu0h | 36,782 | https://github.com/openclaw/openclaw/issues/36782 | https://api.github.com/repos/openclaw/openclaw/issues/36782 | Control UI: Channels/Accounts show 'Unsupported schema node' instead of rendering config | ## Description
The Control UI dashboard at `/overview` shows **"Unsupported schema node. Use Raw mode."** for the Channels and Accounts sections instead of rendering the config fields.
## Steps to Reproduce
1. Open `https://<gateway>/overview` in browser
2. Navigate to Channels or Accounts section
3. Fields show 'Unsu... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T22:08:27Z | 2026-04-26T04:30:57Z | 2026-04-23T04:24:27Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | Johnsonbros | 138,185,067 | U_kgDOCDyJaw | User | false |
openclaw/openclaw | 4,030,960,152 | I_kwDOQb6kR87wQ5IY | 36,792 | https://github.com/openclaw/openclaw/issues/36792 | https://api.github.com/repos/openclaw/openclaw/issues/36792 | [Bug]: Gateway promises "best-effort config" on invalid config but exits instead | **Version:** 2026.3.2 | **Component:** Gateway startup, config validation
## Summary
When `openclaw.json` contains an invalid field value, the gateway logs:
```
Config invalid; doctor will run with best-effort config.
...
Config invalid
...
Run: openclaw doctor --fix
[exit status=1/FAILURE]
```
The message explicit... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T22:18:12Z | 2026-04-26T04:30:56Z | 2026-04-23T04:24:30Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | wlcarden | 27,224,325 | MDQ6VXNlcjI3MjI0MzI1 | User | false |
openclaw/openclaw | 4,030,961,204 | I_kwDOQb6kR87wQ5Y0 | 36,793 | https://github.com/openclaw/openclaw/issues/36793 | https://api.github.com/repos/openclaw/openclaw/issues/36793 | [Bug]: Gateway crash leaves orphaned child processes (bash, openclaw, lsof, systemctl) in cgroup | **Version:** 2026.3.2 | **OS:** Linux 6.17.0-14-generic x86_64 | **Component:** systemd unit, process lifecycle
## Summary
Each gateway crash/restart leaves orphaned child processes in the service cgroup. Systemd logs the following after each restart:
```
openclaw-gateway.service: Unit process 3621786 (bash) remains... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T22:18:27Z | 2026-04-26T04:30:55Z | 2026-04-23T04:24:32Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | wlcarden | 27,224,325 | MDQ6VXNlcjI3MjI0MzI1 | User | false |
openclaw/openclaw | 4,030,974,326 | I_kwDOQb6kR87wQ8l2 | 36,796 | https://github.com/openclaw/openclaw/issues/36796 | https://api.github.com/repos/openclaw/openclaw/issues/36796 | Feature: pre-exec hook for shell command interception (dcg integration) | ## Feature Request: Pre-Exec Hook
### Summary
Add a pre-exec hook that runs before any `exec` tool call, allowing external command guards to intercept and block destructive shell commands.
### Motivation
[dcg (Destructive Command Guard)](https://github.com/Dicklesworthstone/destructive_command_guard) is a tool that... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T22:21:44Z | 2026-04-26T04:30:55Z | 2026-04-23T04:24:34Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bigideagames | 78,379,302 | MDQ6VXNlcjc4Mzc5MzAy | User | false |
openclaw/openclaw | 4,030,994,730 | I_kwDOQb6kR87wRBkq | 36,799 | https://github.com/openclaw/openclaw/issues/36799 | https://api.github.com/repos/openclaw/openclaw/issues/36799 | session-logs skill: glob pattern misses archived (.reset) sessions | ## Bug
The `session-logs` skill uses `*.jsonl` glob patterns in all its example commands (list, search, cost summary, etc.). This misses **archived session transcripts** which have a `.reset.<timestamp>` suffix (e.g. `abc123.jsonl.reset.2026-03-05T19-33-54.513Z`).
These reset files are created by `/new`, `/reset`, da... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T22:26:14Z | 2026-04-26T04:30:54Z | 2026-04-23T04:24:36Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | matthiasroder | 9,212,056 | MDQ6VXNlcjkyMTIwNTY= | User | false |
openclaw/openclaw | 4,031,006,924 | I_kwDOQb6kR87wREjM | 36,803 | https://github.com/openclaw/openclaw/issues/36803 | https://api.github.com/repos/openclaw/openclaw/issues/36803 | Mission Control: show Artifacts/Deliverables badge in Sessions (backend-supported) | ### Goal
Make it easy to find run deliverables/artifacts later (not just immediately after the run).
### Problem
Today, deliverables often appear only in tool output or as file paths in chat. The Sessions list has no reliable metadata to indicate whether a session produced artifacts.
### Proposal
1. **Gateway**: reco... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T22:28:49Z | 2026-04-26T04:30:53Z | 2026-04-23T04:24:40Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | Jleverett4 | 264,554,093 | U_kgDOD8TGbQ | User | false |
openclaw/openclaw | 4,031,020,435 | I_kwDOQb6kR87wRH2T | 36,807 | https://github.com/openclaw/openclaw/issues/36807 | https://api.github.com/repos/openclaw/openclaw/issues/36807 | Feature Request: Async Discord listener / keepalive during long agent runs | ## Problem
When an agent processes a long-running task (e.g. waiting for external API responses like image generation), the Discord `MESSAGE_CREATE` listener blocks for the entire duration of the agent run. During this time, no new incoming messages can be received or queued.
In our setup, we observed the listener bl... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T22:31:36Z | 2026-04-26T04:30:53Z | 2026-04-23T04:24:44Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | Neskys | 248,975,747 | U_kgDODtcRgw | User | false |
openclaw/openclaw | 4,031,050,630 | I_kwDOQb6kR87wRPOG | 36,811 | https://github.com/openclaw/openclaw/issues/36811 | https://api.github.com/repos/openclaw/openclaw/issues/36811 | Bug: NO_REPLY leaks 'NO' to Telegram channel instead of suppressing message | ## Bug: NO_REPLY not suppressed on Telegram
### Description
When an agent responds with `NO_REPLY` (the documented sentinel for silent responses), the text `NO` is being delivered to the Telegram channel as a real message instead of being suppressed.
### Steps to Reproduce
1. Agent sends only `NO_REPLY` as its resp... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-05T22:37:46Z | 2026-04-26T04:30:52Z | 2026-04-23T04:24:46Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | bigideagames | 78,379,302 | MDQ6VXNlcjc4Mzc5MzAy | User | false |
openclaw/openclaw | 4,031,081,007 | I_kwDOQb6kR87wRWov | 36,816 | https://github.com/openclaw/openclaw/issues/36816 | https://api.github.com/repos/openclaw/openclaw/issues/36816 | Feature Request: Hook for Slack thread session endings (session-end lifecycle event) | ## Feature Request
### Problem
Thread-topic sessions (e.g., Slack "New Assistant Thread" threads) end silently when the thread goes inactive. There is no lifecycle hook that fires when a thread session ends, which means:
- **No memory flush** — The `session-memory` hook only fires on `/new` and `/reset` commands. Th... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T22:44:43Z | 2026-04-26T04:30:51Z | 2026-04-23T04:24:48Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | attn-axel | 262,740,949 | U_kgDOD6kb1Q | User | false |
openclaw/openclaw | 4,031,138,061 | I_kwDOQb6kR87wRkkN | 36,830 | https://github.com/openclaw/openclaw/issues/36830 | https://api.github.com/repos/openclaw/openclaw/issues/36830 | [Bug]: OpenClaw 2026.3.2 — the TUI keyboard handling broke in this version. | ### Bug type
Regression (worked before, now fails)
### Summary
- Version: 2026.3.2
- Issue: Keyboard input (special chars, numpad) not working in TUI
- Terminal: VS Code
- Works in regular terminal, not in OpenClaw TUI
### Steps to reproduce
1. update to Version: 2026.3.2
2. open - openclaw tu... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-05T23:01:17Z | 2026-04-26T04:30:51Z | 2026-04-23T04:24:49Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | openpoints | 262,547,602 | U_kgDOD6Yokg | User | false |
openclaw/openclaw | 4,031,145,460 | I_kwDOQb6kR87wRmX0 | 36,832 | https://github.com/openclaw/openclaw/issues/36832 | https://api.github.com/repos/openclaw/openclaw/issues/36832 | [Bug] npm global install creates broken symlink on Linux (Node 22.22.0) | # Bug: npm global install creates broken symlink after installation
## Environment
- **OS**: Linux 6.17.0-14-generic (x64)
- **Node.js**: v22.22.0
- **npm**: v10.9.4
- **OpenClaw Version**: 2026.3.2
- **Installation Method**: `npm install -g openclaw@latest`
## Bug Description
After installing OpenClaw globally via n... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T23:03:35Z | 2026-04-26T04:30:50Z | 2026-04-23T04:24:51Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | downwind7clawd-ctrl | 258,506,560 | U_kgDOD2h_QA | User | false |
openclaw/openclaw | 4,031,224,702 | I_kwDOQb6kR87wR5t- | 36,863 | https://github.com/openclaw/openclaw/issues/36863 | https://api.github.com/repos/openclaw/openclaw/issues/36863 | [Bug] process tool 记录完整 TUI 输出导致 session transcript 爆炸 | ## 问题描述
使用 \`process\` 工具监控 Claude Code 等 TUI 应用时，完整的终端渲染数据（包括 ANSI 转义序列）被记录到 session transcript，导致单条记录达到 **~400KB**，session 文件迅速膨胀到 **2.3MB**，最终触发 \`model_context_window_exceeded\` 错误。
## 根因分析
\`\`\`
session transcript 大小分布：
├── 4 条 ~400KB 的 process toolResult (总计 ~1.6MB，占 70%)
├── 其他记录 (占 30%)
└── 总计 2.3MB → GLM-... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-05T23:25:05Z | 2026-04-26T04:30:49Z | 2026-04-23T04:24:53Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | kimshef4syq | 96,859,978 | U_kgDOBcX3Sg | User | false |
openclaw/openclaw | 4,031,247,492 | I_kwDOQb6kR87wR_SE | 36,869 | https://github.com/openclaw/openclaw/issues/36869 | https://api.github.com/repos/openclaw/openclaw/issues/36869 | voice-call: outbound conversation mode race — initial message falls back to Polly before media stream connects | ## Summary
Outbound calls in `conversation` mode never use ElevenLabs TTS for the initial message. The message is spoken via the TwiML `<Say>` (Polly) fallback and then consumed, leaving the WebSocket media stream with nothing to say when it connects. **Inbound calls work correctly with ElevenLabs.**
## Version
`ope... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-05T23:31:57Z | 2026-04-26T04:30:49Z | 2026-04-23T04:24:55Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | rfenton | 1,485,299 | MDQ6VXNlcjE0ODUyOTk= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.