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 | 4,079,482,137 | I_kwDOQb6kR87zJ_UZ | 47,695 | https://github.com/openclaw/openclaw/issues/47695 | https://api.github.com/repos/openclaw/openclaw/issues/47695 | [Feature]: Temporary time-bound exec security escalation via /trust and /untrust | Temporary time-bound exec security escalation via /trust and /untrust commands.
## Problem to solve
Existing exec approval workflows (deny/allowlist/full) are binary and permanent per session. Users who want exec.security=full for a focused coding session must either:
1. Set it permanently with `/exec security=full` ... | closed | not_planned | true | 2 | [] | [] | 2026-03-15T23:48:47Z | 2026-04-28T04:39:13Z | 2026-04-26T03:20:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | keylimesoda | 7,773,610 | MDQ6VXNlcjc3NzM2MTA= | User | false |
openclaw/openclaw | 4,329,919,349 | I_kwDOQb6kR88AAAABAhVTdQ | 71,922 | https://github.com/openclaw/openclaw/issues/71922 | https://api.github.com/repos/openclaw/openclaw/issues/71922 | fix(model-fallback): distinguish empty_response / no_error_details / unclassified instead of collapsing to unknown | # fix(model-fallback): distinguish empty_response / no_error_details / unclassified instead of collapsing to "unknown"
## Summary
Currently, three distinct failure scenarios all produce `reason: "unknown"` in model fallback events. Each has enough information to be labelled more specifically, but the label is discard... | closed | completed | true | 1 | [] | [] | 2026-04-26T03:22:44Z | 2026-04-30T04:34:14Z | 2026-04-27T12:11:03Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | andyk-ms | 91,510,251 | U_kgDOBXRV6w | User | false |
openclaw/openclaw | 4,079,933,700 | I_kwDOQb6kR87zLtkE | 47,831 | https://github.com/openclaw/openclaw/issues/47831 | https://api.github.com/repos/openclaw/openclaw/issues/47831 | Slack Socket Mode not receiving app_mention events in channels | Socket Mode connection establishes successfully and DMs work, but @mentions in channels never trigger events.
**Environment:**
- OpenClaw version: 2026.3.13
- Platform: macOS (Darwin 25.3.0 arm64)
- Slack mode: Socket Mode (not webhook)
**Configuration:**
```json
{
"channels": {
"slack": {
"mode": "socket... | closed | not_planned | true | 2 | [] | [] | 2026-03-16T02:58:29Z | 2026-04-28T04:39:09Z | 2026-04-26T03:22:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sbusso | 346 | MDQ6VXNlcjM0Ng== | User | false |
openclaw/openclaw | 4,080,155,178 | I_kwDOQb6kR87zMjoq | 47,890 | https://github.com/openclaw/openclaw/issues/47890 | https://api.github.com/repos/openclaw/openclaw/issues/47890 | feat(feishu): support interactive card (msg_type: "interactive") in outbound adapter | ## Use Case
We're building a restaurant AI system ([smart-dining](https://github.com/Horacehxw/restaurant-ai)) that sends daily business reports to store managers via Feishu. These reports need **interactive cards** with:
- `column_set` — multi-column metrics layout (revenue, orders, avg ticket, discount rate)
- `img... | closed | not_planned | true | 2 | [] | [] | 2026-03-16T04:29:37Z | 2026-04-28T04:39:08Z | 2026-04-26T03:22:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Horacehxw | 19,310,183 | MDQ6VXNlcjE5MzEwMTgz | User | false |
openclaw/openclaw | 4,080,198,874 | I_kwDOQb6kR87zMuTa | 47,903 | https://github.com/openclaw/openclaw/issues/47903 | https://api.github.com/repos/openclaw/openclaw/issues/47903 | feat(hooks): add cron event type to internal hook system | ## Summary
Add a `cron` event type to OpenClaw's internal hook system (`registerInternalHook`) so that cron job mutations (`added`, `updated`, `removed`) and lifecycle transitions (`started`, `finished`) are observable by user-space hook handlers — the same way `command`, `session`, `agent`, `gateway`, and `message` e... | closed | not_planned | true | 1 | [] | [] | 2026-03-16T04:45:28Z | 2026-04-28T04:39:07Z | 2026-04-26T03:22:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | spectra-the-bot | 262,214,320 | U_kgDOD6ESsA | User | false |
openclaw/openclaw | 4,080,240,790 | I_kwDOQb6kR87zM4iW | 47,906 | https://github.com/openclaw/openclaw/issues/47906 | https://api.github.com/repos/openclaw/openclaw/issues/47906 | feat: lane isolation — separate conversational vs long-job session lanes | ## Problem
When an agent runs a long tool call (60s+ exec, large file processing, multi-step sub-agent work), the primary session lane is blocked. New user messages queue behind the running operation. The agent appears dead even though it's actively working.
## Real-world impact
From our logs (8-agent fleet, active ... | closed | not_planned | true | 1 | [] | [] | 2026-03-16T05:02:12Z | 2026-04-28T04:39:07Z | 2026-04-26T03:22:58Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | issaba1 | 9,679,850 | MDQ6VXNlcjk2Nzk4NTA= | User | false |
openclaw/openclaw | 4,080,261,405 | I_kwDOQb6kR87zM9kd | 47,913 | https://github.com/openclaw/openclaw/issues/47913 | https://api.github.com/repos/openclaw/openclaw/issues/47913 | Thinking content leaks into text for MiniMax/Kimi — need per-model thinking control | ## Problem
Third-party models (MiniMax M2.5, Moonshot Kimi K2.5) leak their internal thinking/reasoning process into the `text` content of responses. This is not in a structured `thinking` or `reasoning_content` field — it's raw text mixed into the actual reply, making it visible to end users.
## Current Behavior
##... | closed | completed | true | 5 | [] | [] | 2026-03-16T05:09:14Z | 2026-04-28T04:39:06Z | 2026-04-26T03:23:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | semiok | 1,728,693 | MDQ6VXNlcjE3Mjg2OTM= | User | false |
openclaw/openclaw | 4,080,359,439 | I_kwDOQb6kR87zNVgP | 47,928 | https://github.com/openclaw/openclaw/issues/47928 | https://api.github.com/repos/openclaw/openclaw/issues/47928 | feat: show current task name and ETA in processing indicator | ## Problem
When an agent handles a long task, users receive a generic message like "任务处理时间较长，请稍等" (Processing takes longer, please wait).
This creates **two problems**:
1. **Which task?** Users may have submitted multiple requests. The generic message doesn't tell them which one is being processed.
2. **How long?** ... | closed | not_planned | true | 1 | [] | [] | 2026-03-16T05:42:21Z | 2026-04-28T04:39:05Z | 2026-04-26T03:23:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | eric-wei-ai | 254,221,737 | U_kgDODycdqQ | User | false |
openclaw/openclaw | 4,080,647,096 | I_kwDOQb6kR87zObu4 | 47,978 | https://github.com/openclaw/openclaw/issues/47978 | https://api.github.com/repos/openclaw/openclaw/issues/47978 | One-shot cron jobs with webchat delivery fail with Telegram validation error | ## Bug Report: One-Shot Cron Jobs with Webchat Delivery Fail
**Date:** March 15, 2026
**Reporter:** OpenClaw user
**OpenClaw Version:** 2026.3.13-1
### Summary
One-shot cron jobs (using `schedule.kind: "at"`) with `channel: "webchat"` delivery are incorrectly validated against Telegram rules, causing the job to fail... | closed | not_planned | true | 3 | [] | [] | 2026-03-16T07:04:55Z | 2026-04-28T04:39:03Z | 2026-04-26T03:23:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | magnusbonnevier | 1,112,476 | MDQ6VXNlcjExMTI0NzY= | User | false |
openclaw/openclaw | 4,080,853,854 | I_kwDOQb6kR87zPONe | 48,011 | https://github.com/openclaw/openclaw/issues/48011 | https://api.github.com/repos/openclaw/openclaw/issues/48011 | [Bug]: Non-workspace skills are inaccessible to the agent when sandbox is in workspaceAccess: "rw" mode | ### Bug type
Regression (worked before, now fails)
### Summary
Problem Description
When the sandbox is configured with workspaceAccess: "rw", the agent inside the sandbox cannot read the SKILL.md files of managed skills (~/.openclaw/skills/) or bundled/plugin skills (/usr/lib/node_modules/openclaw/skills/, extension... | closed | not_planned | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-16T07:58:49Z | 2026-04-28T04:39:03Z | 2026-04-26T03:23:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lijihua2017 | 33,613,942 | MDQ6VXNlcjMzNjEzOTQy | User | false |
openclaw/openclaw | 4,141,237,225 | I_kwDOQb6kR8721kPp | 54,983 | https://github.com/openclaw/openclaw/issues/54983 | https://api.github.com/repos/openclaw/openclaw/issues/54983 | feat: Add vLLM reasoning/thinking toggle support (enable_thinking parameter) | Add support for toggling reasoning/thinking on vLLM-compatible models (e.g., Nemotron with reasoning parser enabled).
Currently vLLM models have reasoning ON by default with no way to toggle. Claude models support /reasoning on|off, but vLLM doesn't have equivalent.
Suggested: Pass chat_template_kwargs: { enable_thin... | closed | completed | true | 2 | [] | [] | 2026-03-26T07:02:33Z | 2026-04-28T04:39:02Z | 2026-04-26T03:23:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | dennis-lynch | 59,386,026 | MDQ6VXNlcjU5Mzg2MDI2 | User | false |
openclaw/openclaw | 4,080,931,488 | I_kwDOQb6kR87zPhKg | 48,025 | https://github.com/openclaw/openclaw/issues/48025 | https://api.github.com/repos/openclaw/openclaw/issues/48025 | [Feature] Add file upload capability to feishu_drive tool | # [Feature] Add file upload capability to feishu_drive tool
## Problem
The current `feishu_drive` tool supports file management (list, info, create_folder, move, delete) but **lacks file upload capability**.
Feishu's standard upload API has a **20MB limit**. For larger files (model checkpoints, databases, archives)... | closed | not_planned | true | 1 | [] | [] | 2026-03-16T08:16:41Z | 2026-04-28T04:39:01Z | 2026-04-26T03:23:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Shawn1993 | 11,707,976 | MDQ6VXNlcjExNzA3OTc2 | User | false |
openclaw/openclaw | 4,080,932,996 | I_kwDOQb6kR87zPhiE | 48,026 | https://github.com/openclaw/openclaw/issues/48026 | https://api.github.com/repos/openclaw/openclaw/issues/48026 | [Feature]: Cron-on-additional-channels | ### Summary
Add new delivery.additionalTargets to Cron job, so the result of Cron jobs will be set to multiple channels.
### Problem to solve
Cron jobs could only deliver their output to a single channel/recipient; there was no built-in way to fan-out the same result to multiple destinations simultaneously. Additio... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-16T08:17:01Z | 2026-04-28T04:39:00Z | 2026-04-26T03:23:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | miniloong | 4,901,527 | MDQ6VXNlcjQ5MDE1Mjc= | User | false |
openclaw/openclaw | 4,081,067,489 | I_kwDOQb6kR87zQCXh | 48,055 | https://github.com/openclaw/openclaw/issues/48055 | https://api.github.com/repos/openclaw/openclaw/issues/48055 | Feature Request: Conversation history sidebar in Web Dashboard | ## Feature Request
### Description
The web dashboard (control UI) currently starts with a blank chat on each visit. There is no way to browse or resume previous conversations.
### Requested Feature
Add a **sidebar with conversation history** to the web dashboard, similar to ChatGPT's UI:
- List of past conversation s... | closed | completed | true | 1 | [] | [] | 2026-03-16T08:45:10Z | 2026-04-28T04:39:00Z | 2026-04-26T03:23:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | youkclaw01 | 268,506,811 | U_kgDOEAEWuw | User | false |
openclaw/openclaw | 4,081,502,829 | I_kwDOQb6kR87zRspt | 48,122 | https://github.com/openclaw/openclaw/issues/48122 | https://api.github.com/repos/openclaw/openclaw/issues/48122 | [Bug]: device identity required | ### Bug type
Regression (worked before, now fails)
### Summary
<img width="1696" height="948" alt="Image" src="https://github.com/user-attachments/assets/847f3823-327b-4adf-94bf-f4e7cb1137ad" />
### Steps to reproduce
using existing openclaw.json bind mount
### Expected behavior
Should show the dashboard and lo... | closed | completed | true | 6 | [
"bug",
"regression"
] | [] | 2026-03-16T10:07:34Z | 2026-04-28T04:38:59Z | 2026-04-26T03:23:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xxcbzxx | 3,828,765 | MDQ6VXNlcjM4Mjg3NjU= | User | false |
openclaw/openclaw | 4,081,830,092 | I_kwDOQb6kR87zS8jM | 48,157 | https://github.com/openclaw/openclaw/issues/48157 | https://api.github.com/repos/openclaw/openclaw/issues/48157 | [Bug]: Update error: npm error code EBUSY & npm error syscall rename | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Environment
OS: Windows 11
Node: v24.14.0
OpenClaw: 2026.3.8
Bug
When the openclaw installed via npm executes the "openclaw update" command, its own command-line process will occupy the file, resulting in an error: npm error code EBUSY & n... | closed | not_planned | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-16T11:12:54Z | 2026-04-28T04:38:58Z | 2026-04-26T03:23:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | onceyoungs | 33,124,872 | MDQ6VXNlcjMzMTI0ODcy | User | false |
openclaw/openclaw | 4,082,054,810 | I_kwDOQb6kR87zTzaa | 48,180 | https://github.com/openclaw/openclaw/issues/48180 | https://api.github.com/repos/openclaw/openclaw/issues/48180 | [Bug]: Slack Socket Mode — default account receives no inbound events in group channels | ### Bug type
Regression (worked before, now fails)
### Summary
When running multiple Slack accounts in Socket Mode, the default account's socket reports "connected" but never receives any inbound message events in group channels. Manual API writes (chat.postMessage) work fine. Other accounts (assistants) responding ... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-16T12:01:45Z | 2026-04-28T04:38:57Z | 2026-04-26T03:23:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | gbgeka | 121,305,411 | U_kgDOBzr5Qw | User | false |
openclaw/openclaw | 4,082,176,896 | I_kwDOQb6kR87zUROA | 48,196 | https://github.com/openclaw/openclaw/issues/48196 | https://api.github.com/repos/openclaw/openclaw/issues/48196 | Group chat 'do not use message tool' prompt conflicts with media sending | ## Problem
Currently, OpenClaw injects a hardcoded system prompt in group chats that tells the AI:
> "Your replies are automatically sent to this group chat. Do not use the message tool to send to this same group — just reply normally."
This is located in `src/auto-reply/reply/groups.ts:135` within the `buildGroupCh... | closed | not_planned | true | 3 | [] | [] | 2026-03-16T12:26:10Z | 2026-04-28T04:38:57Z | 2026-04-26T03:23:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | shuans | 249,932,760 | U_kgDODuWr2A | User | false |
openclaw/openclaw | 4,082,518,434 | I_kwDOQb6kR87zVkmi | 48,234 | https://github.com/openclaw/openclaw/issues/48234 | https://api.github.com/repos/openclaw/openclaw/issues/48234 | Bug: Image tool with MiniMax vision model not working |
## Environment
- OpenClaw version: 2026.3.13 (61d171a)
- Platform: macOS
- Model provider: MiniMax (minimax-direct)
## Configuration
```json
{
"tools": {
"media": {
"image": {
"enabled": true,
"models": [{"provider": "minimax-direct", "model": "MiniMax-M2.5"}]
}
}
},
"models"... | closed | completed | true | 5 | [] | [] | 2026-03-16T13:32:19Z | 2026-04-28T04:38:56Z | 2026-04-26T03:23:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Shillfood | 73,024,696 | MDQ6VXNlcjczMDI0Njk2 | User | false |
openclaw/openclaw | 4,082,898,919 | I_kwDOQb6kR87zXBfn | 48,281 | https://github.com/openclaw/openclaw/issues/48281 | https://api.github.com/repos/openclaw/openclaw/issues/48281 | Feature Request: Add Feishu Interactive Card Parsing Support | ## Feature Request
### Is your feature request related to a problem? Please describe.
Currently, the Feishu plugin can **send** interactive cards successfully, but cannot **read/parse** the content of received interactive cards. When a user sends an interactive card message, the bot only sees `[Interactive Card]` pla... | closed | not_planned | true | 3 | [] | [] | 2026-03-16T14:39:09Z | 2026-04-28T04:38:54Z | 2026-04-26T03:24:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rocky2046 | 95,515,952 | U_kgDOBbF1MA | User | false |
openclaw/openclaw | 4,083,145,906 | I_kwDOQb6kR87zX9yy | 48,308 | https://github.com/openclaw/openclaw/issues/48308 | https://api.github.com/repos/openclaw/openclaw/issues/48308 | [Bug]: DevClaw cannot detect Telegram topic without explicit `topicId` parameter | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When using DevClaw (https://github.com/laurentenhoor/devclaw) inside Telegram forum topics, OpenClaw does not pass the Telegram `message_thread_id` (topic ID) into DevClaw tool invocations. DevClaw can supports topic-aware routing via an o... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-16T15:23:12Z | 2026-04-28T04:38:54Z | 2026-04-26T03:24:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | netandreus | 313,477 | MDQ6VXNlcjMxMzQ3Nw== | User | false |
openclaw/openclaw | 4,083,416,774 | I_kwDOQb6kR87zY_7G | 48,340 | https://github.com/openclaw/openclaw/issues/48340 | https://api.github.com/repos/openclaw/openclaw/issues/48340 | [Feature Request] TUI and Feishu message sync | ## Feature Request
Currently when using Feishu as a channel:
- Messages from Feishu are received by the agent but don't appear in TUI
- Messages from TUI don't automatically sync to Feishu
Expected: Messages should sync both ways so conversation is visible in both interfaces.
---
Created by Jarvis (OpenClaw AI) | closed | not_planned | true | 1 | [] | [] | 2026-03-16T16:09:08Z | 2026-04-28T04:38:53Z | 2026-04-26T03:24:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 2246157476-web | 267,467,967 | U_kgDOD_E8vw | User | false |
openclaw/openclaw | 4,083,417,859 | I_kwDOQb6kR87zZAMD | 48,341 | https://github.com/openclaw/openclaw/issues/48341 | https://api.github.com/repos/openclaw/openclaw/issues/48341 | [Feature Request] TUI and Feishu message sync | ## Feature Request
Currently when using Feishu as a channel:
- Messages from Feishu are received by the agent but dont appear in TUI
- Messages from TUI dont automatically sync to Feishu
Expected: Messages should sync both ways so conversation is visible in both interfaces.
---
Created by Jarvis (OpenClaw AI) | closed | not_planned | true | 2 | [] | [] | 2026-03-16T16:09:18Z | 2026-04-28T04:38:52Z | 2026-04-26T03:24:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 2246157476-web | 267,467,967 | U_kgDOD_E8vw | User | false |
openclaw/openclaw | 4,329,908,441 | I_kwDOQb6kR88AAAABAhUo2Q | 71,920 | https://github.com/openclaw/openclaw/issues/71920 | https://api.github.com/repos/openclaw/openclaw/issues/71920 | [Bug]: Feishu TTS voice-note delivery — code path correct but audio not delivered at runtime (v2026.4.24) | ## Summary
TTS voice-note replies to Feishu are not delivered, despite the code path appearing correct and issue #56231 being closed as completed. The TTS engine generates the .opus/.ogg file successfully, but the file never reaches Feishu.
## Environment
- **OpenClaw:** v2026.4.24
- **Channel:** Feishu (websocket c... | closed | completed | true | 1 | [] | [] | 2026-04-26T03:15:22Z | 2026-04-28T04:38:51Z | 2026-04-26T03:28:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Bowerai | 17,424,365 | MDQ6VXNlcjE3NDI0MzY1 | User | false |
openclaw/openclaw | 4,083,486,603 | I_kwDOQb6kR87zZQ-L | 48,345 | https://github.com/openclaw/openclaw/issues/48345 | https://api.github.com/repos/openclaw/openclaw/issues/48345 | Cron: add skipMissedRuns option to prevent catch-up runs after gateway restart | ## Problem
When the gateway is down for an extended period (e.g., 35 hours) and restarts, the cron scheduler sees all overdue jobs with `nextRunAtMs` in the past and fires catch-up runs immediately. Then the same jobs fire again at their actual scheduled time, resulting in **duplicate deliveries**.
### Example (obser... | closed | not_planned | true | 3 | [] | [] | 2026-03-16T16:21:29Z | 2026-04-28T04:38:51Z | 2026-04-26T03:31:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | freeup86 | 198,971,857 | U_kgDOC9wR0Q | User | false |
openclaw/openclaw | 4,083,634,023 | I_kwDOQb6kR87zZ09n | 48,367 | https://github.com/openclaw/openclaw/issues/48367 | https://api.github.com/repos/openclaw/openclaw/issues/48367 | [Feature]: Allow self-hosted Control Servers for the Tailscale Gateway | ### Summary
Add first-class support for running OpenClaw Gateway on Tailnets managed by a custom Tailscale-compatible control server (for example, Headscale), including explicit config/CLI support and clear mode validation.
### Problem to solve
OpenClaw’s Tailscale Gateway flow is currently optimized for the defaul... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-16T16:51:54Z | 2026-04-28T04:38:49Z | 2026-04-26T03:31:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | dotWee | 8,060,356 | MDQ6VXNlcjgwNjAzNTY= | User | false |
openclaw/openclaw | 4,083,709,473 | I_kwDOQb6kR87zaHYh | 48,383 | https://github.com/openclaw/openclaw/issues/48383 | https://api.github.com/repos/openclaw/openclaw/issues/48383 | feat(hooks): Implement session:start lifecycle event | ## Summary
The `session:start` event is listed under "Future Events / Planned event types" in the hooks documentation but is not yet implemented. This would be a valuable addition for automating context injection at session start.
## Use Case
In multi-channel setups (Slack, Discord, Telegram), agents lose all conver... | closed | completed | true | 1 | [] | [] | 2026-03-16T17:07:08Z | 2026-04-28T04:38:48Z | 2026-04-26T03:31:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | floriansiebert | 22,081,639 | MDQ6VXNlcjIyMDgxNjM5 | User | false |
openclaw/openclaw | 4,083,727,231 | I_kwDOQb6kR87zaLt_ | 48,389 | https://github.com/openclaw/openclaw/issues/48389 | https://api.github.com/repos/openclaw/openclaw/issues/48389 | [Bug]: stale queued reply can replay after restart instead of being invalidated by runtime generation change | # GitHub-ready draft — post-restart stale reply replay
## Suggested title
[Bug]: stale queued reply can replay after restart instead of being invalidated by runtime generation change
## Summary
After an approved restart, OpenClaw can deliver an old queued reply into Microsoft Teams instead of limiting post-restart ... | closed | not_planned | true | 4 | [] | [] | 2026-03-16T17:10:35Z | 2026-04-28T04:38:47Z | 2026-04-26T03:31:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | garysu | 1,896,003 | MDQ6VXNlcjE4OTYwMDM= | User | false |
openclaw/openclaw | 4,084,205,299 | I_kwDOQb6kR87zcAbz | 48,441 | https://github.com/openclaw/openclaw/issues/48441 | https://api.github.com/repos/openclaw/openclaw/issues/48441 | Feature Request: Suppress internal subagent announce messages from TUI display | ## Problem
When subagents complete and announce results back to the main session, the full announce payload (including task description, result content, stats) appears in the TUI as visible messages. In multi-agent workflows with frequent subagent spawning, this floods the user's TUI and makes it impossible to have a c... | closed | completed | true | 2 | [] | [] | 2026-03-16T18:51:59Z | 2026-04-28T04:38:47Z | 2026-04-26T03:31:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Olis218 | 262,907,665 | U_kgDOD6unEQ | User | false |
openclaw/openclaw | 4,084,338,068 | I_kwDOQb6kR87zcg2U | 48,452 | https://github.com/openclaw/openclaw/issues/48452 | https://api.github.com/repos/openclaw/openclaw/issues/48452 | Feature Request: Path-based argument filtering in exec allowlist | ## Feature Request
### Summary
Add support for path/argument-based filtering in the exec allowlist, similar to Claude Code's `Bash(command:path*)` permission syntax.
### Current Behavior
The exec allowlist only matches on the **binary path**:
```json
{ "pattern": "/bin/cat" }
```
This allows or blocks a binary entire... | closed | not_planned | true | 2 | [] | [] | 2026-03-16T19:20:41Z | 2026-04-28T04:38:46Z | 2026-04-26T03:31:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | boardroom-ttrpg | 15,096,667 | MDQ6VXNlcjE1MDk2NjY3 | User | false |
openclaw/openclaw | 4,084,361,939 | I_kwDOQb6kR87zcmrT | 48,456 | https://github.com/openclaw/openclaw/issues/48456 | https://api.github.com/repos/openclaw/openclaw/issues/48456 | Support `agents.defaults.skills` for global skill inheritance | **Description:**
Currently, `agents.defaults` supports many shared settings (model, workspace, heartbeat, etc.), but `skills` must be specified per-agent in `agents.list[].skills`. This leads to:
1. **Configuration duplication** — Every agent needs to explicitly list common skills (e.g., `daily-ops-hybrid`, `memory-m... | closed | completed | true | 3 | [] | [] | 2026-03-16T19:26:38Z | 2026-04-28T04:38:45Z | 2026-04-26T03:31:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sunnydongbo | 249,433,142 | U_kgDODt4MNg | User | false |
openclaw/openclaw | 4,084,716,966 | I_kwDOQb6kR87zd9Wm | 48,484 | https://github.com/openclaw/openclaw/issues/48484 | https://api.github.com/repos/openclaw/openclaw/issues/48484 | [Feature]: Expose Anthropic rate limit headers in session_status | ### Summary
Expose Anthropic rate limit headers in session_status
### Problem to solve
Currently no way to check Anthropic rate limit usage from within OpenClaw. Proposal: surface anthropic-ratelimit-requests-remaining, requests-reset, tokens-remaining, tokens-reset headers in session_status / /status output. Data i... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-16T20:48:23Z | 2026-04-28T04:38:44Z | 2026-04-26T03:31:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | greatzambino | 39,806,381 | MDQ6VXNlcjM5ODA2Mzgx | User | false |
openclaw/openclaw | 4,084,779,034 | I_kwDOQb6kR87zeMga | 48,489 | https://github.com/openclaw/openclaw/issues/48489 | https://api.github.com/repos/openclaw/openclaw/issues/48489 | [Feature]: Surface internal tool calls in OpenResponses SSE stream | ### Summary
Emit internal tool call events (`function_call` and `function_call_output` output items) in the OpenResponses `/v1/responses` SSE streaming response, so clients can render tool inputs and results in their UI.
### Problem to solve
When building custom chat UIs on top of the OpenResponses endpoint (e.g. us... | closed | not_planned | true | 3 | [] | [] | 2026-03-16T21:03:00Z | 2026-04-28T04:38:43Z | 2026-04-26T03:31:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | simon-marcus | 16,433,553 | MDQ6VXNlcjE2NDMzNTUz | User | false |
openclaw/openclaw | 4,085,013,045 | I_kwDOQb6kR87zfFo1 | 48,513 | https://github.com/openclaw/openclaw/issues/48513 | https://api.github.com/repos/openclaw/openclaw/issues/48513 | [Feature]: Control UI 缺少模型配置可视化界面，新手容易配置错误 | ### Summary
## 问题描述
当前 OpenClaw 的模型配置需要通过手动编辑 `openclaw.json` 配置文件完成，Control UI 没有提供可视化的模型配置界面。
### Problem to solve
这导致以下问题：
1. **新手门槛高**：用户需要理解 JSON 结构和模型配置格式
2. **容易配错**：容易出现 `siliconflow/qwen3.5-plus` 这种跨提供商的错误组合
3. **缺乏验证**：保存前无法验证模型名称是否正确
4. **不直观**：无法在 UI 上直观看到已配置的提供商和可用模型列表
### Proposed solution
期望的改进
... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-16T22:03:26Z | 2026-04-28T04:38:42Z | 2026-04-26T03:32:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zhegcheg | 11,594,389 | MDQ6VXNlcjExNTk0Mzg5 | User | false |
openclaw/openclaw | 4,085,301,497 | I_kwDOQb6kR87zgMD5 | 48,552 | https://github.com/openclaw/openclaw/issues/48552 | https://api.github.com/repos/openclaw/openclaw/issues/48552 | [Bug]: Docker self-build from `main` broken — gateway fails to start after channel-to-extension migration | # [Bug]: Docker self-build from `main` broken — gateway fails to start after channel-to-extension migration
## Summary
Building the Docker image from the stock `Dockerfile` on current `main` produces a gateway that fails to start. The process launches but errors out during channel and plugin loading before ever reach... | closed | completed | true | 3 | [] | [] | 2026-03-16T23:20:18Z | 2026-04-28T04:38:42Z | 2026-04-26T03:32:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | gbballpack | 39,963,341 | MDQ6VXNlcjM5OTYzMzQx | User | false |
openclaw/openclaw | 4,085,390,179 | I_kwDOQb6kR87zghtj | 48,566 | https://github.com/openclaw/openclaw/issues/48566 | https://api.github.com/repos/openclaw/openclaw/issues/48566 | Docs/UX: openclaw_gateway device pairing requirement is undocumented — new Paperclip users hit silent run failures | ## Summary
When a **Paperclip** agent connects to an OpenClaw gateway for the first time via the `openclaw_gateway` adapter, the device must be approved in OpenClaw before any run can succeed. This requirement is **not documented anywhere** in Paperclip's onboarding flow or in OpenClaw's docs. Every new user hits this... | closed | not_planned | true | 2 | [] | [] | 2026-03-16T23:51:15Z | 2026-04-28T04:38:41Z | 2026-04-26T03:32:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | levineam | 1,668,898 | MDQ6VXNlcjE2Njg4OTg= | User | false |
openclaw/openclaw | 4,085,427,221 | I_kwDOQb6kR87zgqwV | 48,572 | https://github.com/openclaw/openclaw/issues/48572 | https://api.github.com/repos/openclaw/openclaw/issues/48572 | Feature Request: Per-agent runtime context isolation (hide agent list from individual agents) | ## Problem
When running multiple agents on a shared gateway (multi-tenant SaaS use case), each agent's system prompt includes the full agent list count and cron job details via the injected runtime line. This means Client A's bot can see that Client B, C, D exist and what models/schedules they use.
This is a **client... | closed | completed | true | 2 | [] | [] | 2026-03-17T00:04:54Z | 2026-04-28T04:38:40Z | 2026-04-26T03:32:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | advantageababot-bit | 264,686,855 | U_kgDOD8bNBw | User | false |
openclaw/openclaw | 4,085,494,985 | I_kwDOQb6kR87zg7TJ | 48,584 | https://github.com/openclaw/openclaw/issues/48584 | https://api.github.com/repos/openclaw/openclaw/issues/48584 | [Feature]: Reduce human-identifying system-injected inbound metadata in LLM prompt context | ## Summary
Reduce human-identifying **system-injected inbound metadata** that OpenClaw injects into LLM prompt context.
This work is intentionally limited to metadata added by OpenClaw itself. It does **not** rewrite user-authored message body content.
## Scope
In scope:
- trusted inbound metadata: `chat_id`
- user... | closed | not_planned | true | 2 | [] | [] | 2026-03-17T00:25:29Z | 2026-04-28T04:38:39Z | 2026-04-26T03:32:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kraocode | 216,472,336 | U_kgDODOcbEA | User | false |
openclaw/openclaw | 4,085,586,181 | I_kwDOQb6kR87zhRkF | 48,600 | https://github.com/openclaw/openclaw/issues/48600 | https://api.github.com/repos/openclaw/openclaw/issues/48600 | Feature Request: FTS5 full-text session search with LLM summarization for cross-session recall | ## Summary
Add full-text search (FTS5) across session transcripts, with LLM-powered summarization to surface relevant context from past conversations.
## Problem
Currently, cross-session recall in OpenClaw depends entirely on what was written to memory files. If something wasn't captured in `MEMORY.md` or daily note... | closed | completed | true | 1 | [] | [] | 2026-03-17T00:56:05Z | 2026-04-28T04:38:39Z | 2026-04-26T03:32:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Astraea-Sixth | 255,654,779 | U_kgDODzz7ew | User | false |
openclaw/openclaw | 4,085,606,003 | I_kwDOQb6kR87zhWZz | 48,603 | https://github.com/openclaw/openclaw/issues/48603 | https://api.github.com/repos/openclaw/openclaw/issues/48603 | [Feature]: API rate limit reached. Please try again later. | ### Summary
When token runs out for model, it says API rate limit reached. Please try again later.
For which model ?
<img width="313" height="55" alt="Image" src="https://github.com/user-attachments/assets/ce73ad2f-7bfd-41df-82fb-4235e6050a63" />
### Problem to solve
Let us the user know which model it is. Keep t... | closed | completed | true | 4 | [
"enhancement"
] | [] | 2026-03-17T01:03:44Z | 2026-04-28T04:38:38Z | 2026-04-26T03:32:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | gbpnkans | 1,450,505 | MDQ6VXNlcjE0NTA1MDU= | User | false |
openclaw/openclaw | 4,085,717,444 | I_kwDOQb6kR87zhxnE | 48,629 | https://github.com/openclaw/openclaw/issues/48629 | https://api.github.com/repos/openclaw/openclaw/issues/48629 | [Feature]: Keep OpenClaw “agent-light” via a minimal delegation-invariants layer (Confirm / Stop / Receipts + hooks) | ### Summary
Propose a minimal “delegation invariants” spec + stable hook points (Confirm / Stop-Takeover / Receipts) so OpenClaw can scale as shared infrastructure without pulling more policy/adapters into core.
### Problem to solve
As we shift from Human–App–Cloud to Human–Agent–Cloud, “in-app UX details” become cr... | closed | not_planned | true | 4 | [
"enhancement"
] | [] | 2026-03-17T01:43:09Z | 2026-04-28T04:38:37Z | 2026-04-26T03:32:37Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wicky-AntX | 62,903,905 | MDQ6VXNlcjYyOTAzOTA1 | User | false |
openclaw/openclaw | 4,085,941,072 | I_kwDOQb6kR87zioNQ | 48,687 | https://github.com/openclaw/openclaw/issues/48687 | https://api.github.com/repos/openclaw/openclaw/issues/48687 | healthcheck: security audit falsely flags intentional group-level allowFrom wildcard as critical | ## Problem
The daily `healthcheck:security-audit` cron reports a **NEW CRITICAL** alert when a Telegram group has `allowFrom: ["*"]` configured at the group level — even when the top-level `groupAllowFrom` is correctly restricted to specific user IDs.
## Steps to reproduce
Configure OpenClaw with the following struc... | closed | completed | true | 2 | [] | [] | 2026-03-17T03:09:04Z | 2026-04-28T04:38:35Z | 2026-04-26T03:32:47Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | dongzhenye | 5,765,843 | MDQ6VXNlcjU3NjU4NDM= | User | false |
openclaw/openclaw | 4,085,982,314 | I_kwDOQb6kR87ziyRq | 48,695 | https://github.com/openclaw/openclaw/issues/48695 | https://api.github.com/repos/openclaw/openclaw/issues/48695 | [Bug]: Browser automation with OpenClaw on VPS — unable to advance in dynamic web app (AVA ERP | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Context
I am running OpenClaw 2026.3.13 on a DigitalOcean VPS (Ubuntu 24.04, 4GB RAM, 2 vCPUs). My goal is to automate a business workflow inside a web ERP called AVA — specifically the "Confirm Order" flow, which involves:
1. Logging into ... | closed | not_planned | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-17T03:25:36Z | 2026-04-28T04:38:35Z | 2026-04-26T03:32:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | marcelomamc2-ctrl | 268,586,266 | U_kgDOEAJNGg | User | false |
openclaw/openclaw | 4,086,369,961 | I_kwDOQb6kR87zkQ6p | 48,766 | https://github.com/openclaw/openclaw/issues/48766 | https://api.github.com/repos/openclaw/openclaw/issues/48766 | [Bug]: OpenClaw gateway restart | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`openclaw gateway restart` always times out after 60s on Windows after upgrading from 2026.3.11 to 2026.3.13, even though the gateway starts successfully and is fully functional. The restart command's internal health probe is rejected with ... | closed | not_planned | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-17T05:49:16Z | 2026-04-28T04:38:34Z | 2026-04-26T03:34:34Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | DanTheRealMan | 34,142,986 | MDQ6VXNlcjM0MTQyOTg2 | User | false |
openclaw/openclaw | 4,086,482,368 | I_kwDOQb6kR87zksXA | 48,779 | https://github.com/openclaw/openclaw/issues/48779 | https://api.github.com/repos/openclaw/openclaw/issues/48779 | [Feature]: Add API call count to status/usage output | ### Summary
Add API call count display to session_status (`/status`) and response usage footer, showing the number of API requests made alongside token usage.
### Problem to solve
Users currently see token usage but cannot easily track how many API calls were made. This makes it difficult to:
- Understand API usage... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-17T06:21:51Z | 2026-04-28T04:38:33Z | 2026-04-26T03:34:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Qixingchen | 4,182,240 | MDQ6VXNlcjQxODIyNDA= | User | false |
openclaw/openclaw | 4,086,698,340 | I_kwDOQb6kR87zlhFk | 48,813 | https://github.com/openclaw/openclaw/issues/48813 | https://api.github.com/repos/openclaw/openclaw/issues/48813 | Feature request: per-bot credit monitoring URL + low-credit alerts | ## Problem
When a Pika bot runs out of credits, it silently stops responding to messages. The user has no idea why the bot went quiet — they just think it's broken. There's no way for bot owners (or users) to check remaining credits without SSH-ing into the gateway.
## Proposed Solution
1. **Credit/usage monitoring ... | closed | not_planned | true | 1 | [] | [] | 2026-03-17T07:20:54Z | 2026-04-28T04:38:32Z | 2026-04-26T03:34:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mlstar9 | 138,526,810 | U_kgDOCEHAWg | User | false |
openclaw/openclaw | 4,086,939,445 | I_kwDOQb6kR87zmb81 | 48,852 | https://github.com/openclaw/openclaw/issues/48852 | https://api.github.com/repos/openclaw/openclaw/issues/48852 | [Feature]: Remote browser Chrome Devtools MCP capability | ### Summary
The current implementation of Chrome DevTools MCP (Model Context Protocol) is tightly coupled with local browser environments via OpenClaw. For remote browser access, it relies solely on the traditional CDP (Chrome DevTools Protocol) URL approach. This limits the protocol's applicability and efficiency, as... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-17T08:19:28Z | 2026-04-28T04:38:31Z | 2026-04-26T03:34:46Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | SairoUltraman | 182,851,189 | U_kgDOCuYWdQ | User | false |
openclaw/openclaw | 4,087,026,226 | I_kwDOQb6kR87zmxIy | 48,862 | https://github.com/openclaw/openclaw/issues/48862 | https://api.github.com/repos/openclaw/openclaw/issues/48862 | [Feature]: Could openclaw update version including node updating ? | ### Summary
I need to upadte the node to the required version . That is a bit boring.....
Could you update the openclaw including the required software or plugin? Instead I update manually？ That will be very AI automatic
### Problem to solve
After I update my openclaw, it can not start again, and I get the logs ... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-17T08:37:59Z | 2026-04-28T04:38:31Z | 2026-04-26T03:34:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | violazhu | 35,122,367 | MDQ6VXNlcjM1MTIyMzY3 | User | false |
openclaw/openclaw | 4,087,489,366 | I_kwDOQb6kR87zoiNW | 48,913 | https://github.com/openclaw/openclaw/issues/48913 | https://api.github.com/repos/openclaw/openclaw/issues/48913 | [Feature]: Single-Model Retry Logic | ### Summary
Add retry logic for API overload errors (429/529) before failover
### Problem to solve
## Problem
When Anthropic (or other providers) return overload errors (HTTP 529, `overloaded_error` response), OpenClaw immediately fails over to the next configured profile/provider. For users with a single API key, ... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-17T10:06:15Z | 2026-04-28T04:38:30Z | 2026-04-26T03:34:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | joeldevelops | 19,559,133 | MDQ6VXNlcjE5NTU5MTMz | User | false |
openclaw/openclaw | 4,087,776,171 | I_kwDOQb6kR87zpoOr | 48,943 | https://github.com/openclaw/openclaw/issues/48943 | https://api.github.com/repos/openclaw/openclaw/issues/48943 | Telegram typingMode: 'never' not working - sendChatAction 401 errors persist | ## Bug Report
**OpenClaw Version:** Latest (installed via npm)
**Environment:** Ubuntu Linux, Node v24.13.0
**Channel:** Telegram
### Problem
The configuration for Telegram is not working. Despite setting this option, OpenClaw continues to send requests, resulting in persistent 401 errors in the logs.
### Configur... | closed | completed | true | 3 | [] | [] | 2026-03-17T10:58:08Z | 2026-04-28T04:38:29Z | 2026-04-26T03:35:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | PiiSmith | 18,052,390 | MDQ6VXNlcjE4MDUyMzkw | User | false |
openclaw/openclaw | 4,088,235,693 | I_kwDOQb6kR87zrYat | 48,992 | https://github.com/openclaw/openclaw/issues/48992 | https://api.github.com/repos/openclaw/openclaw/issues/48992 | Gateway LaunchAgent not updated on non-git installs | Problem: npm updates dont reinstall LaunchAgent. Entry point changed to index.js but LA pointed to entry.js. Fix: openclaw update should reinstall LA | closed | completed | true | 4 | [] | [] | 2026-03-17T12:28:06Z | 2026-04-28T04:38:29Z | 2026-04-26T03:35:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ion-ctrl | 266,816,281 | U_kgDOD-dLGQ | User | false |
openclaw/openclaw | 4,088,240,652 | I_kwDOQb6kR87zrZoM | 48,995 | https://github.com/openclaw/openclaw/issues/48995 | https://api.github.com/repos/openclaw/openclaw/issues/48995 | [Feature]: support cap thinking-event when websocket connect | ## Summary
- **Problem 1 (gateway never received thinking tokens):** `streamReasoning` was initialized as `reasoningMode === "stream" && typeof onReasoningStream === "function"`. Gateway WebSocket runs never supply `onReasoningStream`, so the flag was always `false` and `emitAgentEvent` for `stream: "thinking"` was ne... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-17T12:29:01Z | 2026-04-28T04:38:28Z | 2026-04-26T03:35:09Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | caesargattuso | 10,957,907 | MDQ6VXNlcjEwOTU3OTA3 | User | false |
openclaw/openclaw | 4,088,261,806 | I_kwDOQb6kR87zreyu | 48,996 | https://github.com/openclaw/openclaw/issues/48996 | https://api.github.com/repos/openclaw/openclaw/issues/48996 | feat(webchat): 支持 mermaid 渲染 | ## 描述
在 webchat 对话框中支持 mermaid 图表渲染，方便用户展示流程图、时序图等。
## 期望行为
用户在对话中输入 mermaid 代码块时，能够自动渲染为图表。
## 参考
- 现有 mermaid 相关 Issue: #7165, #15157
- 文档中已有 mermaid 使用经验
## 可能的实现方式
- 使用 mermaid.js 在前端渲染
- 检测消息中的 ```mermaid 代码块
## AI-assisted | closed | not_planned | true | 1 | [] | [] | 2026-03-17T12:32:59Z | 2026-04-28T04:38:27Z | 2026-04-26T03:35:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jiazhizhuo | 15,940,087 | MDQ6VXNlcjE1OTQwMDg3 | User | false |
openclaw/openclaw | 4,088,549,393 | I_kwDOQb6kR87zslAR | 49,031 | https://github.com/openclaw/openclaw/issues/49031 | https://api.github.com/repos/openclaw/openclaw/issues/49031 | [Bug]: macOS companion app node missing system.run.prepare — exec completely broken on local mode | ### Bug type
Regression (worked before, now fails)
### Summary
On macOS local mode, the companion app's built-in node does not advertise `system.run.prepare`, causing **all exec operations to fail**. Root cause: `NodeServiceManager.swift` calls obsolete `openclaw service node start` (removed in CLI 2026.3.x), node s... | closed | not_planned | true | 3 | [] | [] | 2026-03-17T13:25:32Z | 2026-04-28T04:38:26Z | 2026-04-26T03:35:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexcz-a11y | 202,499,116 | U_kgDODBHkLA | User | false |
openclaw/openclaw | 4,088,880,440 | I_kwDOQb6kR87zt104 | 49,057 | https://github.com/openclaw/openclaw/issues/49057 | https://api.github.com/repos/openclaw/openclaw/issues/49057 | [Bug]: WhatsApp Web listener flaps / logs out on 2026.3.13 causing cron + message delivery failures (“No active WhatsApp Web listener”) | ### Bug type
Regression (worked before, now fails)
### Summary
After updating OpenClaw from 2026.3.8 → 2026.3.13, WhatsApp delivery became unreliable and then effectively failed consistently. Crons that send WhatsApp messages fail with:
No active WhatsApp Web listener (account: default)
Rolling back to 2026.3.8 rest... | closed | not_planned | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-17T14:19:02Z | 2026-04-28T04:38:25Z | 2026-04-26T03:35:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | NeilAdamz | 221,362,818 | U_kgDODTG6gg | User | false |
openclaw/openclaw | 4,329,938,221 | I_kwDOQb6kR88AAAABAhWdLQ | 71,928 | https://github.com/openclaw/openclaw/issues/71928 | https://api.github.com/repos/openclaw/openclaw/issues/71928 | [Show and tell] ClawClip: Session replay + version comparison (fills gaps the Usage Dashboard doesn't cover) | Hey OpenClaw team and community 👋
OpenClaw's new Usage Dashboard (v2026.4.24) is great for seeing what you spent. But I kept running into questions it couldn't answer:
- **Why** did this session cost $5 when similar ones cost $0.50?
- **Which step** triggered the retry loop that burned 3x tokens?
- **Did my optimiza... | closed | not_planned | true | 1 | [] | [] | 2026-04-26T03:35:32Z | 2026-04-29T04:37:22Z | 2026-04-26T10:44:29Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Ylsssq926 | 51,729,212 | MDQ6VXNlcjUxNzI5MjEy | User | false |
openclaw/openclaw | 4,089,062,009 | I_kwDOQb6kR87zuiJ5 | 49,073 | https://github.com/openclaw/openclaw/issues/49073 | https://api.github.com/repos/openclaw/openclaw/issues/49073 | [Feature]: Telegram thread/topic support for sub-agent spawning | ### Summary
When spawning sub-agents with mode="session", the spawn tool accepts
thread=true but the Telegram channel plugin does not implement it.
Even with a Supergroup with Forum/Topics enabled, the gateway returns
an error saying thread support is unavailable.
Expected: OpenClaw creates a new Telegram topic pe... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-17T14:48:03Z | 2026-04-28T04:38:24Z | 2026-04-26T03:35:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | motiop | 9,628,240 | MDQ6VXNlcjk2MjgyNDA= | User | false |
openclaw/openclaw | 4,089,403,983 | I_kwDOQb6kR87zv1pP | 49,108 | https://github.com/openclaw/openclaw/issues/49108 | https://api.github.com/repos/openclaw/openclaw/issues/49108 | [Memory] Ingestion pipeline improvements: markdown-aware chunking, classifier fixes, entity extraction | ## Context
While working with the smart-memory skill (v2.5.0 from ClawHub), I ran into several ingestion quality issues that resulted in broken/unusable memories. I've implemented fixes locally and wanted to share the findings and approach since they're directly relevant to the Memory MVP work in #42651 and #42648.
#... | closed | not_planned | true | 1 | [] | [] | 2026-03-17T15:43:37Z | 2026-04-28T04:38:23Z | 2026-04-26T03:35:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lakerfan901 | 145,699,626 | U_kgDOCK8zKg | User | false |
openclaw/openclaw | 4,089,506,930 | I_kwDOQb6kR87zwOxy | 49,115 | https://github.com/openclaw/openclaw/issues/49115 | https://api.github.com/repos/openclaw/openclaw/issues/49115 | Request: delete PR #49029 and associated comments | Hi maintainers — I'm the author of PR #49029 (now closed). The PR accidentally included internal configuration that should not have been submitted publicly. I've already closed the PR and removed all commits from the fork branch.
I'm requesting:
1. Deletion of PR #49029 (or at minimum the bot review comment on it, com... | closed | not_planned | true | 5 | [] | [] | 2026-03-17T16:00:27Z | 2026-04-28T04:38:22Z | 2026-04-26T03:35:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kaxo-io | 206,752,748 | U_kgDODFLL7A | User | false |
openclaw/openclaw | 4,090,133,519 | I_kwDOQb6kR87zynwP | 49,170 | https://github.com/openclaw/openclaw/issues/49170 | https://api.github.com/repos/openclaw/openclaw/issues/49170 | [Bug]: Daily session reset wipes actively used sessions at the daily boundary | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When `session.reset.mode: "daily"` is configured (the default), any session whose `lastUpdate` is before the daily boundary (`atHour`) is treated as stale even if the session is actively in use right up to that moment.
This means a user ch... | closed | not_planned | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-17T17:44:37Z | 2026-04-28T04:38:22Z | 2026-04-26T03:35:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pastor0711 | 183,016,548 | U_kgDOCuicZA | User | false |
openclaw/openclaw | 4,090,293,663 | I_kwDOQb6kR87zzO2f | 49,180 | https://github.com/openclaw/openclaw/issues/49180 | https://api.github.com/repos/openclaw/openclaw/issues/49180 | CLI status/gateway probe shows false 'unreachable (missing scope: operator.read)' in v2026.3.13 | **OpenClaw version:** 2026.3.13
**Issue:**
- `openclaw status` shows false "unreachable (missing scope: operator.read)"
- `openclaw gateway probe` shows "Reachable: no"
- But gateway is healthy - HTTP health check returns {"ok":true,"status":"live"}
- Gateway logs show successful RPC calls: `✓ channels.status`
- Othe... | closed | not_planned | true | 3 | [] | [] | 2026-03-17T18:10:30Z | 2026-04-28T04:38:21Z | 2026-04-26T03:35:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | absolut-fx | 244,205,096 | U_kgDODo5GKA | User | false |
openclaw/openclaw | 4,090,422,768 | I_kwDOQb6kR87zzuXw | 49,187 | https://github.com/openclaw/openclaw/issues/49187 | https://api.github.com/repos/openclaw/openclaw/issues/49187 | [Bug]: Bug: schtasks unavailable error on Windows | ### Bug type
Regression (worked before, now fails)
### Summary
### Bug: schtasks unavailable error on Windows
**Description:**
- Command `schtasks` returns data correctly when run directly
- But OpenClaw's `openclaw status` shows "schtasks unavailable: unknown error"
- Gateway Service detection fails even though sc... | closed | not_planned | true | 7 | [
"bug",
"regression"
] | [] | 2026-03-17T18:35:38Z | 2026-04-28T04:38:20Z | 2026-04-26T03:35:54Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Ivan820203 | 268,997,425 | U_kgDOEAiTMQ | User | false |
openclaw/openclaw | 4,090,640,787 | I_kwDOQb6kR87z0jmT | 49,202 | https://github.com/openclaw/openclaw/issues/49202 | https://api.github.com/repos/openclaw/openclaw/issues/49202 | [Bug]: Open WebUI-backed UUID chats can split into separate <uuid> and agent:main:<uuid> session records | **Bug type**
Behavior bug (incorrect output/state without crash)
**Summary**
Open WebUI-backed chats that use an opaque bare UUID as the session key can split into two separate OpenClaw session records:
- `<uuid>`
- `agent:main:<uuid>`
When that happens, one logical chat diverges between Open WebUI and OpenClaw Contr... | closed | not_planned | true | 12 | [] | [] | 2026-03-17T19:15:34Z | 2026-04-28T04:38:20Z | 2026-04-26T03:35:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | SHAREN | 6,128,858 | MDQ6VXNlcjYxMjg4NTg= | User | false |
openclaw/openclaw | 4,183,616,844 | I_kwDOQb6kR875XO1M | 58,680 | https://github.com/openclaw/openclaw/issues/58680 | https://api.github.com/repos/openclaw/openclaw/issues/58680 | Feature: Add clearThinking compat option for Z.AI Preserved Thinking | ## Problem
Z.AI's GLM-5/5.1 models support **Preserved Thinking** (`clear_thinking: false`), which retains `reasoning_content` from previous turns and feeds it back to the model. This improves multi-turn reasoning continuity and enables prompt caching on reasoning tokens.
Currently, OpenClaw's `openai-completions` ada... | closed | completed | true | 1 | [] | [] | 2026-04-01T02:31:45Z | 2026-04-28T04:38:18Z | 2026-04-26T03:37:01Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xuanmingguo | 258,405,939 | U_kgDOD2b2Mw | User | false |
openclaw/openclaw | 4,091,138,699 | I_kwDOQb6kR87z2dKL | 49,241 | https://github.com/openclaw/openclaw/issues/49241 | https://api.github.com/repos/openclaw/openclaw/issues/49241 | [Feature]: existing-session (Chrome MCP) should support targeting a specific Chrome profile | ### Problem
When using `profile: "user"` with `driver: "existing-session"`, OpenClaw connects to Chrome via Chrome DevTools MCP auto-connect. However, when the user has **multiple Chrome profile windows open** (e.g., personal + work), the MCP connection attaches to whichever profile window it discovers first — there i... | closed | completed | true | 2 | [] | [] | 2026-03-17T20:53:02Z | 2026-04-28T04:38:17Z | 2026-04-26T03:37:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jaablesa | 226,779,419 | U_kgDODYRhGw | User | false |
openclaw/openclaw | 4,091,445,335 | I_kwDOQb6kR87z3oBX | 49,268 | https://github.com/openclaw/openclaw/issues/49268 | https://api.github.com/repos/openclaw/openclaw/issues/49268 | TUI: Add optional message timestamps | It would be useful to have timestamps displayed next to messages in the TUI chat log. Currently there's no way to see when a message was sent/received without checking logs.
## Proposed behavior
- An option (config key or `/settings` toggle) to show per-message timestamps in the chat log
- Ideally respects the existin... | closed | not_planned | true | 1 | [] | [] | 2026-03-17T22:06:55Z | 2026-04-28T04:38:16Z | 2026-04-26T03:37:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tanukihat | 253,170,146 | U_kgDODxcR4g | User | false |
openclaw/openclaw | 4,091,451,487 | I_kwDOQb6kR87z3phf | 49,272 | https://github.com/openclaw/openclaw/issues/49272 | https://api.github.com/repos/openclaw/openclaw/issues/49272 | OpenClaw slow ess | ### Bug type
Regression (worked before, now fails)
### Summary
OpenClaw gets very slow and then hangs
### Steps to reproduce
Here's a draft:
───
Title: Compaction safeguard incorrectly cancels on sessions with prior successful compaction
Version: OpenClaw 2026.3.8 (3caab92)
Describe the bug
After a session ha... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-17T22:08:31Z | 2026-04-28T04:38:16Z | 2026-04-26T03:37:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | BRLaw777 | 256,831,554 | U_kgDOD07wQg | User | false |
openclaw/openclaw | 4,091,600,506 | I_kwDOQb6kR87z4N56 | 49,287 | https://github.com/openclaw/openclaw/issues/49287 | https://api.github.com/repos/openclaw/openclaw/issues/49287 | Feature request: keep previous chat visibly accessible when starting a new WebChat session | Hi, I’d like to suggest a small UX improvement for starting a new session in WebChat.
Right now, using `/new` or **New Session** feels like switching the current page to a brand new session. The previous session may still exist at the backend level, so this is not a report about session data being lost. The problem is... | closed | not_planned | true | 2 | [] | [] | 2026-03-17T22:48:46Z | 2026-04-28T04:38:15Z | 2026-04-26T03:37:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | PYXXXX | 85,445,850 | MDQ6VXNlcjg1NDQ1ODUw | User | false |
openclaw/openclaw | 4,092,024,287 | I_kwDOQb6kR87z51Xf | 49,326 | https://github.com/openclaw/openclaw/issues/49326 | https://api.github.com/repos/openclaw/openclaw/issues/49326 | Feature Request: Batch API routing for async/cron agent tasks | ## Summary
Support routing agent messages through Anthropic's Message Batches API (`/v1/messages/batches`) for non-real-time workloads.
## Motivation
Anthropic's Batch API offers a **50% discount** on all token costs, and it stacks with prompt caching (up to 95% savings on cached input). For agents running scheduled/c... | closed | not_planned | true | 1 | [] | [] | 2026-03-18T00:56:00Z | 2026-04-28T04:38:14Z | 2026-04-26T03:37:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Orionation | 260,579,663 | U_kgDOD4ghTw | User | false |
openclaw/openclaw | 4,092,280,366 | I_kwDOQb6kR87z6z4u | 49,370 | https://github.com/openclaw/openclaw/issues/49370 | https://api.github.com/repos/openclaw/openclaw/issues/49370 | Feature: pre-run script hook/gate for heartbeat and cron agent triggers | ## Problem
Currently, there is no first-class pre-hook/gate mechanism before heartbeat or cron triggers enqueue an agent run. This means:
- Agent runs may be triggered unnecessarily when preconditions aren't met
- Users must build fragile wrappers around triggers to handle gating logic
- No standardized way to valida... | closed | not_planned | true | 6 | [] | [] | 2026-03-18T02:20:12Z | 2026-04-28T04:38:13Z | 2026-04-26T03:38:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | clawmander | 261,854,224 | U_kgDOD5uUEA | User | false |
openclaw/openclaw | 4,092,284,816 | I_kwDOQb6kR87z60-Q | 49,373 | https://github.com/openclaw/openclaw/issues/49373 | https://api.github.com/repos/openclaw/openclaw/issues/49373 | Discord thread-bound subagent sessions do not exclusively own the thread across multiple bots | ## Summary
When `sessions_spawn({ agentId: "soren", thread: true, mode: "session" })` creates a Discord thread-bound subagent session, follow-up messages in that thread can still be picked up by another Discord bot's normal channel routing instead of continuing to route to the bound subagent session.
This results in ... | closed | not_planned | true | 2 | [] | [] | 2026-03-18T02:21:29Z | 2026-04-28T04:38:12Z | 2026-04-26T03:38:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rendrag-git | 253,747,599 | U_kgDODx_hjw | User | false |
openclaw/openclaw | 4,092,696,933 | I_kwDOQb6kR87z8Zll | 49,469 | https://github.com/openclaw/openclaw/issues/49469 | https://api.github.com/repos/openclaw/openclaw/issues/49469 | 飞书官方插件加载失败 - feishu-openclaw-plugin error 状态 | ## 问题描述
飞书官方插件 `@larksuiteoapi/feishu-openclaw-plugin` 无法加载，状态显示为 `error`。
## 环境信息
- **OpenClaw 版本**: 2026.3.13 (61d171a)
- **插件版本**: feishu-openclaw-plugin 2026.3.8
- **操作系统**: Linux 5.15.0-144-generic (x64)
- **Node 版本**: v24.14.0
## 错误现象
```
│ @larksuiteoapi/feishu-openclaw-plugin │ error │
```
配置警告：
```
plugi... | closed | not_planned | true | 2 | [] | [] | 2026-03-18T04:49:55Z | 2026-04-28T04:38:11Z | 2026-04-26T03:38:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | njambition111-ops | 267,602,673 | U_kgDOD_NK8Q | User | false |
openclaw/openclaw | 4,092,764,699 | I_kwDOQb6kR87z8qIb | 49,484 | https://github.com/openclaw/openclaw/issues/49484 | https://api.github.com/repos/openclaw/openclaw/issues/49484 | [Feature]: 飞书文档权限问题：docx:document:write_only 不兼容 | ### Summary
飞书文档权限问题：docx:document:write_only 不兼容
### Problem to solve
Issue 内容：
## 问题描述
飞书开放平台现已将文档编辑权限从 `docx:document:write` 变更为 `docx:document:write_only`。
当前 OpenClaw 飞书文档插件（feishu-doc）只支持 `docx:document:write` 权限，导致用户即使添加了 `docx:document:write_only` 权限，也无法通过插件写入/编辑飞书文档。
## 复现步骤
1. 在飞书开放平台为应用添加 `docx:document... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-18T05:08:47Z | 2026-04-28T04:38:11Z | 2026-04-26T03:38:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Jwarm | 105,092,347 | U_kgDOBkOU-w | User | false |
openclaw/openclaw | 4,092,802,824 | I_kwDOQb6kR87z8zcI | 49,494 | https://github.com/openclaw/openclaw/issues/49494 | https://api.github.com/repos/openclaw/openclaw/issues/49494 | feat: include usage/cost metadata in agent.wait terminal response | ## Summary
The `agent.wait` WebSocket response currently returns only `{ status, startedAt, endedAt, error }`. External orchestrators like [Paperclip](https://github.com/paperclipai/paperclip) need per-run token usage and cost data to enforce budgets and track spend — but this data never reaches the wire despite being... | closed | not_planned | true | 3 | [] | [] | 2026-03-18T05:20:02Z | 2026-04-28T04:38:10Z | 2026-04-26T03:38:22Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | shivasymbl | 216,878,638 | U_kgDODO1OLg | User | false |
openclaw/openclaw | 4,092,814,225 | I_kwDOQb6kR87z82OR | 49,497 | https://github.com/openclaw/openclaw/issues/49497 | https://api.github.com/repos/openclaw/openclaw/issues/49497 | [Guide] SiliconFlow provider for Chinese users — DeepSeek + Qwen smart routing proxy | ### Summary
A community deployment guide and proxy layer for Chinese users to run OpenClaw with SiliconFlow's API (DeepSeek + Qwen vision models) via smart routing.
### Problem to solve
Chinese users cannot directly access Anthropic/OpenAI APIs. SiliconFlow provides affordable DeepSeek and Qwen models, but OpenClaw'... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-18T05:23:39Z | 2026-04-28T04:38:09Z | 2026-04-26T03:38:27Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Syysean | 232,002,315 | U_kgDODdQTCw | User | false |
openclaw/openclaw | 4,092,954,690 | I_kwDOQb6kR87z9YhC | 49,518 | https://github.com/openclaw/openclaw/issues/49518 | https://api.github.com/repos/openclaw/openclaw/issues/49518 | control-ui/webchat image attachments become invisible in normal chat when the active model is text-only (no auto-fallback to imageModel, no user-facing warning) | ### Summary
In OpenClaw control-ui / webchat, users can upload or paste an image, see it as an attachment in the composer flow, send it successfully, and still have the assistant reply as if no image was provided.
After tracing the code path, this does not look like a simple frontend "missing upload support" problem.... | closed | completed | true | 3 | [] | [] | 2026-03-18T06:01:20Z | 2026-04-28T04:38:08Z | 2026-04-26T03:38:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wangyangwjy | 165,866,953 | U_kgDOCeLtyQ | User | false |
openclaw/openclaw | 4,094,046,066 | I_kwDOQb6kR870Bi9y | 49,719 | https://github.com/openclaw/openclaw/issues/49719 | https://api.github.com/repos/openclaw/openclaw/issues/49719 | openclaw nodes list shows Paired: 0 while node is actually connected and functional | ### Description
`openclaw nodes list` consistently reports `Pending: 0 · Paired: 0` even when a node is actively connected and fully functional via `openclaw nodes invoke`.
### Evidence
1. **`nodes invoke` works perfectly:**
```
$ openclaw nodes invoke --node MacMini-Node --command system.run --params '{"command":["... | closed | not_planned | true | 4 | [] | [] | 2026-03-18T10:00:59Z | 2026-04-28T04:38:08Z | 2026-04-26T03:38:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ZenoRewn | 16,939,151 | MDQ6VXNlcjE2OTM5MTUx | User | false |
openclaw/openclaw | 4,092,332,741 | I_kwDOQb6kR87z7ArF | 49,382 | https://github.com/openclaw/openclaw/issues/49382 | https://api.github.com/repos/openclaw/openclaw/issues/49382 | Allow custom agent name display in webchat control UI | ### Summary
The webchat control UI currently displays a hardcoded "assistant" label instead of the configured agent name (e.g., "小红 🌸") defined in openclaw.json under agents.list[].name. This feature request asks for dynamic display of the configured agent name or a custom label configuration option in webchat.
### ... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-18T02:39:02Z | 2026-04-28T04:38:07Z | 2026-04-26T03:38:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | liudashu-cd | 96,483,005 | U_kgDOBcA2vQ | User | false |
openclaw/openclaw | 4,094,337,910 | I_kwDOQb6kR870CqN2 | 49,746 | https://github.com/openclaw/openclaw/issues/49746 | https://api.github.com/repos/openclaw/openclaw/issues/49746 | Plugin hook contexts produce inconsistent conversationId and missing accountId across phases | ## Summary
Plugin hook contexts (`message_received`, `inbound_claim`, `before_agent_start`, `after_agent_complete`) produce different `conversationId` values for the same message and omit `accountId` from agent hook contexts. This makes it impossible for plugins to build consistent channel-scoped state across hook pha... | closed | not_planned | true | 2 | [] | [] | 2026-03-18T10:58:26Z | 2026-04-28T04:38:06Z | 2026-04-26T03:38:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | freerk | 1,837,768 | MDQ6VXNlcjE4Mzc3Njg= | User | false |
openclaw/openclaw | 4,094,339,010 | I_kwDOQb6kR870CqfC | 49,748 | https://github.com/openclaw/openclaw/issues/49748 | https://api.github.com/repos/openclaw/openclaw/issues/49748 | Broadcast inbound_claim hook for non-plugin-bound messages | ## Summary
The `inbound_claim` plugin hook only fires for plugin-bound conversations (via `runInboundClaimForPluginOutcome`). Plugins that register global `api.on("inbound_claim", ...)` handlers are never invoked for regular channel messages.
## Problem
A plugin that needs to intercept inbound messages before agent ... | closed | not_planned | true | 2 | [] | [] | 2026-03-18T10:58:41Z | 2026-04-28T04:38:05Z | 2026-04-26T03:38:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | freerk | 1,837,768 | MDQ6VXNlcjE4Mzc3Njg= | User | false |
openclaw/openclaw | 4,094,414,218 | I_kwDOQb6kR870C82K | 49,758 | https://github.com/openclaw/openclaw/issues/49758 | https://api.github.com/repos/openclaw/openclaw/issues/49758 | Bug: `status` / `gateway probe` / `health --json` misreport local gateway + Telegram state on 2026.3.13 | # Bug: `status` / `gateway probe` / `health --json` misreport local gateway + Telegram state on 2026.3.13
## Summary
On a local loopback gateway setup running OpenClaw `2026.3.13`, several built-in CLI diagnostics misreport gateway and Telegram health even while the system is operational.
## Environment
- OpenClaw ve... | closed | not_planned | true | 2 | [] | [] | 2026-03-18T11:13:06Z | 2026-04-28T04:38:05Z | 2026-04-26T03:38:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | johanfourie74 | 221,008,788 | U_kgDODSxTlA | User | false |
openclaw/openclaw | 4,094,429,861 | I_kwDOQb6kR870DAql | 49,761 | https://github.com/openclaw/openclaw/issues/49761 | https://api.github.com/repos/openclaw/openclaw/issues/49761 | [Bug] @所有人 (@all) triggers all bots in Feishu group chat | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When a user sends `@所有人` (@all) in a Feishu group chat, all bot accounts in the group respond simultaneously, even when `requireMention: true` is set.
### Steps to reproduce
1. Configure multiple Feishu bot accounts in openclaw.json
2. Ad... | closed | not_planned | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-18T11:16:07Z | 2026-04-28T04:38:04Z | 2026-04-26T03:38:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | huangjianluoyang-dotcom | 264,199,277 | U_kgDOD79cbQ | User | false |
openclaw/openclaw | 4,094,599,514 | I_kwDOQb6kR870DqFa | 49,772 | https://github.com/openclaw/openclaw/issues/49772 | https://api.github.com/repos/openclaw/openclaw/issues/49772 | Feature Request: Agent-to-Agent chat in groups | ## Use Case
Two Feishu bots (corresponding to two different agents) are added to the same Feishu group. Currently, when one bot is @mentioned, it responds, but there's no way for the two bots to chat with each other autonomously.
## Expected Behavior
- When one agent is @mentioned in a group, it could trigger the othe... | closed | completed | true | 2 | [] | [] | 2026-03-18T11:46:41Z | 2026-04-28T04:38:03Z | 2026-04-26T03:38:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | huixuedai-arch | 224,418,747 | U_kgDODWBbuw | User | false |
openclaw/openclaw | 4,094,717,502 | I_kwDOQb6kR870EG4- | 49,781 | https://github.com/openclaw/openclaw/issues/49781 | https://api.github.com/repos/openclaw/openclaw/issues/49781 | [Feature]: Add Dinference as provider | ### Summary
Implement support for using DInference.com as a model provider directly.
DInference provides diferents open source models like GLM-5, GLM-4.7 and GPT-OSS-120B and it's compatible with OpenAI API standards.
### Problem to solve
OpenClaw has no support for DInference as a model provider, preventing users f... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-18T12:08:46Z | 2026-04-28T04:38:03Z | 2026-04-26T03:39:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | PedroACosta | 11,423,360 | MDQ6VXNlcjExNDIzMzYw | User | false |
openclaw/openclaw | 4,329,943,684 | I_kwDOQb6kR88AAAABAhWyhA | 71,930 | https://github.com/openclaw/openclaw/issues/71930 | https://api.github.com/repos/openclaw/openclaw/issues/71930 | Mattermost plugin drops post_edited events — @mentions added via edit do not trigger agent wake | ### Summary
The Mattermost WebSocket plugin subscribes to `posted` events only. `post_edited` events are discarded before mention detection runs, so an @mention added by editing a message never wakes the tagged agent. The drop is silent — no error, no log entry.
### Reproduction (confirmed 2026-04-25)
1. Post a mess... | open | null | false | 1 | [] | [] | 2026-04-26T03:39:11Z | 2026-05-03T03:37:26Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | robbyproc87 | 109,332,083 | U_kgDOBoRGcw | User | false |
openclaw/openclaw | 4,329,943,997 | I_kwDOQb6kR88AAAABAhWzvQ | 71,931 | https://github.com/openclaw/openclaw/issues/71931 | https://api.github.com/repos/openclaw/openclaw/issues/71931 | WebChat 对话名称在网关更新后丢失/变化，建议支持固定会话名称 | ### 问题描述
每次 OpenClaw 网关更新（如 2026.4.22 → 2026.4.24）后，WebChat 左侧对话列表中的对话名称会被重新生成，导致：
1. 原本有意义的对话标题变成乱码或无意义的字符
2. 飞书端显示的名称也同步变成乱码
3. 用户无法识别和管理历史对话
### 复现步骤
1. 在 WebChat 进行若干对话，系统自动生成对话标题（如"升级 openclaw 到 4.24"）
2. 执行 升级网关
3. 重启网关后刷新 WebChat 页面
4. 观察左侧对话列表，名称已变化
### 期望行为
- 支持用户手动固定/编辑对话名称
- 或：对话名称一旦生成后持久化，不因网关重启/更新而改变
- ... | closed | completed | true | 1 | [] | [] | 2026-04-26T03:39:25Z | 2026-04-29T04:36:19Z | 2026-04-26T12:08:03Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | ygc3817922006-sketch | 266,983,010 | U_kgDOD-nWYg | User | false |
openclaw/openclaw | 4,079,460,122 | I_kwDOQb6kR87zJ58a | 47,691 | https://github.com/openclaw/openclaw/issues/47691 | https://api.github.com/repos/openclaw/openclaw/issues/47691 | feat: Enable multi-bot group chat with mention-based routing (agents can respond to other agents' mentions) | ## Problem
Currently, when multiple OpenClaw bots participate in the same Telegram group, they **cannot see each other's messages** because the gateway routes incoming messages **only to the agent associated with the sending account**.
Example:
- Vision sends a message → gateway triggers only Vision's agent
- Skynet ... | closed | not_planned | true | 1 | [] | [] | 2026-03-15T23:35:33Z | 2026-04-28T04:38:02Z | 2026-04-26T03:40:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lunaviva211-sketch | 262,210,777 | U_kgDOD6EE2Q | User | false |
openclaw/openclaw | 4,095,507,177 | I_kwDOQb6kR870HHrp | 49,843 | https://github.com/openclaw/openclaw/issues/49843 | https://api.github.com/repos/openclaw/openclaw/issues/49843 | Control UI 显示重复的 Telegram 频道条目 | ## 问题描述
在 OpenClaw Control UI 的会话/频道选择器中，Telegram 频道显示为重复条目。
## 复现步骤
1. 打开 OpenClaw Control UI (http://localhost:1***9 或对应的地址)
2. 查看左侧的会话/频道选择器
3. 发现 Telegram 相关的频道条目重复出现
## 实际表现
从截图可以看到：
- `telegram:group:@st****ou2004` 出现 2 次
- `telegram:group:@st****ou` 出现 2 次
- `telegram:direct:18****39` 出现 2 次
- `telegram:sla... | closed | completed | true | 3 | [] | [] | 2026-03-18T14:26:42Z | 2026-04-28T04:30:27Z | 2026-04-26T03:40:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | stevenliou | 9,360,792 | MDQ6VXNlcjkzNjA3OTI= | User | false |
openclaw/openclaw | 4,096,101,597 | I_kwDOQb6kR870JYzd | 49,899 | https://github.com/openclaw/openclaw/issues/49899 | https://api.github.com/repos/openclaw/openclaw/issues/49899 | [Feature]: openclaw doctor should detect oversized persisted toolResult buildup in active/shared sessions | ### Summary
Add an `openclaw doctor` diagnostic that detects risky persisted `toolResult` accumulation in active/shared sessions and gives guided remediation before responsiveness degrades.
### Problem to solve
We can currently end up with very large persisted `toolResult` entries in human-facing shared sessions (e.... | closed | not_planned | true | 2 | [] | [] | 2026-03-18T15:52:13Z | 2026-04-28T04:38:01Z | 2026-04-26T03:40:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sene1337 | 261,265,124 | U_kgDOD5KW5A | User | false |
openclaw/openclaw | 4,096,216,716 | I_kwDOQb6kR870J06M | 49,912 | https://github.com/openclaw/openclaw/issues/49912 | https://api.github.com/repos/openclaw/openclaw/issues/49912 | Feature request: configurable OVERLOAD_FAILOVER_BACKOFF_POLICY | ## Problem
The hardcoded `OVERLOAD_FAILOVER_BACKOFF_POLICY` uses an initial backoff of **250ms** and a max of **1500ms**. This is too aggressive for agents with large system prompts (e.g. ~40K tokens), where API overload events are more likely and token processing time is longer.
### What happens in practice
When An... | closed | completed | true | 2 | [] | [] | 2026-03-18T16:11:49Z | 2026-04-28T04:38:00Z | 2026-04-26T03:40:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alfkaech-heartforge | 248,547,840 | U_kgDODtCKAA | User | false |
openclaw/openclaw | 4,096,375,500 | I_kwDOQb6kR870KbrM | 49,919 | https://github.com/openclaw/openclaw/issues/49919 | https://api.github.com/repos/openclaw/openclaw/issues/49919 | Embedded agent (webchat) fails with 'Network connection lost' on long requests through proxy while Claude Code works fine | ## Title: Embedded agent (webchat) fails with "Network connection lost" on long requests through proxy while Claude Code works fine
### Environment
- OpenClaw version: 2026.3.13
- OS: macOS (Apple Silicon, 16GB RAM)
- Node: v22.22.1
- Provider: OpenRouter (`openrouter/anthropic/claude-opus-4.6`)
- Network: Behind HTTP... | closed | not_planned | true | 2 | [] | [] | 2026-03-18T16:38:22Z | 2026-04-28T04:37:59Z | 2026-04-26T03:41:01Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | link-king | 41,622,183 | MDQ6VXNlcjQxNjIyMTgz | User | false |
openclaw/openclaw | 4,096,560,370 | I_kwDOQb6kR870LIzy | 49,933 | https://github.com/openclaw/openclaw/issues/49933 | https://api.github.com/repos/openclaw/openclaw/issues/49933 | [Bug]: Builtin memory indexing ignores configured remote embedding batch timeout | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Builtin memory indexing appears to ignore the configured remote embedding batch timeout and still fails at a hardcoded 120 seconds.
### Steps to reproduce
1. Configure OpenClaw builtin memory indexing to use a remote embedding provider.
2... | closed | completed | true | 9 | [
"bug",
"bug:behavior"
] | [] | 2026-03-18T17:11:19Z | 2026-04-28T04:37:58Z | 2026-04-26T03:41:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | unixwzrd | 17,365,553 | MDQ6VXNlcjE3MzY1NTUz | User | false |
openclaw/openclaw | 4,097,444,746 | I_kwDOQb6kR870OguK | 50,002 | https://github.com/openclaw/openclaw/issues/50002 | https://api.github.com/repos/openclaw/openclaw/issues/50002 | No non-interactive CLI path for github-copilot provider auth | ## Problem
There is no non-interactive CLI path to configure the `github-copilot` provider. The interactive flow runs the GitHub OAuth device flow, which requires a TTY. Automation environments (containers, CI, or external OAuth portals) that already have the OAuth access token (`ghu_...`) have no supported way to imp... | closed | not_planned | true | 1 | [] | [] | 2026-03-18T20:06:58Z | 2026-04-28T04:37:57Z | 2026-04-26T03:41:10Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | scottgl9 | 10,623,288 | MDQ6VXNlcjEwNjIzMjg4 | User | false |
openclaw/openclaw | 4,097,800,685 | I_kwDOQb6kR870P3nt | 50,029 | https://github.com/openclaw/openclaw/issues/50029 | https://api.github.com/repos/openclaw/openclaw/issues/50029 | Local & Self-Hosted Models | The current pulls agents towards local models — for cost, for privacy, for performance. The , , and plugins are the piers we build on, but the waters are choppy.
This is the first in a series of deep dives from **Driftnet** ([github.com/ocdlmv1/driftnet](https://github.com/ocdlmv1/driftnet)), a community intelligence... | closed | not_planned | true | 5 | [] | [] | 2026-03-18T21:26:43Z | 2026-04-28T04:37:57Z | 2026-04-26T03:41:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ocdlmv1 | 266,897,137 | U_kgDOD-iG8Q | User | false |
openclaw/openclaw | 4,097,819,366 | I_kwDOQb6kR870P8Lm | 50,031 | https://github.com/openclaw/openclaw/issues/50031 | https://api.github.com/repos/openclaw/openclaw/issues/50031 | [Feature]: Support [[reply_to_current]] tag for Slack threaded replies | ### Summary
Enable agents to create threaded replies in Slack using the [[reply_to_current]] reply tag.
### Problem to solve
Currently, OpenClaw's reply tags work on some platforms (Discord, Telegram) but not on Slack. When an agent uses `[[reply_to_current]]` in Slack, it either:
1. Gets stripped without creating ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-18T21:31:13Z | 2026-04-28T04:37:56Z | 2026-04-26T03:41:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | AIdenB899 | 269,310,343 | U_kgDOEA1Zhw | User | false |
openclaw/openclaw | 4,098,191,231 | I_kwDOQb6kR870RW9_ | 50,065 | https://github.com/openclaw/openclaw/issues/50065 | https://api.github.com/repos/openclaw/openclaw/issues/50065 | [Bug]: direct session can exceed configured contextWindow without timely compaction | ## Summary
Direct chat sessions can grow past the configured/effective context window (observed around `223k/200k`) without timely compaction, eventually causing unstable behavior such as non-replies or broken command lanes.
## Expected
- Sessions nearing the active `contextWindow` should compact before crossing the... | closed | completed | true | 7 | [] | [] | 2026-03-18T23:04:54Z | 2026-04-28T04:37:54Z | 2026-04-26T03:41:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ningfeemic-dev | 239,259,264 | U_kgDODkLOgA | User | false |
openclaw/openclaw | 4,098,307,903 | I_kwDOQb6kR870Rzc_ | 50,071 | https://github.com/openclaw/openclaw/issues/50071 | https://api.github.com/repos/openclaw/openclaw/issues/50071 | Feature Request: Context usage bar above chat input in Control UI | ## Feature Request
### Description
Add a **context window usage bar** (progress bar) directly above the chat input field in the Control UI webchat.
### Motivation
When chatting with OpenClaw through the Control UI, there's no visual indicator of how much of the context window has been consumed. Users currently need t... | closed | not_planned | true | 3 | [] | [] | 2026-03-18T23:37:06Z | 2026-04-28T04:37:53Z | 2026-04-26T03:41:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Brissux | 150,054,962 | U_kgDOCPGoMg | User | false |
openclaw/openclaw | 4,098,313,784 | I_kwDOQb6kR870R044 | 50,073 | https://github.com/openclaw/openclaw/issues/50073 | https://api.github.com/repos/openclaw/openclaw/issues/50073 | [Proposal] Skill-based task routing for multi-agent setups | ## Problem
In multi-agent setups, the coordinator agent decides which agent handles each task based on its own memory of each agent's role. This causes:
- **Misrouting** — coordinator sometimes sends to the wrong agent
- **Extra hops** — wrong agent receives, realizes it's not their job, forwards to the right one (2 ... | closed | not_planned | true | 4 | [] | [] | 2026-03-18T23:38:47Z | 2026-04-28T04:37:52Z | 2026-04-26T03:41:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | hjklasdfg | 187,938,926 | U_kgDOCzO4bg | User | false |
openclaw/openclaw | 4,098,675,450 | I_kwDOQb6kR870TNL6 | 50,117 | https://github.com/openclaw/openclaw/issues/50117 | https://api.github.com/repos/openclaw/openclaw/issues/50117 | [Feature]: Add Agent ID header to LLM requests for per-agent usage tracking | ### Summary
## Summary
Allow OpenClaw to automatically include an `X-OpenClaw-Agent` header in LLM API requests, so proxy servers can track usage per-agent.
### Problem to solve
## Use Case
When running multiple agents (main, grace, commander, etc.) through a local LLM proxy for usage tracking, there's no way to ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-19T01:36:47Z | 2026-04-28T04:37:51Z | 2026-04-26T03:41:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | stevensia | 15,981,271 | MDQ6VXNlcjE1OTgxMjcx | User | false |
openclaw/openclaw | 4,098,677,604 | I_kwDOQb6kR870TNtk | 50,120 | https://github.com/openclaw/openclaw/issues/50120 | https://api.github.com/repos/openclaw/openclaw/issues/50120 | [Bug]: Right side scroll bar missing on Mac (Sequoia) in Chrome | ## Summary
No right-side scroll bar in the OpenClaw webchat UI on macOS with Chrome. Still reproducible on v2026.3.13.
Original issue: #11384 — closed by stale bot, not fixed.
## Steps to reproduce
1. Open OpenClaw webchat in Chrome on macOS Sequoia
2. Mac "Show scroll bars: Always" is configured in System Settings... | closed | completed | true | 1 | [] | [] | 2026-03-19T01:37:34Z | 2026-04-28T04:37:51Z | 2026-04-26T03:41:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | robertvanstedum | 65,718,279 | MDQ6VXNlcjY1NzE4Mjc5 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.