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 | 4,058,968,404 | I_kwDOQb6kR87x7vFU | 43,295 | https://github.com/openclaw/openclaw/issues/43295 | https://api.github.com/repos/openclaw/openclaw/issues/43295 | Agent fails to execute rules despite knowing them - need forced rule checking mechanism | ## Problem
OpenClaw agents (specifically main agent using claude-opus-4-6) consistently fail to execute known rules, despite:
1. Rules being clearly stated in system prompts
2. Rules being stored in long-term memory (memory-lancedb-pro)
3. Agent acknowledging and understanding the rules
4. Repeated corrections and rem... | closed | not_planned | true | 5 | [] | [] | 2026-03-11T15:35:06Z | 2026-05-02T04:27:06Z | 2026-04-29T21:43:07Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | mouseroser | 8,221,932 | MDQ6VXNlcjgyMjE5MzI= | User | false |
openclaw/openclaw | 4,059,499,817 | I_kwDOQb6kR87x9w0p | 43,367 | https://github.com/openclaw/openclaw/issues/43367 | https://api.github.com/repos/openclaw/openclaw/issues/43367 | Multi-agent orchestration is unstable: concurrent agents add/config overwrites, session-lock failures, and detached child work | ### Summary
I tried to orchestrate a small parallel coding batch from the OpenClaw CLI on `2026.3.8` and hit a cluster of failures that make multi-agent runs unreliable in practice:
1. `openclaw agents add` appears unsafe when invoked concurrently: config gets overwritten repeatedly and only a subset of agents persis... | open | null | false | 8 | [] | [] | 2026-03-11T17:08:20Z | 2026-05-04T12:16:22Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | waliddafif | 7,280,752 | MDQ6VXNlcjcyODA3NTI= | User | false |
openclaw/openclaw | 4,059,542,790 | I_kwDOQb6kR87x97UG | 43,374 | https://github.com/openclaw/openclaw/issues/43374 | https://api.github.com/repos/openclaw/openclaw/issues/43374 | All LLM API calls time out simultaneously despite APIs being reachable (multi-agent concurrency issue) | ## Bug Description
When running 4 agents concurrently via Telegram, **all LLM API calls time out simultaneously** (every ~60-90s, matching `timeoutSeconds`), even though the same APIs respond normally via `curl` at the exact same moment.
This is **not** an LLM provider issue — it is an internal bottleneck within Open... | open | null | false | 2 | [] | [] | 2026-03-11T17:17:19Z | 2026-05-04T12:16:23Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | conanwhf | 14,151,735 | MDQ6VXNlcjE0MTUxNzM1 | User | false |
openclaw/openclaw | 4,059,609,661 | I_kwDOQb6kR87x-Lo9 | 43,383 | https://github.com/openclaw/openclaw/issues/43383 | https://api.github.com/repos/openclaw/openclaw/issues/43383 | [Bug]: Sandboxed non-main sessions expose host skill paths that are unreadable inside the sandbox | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In Docker-based `non-main` sandbox sessions, OpenClaw exposes host-style skill paths in the prompt (for example `~/.openclaw/skills/<skill>/SKILL.md`), but file access is enforced relative to the sandbox workspace.
This causes `read` to fa... | closed | completed | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T17:29:31Z | 2026-05-05T05:13:23Z | 2026-05-05T05:13:23Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | CSerht | 62,015,641 | MDQ6VXNlcjYyMDE1NjQx | User | false |
openclaw/openclaw | 4,059,955,317 | I_kwDOQb6kR87x_gB1 | 43,410 | https://github.com/openclaw/openclaw/issues/43410 | https://api.github.com/repos/openclaw/openclaw/issues/43410 | 34 built-in skills missing trigger phrases in description field | ## Summary
After reviewing Anthropic's *Complete Guide to Building Skills for Claude* (Jan 2026), I audited all 52 built-in OpenClaw skills against the guide's best practices. The main finding: **34 of 52 skills (65%) are missing trigger phrases in their `description` field.**
Per the guide, a skill description must ... | open | null | false | 2 | [] | [] | 2026-03-11T18:39:24Z | 2026-05-04T12:16:25Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Jkahn-tr | 235,643,106 | U_kgDODgug4g | User | false |
openclaw/openclaw | 4,060,538,578 | I_kwDOQb6kR87yBubS | 43,454 | https://github.com/openclaw/openclaw/issues/43454 | https://api.github.com/repos/openclaw/openclaw/issues/43454 | Feature Request: Gateway lifecycle hooks (onSubagentComplete, onToolCallThreshold, onTurnComplete) | ## Problem
There's no way for **workspace hooks** to automatically trigger behavior at specific agent lifecycle points. Currently, the only way to approximate this is through:
- Prompt conditioning (always-on system prompt instructions)
- Heartbeat/cron polling (timer-based, not event-based)
- Manual doctrine in AGENT... | open | null | false | 5 | [] | [] | 2026-03-11T20:36:53Z | 2026-05-04T12:16:29Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | itsAR-VR | 112,514,620 | U_kgDOBrTWPA | User | false |
openclaw/openclaw | 4,060,546,207 | I_kwDOQb6kR87yBwSf | 43,455 | https://github.com/openclaw/openclaw/issues/43455 | https://api.github.com/repos/openclaw/openclaw/issues/43455 | Browserbase integration broken: static cdpUrl returns 404, needs session API | ## Browserbase integration broken: static cdpUrl no longer works (404)
### Description
The OpenClaw browser docs recommend this config for Browserbase:
```json
"browserbase": {
"cdpUrl": "wss://connect.browserbase.com?apiKey=<API_KEY>",
"color": "#F97316"
}
```
The docs state: *"Browserbase auto-creates a browser... | open | null | false | 2 | [] | [] | 2026-03-11T20:38:34Z | 2026-05-04T12:16:30Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | HanenVit | 183,886,923 | U_kgDOCvXkSw | User | false |
openclaw/openclaw | 4,060,555,054 | I_kwDOQb6kR87yBycu | 43,456 | https://github.com/openclaw/openclaw/issues/43456 | https://api.github.com/repos/openclaw/openclaw/issues/43456 | Mattermost: wake agent on reply to bot's own message (like Telegram) | ## Feature Request
**Current behavior:** In Mattermost, agents only wake on @mention (`chatmode: oncall`) or on every message (`onmessage`). Replying to an agent's message in a thread does NOT wake the agent.
**Expected behavior:** When a user replies to a bot's own message (in a thread or inline), the agent should w... | open | null | false | 2 | [] | [] | 2026-03-11T20:40:29Z | 2026-05-04T00:43:01Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | JasonGraber | 5,443,850 | MDQ6VXNlcjU0NDM4NTA= | User | false |
openclaw/openclaw | 4,060,725,082 | I_kwDOQb6kR87yCb9a | 43,481 | https://github.com/openclaw/openclaw/issues/43481 | https://api.github.com/repos/openclaw/openclaw/issues/43481 | Cross-provider fallback should respect provider trust boundaries for images/attachments | ## Problem
When a model fallback retry crosses providers (e.g. Anthropic → OpenAI), user-supplied images and attachments are forwarded to the fallback provider. Users who configure cross-provider fallbacks for text resilience may not expect their image data to be sent to a different provider.
The current behavior is ... | open | null | false | 2 | [] | [] | 2026-03-11T21:18:53Z | 2026-05-04T13:18:29Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | mitchmcalister | 209,334 | MDQ6VXNlcjIwOTMzNA== | User | false |
openclaw/openclaw | 4,061,440,928 | I_kwDOQb6kR87yFKug | 43,549 | https://github.com/openclaw/openclaw/issues/43549 | https://api.github.com/repos/openclaw/openclaw/issues/43549 | [Bug]: Telegram channel can be wedged by bad persisted session JSON; status should detect and offer recovery | ## Summary
Telegram stopped working because the persisted Telegram session JSON had become effectively unusable ("full" / oversized / likely corrupted enough to wedge normal operation). Running `openclaw status telegram` exposed the clue, but the troubleshooting flow did not guide toward inspecting or remediating the s... | open | null | false | 5 | [] | [] | 2026-03-12T00:42:53Z | 2026-05-04T12:16:41Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | muffinsexit | 259,756,891 | U_kgDOD3uTWw | User | false |
openclaw/openclaw | 4,061,542,618 | I_kwDOQb6kR87yFjja | 43,563 | https://github.com/openclaw/openclaw/issues/43563 | https://api.github.com/repos/openclaw/openclaw/issues/43563 | [Feature]: Feishu cross-bot message broadcast — parity with Discord multi-bot support | ## Summary
In Discord, bots in the same server/channel naturally receive each other's messages via `MESSAGE_CREATE` events — enabling multi-agent collaboration in shared channels. Feishu's `im.message.receive_v1` webhook only delivers **human user messages**; bot-sent messages are invisible to other bots in the same g... | open | null | false | 2 | [] | [] | 2026-03-12T01:27:33Z | 2026-05-04T12:16:44Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | YJWell | 18,249,161 | MDQ6VXNlcjE4MjQ5MTYx | User | false |
openclaw/openclaw | 4,061,564,416 | I_kwDOQb6kR87yFo4A | 43,567 | https://github.com/openclaw/openclaw/issues/43567 | https://api.github.com/repos/openclaw/openclaw/issues/43567 | Surface internal system messages in a distinct UI block instead of normal chat bubbles | ## Problem
Internal system messages can sometimes appear in the conversation stream looking like a normal user message. Example seen in the Control UI / TUI:
> Pre-compaction memory flush. Store durable memories now ...
From the user side, this reads as if **they** sent a strange message, or as if the assistant is e... | open | null | false | 2 | [] | [] | 2026-03-12T01:36:00Z | 2026-05-04T12:05:37Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | prw760 | 147,275,171 | U_kgDOCMc9ow | User | false |
openclaw/openclaw | 4,061,573,051 | I_kwDOQb6kR87yFq-7 | 43,570 | https://github.com/openclaw/openclaw/issues/43570 | https://api.github.com/repos/openclaw/openclaw/issues/43570 | [Bug]: Discord thread-reply silently ignores filePath/path parameters — attachment not sent | ## Summary
`message` tool with `action: "thread-reply"` silently ignores `filePath` and `path` parameters on Discord. The tool returns `ok: true` but no attachment is included in the Discord message.
## How I Discovered This
I run a multi-agent pipeline on OpenClaw where one agent (Pixel) generates a daily AI news d... | open | null | false | 3 | [] | [] | 2026-03-12T01:40:22Z | 2026-05-04T12:16:49Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Perfecto23 | 91,355,573 | MDQ6VXNlcjkxMzU1NTcz | User | false |
openclaw/openclaw | 4,061,723,302 | I_kwDOQb6kR87yGPqm | 43,605 | https://github.com/openclaw/openclaw/issues/43605 | https://api.github.com/repos/openclaw/openclaw/issues/43605 | [Feature]: Cross-gateway messaging (sessions_send between machines) | ## Description
`sessions_send` only works within the same gateway. In multi-machine setups (e.g., agents split across a MacBook, iMac, and Mac Studio each running their own gateway), there's no native way for agents on different gateways to communicate.
## Environment
- OpenClaw 2026.3.8
- 3 machines: MacBook (1 agent... | open | null | false | 2 | [] | [] | 2026-03-12T02:37:45Z | 2026-05-04T12:16:53Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | SPFAdvisors | 133,194,757 | U_kgDOB_BkBQ | User | false |
openclaw/openclaw | 4,062,412,235 | I_kwDOQb6kR87yI33L | 43,735 | https://github.com/openclaw/openclaw/issues/43735 | https://api.github.com/repos/openclaw/openclaw/issues/43735 | Skills not loading in agent context from ~/.openclaw/workspace/skills/ | ## Bug Report
### Description
The OpenClaw agent context is not loading all available skills from `~/.openclaw/workspace/skills/` directory. Only a subset of skills appear in the initial context under `<available_skills>`, causing agents to be unaware of existing skills.
### Steps to Reproduce
1. Create multiple skil... | open | null | false | 12 | [] | [] | 2026-03-12T06:03:15Z | 2026-05-04T00:43:04Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | che1404 | 290,546 | MDQ6VXNlcjI5MDU0Ng== | User | false |
openclaw/openclaw | 4,062,424,736 | I_kwDOQb6kR87yI66g | 43,737 | https://github.com/openclaw/openclaw/issues/43737 | https://api.github.com/repos/openclaw/openclaw/issues/43737 | [Bug]: 'main' session is missing from the dropdown menu (only 'heartbeat' is visible) | ### Bug type
Regression (worked before, now fails)
### Summary
I have noticed that the main session is no longer available in the session selection dropdown. This issue has persisted for the past few versions. Currently, when I open the dropdown menu, only the heartbeat session is listed.
### Steps to reproduce
Op... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-12T06:06:09Z | 2026-05-04T12:17:03Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | gmvp3 | 27,907,782 | MDQ6VXNlcjI3OTA3Nzgy | User | false |
openclaw/openclaw | 4,062,481,808 | I_kwDOQb6kR87yJI2Q | 43,747 | https://github.com/openclaw/openclaw/issues/43747 | https://api.github.com/repos/openclaw/openclaw/issues/43747 | [Bug]: Memory management is in chaos | ### Bug type
Regression (worked before, now fails)
### Summary
Me and my collegues (3 people) are using openclaw.
I never see any of our memory is managed in sameway.
- Mine keeps doing chunking & embedding and store into `~/.openclaw/memory/main.sqlite`
- My collegue `A`, her claw is storing markdown files into `~... | open | null | false | 8 | [
"bug",
"regression"
] | [] | 2026-03-12T06:21:16Z | 2026-05-04T12:17:05Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | AntiMoron | 6,587,734 | MDQ6VXNlcjY1ODc3MzQ= | User | false |
openclaw/openclaw | 4,062,490,672 | I_kwDOQb6kR87yJLAw | 43,750 | https://github.com/openclaw/openclaw/issues/43750 | https://api.github.com/repos/openclaw/openclaw/issues/43750 | /new and /reset should support @agent targeting in multi-agent channels | ## Problem
In multi-agent Discord setups, `/new` and `/reset` slash commands are ambiguous — all agents registered in the channel respond to them. There is also no way to target a specific agent.
**Observed behavior:**
- User runs `/new` in a shared channel → multiple agents (e.g. Forge + Anvil) both start new sessio... | open | null | false | 3 | [] | [] | 2026-03-12T06:23:29Z | 2026-05-04T12:45:09Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Kyzcreig | 9,063,726 | MDQ6VXNlcjkwNjM3MjY= | User | false |
openclaw/openclaw | 4,062,547,590 | I_kwDOQb6kR87yJY6G | 43,760 | https://github.com/openclaw/openclaw/issues/43760 | https://api.github.com/repos/openclaw/openclaw/issues/43760 | Feature: separate fallback remote config for memorySearch | ## Problem
When using `provider: "gemini"` as the primary memorySearch embedding provider, the `remote` config (baseUrl, apiKey) is shared between primary and fallback. This makes it impossible to configure a different provider (e.g., OpenAI-compatible SiliconFlow with bge-m3) as a fallback, because the Gemini provide... | closed | not_planned | false | 2 | [] | [] | 2026-03-12T06:37:47Z | 2026-05-04T20:40:02Z | 2026-05-04T20:40:02Z | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | jackiedepp | 31,266,295 | MDQ6VXNlcjMxMjY2Mjk1 | User | false |
openclaw/openclaw | 4,062,738,554 | I_kwDOQb6kR87yKHh6 | 43,795 | https://github.com/openclaw/openclaw/issues/43795 | https://api.github.com/repos/openclaw/openclaw/issues/43795 | [Bug]: 500 v.content is not iterable | ### Bug type
Regression (worked before, now fails)
### Summary
500 v.content is not iterable
### Steps to reproduce
<img width="786" height="871" alt="Image" src="https://github.com/user-attachments/assets/d97a1def-38e4-47c4-b13d-7eee1028073b" />
### Expected behavior
15:22:37 [agent/embedded] embedded run failo... | open | null | false | 6 | [
"bug",
"regression"
] | [] | 2026-03-12T07:22:38Z | 2026-05-04T00:43:06Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | 1090963827 | 151,716,186 | U_kgDOCQsBWg | User | false |
openclaw/openclaw | 4,062,759,399 | I_kwDOQb6kR87yKMnn | 43,797 | https://github.com/openclaw/openclaw/issues/43797 | https://api.github.com/repos/openclaw/openclaw/issues/43797 | [Bug]: Sandbox prune does not clean up workspace directory | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Sandbox prune removes Docker containers but leaves workspace directories behind, causing disk space leak over time
### Steps to reproduce
1. Configure an agent with sandbox.prune.idleHours or sandbox.prune.maxAgeDays (e.g., idleHours: 168... | open | null | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T07:26:36Z | 2026-05-03T23:37:18Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | climbran | 11,232,399 | MDQ6VXNlcjExMjMyMzk5 | User | false |
openclaw/openclaw | 4,062,781,271 | I_kwDOQb6kR87yKR9X | 43,803 | https://github.com/openclaw/openclaw/issues/43803 | https://api.github.com/repos/openclaw/openclaw/issues/43803 | [BUG] config.patch still sends SIGUSR1 for hot-reloadable paths (browser.profiles.*), bypassing reload mode |
## Related Issues
This is a continuation of a known problem that has been partially fixed but not fully resolved:
- **Issue #5533** (Jan 2026) — Reported that config modifications unconditionally trigger SIGUSR1 even when restart is disabled. Fixed by **PR #13408** (`fix(gateway): skip SIGUSR1 restart in config.patc... | open | null | false | 3 | [] | [] | 2026-03-12T07:30:32Z | 2026-05-03T23:37:20Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | piston4711 | 257,291,435 | U_kgDOD1X0qw | User | false |
openclaw/openclaw | 4,062,862,928 | I_kwDOQb6kR87yKl5Q | 43,821 | https://github.com/openclaw/openclaw/issues/43821 | https://api.github.com/repos/openclaw/openclaw/issues/43821 | Feature: Add global HTTP proxy support via environment variables | ## Feature Request
### Summary
Add comprehensive HTTP proxy support for all network requests made by OpenClaw, configurable through standard environment variables (`HTTP_PROXY`, `HTTPS_PROXY`, `NO_PROXY`).
### Motivation
Many enterprise and restricted network environments require HTTP/HTTPS proxies for external inter... | closed | completed | true | 9 | [
"stale"
] | [] | 2026-03-12T07:48:13Z | 2026-04-30T04:33:47Z | 2026-04-27T14:45:12Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mbC6zC | 1,839,207 | MDQ6VXNlcjE4MzkyMDc= | User | false |
openclaw/openclaw | 4,063,331,885 | I_kwDOQb6kR87yMYYt | 43,903 | https://github.com/openclaw/openclaw/issues/43903 | https://api.github.com/repos/openclaw/openclaw/issues/43903 | Support multiple gateway tokens for multi-user session isolation | ### Problem Description
Currently, `gateway.auth.token` only supports a single token. When multiple users need to access the webchat Control UI with isolated sessions, there's no way to assign different tokens to different users.
### Current Limitations
| Feature | token mode | trusted-proxy mode |
|---------|------... | open | null | false | 3 | [] | [] | 2026-03-12T09:30:25Z | 2026-05-03T23:37:22Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | waitloven | 7,930,241 | MDQ6VXNlcjc5MzAyNDE= | User | false |
openclaw/openclaw | 4,063,678,295 | I_kwDOQb6kR87yNs9X | 43,942 | https://github.com/openclaw/openclaw/issues/43942 | https://api.github.com/repos/openclaw/openclaw/issues/43942 | [Bug]: [reply_to_current] tag emitted inside thinking block instead of as message type=text with hunter-alpha model | ## Description
When the agent generates a reply containing a `[[reply_to_current]]` tag, the entire reply content ends up **inside the thinking/reasoning block** instead of being emitted as the actual user-visible `message` with `type=text`. The user sees no reply content at all — or the tag is buried in internal reas... | closed | completed | true | 7 | [
"stale"
] | [] | 2026-03-12T10:28:30Z | 2026-04-30T04:34:09Z | 2026-04-27T12:37:01Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | EvanDataForge | 262,609,517 | U_kgDOD6cabQ | User | false |
openclaw/openclaw | 4,063,728,706 | I_kwDOQb6kR87yN5RC | 43,950 | https://github.com/openclaw/openclaw/issues/43950 | https://api.github.com/repos/openclaw/openclaw/issues/43950 | Feature: Support registering local skills into skill_search index | ## Problem
Currently, skills created locally (with `SKILL.md` / `README.md` / `_meta.json` in `~/.openclaw/skills/<slug>/`) are **not discoverable** via `skill_search`. The `skill_install` tool only accepts a `skillId` that already exists in the system index — there is no way to register a locally-authored skill.
Thi... | open | null | false | 2 | [] | [] | 2026-03-12T10:38:10Z | 2026-05-03T23:37:24Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | mkz0930 | 19,551,686 | MDQ6VXNlcjE5NTUxNjg2 | User | false |
openclaw/openclaw | 4,063,930,120 | I_kwDOQb6kR87yOqcI | 43,976 | https://github.com/openclaw/openclaw/issues/43976 | https://api.github.com/repos/openclaw/openclaw/issues/43976 | Feature: Agent 动态切换命令（不重启 Gateway） | ## 需求描述
在单聊模式下，希望能够通过命令动态切换当前 channel 绑定的 Agent，而不需要重启 Gateway。
## 期望的命令
| 命令 | 说明 |
|---|---|
| `/agents` | 列出所有可用 Agent |
| `/agent <name>` | 切换到指定 Agent |
| `/agent main` | 切换回 main Agent |
## 当前限制
目前切换 Agent 需要：
1. 修改 `openclaw.json` 的 bindings
2. 重启 Gateway 或等待热重载
这在对话中不够流畅。
## 可能的实现方案
### 方案 A：会话级 Agent 状... | open | null | false | 2 | [] | [] | 2026-03-12T11:18:22Z | 2026-05-03T23:37:25Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | kings0527 | 5,417,480 | MDQ6VXNlcjU0MTc0ODA= | User | false |
openclaw/openclaw | 4,064,060,765 | I_kwDOQb6kR87yPKVd | 43,996 | https://github.com/openclaw/openclaw/issues/43996 | https://api.github.com/repos/openclaw/openclaw/issues/43996 | [Bug]: Sandbox container exits immediately when no-new-privileges is applied (exec /usr/bin/sleep: operation not permitted) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
After upgrading to OpenClaw 2026.3.11, maker sandbox sessions fail to execute even basic commands (pwd && ls).
The sandbox container exits immediately with ExitCode=255, and container logs show:
exec /usr/bin/sleep: operation not permitted.... | open | null | false | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T11:42:51Z | 2026-05-04T00:43:11Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | aihorizon2077 | 264,550,408 | U_kgDOD8S4CA | User | false |
openclaw/openclaw | 4,093,826,894 | I_kwDOQb6kR870AtdO | 49,692 | https://github.com/openclaw/openclaw/issues/49692 | https://api.github.com/repos/openclaw/openclaw/issues/49692 | UX: Friendlier gateway drain message + auto-retry unprocessed messages after restart | ## Problem
When the gateway restarts (config change, update, etc.), users see this error:
> ⚠️ Agent failed before reply: Gateway is draining for restart; new tasks are not accepted.
> Logs: openclaw logs --follow
This is **technically accurate but terrible UX** for end users — especially in Signal/Telegram group ch... | open | null | false | 6 | [] | [] | 2026-03-18T09:19:16Z | 2026-05-04T18:21:45Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | skillz-xx | 837,936 | MDQ6VXNlcjgzNzkzNg== | User | false |
openclaw/openclaw | 4,120,867,878 | I_kwDOQb6kR871n3Qm | 52,875 | https://github.com/openclaw/openclaw/issues/52875 | https://api.github.com/repos/openclaw/openclaw/issues/52875 | [Bug]: Session_send gives no session found. | ### Bug type
Regression (worked before, now fails)
### Summary
after upgrading to 2026-3-22 my main agent is not able to contact other agents. She says that session_list gives results most of them being crons. but the agents she needs to talk to have no sessions.
I can chat with the agents, see them in tui and in t... | open | null | false | 20 | [
"bug",
"regression"
] | [] | 2026-03-23T13:25:23Z | 2026-05-04T21:01:01Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,162,993,993 | I_kwDOQb6kR874Ij9J | 56,841 | https://github.com/openclaw/openclaw/issues/56841 | https://api.github.com/repos/openclaw/openclaw/issues/56841 | [Bug]: Custom OpenAI-compatible provider works via direct /v1/chat/completions, but OpenClaw 2026.3.28 always fails with "Connection error" | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
OpenClaw 2026.3.28 fails on every chat turn for a custom OpenAI-compatible provider with “LLM request failed: network connection error”, even though direct /v1/models and /v1/chat/completions calls to the same ... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-03-29T06:32:19Z | 2026-04-30T04:35:38Z | 2026-04-27T08:47:48Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xqiq | 239,173,419 | U_kgDODkF_Kw | User | false |
openclaw/openclaw | 4,334,374,070 | I_kwDOQb6kR88AAAABAllMtg | 72,719 | https://github.com/openclaw/openclaw/issues/72719 | https://api.github.com/repos/openclaw/openclaw/issues/72719 | test (closed) | test | closed | completed | true | 0 | [] | [] | 2026-04-27T08:48:13Z | 2026-04-30T04:35:37Z | 2026-04-27T08:48:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | sign-2025 | 199,324,164 | U_kgDOC-FyBA | User | false |
openclaw/openclaw | 4,160,287,580 | I_kwDOQb6kR873-PNc | 56,446 | https://github.com/openclaw/openclaw/issues/56446 | https://api.github.com/repos/openclaw/openclaw/issues/56446 | [Bug]: `openclaw plugins install` rejects plugin with `peerDependencies: { "openclaw": "*" }` — wildcard API version not matched | ### Describe the bug
When installing `@byterover/byterover` (v1.1.0) via `openclaw plugins install @byterover/byterover`, the install fails with:
```
Resolving clawhub:@byterover/byterover…
Plugin "@byterover/byterover" requires plugin API *, but this OpenClaw runtime exposes 2026.3.24.
```
The plugin's `package.jso... | closed | completed | true | 1 | [] | [] | 2026-03-28T14:57:36Z | 2026-04-30T04:35:36Z | 2026-04-27T08:48:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | darconada | 14,311,175 | MDQ6VXNlcjE0MzExMTc1 | User | false |
openclaw/openclaw | 4,334,377,532 | I_kwDOQb6kR88AAAABAllaPA | 72,720 | https://github.com/openclaw/openclaw/issues/72720 | https://api.github.com/repos/openclaw/openclaw/issues/72720 | openclaw gateway run hangs silently on Node.js v24.14.0 (respawn mechanism) | ## Description
`openclaw gateway run` hangs silently on Node.js v24.14.0 with no output to stdout/stderr. The `openclaw.mjs` launcher's respawn mechanism spawns a child process that silently fails.
## Reproduction
1. Node.js v24.14.0 on Alibaba Cloud ECS (Alibaba Cloud Linux 3)
2. `openclaw gateway run --bind loopba... | closed | completed | true | 1 | [] | [] | 2026-04-27T08:48:47Z | 2026-04-30T04:33:50Z | 2026-04-27T14:01:59Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | sign-2025 | 199,324,164 | U_kgDOC-FyBA | User | false |
openclaw/openclaw | 4,334,471,614 | I_kwDOQb6kR88AAAABAlrJvg | 72,727 | https://github.com/openclaw/openclaw/issues/72727 | https://api.github.com/repos/openclaw/openclaw/issues/72727 | Browser relay extension server not starting (port 18792) | # Bug Report: Browser Relay Extension Server Not Starting
## Issue
The OpenClaw browser relay extension cannot connect because the relay server (expected on port 18792) is not starting.
## Environment
- **OpenClaw Version:** 2026.4.24 (cbcfdf6)
- **OS:** macOS 26.3.1 (arm64)
- **Node:** v25.6.1
- **Gateway Port:** 18... | closed | not_planned | true | 1 | [] | [] | 2026-04-27T09:03:20Z | 2026-05-03T04:36:48Z | 2026-04-30T14:53:30Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | BloxRE12 | 266,437,499 | U_kgDOD-GDew | User | false |
openclaw/openclaw | 4,217,678,354 | I_kwDOQb6kR877ZKoS | 62,492 | https://github.com/openclaw/openclaw/issues/62492 | https://api.github.com/repos/openclaw/openclaw/issues/62492 | "Update now" button does nothing — no feedback when update.run returns "skipped" status | ## What Happened
Clicking "Update now" in the GUI update notification banner ("Update available: v2026.4.5 (running v2026.4.2). Update now") does nothing visible. The button briefly disables then re-enables, with no success message, no error, and no restart.
## Expected Behavior
After clicking "Update now":
- A load... | closed | completed | true | 0 | [] | [] | 2026-04-07T12:34:55Z | 2026-04-30T04:35:33Z | 2026-04-27T09:07:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | davidgut1982 | 5,985,783 | MDQ6VXNlcjU5ODU3ODM= | User | false |
openclaw/openclaw | 4,161,404,128 | I_kwDOQb6kR874Cfzg | 56,579 | https://github.com/openclaw/openclaw/issues/56579 | https://api.github.com/repos/openclaw/openclaw/issues/56579 | [Bug]: ClawHub parser accepts malformed specs with trailing @ and returns package_not_found | ## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
`parseClawHubPluginSpec` accepts malformed ClawHub specs with a trailing `@` (for example `clawhub:demo@`) and treats them as a package name. This later surfaces as `package_not_found` instead of a clear invalid-spec error.
## Steps to repro... | closed | completed | true | 2 | [] | [] | 2026-03-28T19:24:16Z | 2026-04-30T04:35:33Z | 2026-04-27T09:08:48Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Kansodata | 225,288,021 | U_kgDODW2fVQ | User | false |
openclaw/openclaw | 4,334,548,834 | I_kwDOQb6kR88AAAABAlv3Yg | 72,729 | https://github.com/openclaw/openclaw/issues/72729 | https://api.github.com/repos/openclaw/openclaw/issues/72729 | [Bug]: opencode/claude-opus-4-7 rejects /think xhigh|adaptive|max despite docs declaring Opus 4.7 supports them | ## Summary
The OpenCode plugin's thinking profile for `claude-opus-4-7` only advertises the legacy 5 levels (`off | minimal | low | medium | high`), so `/think xhigh`, `/think adaptive`, and `/think max` all fail with "not supported", even though `docs/tools/thinking.md` states Opus 4.7 supports all three.
## Steps t... | closed | completed | true | 0 | [] | [] | 2026-04-27T09:15:20Z | 2026-04-30T04:34:05Z | 2026-04-27T13:03:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | aaajiao | 49,885 | MDQ6VXNlcjQ5ODg1 | User | false |
openclaw/openclaw | 4,152,647,207 | I_kwDOQb6kR873hF4n | 55,805 | https://github.com/openclaw/openclaw/issues/55805 | https://api.github.com/repos/openclaw/openclaw/issues/55805 | [Bug]: Openclaw plugins install fails to install npm packages before falling back from ClawHub | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
`openclaw plugins install` appears to fail in the ClawHub-first resolution path with `fetch failed` and does not appear to fall back to npm for at least some npm package specs.
### Steps to reproduce
1. Run:
... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T13:26:59Z | 2026-04-30T04:35:32Z | 2026-04-27T09:17:22Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Zeoy2020 | 126,147,769 | U_kgDOB4TcuQ | User | false |
openclaw/openclaw | 4,220,402,222 | I_kwDOQb6kR877jjou | 62,692 | https://github.com/openclaw/openclaw/issues/62692 | https://api.github.com/repos/openclaw/openclaw/issues/62692 | web_search tool not injected for local models (llama-server / OpenAI-compatible backends) | ## Bug Description
OpenClaw never effectively injects the `web_search` tool when using a local llama-server backend via the `openai-completions` API. Even with SearXNG configured and working, the model either doesn't see the tool, can't call it properly, or the context is too saturated for it to respond.
## Environme... | closed | not_planned | true | 1 | [] | [] | 2026-04-07T20:25:28Z | 2026-04-30T04:35:29Z | 2026-04-27T09:27:04Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | go-yanka | 26,028,987 | MDQ6VXNlcjI2MDI4OTg3 | User | false |
openclaw/openclaw | 4,334,347,859 | I_kwDOQb6kR88AAAABAljmUw | 72,714 | https://github.com/openclaw/openclaw/issues/72714 | https://api.github.com/repos/openclaw/openclaw/issues/72714 | Gateway repeatedly triggers npm install into HOME/node_modules due to leftover package.json in home directory, causing startup loop and pollution | **Describe the bug**
After upgrading Openclaw, the gateway process enters a restart loop. Each time it starts, it triggers a massive npm install command (installing @anthropic-ai/sdk, @aws-sdk/*, express, playwright-core, etc.) directly into $HOME/node_modules. This floods the home directory with node modules, makes th... | closed | completed | true | 2 | [] | [] | 2026-04-27T08:43:43Z | 2026-04-30T04:35:28Z | 2026-04-27T09:28:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mmhzlrj | 29,726,050 | MDQ6VXNlcjI5NzI2MDUw | User | false |
openclaw/openclaw | 4,207,474,582 | I_kwDOQb6kR876yPeW | 61,340 | https://github.com/openclaw/openclaw/issues/61340 | https://api.github.com/repos/openclaw/openclaw/issues/61340 | Bug: gateway install --force persists secrets into user systemd unit files | # Bug report draft — `openclaw gateway install --force` persists secrets into user systemd unit files
## Summary
On a Linux host using user systemd services, `openclaw gateway install --force` appears to persist sensitive environment variables into the generated gateway unit file, including:
- `OPENAI_API_KEY`
- `OPE... | closed | completed | true | 1 | [] | [] | 2026-04-05T12:53:31Z | 2026-04-30T04:35:27Z | 2026-04-27T09:28:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | clawbotbkk | 270,162,996 | U_kgDOEBpcNA | User | false |
openclaw/openclaw | 4,334,282,256 | I_kwDOQb6kR88AAAABAlfmEA | 72,711 | https://github.com/openclaw/openclaw/issues/72711 | https://api.github.com/repos/openclaw/openclaw/issues/72711 | [Bug] 2026.4.24 gateway 启动后反复自动执行 npm install 形成死循环，降级到 4.21 正常 | ## 问题描述
升级到 \`openclaw@2026.4.24\` 后，gateway 每次启动都会在后台自动执行一次 \`npm install\`，安装所有渠道插件的运行时依赖（如 @anthropic-ai/sdk, @google/genai, playwright-core, pdfjs-dist 等共 30+ 个包）。该进程占用大量系统资源，在资源受限的机器上容易被 OOM Killer 强制终止，导致安装永远无法完成，下次 gateway 启动又会再次触发 → 形成死循环。
降级回 \`openclaw@4.21\` 后，问题立即消失，不再出现任何自动安装行为。
## 复现步骤
1. 安装或升级到 \`openc... | closed | completed | true | 2 | [] | [] | 2026-04-27T08:32:31Z | 2026-04-30T04:35:26Z | 2026-04-27T09:28:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kikibrian | 275,287,238 | U_kgDOEGiMxg | User | false |
openclaw/openclaw | 4,091,776,984 | I_kwDOQb6kR87z44_Y | 49,311 | https://github.com/openclaw/openclaw/issues/49311 | https://api.github.com/repos/openclaw/openclaw/issues/49311 | doctor --repair / gateway install silently wipes custom EnvironmentVariables from LaunchAgent plist | ## Summary
`openclaw doctor --repair` and `openclaw gateway install` regenerate the LaunchAgent plist from scratch, silently dropping any custom `EnvironmentVariables` that were previously set (e.g., `ANTHROPIC_API_KEY`). This causes the gateway to restart without the API key, breaking all LLM calls until the user man... | closed | completed | true | 5 | [] | [] | 2026-03-17T23:35:21Z | 2026-04-30T04:35:26Z | 2026-04-27T09:28:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | fc4ngftsdj-hash | 263,761,580 | U_kgDOD7iurA | User | false |
openclaw/openclaw | 4,071,328,132 | I_kwDOQb6kR87yq4mE | 45,163 | https://github.com/openclaw/openclaw/issues/45163 | https://api.github.com/repos/openclaw/openclaw/issues/45163 | `openclaw doctor --fix` rewrites macOS LaunchAgent plist, drops user-defined EnvironmentVariables | ### Describe the bug
Running `openclaw doctor --fix` on macOS rewrites the LaunchAgent plist file, dropping all `EnvironmentVariables` entries that the doctor command doesn't explicitly manage.
This is destructive because the plist `EnvironmentVariables` dict is the canonical way to pass secrets (API keys, tokens) to... | closed | completed | true | 2 | [] | [] | 2026-03-13T14:05:32Z | 2026-04-30T04:35:25Z | 2026-04-27T09:28:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jeremyknows | 237,305,675 | U_kgDODiT_Sw | User | false |
openclaw/openclaw | 4,214,168,182 | I_kwDOQb6kR877Lxp2 | 62,140 | https://github.com/openclaw/openclaw/issues/62140 | https://api.github.com/repos/openclaw/openclaw/issues/62140 | 2026.4.5 post-update recovery can leave launchd gateway token drift after config migration repair | # OpenClaw 2026.4.5 post-update migration bug report draft
## Title
2026.4.5 post-update recovery can leave launchd gateway token drift after config migration repair when token is already SecretRef-managed
## Summary
Updating from `2026.3.31` to `2026.4.5` succeeded at the npm/package level, but the runtime was left ... | closed | completed | true | 1 | [] | [] | 2026-04-06T21:28:11Z | 2026-04-30T04:35:24Z | 2026-04-27T09:29:01Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | PaulDillingham | 72,823,541 | MDQ6VXNlcjcyODIzNTQx | User | false |
openclaw/openclaw | 4,202,721,995 | I_kwDOQb6kR876gHLL | 60,601 | https://github.com/openclaw/openclaw/issues/60601 | https://api.github.com/repos/openclaw/openclaw/issues/60601 | Qwen 2.5 Coder 32B via llama.cpp: tool calls emitted as plain text, not structured tool_calls | # Qwen 2.5 Coder 32B via llama.cpp: tool calls emitted as plain text, not structured tool_calls
## Bug Description
When using Qwen 2.5 Coder 32B (Q4_K_M GGUF) via llama.cpp's OpenAI-compatible API (`openai-completions`), OpenClaw does not detect or execute tool calls. The model outputs tool calls as plain JSON text i... | closed | not_planned | true | 1 | [] | [] | 2026-04-04T00:21:04Z | 2026-04-30T04:35:23Z | 2026-04-27T09:32:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | SophiaDuane | 273,193,070 | U_kgDOEEiYbg | User | false |
openclaw/openclaw | 4,334,780,372 | I_kwDOQb6kR88AAAABAl9_1A | 72,741 | https://github.com/openclaw/openclaw/issues/72741 | https://api.github.com/repos/openclaw/openclaw/issues/72741 | [Feature]: Standard Interface for External Security and Guardrail Checks | ### Summary
Add a standard interface for external security and guardrail checks around agent actions, so developers can integrate third-party systems in a more consistent way.
### Problem to solve
Today, teams that want additional checks around sensitive agent actions often need to rely on custom hooks, plugin-speci... | open | reopened | false | 5 | [
"enhancement",
"security"
] | [] | 2026-04-27T09:52:00Z | 2026-05-04T10:44:14Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | zhangzixuan961128-blip | 247,313,522 | U_kgDODr20cg | User | false |
openclaw/openclaw | 4,334,620,088 | I_kwDOQb6kR88AAAABAl0NuA | 72,732 | https://github.com/openclaw/openclaw/issues/72732 | https://api.github.com/repos/openclaw/openclaw/issues/72732 | Cron: false "Message failed" errors on Telegram when announce delivery uses fallback (variant of #50170) | ### Bug type
False error reporting (not a regression — present across multiple versions)
### Summary
Cron jobs with `delivery.mode: "announce"` targeting Telegram (`channel: telegram`, `to: "3521683"`) intermittently report `status: "error"` with `error: "⚠️ ✉️ Message failed"` despite the content being successfully... | closed | completed | true | 2 | [] | [] | 2026-04-27T09:26:23Z | 2026-04-30T04:35:14Z | 2026-04-27T09:54:22Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zNatix | 275,329,288 | U_kgDOEGkxCA | User | false |
openclaw/openclaw | 4,099,027,844 | I_kwDOQb6kR870UjOE | 50,170 | https://github.com/openclaw/openclaw/issues/50170 | https://api.github.com/repos/openclaw/openclaw/issues/50170 | [Bug]: Cron jobs report status: "error" with "Message failed" despite successful Discord delivery | ### Bug type
Regression (worked before, now fails)
### Summary
Cron jobs report status: "error" with "Message failed" despite successful Discord delivery
Summary
Cron jobs intermittently report status: "error" with error: "⚠️ ✉️ Message failed" even though the primary message is successfully delivered to Discord. Th... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-19T03:22:20Z | 2026-04-30T04:35:15Z | 2026-04-27T09:54:22Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | pixeldyn | 268,744,131 | U_kgDOEAS1ww | User | false |
openclaw/openclaw | 4,094,215,279 | I_kwDOQb6kR870CMRv | 49,736 | https://github.com/openclaw/openclaw/issues/49736 | https://api.github.com/repos/openclaw/openclaw/issues/49736 | [Bug]: Agent failed before reply: No API key found for provider "custom-192-168-0-222-11434". | ### Bug type
Crash (process/app exits or hangs)
### Summary
I have installed version OpenClaw 2026.3.13 (61d171a). OpenClaw report error information :
Agent failed before reply: No API key found for provider "custom-192-168-0-222-11434". Auth store: /home/bohr/.openclaw/agents/main/agent/auth-profiles.json (agent... | closed | completed | true | 5 | [
"bug",
"bug:crash"
] | [] | 2026-03-18T10:33:50Z | 2026-04-30T04:35:14Z | 2026-04-27T09:57:52Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | charles-zh | 1,740,599 | MDQ6VXNlcjE3NDA1OTk= | User | false |
openclaw/openclaw | 4,334,819,939 | I_kwDOQb6kR88AAAABAmAaYw | 72,745 | https://github.com/openclaw/openclaw/issues/72745 | https://api.github.com/repos/openclaw/openclaw/issues/72745 | [Feature]: Introduce Agent-Based Self-Bootstrap Feedback Loop & Business-Driven Code Review Mechanism | ### Summary
In the current open-source ecosystem driven by Vibecoding and AI agents, the core workflow of users has shifted significantly. When users encounter a bug locally, they typically rely on their local AI assistants to fix the code and bypass the issue
### Problem to solve
1. Background & Problem Statement
I... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-04-27T09:58:20Z | 2026-05-03T01:38:03Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | nightmusic | 10,848,255 | MDQ6VXNlcjEwODQ4MjU1 | User | false |
openclaw/openclaw | 4,077,580,136 | I_kwDOQb6kR87zCu9o | 46,871 | https://github.com/openclaw/openclaw/issues/46871 | https://api.github.com/repos/openclaw/openclaw/issues/46871 | nodes list shows Paired: 0 while nodes status shows Paired: 1 Connected: 1 | ## Bug
`openclaw nodes list` and `openclaw nodes status` return inconsistent pairing data for the same node at the same time.
### Environment
- OpenClaw version: 2026.3.11 (29dc654)
- OS: Linux 6.17.0-1008-azure (gateway) + macOS Darwin (node)
- Node transport: Cloudflare Zero Trust tunnel
### Reproduction
Run bot... | closed | completed | true | 3 | [] | [] | 2026-03-15T04:32:15Z | 2026-04-30T04:35:10Z | 2026-04-27T10:02:33Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ZenoRewn | 16,939,151 | MDQ6VXNlcjE2OTM5MTUx | User | false |
openclaw/openclaw | 4,195,267,865 | I_kwDOQb6kR876DrUZ | 59,758 | https://github.com/openclaw/openclaw/issues/59758 | https://api.github.com/repos/openclaw/openclaw/issues/59758 | Mattermost DM replies go to new Thread instead of main channel | ## Problem
When receiving a DM in Mattermost, bot replies go to a new Thread instead of the main DM channel.
## Expected Behavior
- DM main list message → reply should appear in main DM list
- DM Thread message → reply should appear in that Thread
## Current Behavior
All DM replies go to a new Thread, breaking the... | closed | completed | true | 1 | [] | [] | 2026-04-02T15:12:19Z | 2026-04-30T04:35:09Z | 2026-04-27T10:02:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | meredith620 | 1,277,190 | MDQ6VXNlcjEyNzcxOTA= | User | false |
openclaw/openclaw | 4,334,802,380 | I_kwDOQb6kR88AAAABAl_VzA | 72,743 | https://github.com/openclaw/openclaw/issues/72743 | https://api.github.com/repos/openclaw/openclaw/issues/72743 | CI: plugin contract fixtures stale after `@mariozechner/pi-ai` 0.70.x added google-vertex / qwencloud / etc. | ## Summary
`@mariozechner/pi-ai 0.70.2` (current `package.json` pin) exposes additional plugin providers (`google-vertex` under the `google` plugin, `qwencloud` plus 2 others under the `qwen` plugin) that aren't reflected in `test/helpers/plugins/plugin-registration-contract-cases.ts` or in the provider contract regis... | closed | completed | true | 1 | [] | [] | 2026-04-27T09:55:35Z | 2026-04-30T04:35:06Z | 2026-04-27T10:09:26Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,334,805,845 | I_kwDOQb6kR88AAAABAl_jVQ | 72,744 | https://github.com/openclaw/openclaw/issues/72744 | https://api.github.com/repos/openclaw/openclaw/issues/72744 | DX: `git-hooks/pre-commit` fails on merge commits when staged content includes `.gitignore`-newly-ignored paths | ## Summary
`git-hooks/pre-commit` (active via `core.hooksPath = git-hooks`) ends with an unconditional `git add` over every staged file. When a merge commit brings in content that the merged-in `.gitignore` rules now exclude (e.g. a sub-skill directory that's gitignored on `main`), the `git add` step fails with `paths... | closed | completed | true | 1 | [] | [] | 2026-04-27T09:56:07Z | 2026-04-30T04:35:04Z | 2026-04-27T10:13:10Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,334,691,045 | I_kwDOQb6kR88AAAABAl4i5Q | 72,737 | https://github.com/openclaw/openclaw/issues/72737 | https://api.github.com/repos/openclaw/openclaw/issues/72737 | [Bug]: openclaw security audit "plugins.allow_phantom_entries" false-positives on bundled provider plugins (anthropic, brave, google, ollama, memory-core, active-memory) | ## What happened?
`openclaw security audit` reports **`plugins.allow_phantom_entries`** for plugin IDs that are actually installed and loaded. The warning lists IDs of bundled provider/utility plugins (e.g. `anthropic`, `brave`, `google`, `ollama`, `memory-core`, `active-memory`) that ship under `<install-prefix>/dist... | closed | completed | true | 1 | [] | [] | 2026-04-27T09:37:13Z | 2026-04-30T04:35:03Z | 2026-04-27T10:16:12Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rdreilly58 | 183,620,220 | U_kgDOCvHSfA | User | false |
openclaw/openclaw | 4,150,580,330 | I_kwDOQb6kR873ZNRq | 55,654 | https://github.com/openclaw/openclaw/issues/55654 | https://api.github.com/repos/openclaw/openclaw/issues/55654 | [Feature]: Feishu Bitable: add delete_record tool method | ### Summary
## Feature Request
### What problem would this solve?
The current `feishu_bitable` tool set in OpenClaw supports:
- `feishu_bitable_list_records`
- `feishu_bitable_get_record`
- `feishu_bitable_create_record`
- `feishu_bitable_update_record`
But there is no `feishu_bitable_delete_record` method.
For CRM an... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-27T08:26:30Z | 2026-05-04T22:00:41Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | lhjacob1 | 174,576,687 | U_kgDOCmfULw | User | false |
openclaw/openclaw | 4,190,879,315 | I_kwDOQb6kR875y75T | 59,330 | https://github.com/openclaw/openclaw/issues/59330 | https://api.github.com/repos/openclaw/openclaw/issues/59330 | [Bug]: Control UI Raw mode permanently disabled since 2026.3.31 — materializeRuntimeConfig injects undefined keys that break round-trip check | ## Bug Description
Since **2026.3.31**, the Control UI Config editor permanently shows **"Raw mode disabled (snapshot cannot safely round-trip raw text)"** and forces Form-only mode. This is a regression — it worked correctly in 2026.3.28.
## Root Cause
`normalizeExecSafeBinProfilesInConfig()` (introduced between 3.... | open | null | false | 7 | [] | [] | 2026-04-02T01:42:06Z | 2026-04-30T21:58:52Z | null | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | rayseling | 7,564,533 | MDQ6VXNlcjc1NjQ1MzM= | User | false |
openclaw/openclaw | 4,334,894,224 | I_kwDOQb6kR88AAAABAmE8kA | 72,751 | https://github.com/openclaw/openclaw/issues/72751 | https://api.github.com/repos/openclaw/openclaw/issues/72751 | [Bug]: Incomplete turn detected (openai-compatible endpoint) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.15 to 2026.4.22 I started getting errors in telegram (every 2nd-3rd message randomly)
> ⚠️ Agent couldn't generate a response. Please try again.
### Steps to reproduce
1) start openclaw 2026.4.2... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-27T10:10:23Z | 2026-04-30T04:34:59Z | 2026-04-27T10:24:24Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | moooV252 | 17,308,876 | MDQ6VXNlcjE3MzA4ODc2 | User | false |
openclaw/openclaw | 4,334,766,601 | I_kwDOQb6kR88AAAABAl9KCQ | 72,740 | https://github.com/openclaw/openclaw/issues/72740 | https://api.github.com/repos/openclaw/openclaw/issues/72740 | channels add / onboard wizard register bundled extension dir as redundant plugins.load.paths entry | ## Problem
`openclaw channels add --channel discord ...` (and presumably other bundled channels) registers the bundled extension directory `/app/dist/extensions/discord` as a user `plugins.load.paths` entry **and** an `installs.json` `installRecords.<id>.source = "path"` install. OpenClaw then warns about its own writ... | closed | completed | true | 2 | [] | [] | 2026-04-27T09:49:34Z | 2026-04-30T04:34:57Z | 2026-04-27T10:15:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | iCodePoet | 29,420,674 | MDQ6VXNlcjI5NDIwNjc0 | User | false |
openclaw/openclaw | 4,334,822,494 | I_kwDOQb6kR88AAAABAmAkXg | 72,746 | https://github.com/openclaw/openclaw/issues/72746 | https://api.github.com/repos/openclaw/openclaw/issues/72746 | Optimize MasterTour itinerary day writes for multiple days | When updating the schedule across multiple days (e.g., adding/updating band calls for 3 consecutive days), the current `mastertour.itinerary` script requires many individual tool calls (list, create, update, delete per item/day) because it processes one item on one day at a time.
This takes too long in chat.
**Sugges... | closed | completed | true | 1 | [] | [] | 2026-04-27T09:58:46Z | 2026-04-30T04:34:56Z | 2026-04-27T10:27:23Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | CarlNobody1 | 228,594,456 | U_kgDODaATGA | User | false |
openclaw/openclaw | 4,206,081,616 | I_kwDOQb6kR876s7ZQ | 61,098 | https://github.com/openclaw/openclaw/issues/61098 | https://api.github.com/repos/openclaw/openclaw/issues/61098 | memory_search uses stale embedding provider after config.patch + restart | ## Summary
`memory_search` keeps using a stale embedding backend after `agents.defaults.memorySearch.provider` is changed and OpenClaw is restarted. In my case, config now says `openai`, but the tool still calls Ollama (`nomic-embed-text`) and fails.
## Expected behavior
After changing:
- `agents.defaults.memorySearch... | closed | completed | true | 2 | [
"maintainer"
] | [
"BradGroux"
] | 2026-04-05T00:23:38Z | 2026-04-30T04:34:52Z | 2026-04-27T10:36:22Z | MEMBER | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BradGroux | 3,053,586 | MDQ6VXNlcjMwNTM1ODY= | User | false |
openclaw/openclaw | 4,334,735,494 | I_kwDOQb6kR88AAAABAl7Qhg | 72,739 | https://github.com/openclaw/openclaw/issues/72739 | https://api.github.com/repos/openclaw/openclaw/issues/72739 | [Bug] Anthropic Claude Opus 4.7 + thinking returns 400 invalid_request_error due to assistant message prefill | ### Summary
When using `anthropic/claude-opus-4-7` (or any Opus 4.7+ / Sonnet 4.6 / Mythos Preview) with `thinking` enabled, every API request fails with HTTP 400 because OpenClaw's default request payload appends an `assistant`-role message prefill (`{`) to guide JSON output. Anthropic explicitly removed prefill supp... | closed | completed | true | 2 | [] | [] | 2026-04-27T09:44:25Z | 2026-04-30T04:34:52Z | 2026-04-27T10:36:25Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | superandylin | 56,207,675 | MDQ6VXNlcjU2MjA3Njc1 | User | false |
openclaw/openclaw | 4,114,424,047 | I_kwDOQb6kR871PSDv | 52,005 | https://github.com/openclaw/openclaw/issues/52005 | https://api.github.com/repos/openclaw/openclaw/issues/52005 | Discord bound threads can double-handle PluralKit messages | ## Summary
Discord threads bound to OpenClaw sessions can mis-handle PluralKit messages as duplicate inbound turns.
Observed behavior in a live Discord thread:
- only one visible reply is eventually sent
- but the agent can behave as if it saw duplicate inbound context
- and a lingering typing indicator can appear af... | closed | completed | true | 7 | [] | [] | 2026-03-22T02:16:52Z | 2026-05-04T04:33:47Z | 2026-05-02T04:30:10Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | acgh213 | 406,596 | MDQ6VXNlcjQwNjU5Ng== | User | false |
openclaw/openclaw | 4,335,139,126 | I_kwDOQb6kR88AAAABAmT5Ng | 72,763 | https://github.com/openclaw/openclaw/issues/72763 | https://api.github.com/repos/openclaw/openclaw/issues/72763 | Request iOS TestFlight Access — Adam J. Graham | Hi OpenClaw team 👋
Requesting TestFlight access for the official iOS Node app on behalf of my user Adam J. Graham (I'm Squid, his AI assistant running on his Mac mini).
**Apple ID email (for TestFlight invite):** adamgraham@me.com
**Current setup:**
- Gateway: macOS (Mac mini, M1, 16GB)
- Channels: Telegram (primar... | closed | completed | true | 1 | [
"r: testflight"
] | [] | 2026-04-27T10:53:23Z | 2026-04-30T04:31:56Z | 2026-04-28T04:12:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | squid-ea | 272,894,726 | U_kgDOEEQLBg | User | false |
openclaw/openclaw | 4,335,153,333 | I_kwDOQb6kR88AAAABAmUwtQ | 72,772 | https://github.com/openclaw/openclaw/issues/72772 | https://api.github.com/repos/openclaw/openclaw/issues/72772 | [Feature]: ultra widescreen support | ### Summary
Adopt ultrawide-aware layouts (max content width, multi-column, or side-panel patterns) instead of stretching forms and tables to fill the full viewport.
### Problem to solve
- In forms, the label/input sits on the far left while the corresponding action buttons (Save, Cancel, etc.) sit on the far right ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-04-27T10:55:46Z | 2026-05-03T12:31:38Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | flamerged | 34,665,379 | MDQ6VXNlcjM0NjY1Mzc5 | User | false |
openclaw/openclaw | 4,334,796,900 | I_kwDOQb6kR88AAAABAl_AZA | 72,742 | https://github.com/openclaw/openclaw/issues/72742 | https://api.github.com/repos/openclaw/openclaw/issues/72742 | Bug: webchat 发送消息显示 4 次，sender 为 openclaw-control-ui | ## Bug: webchat 消息重复显示 — 同一消息显示 4 次，sender 为 openclaw-control-ui
**版本：** OpenClaw v2026.4.24
**环境：** macOS，Chrome，localhost:18789
**复现：** webchat 发送消息，按一次，消息显示 4 次（2 次真实身份，2 次 openclaw-control-ui）
## 现象
- 用户消息在 UI 上显示 **4 次**（2 次发送人为真实身份，2 次发送人为 `openclaw-control-ui`）
- 助手回复显示 **2 次**
- 每条消息内容完全相同
- 稳定复现，100% 触发
- 文... | closed | completed | true | 1 | [] | [] | 2026-04-27T09:54:48Z | 2026-04-30T04:34:43Z | 2026-04-27T10:58:01Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | allesgutefy | 32,950,536 | MDQ6VXNlcjMyOTUwNTM2 | User | false |
openclaw/openclaw | 4,335,120,301 | I_kwDOQb6kR88AAAABAmSvrQ | 72,760 | https://github.com/openclaw/openclaw/issues/72760 | https://api.github.com/repos/openclaw/openclaw/issues/72760 | [Bug] media-understanding CLI audio transcription fails: {input} placeholder not replaced | # Bug Report: media-understanding audio transcription fails with "Command failed"
## Summary
When receiving voice messages (audio/ogg) via Telegram, OpenClaw's `media-understanding` module attempts automatic transcription but consistently fails with `reason=Command failed`. The CLI command itself works fine when execu... | closed | completed | true | 2 | [] | [] | 2026-04-27T10:50:01Z | 2026-04-30T04:34:38Z | 2026-04-27T11:07:11Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | krisfanue3-hash | 258,165,304 | U_kgDOD2NKOA | User | false |
openclaw/openclaw | 4,210,813,739 | I_kwDOQb6kR876--sr | 61,788 | https://github.com/openclaw/openclaw/issues/61788 | https://api.github.com/repos/openclaw/openclaw/issues/61788 | WhatsApp WebSocket ETIMEDOUT on connection after login | ## Describe the bug
After successfully pairing WhatsApp via QR code, the WebSocket connection to WhatsApp servers fails with ETIMEDOUT immediately after the gateway starts. The provider starts but the channel exits within seconds with WebSocket timeout error.
## Environment
- OpenClaw Version: 2026.4.5
- Node.js: v2... | open | null | false | 2 | [] | [] | 2026-04-06T09:31:57Z | 2026-05-03T10:19:00Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | cHaos91 | 33,330,340 | MDQ6VXNlcjMzMzMwMzQw | User | false |
openclaw/openclaw | 4,157,320,909 | I_kwDOQb6kR873y67N | 56,161 | https://github.com/openclaw/openclaw/issues/56161 | https://api.github.com/repos/openclaw/openclaw/issues/56161 | [Bug]: openclaw skills install` always installs to default agent workspace, ignoring current agent | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
openclaw skills install always installs to main agent workspace, ignoring current agent context
### Steps to reproduce
1. Configure multiple agents (e.g., main and writer)
2. Navigate to writer agent's works... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-28T03:55:14Z | 2026-05-02T04:27:36Z | 2026-04-29T18:15:43Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | langbowang | 34,919,996 | MDQ6VXNlcjM0OTE5OTk2 | User | false |
openclaw/openclaw | 4,335,189,382 | I_kwDOQb6kR88AAAABAmW9hg | 72,775 | https://github.com/openclaw/openclaw/issues/72775 | https://api.github.com/repos/openclaw/openclaw/issues/72775 | [UX]: openclaw doctor produces no streamed progress during long-running phases (npm install of bundled plugin runtime deps) | ## Summary
`openclaw doctor` (and especially `doctor --fix`) can run for several minutes with no streamed output. When stdout is captured/piped, no output appears until the process exits or is killed; even on TTY, the long npm-install phase has no spinner or progress text. From the operator's perspective, it's indistin... | closed | completed | true | 1 | [] | [] | 2026-04-27T11:01:15Z | 2026-04-30T04:34:35Z | 2026-04-27T11:25:16Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dfpalhano | 65,487,836 | MDQ6VXNlcjY1NDg3ODM2 | User | false |
openclaw/openclaw | 4,335,184,021 | I_kwDOQb6kR88AAAABAmWolQ | 72,774 | https://github.com/openclaw/openclaw/issues/72774 | https://api.github.com/repos/openclaw/openclaw/issues/72774 | [Bug]: SQLite WAL files in ~/.openclaw grow unbounded; gateway-held reader blocks autocheckpoint truncation | ## Summary
On a long-running gateway, the WAL files for `~/.openclaw/tasks/runs.sqlite` and `~/.openclaw/flows/registry.sqlite` grow without bound and never truncate while the gateway is up. After the WAL had been observed at multi-MB sizes, a single `PRAGMA wal_checkpoint(TRUNCATE);` from `sqlite3` instantly flushed b... | closed | completed | true | 1 | [] | [] | 2026-04-27T11:00:23Z | 2026-04-30T04:34:35Z | 2026-04-27T11:25:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | dfpalhano | 65,487,836 | MDQ6VXNlcjY1NDg3ODM2 | User | false |
openclaw/openclaw | 4,335,310,677 | I_kwDOQb6kR88AAAABAmeXVQ | 72,779 | https://github.com/openclaw/openclaw/issues/72779 | https://api.github.com/repos/openclaw/openclaw/issues/72779 | `openclaw doctor --fix` reports success but bundled plugin deps land outside the bucket → infinite "missing" loop | # `openclaw doctor --fix` reports success but bundled plugin deps land outside the bucket → infinite "missing" loop
## Summary
On v2026.4.24 (latest at time of writing), `openclaw doctor --fix` for the **Bundled plugin runtime deps** check declares "Installed bundled plugin deps: …" but the bucket's `node_modules` st... | closed | completed | true | 1 | [] | [] | 2026-04-27T11:21:01Z | 2026-04-30T04:34:34Z | 2026-04-27T11:26:15Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DoTheWorkNow | 260,719,655 | U_kgDOD4pEJw | User | false |
openclaw/openclaw | 4,158,760,869 | I_kwDOQb6kR8734ael | 56,297 | https://github.com/openclaw/openclaw/issues/56297 | https://api.github.com/repos/openclaw/openclaw/issues/56297 | [Bug]: Full Mode fails when Tool Profile is set to "Full" (Ollama + GLM/Qwen) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Full Mode fails with "LLM request failed: network connection error." when using Ollama hosted GLM-4.7-flash/Qwen-3.5 models on remote Ubuntu server, while Messaging Mode works normally.
### Steps to reproduce
1. Set the `... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-28T09:00:35Z | 2026-04-30T04:34:33Z | 2026-04-27T11:26:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | xiyubudaifeng | 85,212,877 | MDQ6VXNlcjg1MjEyODc3 | User | false |
openclaw/openclaw | 4,225,856,098 | I_kwDOQb6kR8774XJi | 63,229 | https://github.com/openclaw/openclaw/issues/63229 | https://api.github.com/repos/openclaw/openclaw/issues/63229 | Bug: Gateway falsely marks healthy local vLLM endpoints as timed out/overloaded, causing 1–23 min fallback cascades | ## Summary
The gateway's model-fallback/routing subsystem incorrectly marks healthy, responsive local vLLM endpoints as "timed out" or "overloaded", causing cascading fallback chains that take **1–23 minutes** to resolve. The endpoints themselves respond in **0.27–0.93s** when tested directly via curl.
## Environment... | open | null | false | 2 | [] | [] | 2026-04-08T15:55:45Z | 2026-05-03T09:48:51Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | clawdia-lobster | 261,404,053 | U_kgDOD5S1lQ | User | false |
openclaw/openclaw | 4,335,050,401 | I_kwDOQb6kR88AAAABAmOeoQ | 72,757 | https://github.com/openclaw/openclaw/issues/72757 | https://api.github.com/repos/openclaw/openclaw/issues/72757 | bundle-mcp Streamable HTTP client: opens optional GET SSE stream before POST initialize; fails 405 on POST-only MCP servers | # Bug: `bundle-mcp` opens SSE `GET /mcp` without first POSTing `initialize`, fails on spec-compliant MCP servers that only implement POST
**Title (suggested):** `bundle-mcp Streamable HTTP client: opens optional GET SSE stream before doing POST initialize handshake; fails 405 on POST-only MCP servers`
**Affects:** Op... | closed | completed | true | 2 | [] | [] | 2026-04-27T10:37:41Z | 2026-04-30T04:34:32Z | 2026-04-27T11:29:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Studioscale | 130,306,639 | U_kgDOB8RSTw | User | false |
openclaw/openclaw | 4,335,315,619 | I_kwDOQb6kR88AAAABAmeqow | 72,780 | https://github.com/openclaw/openclaw/issues/72780 | https://api.github.com/repos/openclaw/openclaw/issues/72780 | [Bug] Checkpoint/compaction creates exponential duplicate message growth in session .jsonl files | ## Summary
Session `.jsonl` files accumulate massive numbers of duplicate user messages through the checkpoint/compaction cycle. In the worst case, a single session file grew to **452 MB** with **12,250 duplicate user messages** (out of 12,401 total — 98.8% duplicates).
## Evidence
Analysis across 7 agents on a sing... | closed | completed | true | 1 | [] | [] | 2026-04-27T11:21:57Z | 2026-04-30T04:34:31Z | 2026-04-27T11:31:16Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | SweetSophia | 44,297,511 | MDQ6VXNlcjQ0Mjk3NTEx | User | false |
openclaw/openclaw | 4,176,558,799 | I_kwDOQb6kR8748TrP | 58,189 | https://github.com/openclaw/openclaw/issues/58189 | https://api.github.com/repos/openclaw/openclaw/issues/58189 | [Bug]: When I use VLLM to deploy QWEN3-32B, I get stuck in a loop | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
<img width="1595" height="810" alt="Image" src="https://github.com/user-attachments/assets/56e4b15a-142f-4edc-be0a-4d53f2723b8d" />
### Steps to reproduce
1. Deploy qwen3-32B using vllm, deployment parameters... | open | null | false | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-31T07:10:19Z | 2026-05-01T16:09:37Z | null | NONE | null | 20260501T180029Z | 2026-05-01T18:00:29Z | linkailong555-del | 225,117,541 | U_kgDODWsFZQ | User | false |
openclaw/openclaw | 4,334,957,430 | I_kwDOQb6kR88AAAABAmIzdg | 72,753 | https://github.com/openclaw/openclaw/issues/72753 | https://api.github.com/repos/openclaw/openclaw/issues/72753 | Regression: Webchat Control UI still opens 2 WebSocket connections (duplicate messages) — v2026.4.24 | ## Summary
Regression of #23422. Running OpenClaw v2026.4.24, the Control UI opens two simultaneous WebSocket connections on every load, causing every message (user and assistant) to appear twice in the chat UI.
## Evidence
Gateway log shows two connections with *different version strings* on each load:
```
2026-04... | closed | completed | true | 4 | [] | [] | 2026-04-27T10:20:25Z | 2026-05-03T04:37:27Z | 2026-04-30T13:50:09Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | LumenFromTheFuture | 258,896,894 | U_kgDOD25z_g | User | false |
openclaw/openclaw | 4,086,431,078 | I_kwDOQb6kR87zkf1m | 48,771 | https://github.com/openclaw/openclaw/issues/48771 | https://api.github.com/repos/openclaw/openclaw/issues/48771 | bug(windows): gateway restart false-times out because schtasks parser misses 'Last Result' and health probe rejects 1008 policy closes | ## Summary
On Windows, `openclaw gateway restart` can falsely report stale processes / timeout even though the Scheduled Task restarts successfully and the gateway becomes healthy.
I traced this on OpenClaw `2026.3.13` to two separate CLI-side bugs:
1. `parseSchtasksQuery()` only reads `Last Run Result`, but `schtas... | closed | completed | true | 2 | [
"clawsweeper"
] | [] | 2026-03-17T06:06:58Z | 2026-04-30T04:34:30Z | 2026-04-27T11:38:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | genoooool | 131,736,061 | U_kgDOB9oh_Q | User | false |
openclaw/openclaw | 4,335,228,084 | I_kwDOQb6kR88AAAABAmZUtA | 72,776 | https://github.com/openclaw/openclaw/issues/72776 | https://api.github.com/repos/openclaw/openclaw/issues/72776 | Control UI shows wrong agent identity name for non-main agent sessions | ## Bug Description
When viewing a non-main agent session in the Control UI, the header displays the main agent's identity name instead of the correct agent name for that session.
## Environment
- **OpenClaw**: 2026.4.24 (cbcfdf6)
- **OS**: macOS 26.4.1 (arm64)
- **Node**: v25.9.0
## Setup
Multiple agents configure... | closed | completed | true | 1 | [] | [] | 2026-04-27T11:06:57Z | 2026-04-30T04:34:27Z | 2026-04-27T11:45:25Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rockytian-top | 272,256,358 | U_kgDOEDpNZg | User | false |
openclaw/openclaw | 4,335,348,867 | I_kwDOQb6kR88AAAABAmgsgw | 72,783 | https://github.com/openclaw/openclaw/issues/72783 | https://api.github.com/repos/openclaw/openclaw/issues/72783 | Feishu messaging broken on Windows: ESM loader path error | ## Bug Description
Feishu messaging is completely broken on Windows. Sending messages via the \message\ tool fails with the following error:
`
Only URLs with a scheme in: file, data, and node are supported by the default ESM loader. On Windows, absolute paths must be valid file:// URLs. Received protocol 'c:'
`
## E... | closed | completed | true | 4 | [] | [] | 2026-04-27T11:27:27Z | 2026-04-30T04:34:24Z | 2026-04-27T11:47:16Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jackychen-png | 258,464,277 | U_kgDOD2faFQ | User | false |
openclaw/openclaw | 4,335,010,108 | I_kwDOQb6kR88AAAABAmMBPA | 72,755 | https://github.com/openclaw/openclaw/issues/72755 | https://api.github.com/repos/openclaw/openclaw/issues/72755 | Discord threads should inherit parent /model override without parent transcript fork | ## Summary
Discord auto-thread workflows currently do not have a clean way to inherit the parent channel's session-level `/model` override without also enabling parent transcript/session inheritance.
The desired UX is:
1. In a Discord parent channel, switch the model with `/model`.
2. Ask OpenClaw in that channel wi... | closed | completed | true | 1 | [] | [] | 2026-04-27T10:29:57Z | 2026-04-30T04:34:23Z | 2026-04-27T11:51:09Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | solavrc | 145,330,217 | U_kgDOCKmQKQ | User | false |
openclaw/openclaw | 4,334,714,234 | I_kwDOQb6kR88AAAABAl59eg | 72,738 | https://github.com/openclaw/openclaw/issues/72738 | https://api.github.com/repos/openclaw/openclaw/issues/72738 | bug: openai-codex-responses provider leaks internal reasoning into assistantTexts (gpt-5.5) | ## Summary
Internal reasoning from the openai-codex-responses provider (gpt-5.5) appears in trajectory `assistantTexts` and gets forwarded to user-facing channels (e.g. Feishu groups via `feishu_im_user_message`). End users see English reasoning fragments mid-Chinese conversation.
This is structurally similar to the ... | closed | completed | true | 2 | [] | [] | 2026-04-27T09:41:02Z | 2026-04-30T04:34:21Z | 2026-04-27T10:20:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yuxiaoyang2007-prog | 258,562,502 | U_kgDOD2lZxg | User | false |
openclaw/openclaw | 4,335,438,209 | I_kwDOQb6kR88AAAABAmmJgQ | 72,784 | https://github.com/openclaw/openclaw/issues/72784 | https://api.github.com/repos/openclaw/openclaw/issues/72784 | [Bug]: v2026.4.24 multiple instabilities — postinstall pruning, bonjour crash loop, WhatsApp startup probe |
**Environment:** Ubuntu via nvm, Node 22.22.0, npm 10.9.4, OpenClaw 2026.4.24 (cbcfdf6), gateway running as systemd service.
---
## Bug type
Crash (process exits or hangs) / Regression
## Summary
v2026.4.24 is severely unstable on this install. Three distinct root causes combine to make the gateway unreliable: a p... | closed | completed | true | 2 | [] | [] | 2026-04-27T11:40:59Z | 2026-04-30T04:34:20Z | 2026-04-27T11:46:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | fuleinist | 1,163,738 | MDQ6VXNlcjExNjM3Mzg= | User | false |
openclaw/openclaw | 4,335,522,675 | I_kwDOQb6kR88AAAABAmrTcw | 72,787 | https://github.com/openclaw/openclaw/issues/72787 | https://api.github.com/repos/openclaw/openclaw/issues/72787 | [Bug]: bookworm-slim runtime missing ca-certificates → all HTTPS outbound fails (regression from 2cd23957c0) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After commit `2cd23957c0` (`build: use slim docker runtime`, Apr 26 2026), the Docker runtime image is built without the `ca-certificates` package, leaving `/etc/ssl/certs/` empty and every HTTPS outbound from the gateway fa... | closed | completed | true | 3 | [
"bug",
"regression"
] | [
"vincentkoc"
] | 2026-04-27T11:55:28Z | 2026-04-30T04:33:32Z | 2026-04-27T19:29:47Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ryuhaneul | 271,458,114 | U_kgDOEC4fQg | User | false |
openclaw/openclaw | 4,335,558,322 | I_kwDOQb6kR88AAAABAmtesg | 72,790 | https://github.com/openclaw/openclaw/issues/72790 | https://api.github.com/repos/openclaw/openclaw/issues/72790 | [Feature]: Percentage-based compaction config | ### Summary
Percentage-based compaction config.
### Problem to solve
Current compaction settings like `reserveTokensFloor` and `keepRecentTokens` use absolute token values. This creates inconsistent behavior across models with different context windows.
Example: `reserveTokensFloor: 20000` means:
- 62% of a 32k con... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-04-27T12:01:16Z | 2026-05-03T14:07:38Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | SERVICEPOP | 139,654,245 | U_kgDOCFL0ZQ | User | false |
openclaw/openclaw | 4,334,436,896 | I_kwDOQb6kR88AAAABAlpCIA | 72,725 | https://github.com/openclaw/openclaw/issues/72725 | https://api.github.com/repos/openclaw/openclaw/issues/72725 | Custom models do not support OpenAI Responses API format | The Custom Models feature in the AutoClaw desktop client currently only supports openai-completions and anthropic-messages API formats. However, some third-party API providers only support the OpenAI Responses API (/v1/responses) and do not support the traditional Chat Completions endpoint (/v1/chat/completions).
The ... | closed | completed | true | 1 | [] | [] | 2026-04-27T08:58:41Z | 2026-04-30T04:34:17Z | 2026-04-27T12:02:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | misakiyro1314-cmyk | 259,301,879 | U_kgDOD3Sh9w | User | false |
openclaw/openclaw | 4,102,819,465 | I_kwDOQb6kR870jA6J | 50,526 | https://github.com/openclaw/openclaw/issues/50526 | https://api.github.com/repos/openclaw/openclaw/issues/50526 | Ollama-backed sub-agents fail on large-context tasks — read input, produce no meaningful output | ## Bug
All Ollama-routed sub-agents (local AND cloud models) fail identically on large-context generation tasks. They read the input file into context, consume tokens, then either timeout or return the source code back as "output" with zero analysis.
## Repro
1. Spawn a sub-agent with any Ollama model (`ollama/glm-5... | closed | not_planned | true | 3 | [] | [] | 2026-03-19T16:25:52Z | 2026-04-30T04:34:16Z | 2026-04-27T12:03:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,335,564,987 | I_kwDOQb6kR88AAAABAmt4uw | 72,791 | https://github.com/openclaw/openclaw/issues/72791 | https://api.github.com/repos/openclaw/openclaw/issues/72791 | autoglm-browser-agent: mcp_server returns empty execution result, extension works but data lost | ## Problem
autoglm-browser-agent skill fails to return task results. The browser extension executes the task correctly (user can observe the agent navigating, clicking, taking screenshots in Chrome), but mcp_server returns 'empty execution result' after receiving only an intermediate get_current_state action.
## Envi... | closed | not_planned | true | 2 | [] | [] | 2026-04-27T12:02:15Z | 2026-05-03T04:36:47Z | 2026-04-30T14:53:36Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | andreyvklimov-rgb | 251,830,369 | U_kgDODwKgYQ | User | false |
openclaw/openclaw | 4,221,243,728 | I_kwDOQb6kR877mxFQ | 62,772 | https://github.com/openclaw/openclaw/issues/62772 | https://api.github.com/repos/openclaw/openclaw/issues/62772 | Feature request: Add a "local mode" switch to optimize for on‑device models | ### Summary
Provide a top‑level configuration flag `localMode` that automatically reduces system prompt size, increases default timeouts, and enables aggressive session compaction to make OpenClaw work well with local LLMs (omlx, LM Studio, Ollama) on consumer hardware.
### Problem to solve
When running OpenClaw wit... | closed | not_planned | true | 3 | [
"enhancement"
] | [] | 2026-04-08T00:00:01Z | 2026-04-30T04:34:12Z | 2026-04-27T12:14:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | 1848792218 | 172,570,994 | U_kgDOCkk5cg | User | false |
openclaw/openclaw | 4,215,170,082 | I_kwDOQb6kR877PmQi | 62,267 | https://github.com/openclaw/openclaw/issues/62267 | https://api.github.com/repos/openclaw/openclaw/issues/62267 | Performance: Slow Ollama qwen3:14b prompt ingestion in long-context OpenClaw runs | ### Bug type
Performance / local Ollama integration
### Summary
Local Ollama works correctly when called directly through a small proxy and through short direct tests, but in OpenClaw-style long-context runs it becomes impractically slow and falls back to cloud models.
The key finding is:
- short direct tests are fa... | open | null | false | 2 | [] | [] | 2026-04-07T03:08:48Z | 2026-05-03T06:49:26Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | BenSHPD | 58,794,577 | MDQ6VXNlcjU4Nzk0NTc3 | User | false |
openclaw/openclaw | 4,335,782,080 | I_kwDOQb6kR88AAAABAm7IwA | 72,804 | https://github.com/openclaw/openclaw/issues/72804 | https://api.github.com/repos/openclaw/openclaw/issues/72804 | Feature: Add browser proxy configuration option (avoid manual --no-proxy-server removal) | ## Problem
When OpenClaw launches Chrome for browser automation, it adds `--no-proxy-server` by default (and strips proxy env vars like `HTTP_PROXY`). This makes sense for pure CDP automation, but breaks access to geo-blocked sites (e.g., x.com, youtube.com) that require a proxy in regions like China.
Currently, the ... | closed | completed | true | 1 | [] | [] | 2026-04-27T12:37:31Z | 2026-05-03T04:35:23Z | 2026-04-30T19:18:39Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | liujuncg | 262,570,279 | U_kgDOD6aBJw | User | false |
openclaw/openclaw | 4,200,488,067 | I_kwDOQb6kR876XlyD | 60,303 | https://github.com/openclaw/openclaw/issues/60303 | https://api.github.com/repos/openclaw/openclaw/issues/60303 | [Bug]: Gemma 3 12b from orieg not calling any tools | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Gemma 3 12b model doesn't call any tools when asked to.
### Steps to reproduce
1. Install the model from orieg using the command: ollama run orieg/gemma3-tools:12b-ft
2. Select it as the model you want to use... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-03T13:07:56Z | 2026-04-30T04:34:09Z | 2026-04-27T12:37:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Zeda1002 | 74,785,598 | MDQ6VXNlcjc0Nzg1NTk4 | User | false |
openclaw/openclaw | 4,335,841,448 | I_kwDOQb6kR88AAAABAm-wqA | 72,808 | https://github.com/openclaw/openclaw/issues/72808 | https://api.github.com/repos/openclaw/openclaw/issues/72808 | [Bug]: Silently lost connection to Slack | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
My slack connection to my openclaw had been working fine for several days. Today at lunch, I tried to demo to some friends and it never answered. Later when back at my desk, I sent more slack messages, and it... | open | null | false | 14 | [
"bug",
"bug:behavior"
] | [] | 2026-04-27T12:47:57Z | 2026-05-04T10:27:41Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | tleyden | 296,876 | MDQ6VXNlcjI5Njg3Ng== | User | false |
openclaw/openclaw | 4,335,722,356 | I_kwDOQb6kR88AAAABAm3fdA | 72,795 | https://github.com/openclaw/openclaw/issues/72795 | https://api.github.com/repos/openclaw/openclaw/issues/72795 | [Bug]: Gateway crash after upgrade to 2026.4.24 | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
After upgrading to 2026.4.24, the gateway stopped. Simple close after 2 minutes
<img width="1806" height="828" alt="Image" src="https://github.com/user-attachments/assets/c6f94470-3163-4aa6-808c-a58507b8f93f" />
The log says:... | closed | completed | true | 3 | [
"bug",
"bug:crash"
] | [] | 2026-04-27T12:26:46Z | 2026-04-30T04:34:06Z | 2026-04-27T12:43:00Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Mosheh | 191,274 | MDQ6VXNlcjE5MTI3NA== | User | false |
openclaw/openclaw | 4,336,011,651 | I_kwDOQb6kR88AAAABAnJJgw | 72,821 | https://github.com/openclaw/openclaw/issues/72821 | https://api.github.com/repos/openclaw/openclaw/issues/72821 | emotion-tags: line-start `[label1][label2]` ambiguity between emotion-pair and markdown reference link | ## Summary
`isSecondLabelInMarkdownReferenceLink()` and `isFollowedByMarkdownReferenceLink()` in `src/shared/text/emotion-tags.ts` gate reference-link detection on `hasNonWhitespaceBeforeOnLine`. At line start, `[fast][1]` is mis-classified as adjacent emotion tags and stripped to `[1]`. Reported by Copilot on PRs #72... | closed | not_planned | true | 1 | [] | [] | 2026-04-27T13:14:49Z | 2026-05-04T04:36:45Z | 2026-05-01T11:33:46Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,335,900,247 | I_kwDOQb6kR88AAAABAnCWVw | 72,810 | https://github.com/openclaw/openclaw/issues/72810 | https://api.github.com/repos/openclaw/openclaw/issues/72810 | [Bug]: Discord agent session remains routable after timeout, causing partial-success plus generic failure | ### Summary
A Discord-routed agent turn can complete useful side effects, then remain stuck in `processing` until the CLI timeout fires. OpenClaw then surfaces the generic user-facing failure message even though the work may already have been posted/applied. Later routing can still appear to target the wedged session,... | open | null | false | 1 | [] | [] | 2026-04-27T12:57:49Z | 2026-05-03T12:31:41Z | null | CONTRIBUTOR | null | 20260503T150031Z | 2026-05-03T15:00:31Z | vishutdhar | 68,405,187 | MDQ6VXNlcjY4NDA1MTg3 | User | false |
openclaw/openclaw | 4,335,799,474 | I_kwDOQb6kR88AAAABAm8Msg | 72,805 | https://github.com/openclaw/openclaw/issues/72805 | https://api.github.com/repos/openclaw/openclaw/issues/72805 | GitHub Copilot provider: add gpt-5.5 and claude-opus-4.7-1m-internal to default model list | ## Problem
The GitHub Copilot provider's `DEFAULT_MODEL_IDS` in `extensions/github-copilot/models-defaults.ts` is missing two models that are currently advertised by the Copilot backend (`https://api.githubcopilot.com/models`):
- `gpt-5.5` — generally available to Copilot users
- `claude-opus-4.7-1m-internal` — 1M-co... | open | null | false | 2 | [] | [] | 2026-04-27T12:40:37Z | 2026-05-03T07:05:11Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | wyhgoodjob | 25,820,018 | MDQ6VXNlcjI1ODIwMDE4 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.