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 | 4,263,205,141 | I_kwDOQb6kR87-G1kV | 66,646 | https://github.com/openclaw/openclaw/issues/66646 | https://api.github.com/repos/openclaw/openclaw/issues/66646 | [Bug]: Session file lock errors cascade through model fallback chain | ## Description
When a session file is locked (e.g., concurrent cron + dispatch hitting the same agent), the gateway treats the lock timeout as a model failure and cascades through the entire fallback chain. Each fallback attempt hits the same lock, wasting 10s × N models before failing entirely.
## Expected behavior
... | open | null | false | 5 | [] | [] | 2026-04-14T15:44:01Z | 2026-05-03T07:44:44Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | pescaohq-web | 257,331,994 | U_kgDOD1aTGg | User | false |
openclaw/openclaw | 4,263,219,693 | I_kwDOQb6kR87-G5Ht | 66,647 | https://github.com/openclaw/openclaw/issues/66647 | https://api.github.com/repos/openclaw/openclaw/issues/66647 | [Bug]: Telegram: Binary file content injected into prompt via msg.caption causes token explosion | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When users send binary files (e.g., .mobi e-books) via Telegram, the file's binary content is incorrectly injected into the LLM prompt through msg.caption, causing massive token explosions and context overflow errors.
### S... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-14T15:46:24Z | 2026-04-17T04:22:43Z | 2026-04-14T16:44:52Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | hi-o-id | 114,895,964 | U_kgDOBtksXA | User | false |
openclaw/openclaw | 4,263,230,176 | I_kwDOQb6kR87-G7rg | 66,648 | https://github.com/openclaw/openclaw/issues/66648 | https://api.github.com/repos/openclaw/openclaw/issues/66648 | [Bug] System-level "Exec completed" notifications leak into unrelated webchat sessions | # [Bug] System-level "Exec completed/failed" notifications leak into unrelated webchat sessions
## Summary
System-generated notifications for background exec completions ("Exec completed (session-id, code N)", "Exec failed (session, signal SIGTERM)") are appearing in webchat sessions that did not initiate the exec. T... | closed | completed | true | 3 | [] | [] | 2026-04-14T15:48:01Z | 2026-04-28T04:54:48Z | 2026-04-25T17:49:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Valkster70 | 129,894,487 | U_kgDOB74IVw | User | false |
openclaw/openclaw | 4,263,325,705 | I_kwDOQb6kR87-HTAJ | 66,654 | https://github.com/openclaw/openclaw/issues/66654 | https://api.github.com/repos/openclaw/openclaw/issues/66654 | [Feature]: Time-Travel Editing: Built-in Git System for Safe Rollbacks & Code Versioning | ### Summary
Introduce an integrated Git-like version control system within OpenCLAW to track, review, and revert changes made by the AI.
### Problem to solve
Currently, when OpenCLAW modifies files—especially code—there is no native mechanism to track changes or revert to a previous state. This creates friction and ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-14T16:04:24Z | 2026-04-28T04:54:47Z | 2026-04-25T17:50:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexandre-leng | 72,550,137 | MDQ6VXNlcjcyNTUwMTM3 | User | false |
openclaw/openclaw | 4,263,338,711 | I_kwDOQb6kR87-HWLX | 66,657 | https://github.com/openclaw/openclaw/issues/66657 | https://api.github.com/repos/openclaw/openclaw/issues/66657 | TypeError: Cannot read properties of undefined (reading 'trim') on Feishu group messages with empty content | ## Description
When a user sends a Feishu group message that contains only an @mention (no text body), or certain other message types with empty/undefined content, the bot fails to dispatch the message and throws a `TypeError`.
## Error Log
```
feishu[<account>]: failed to dispatch message: TypeError: Cannot read pr... | open | null | false | 2 | [] | [] | 2026-04-14T16:06:39Z | 2026-05-03T08:25:04Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | rZeR0y | 262,516,383 | U_kgDOD6Wunw | User | false |
openclaw/openclaw | 4,263,346,140 | I_kwDOQb6kR87-HX_c | 66,659 | https://github.com/openclaw/openclaw/issues/66659 | https://api.github.com/repos/openclaw/openclaw/issues/66659 | heartbeat.isolatedSession: true carries forward session entry metadata across ticks — appears cumulative despite new sessionId | **Version:** 2026.4.10 (commit `44e5b62`)
## Summary
`heartbeat.isolatedSession: true` is documented to create a fresh isolated session per heartbeat tick. In practice, while a new `sessionId` is minted each tick, the stable `:heartbeat` session entry carries forward cumulative token metadata (`inputTokens`, `outputT... | closed | completed | true | 1 | [] | [] | 2026-04-14T16:07:52Z | 2026-04-30T04:51:44Z | 2026-04-25T05:17:11Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | vebutton | 12,551,601 | MDQ6VXNlcjEyNTUxNjAx | User | false |
openclaw/openclaw | 4,263,365,989 | I_kwDOQb6kR87-Hc1l | 66,661 | https://github.com/openclaw/openclaw/issues/66661 | https://api.github.com/repos/openclaw/openclaw/issues/66661 | [Bug]: `api.registerHttpRoute()` does not register routes in external plugins - returns 404 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When calling api.registerHttpRoute() in an external plugin's register() function, the HTTP route is not actually registered with the HTTP server. Requests to the route return 404 Not Found, even though the regi... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-14T16:12:01Z | 2026-04-27T04:55:17Z | 2026-04-24T21:45:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | WhiteGiverMa | 152,406,589 | U_kgDOCRWKPQ | User | false |
openclaw/openclaw | 4,263,394,160 | I_kwDOQb6kR87-Hjtw | 66,666 | https://github.com/openclaw/openclaw/issues/66666 | https://api.github.com/repos/openclaw/openclaw/issues/66666 | Feature Request: hook for reply interception | # Feature Request: `before_delivery` hook for reply interception
## Problem
Currently, OpenClaw extensions can only access **user input** via the `before_agent_reply` hook, but cannot inspect or intercept the **assistant's generated reply** before it's delivered to the user.
This makes it impossible to implement mec... | closed | completed | true | 1 | [] | [] | 2026-04-14T16:17:45Z | 2026-04-28T04:54:47Z | 2026-04-25T17:50:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | gan1070485335-ctrl | 251,839,250 | U_kgDODwLDEg | User | false |
openclaw/openclaw | 4,263,395,155 | I_kwDOQb6kR87-Hj9T | 66,667 | https://github.com/openclaw/openclaw/issues/66667 | https://api.github.com/repos/openclaw/openclaw/issues/66667 | Feature Request: before_delivery hook for reply interception | # Feature Request: `before_delivery` hook for reply interception
## Problem
Currently, OpenClaw extensions can only access **user input** via the `before_agent_reply` hook, but cannot inspect or intercept the **assistant's generated reply** before it's delivered to the user.
This makes it impossible to implement mec... | closed | completed | true | 1 | [] | [] | 2026-04-14T16:17:57Z | 2026-04-28T04:54:46Z | 2026-04-25T17:50:25Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | gan1070485335-ctrl | 251,839,250 | U_kgDODwLDEg | User | false |
openclaw/openclaw | 4,263,410,944 | I_kwDOQb6kR87-Hn0A | 66,668 | https://github.com/openclaw/openclaw/issues/66668 | https://api.github.com/repos/openclaw/openclaw/issues/66668 | [Bug]: plugins uninstall does not remove extension files for --link or manually installed plugins | ## Summary
`openclaw plugins uninstall <id>` does not remove the extension directory files when the plugin was installed via `--link` or manually copied to the extensions folder.
## Steps to reproduce
1. Install a plugin using `--link` or manually copy to `~/.openclaw/extensions/<id>`:
```bash
openclaw plugins i... | closed | not_planned | true | 2 | [] | [] | 2026-04-14T16:21:08Z | 2026-04-28T04:36:03Z | 2026-04-26T04:00:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | definitiontv | 2,871,261 | MDQ6VXNlcjI4NzEyNjE= | User | false |
openclaw/openclaw | 4,263,415,942 | I_kwDOQb6kR87-HpCG | 66,669 | https://github.com/openclaw/openclaw/issues/66669 | https://api.github.com/repos/openclaw/openclaw/issues/66669 | Control UI (webchat) does not render user-sent image attachments in chat interface | ## Bug Description
When sending image attachments via the Control UI (webchat) chat interface, the images are not rendered/displayed in the chat history. The user message appears as an empty/blank bubble instead of showing the uploaded image.
## Environment
- OpenClaw version: latest (npm install)
- Interface: Contr... | closed | completed | true | 1 | [] | [] | 2026-04-14T16:22:05Z | 2026-04-19T04:23:04Z | 2026-04-16T14:37:58Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | woai36D | 269,974,840 | U_kgDOEBd9OA | User | false |
openclaw/openclaw | 4,263,433,084 | I_kwDOQb6kR87-HtN8 | 66,670 | https://github.com/openclaw/openclaw/issues/66670 | https://api.github.com/repos/openclaw/openclaw/issues/66670 | [Feature] sessions_spawn should populate agent_id in task_runs SQLite | ## Description
When a task is spawned via `sessions_spawn` from a parent agent session (e.g., main), the `agent_id` field in `~/.openclaw/tasks/runs.sqlite` is left empty/null, even when the task label and context clearly identify a target agent (e.g., `pepper-mc-lotE3`).
This makes it impossible for plugins reading ... | open | null | false | 2 | [] | [] | 2026-04-14T16:25:05Z | 2026-05-02T20:42:55Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | Neomail2 | 120,675,713 | U_kgDOBzFdgQ | User | false |
openclaw/openclaw | 4,263,465,074 | I_kwDOQb6kR87-H1By | 66,672 | https://github.com/openclaw/openclaw/issues/66672 | https://api.github.com/repos/openclaw/openclaw/issues/66672 | [Feature]: Safe AI Workflow Modes: Ask, Plan, and Modify for Full Control & Transparency | ### Summary
Introduce three distinct modes in OpenCLAW: Ask (no changes), Plan (preview actions), and Modify (execute changes).
### Problem to solve
Currently, OpenCLAW tends to act directly on files and systems, which creates a gap between thinking and execution. In many real-world scenarios, users first need to ex... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-04-14T16:31:37Z | 2026-05-02T20:42:56Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | alexandre-leng | 72,550,137 | MDQ6VXNlcjcyNTUwMTM3 | User | false |
openclaw/openclaw | 4,263,488,488 | I_kwDOQb6kR87-H6vo | 66,674 | https://github.com/openclaw/openclaw/issues/66674 | https://api.github.com/repos/openclaw/openclaw/issues/66674 | [Bug]: openai-codex/gpt-5.4 direct CLI infer returns HTML rawError but surfaces as "DNS lookup failed" on 2026.4.14 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After updating to OpenClaw 2026.4.14, direct CLI inference on `openai-codex/gpt-5.4` fails even for a tiny prompt. The raw failure returned by the provider/runtime is HTML, but OpenClaw surfaces it as `LLM request failed: DN... | closed | completed | true | 8 | [
"bug"
] | [] | 2026-04-14T16:36:08Z | 2026-04-19T04:23:16Z | 2026-04-16T11:30:07Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | firstyanko | 276,060,677 | U_kgDOEHRaBQ | User | false |
openclaw/openclaw | 4,263,516,914 | I_kwDOQb6kR87-IBry | 66,675 | https://github.com/openclaw/openclaw/issues/66675 | https://api.github.com/repos/openclaw/openclaw/issues/66675 | openclaw gateway restart can return false failure after healthy systemd restart | ## Summary
`openclaw gateway restart` can return a false failure on Linux/systemd even when the gateway restarts successfully and is immediately healthy afterward.
## Environment
- OpenClaw: `2026.4.12` and still reproducible after updating to `2026.4.14`
- Platform: Linux + systemd user service
- Gateway bind: loop... | open | null | false | 1 | [] | [] | 2026-04-14T16:42:04Z | 2026-05-02T22:58:43Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | pfrederiksen | 908,172 | MDQ6VXNlcjkwODE3Mg== | User | false |
openclaw/openclaw | 4,263,518,634 | I_kwDOQb6kR87-ICGq | 66,676 | https://github.com/openclaw/openclaw/issues/66676 | https://api.github.com/repos/openclaw/openclaw/issues/66676 | Telegram inline button callback feedback: answerCallbackQuery not called | ## Summary
When a user clicks an inline button in Telegram, the bot does not call answerCallbackQuery to dismiss the loading state. This leaves the spinning indicator visible for up to 1 minute, creating a poor user experience.
## Current behavior
- User clicks inline button
- Telegram sends callback_query to bot
- Op... | closed | completed | true | 0 | [] | [] | 2026-04-14T16:42:28Z | 2026-04-17T04:22:44Z | 2026-04-14T16:44:45Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Erxpria1 | 210,373,588 | U_kgDODIoL1A | User | false |
openclaw/openclaw | 4,263,537,071 | I_kwDOQb6kR87-IGmv | 66,677 | https://github.com/openclaw/openclaw/issues/66677 | https://api.github.com/repos/openclaw/openclaw/issues/66677 | [Bug]: Select channel (QuickStart) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
(OpenClaw 2026.4.14 (323493f))!
Select channel (QuickStart)
Skip for now
TypeError: Cannot read properties of undefined (reading 'trim')
### Steps to reproduce
(OpenClaw 2026.4.14 (323493f))!
Select channel (QuickStart)
... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-04-14T16:46:16Z | 2026-04-19T04:23:25Z | 2026-04-16T06:25:49Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | huohuohahi | 156,270,756 | U_kgDOCVCApA | User | false |
openclaw/openclaw | 4,263,543,615 | I_kwDOQb6kR87-IIM_ | 66,679 | https://github.com/openclaw/openclaw/issues/66679 | https://api.github.com/repos/openclaw/openclaw/issues/66679 | [Bug]: On OpenClaw 2026.4.14 (a88c6f0), Telegram and Slack both respond conversationally, but chat-triggered tools/actions often do not execute. | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On OpenClaw 2026.4.14, Telegram and Slack both respond normally in chat, but requests that should trigger tools/actions often remain purely conversational and do not actually execute.
This does not seem to be a basic auth, ... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-14T16:47:26Z | 2026-04-28T04:54:45Z | 2026-04-25T17:50:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Aredes228 | 68,281,093 | MDQ6VXNlcjY4MjgxMDkz | User | false |
openclaw/openclaw | 4,263,561,835 | I_kwDOQb6kR87-IMpr | 66,681 | https://github.com/openclaw/openclaw/issues/66681 | https://api.github.com/repos/openclaw/openclaw/issues/66681 | Health-monitor causes gateway crash: TypeError Cannot read properties of undefined (reading 'info') | ## Summary
When the health-monitor attempts to restart a channel that has stopped, it crashes the gateway with an unhandled promise rejection.
## Expected Behavior
The health-monitor should handle missing log objects gracefully and not crash the gateway.
## Actual Behavior
Gateway crashes with:
```
[openclaw] Unha... | closed | not_planned | true | 1 | [] | [] | 2026-04-14T16:51:10Z | 2026-04-28T04:54:44Z | 2026-04-25T17:50:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | WhiteGiverMa | 152,406,589 | U_kgDOCRWKPQ | User | false |
openclaw/openclaw | 4,263,563,228 | I_kwDOQb6kR87-IM_c | 66,682 | https://github.com/openclaw/openclaw/issues/66682 | https://api.github.com/repos/openclaw/openclaw/issues/66682 | Telegram: commands.native not registering setMyCommands at startup (2026.4.14) | ## Description
On OpenClaw 2026.4.14 (macOS, LaunchAgent, grammY long-polling), `commands.native: "auto"` (default) and `commands.native: true` (explicit) both fail to register native Telegram bot commands via `setMyCommands` at gateway startup.
## Steps to Reproduce
1. Fresh OpenClaw 2026.4.14 install with Telegram... | closed | completed | true | 3 | [] | [] | 2026-04-14T16:51:29Z | 2026-04-17T04:21:58Z | 2026-04-15T02:32:26Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | romarconcepts | 260,996,509 | U_kgDOD459nQ | User | false |
openclaw/openclaw | 4,263,572,995 | I_kwDOQb6kR87-IPYD | 66,683 | https://github.com/openclaw/openclaw/issues/66683 | https://api.github.com/repos/openclaw/openclaw/issues/66683 | Subagent completions can resolve before parent reply delivery succeeds | When a subagent completion is surfaced through the parent agent's normal reply path, the run can be marked resolved before the user-facing reply has actually been delivered.
In practice, reply payload construction is currently close enough to resolution in some parent followup paths, but "reply was built" and "reply w... | closed | completed | true | 1 | [] | [] | 2026-04-14T16:53:45Z | 2026-04-30T04:40:27Z | 2026-04-26T05:06:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | bill492 | 257,063,863 | U_kgDOD1J7tw | User | false |
openclaw/openclaw | 4,263,578,364 | I_kwDOQb6kR87-IQr8 | 66,684 | https://github.com/openclaw/openclaw/issues/66684 | https://api.github.com/repos/openclaw/openclaw/issues/66684 | Security scan findings: remote script execution and credential-pattern hits | Hi, I ran a static ClawGuard scan on this repository and it flagged a few high-risk patterns worth reviewing.
Examples from the scan:
- `AGENTS.md:286` — ClawHavoc-style OpenClaw state targeting
- `CHANGELOG.md:3955` — credential-harvesting style pattern
- `Dockerfile:50` — remote script execution via curl/wget pipe
-... | open | null | false | 1 | [] | [] | 2026-04-14T16:54:54Z | 2026-05-02T21:49:50Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | ClawSafe | 274,975,746 | U_kgDOEGPMAg | User | false |
openclaw/openclaw | 4,263,585,460 | I_kwDOQb6kR87-ISa0 | 66,686 | https://github.com/openclaw/openclaw/issues/66686 | https://api.github.com/repos/openclaw/openclaw/issues/66686 | [Bug] Memory search over-generalization causes incorrect handling of system state verification questions | ## Bug Description
**System-level bug**: Over-generalized memory search rules cause incorrect handling of system state verification questions.
## Steps to Reproduce
1. Ask the assistant: "Do I have CTO and CFO configured?" (or in Chinese: "目前我有一个cto和cfo对吧")
2. **Expected behavior**: Direct system verification via `ope... | open | null | false | 1 | [] | [] | 2026-04-14T16:56:30Z | 2026-05-02T21:25:36Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | huangchaochun | 225,994,021 | U_kgDODXhlJQ | User | false |
openclaw/openclaw | 4,263,598,328 | I_kwDOQb6kR87-IVj4 | 66,688 | https://github.com/openclaw/openclaw/issues/66688 | https://api.github.com/repos/openclaw/openclaw/issues/66688 | [Bug]: `memory index` fails with `Unknown memory embedding provider: ollama` | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
`openclaw memory index --agent robotik --force` fails when `agents.defaults.memorySearch.provider` is set to `"ollama"`.
Error:
```text
Memory index failed (robotik): Unknown memory embedding provider: ollama
```
### Steps... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-14T16:59:17Z | 2026-04-28T04:54:44Z | 2026-04-25T17:50:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ojanovsky | 17,985,977 | MDQ6VXNlcjE3OTg1OTc3 | User | false |
openclaw/openclaw | 4,263,614,722 | I_kwDOQb6kR87-IZkC | 66,690 | https://github.com/openclaw/openclaw/issues/66690 | https://api.github.com/repos/openclaw/openclaw/issues/66690 | [Bug]: Sandboxed agent cannot reach browser CDP — 127.0.0.1 hardcoded in ensureSandboxBrowser | ### Summary
When `agents.defaults.sandbox.mode: "all"`, the sandboxed agent container cannot reach the browser's CDP endpoint. `ensureSandboxBrowser` maps the CDP port to the **host's** `127.0.0.1`, which is the host loopback. From inside a Docker-sandboxed agent container, `127.0.0.1` resolves to the **container's ow... | closed | completed | true | 1 | [] | [] | 2026-04-14T17:02:57Z | 2026-04-17T04:22:36Z | 2026-04-14T20:50:30Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | kpiyush88 | 37,152,866 | MDQ6VXNlcjM3MTUyODY2 | User | false |
openclaw/openclaw | 4,263,626,818 | I_kwDOQb6kR87-IchC | 66,691 | https://github.com/openclaw/openclaw/issues/66691 | https://api.github.com/repos/openclaw/openclaw/issues/66691 | v2026.4.14 regression: models.providers.*.request.allowPrivateNetwork no longer applied for audio transcription | ## Summary
After upgrading to v2026.4.14, voice message transcription via OpenAI-compatible self-hosted STT endpoints (e.g. Parakeet on a LAN IP) fails with an SSRF block, even when `models.providers.<id>.request.allowPrivateNetwork: true` is set in config. This worked correctly in v2026.4.12 and v2026.4.13.
## Error... | closed | completed | true | 1 | [] | [] | 2026-04-14T17:05:31Z | 2026-04-17T04:21:45Z | 2026-04-15T02:35:38Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jhsmith409 | 198,252,919 | U_kgDOC9EZdw | User | false |
openclaw/openclaw | 4,263,654,835 | I_kwDOQb6kR87-IjWz | 66,693 | https://github.com/openclaw/openclaw/issues/66693 | https://api.github.com/repos/openclaw/openclaw/issues/66693 | [Bug]: While onboarding:TypeError: Cannot read properties of undefined (reading 'trim') | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Summary
OpenClaw 2026.4.14 (323493f) crashes during onboarding immediately after “Select channel (QuickStart)” with:
TypeError: Cannot read properties of undefined (reading 'trim')
Environment
- OpenClaw: 2026.4.14 (3234... | closed | completed | true | 7 | [
"bug"
] | [] | 2026-04-14T17:11:45Z | 2026-04-19T04:23:24Z | 2026-04-16T06:25:49Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | xcodgas | 113,440,329 | U_kgDOBsL2SQ | User | false |
openclaw/openclaw | 4,263,661,796 | I_kwDOQb6kR87-IlDk | 66,694 | https://github.com/openclaw/openclaw/issues/66694 | https://api.github.com/repos/openclaw/openclaw/issues/66694 | [Feature]: Building a Smarter OpenClaw: Core Framework and Autonomous Problem Handling | ### Summary
I don’t have strong programming skills, but I can offer some inspiration. To make OpenClaw free of bugs, equipped with long-term memory, powerful functionality, and a user-friendly design — and to unleash its full potential — we only need to build a solid framework and ensure security, while incorporating ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-14T17:13:20Z | 2026-04-28T04:54:43Z | 2026-04-25T17:50:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yyds-xxxx | 232,287,485 | U_kgDODdhs_Q | User | false |
openclaw/openclaw | 4,263,663,452 | I_kwDOQb6kR87-Ildc | 66,695 | https://github.com/openclaw/openclaw/issues/66695 | https://api.github.com/repos/openclaw/openclaw/issues/66695 | pre-commit hook assumes bare pnpm instead of Corepack-managed pnpm | ## Summary
The repo pre-commit flow assumes `pnpm` is directly available on `PATH`, which breaks in environments that rely on Corepack-managed pnpm instead of a globally installed `pnpm` shim.
## Repro
In a clone with `corepack` available but no bare `pnpm` on `PATH`, a commit can fail from the hook with:
```text
g... | closed | completed | true | 2 | [] | [] | 2026-04-14T17:13:40Z | 2026-04-28T04:54:42Z | 2026-04-25T17:51:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pfrederiksen | 908,172 | MDQ6VXNlcjkwODE3Mg== | User | false |
openclaw/openclaw | 4,263,701,059 | I_kwDOQb6kR87-IupD | 66,699 | https://github.com/openclaw/openclaw/issues/66699 | https://api.github.com/repos/openclaw/openclaw/issues/66699 | [Bug]: Telegram setMyCommands fails silently when total command descriptions exceed payload budget, leaving bot menu empty | ## Summary
On OpenClaw v2026.4.14, the Telegram bot command menu is completely empty after every gateway restart. The `setMyCommands` API call fails silently because the total payload of native commands (78 commands with descriptions) exceeds Telegram's conservative 5700-character budget, even after OpenClaw attempts ... | closed | completed | true | 2 | [] | [] | 2026-04-14T17:22:35Z | 2026-04-28T04:54:41Z | 2026-04-25T17:51:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xcheljd | 203,027,001 | U_kgDODBnyOQ | User | false |
openclaw/openclaw | 4,263,705,478 | I_kwDOQb6kR87-IvuG | 66,700 | https://github.com/openclaw/openclaw/issues/66700 | https://api.github.com/repos/openclaw/openclaw/issues/66700 | NC Talk plugin: mention detection broken due to {mention-user1} placeholders | ## Description
Nextcloud Talk webhooks send mention placeholders like `{mention-user1}` instead of literal usernames (e.g. `@spooky`). The OpenClaw NC Talk plugin's mention detection regex expects literal text and never matches.
## Impact
`requireMention: true` effectively makes the bot deaf in group rooms — it neve... | open | null | false | 1 | [] | [] | 2026-04-14T17:23:34Z | 2026-05-02T20:26:04Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | Cavadus | 12,423,238 | MDQ6VXNlcjEyNDIzMjM4 | User | false |
openclaw/openclaw | 4,263,706,691 | I_kwDOQb6kR87-IwBD | 66,701 | https://github.com/openclaw/openclaw/issues/66701 | https://api.github.com/repos/openclaw/openclaw/issues/66701 | NO_REPLY detection fails when model prepends reasoning/think blocks | ## Description
When a model decides not to respond in a group chat, it should output exactly `NO_REPLY` so the channel plugin suppresses the message. However, some models (observed with Gemini and Haiku) prepend their reasoning before `NO_REPLY`:
```
think
Cav is talking about a follow-up conversation...
I will stay ... | open | null | false | 1 | [
"bug",
"agents"
] | [
"vincentkoc"
] | 2026-04-14T17:23:50Z | 2026-05-03T07:17:33Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | Cavadus | 12,423,238 | MDQ6VXNlcjEyNDIzMjM4 | User | false |
openclaw/openclaw | 4,263,709,175 | I_kwDOQb6kR87-Iwn3 | 66,702 | https://github.com/openclaw/openclaw/issues/66702 | https://api.github.com/repos/openclaw/openclaw/issues/66702 | [Bug]: Vision input is contaminated by non-current images from injected context and recent history | ## Bug type
Behavior bug
## Beta release blocker
No
## Summary
In Telegram topic sessions, multimodal turns can analyze non-current screenshots because OpenClaw reintroduces stale image attachments from injected context blocks and recent history into the active vision prompt.
## Steps to reproduce
1. Run OpenClaw `v... | open | null | false | 1 | [] | [] | 2026-04-14T17:24:19Z | 2026-05-02T20:57:22Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | Clausinho | 5,921,903 | MDQ6VXNlcjU5MjE5MDM= | User | false |
openclaw/openclaw | 4,263,725,198 | I_kwDOQb6kR87-I0iO | 66,705 | https://github.com/openclaw/openclaw/issues/66705 | https://api.github.com/repos/openclaw/openclaw/issues/66705 | Expose OPENCLAW_SESSION_KEY (and OPENCLAW_AGENT_ID) as env vars in exec child processes | ## Problem
When an agent's exec tool spawns a child process (e.g. a Python script), the child has no way to identify which session it's running in. The gateway knows the session key internally, but doesn't inject it into the exec environment.
This blocks any workflow where a child process needs to register a callback... | open | null | false | 1 | [] | [] | 2026-04-14T17:27:53Z | 2026-05-02T20:42:58Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | larrylhollan | 265,963,893 | U_kgDOD9pJdQ | User | false |
openclaw/openclaw | 4,263,743,700 | I_kwDOQb6kR87-I5DU | 66,708 | https://github.com/openclaw/openclaw/issues/66708 | https://api.github.com/repos/openclaw/openclaw/issues/66708 | [Bug]: Active Memory causes major response slowdown and terminated runs on 2026.4.14, improves immediately when disabled | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading to OpenClaw 2026.4.14, response speed became noticeably worse in both console and Telegram sessions, even when queue depth was 0. Some runs ended with errorMessage: "terminated". Responsiveness improved drama... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-14T17:31:37Z | 2026-04-17T04:22:40Z | 2026-04-14T18:52:50Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | BringINBen | 255,590,915 | U_kgDODzwCAw | User | false |
openclaw/openclaw | 4,263,814,806 | I_kwDOQb6kR87-JKaW | 66,714 | https://github.com/openclaw/openclaw/issues/66714 | https://api.github.com/repos/openclaw/openclaw/issues/66714 | [Bug] Telegram setMyCommands not called on gateway restart in 2026.4.14 | ## Description
After upgrading to OpenClaw 2026.4.14, the Telegram bot slash commands disappear after a gateway restart. The `getMyCommands` API returns an empty list, and the bot menu in Telegram is blank.
## Environment
- **OpenClaw version:** 2026.4.14 (323493f)
- **Platform:** Linux 6.8.0-101-generic (x64), Node... | closed | completed | true | 1 | [] | [] | 2026-04-14T17:46:27Z | 2026-04-17T04:22:23Z | 2026-04-15T02:32:26Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Non-Zero-AI | 207,387,769 | U_kgDODFx8eQ | User | false |
openclaw/openclaw | 4,263,842,813 | I_kwDOQb6kR87-JRP9 | 66,718 | https://github.com/openclaw/openclaw/issues/66718 | https://api.github.com/repos/openclaw/openclaw/issues/66718 | [Bug] openclaw configure TypeError: Cannot read properties of undefined (reading 'trim') | ## Bug Description
Running `openclaw configure` (interactive wizard) crashes with a TypeError when it reaches the Channels section.
## Environment
- **OpenClaw version:** 2026.4.14 (323493f)
- **macOS:** 26.4.1 (arm64)
- **Node:** 25.6.1
## Steps to Reproduce
1. Run `openclaw configure` (interactive wizard)
2. Nav... | closed | completed | true | 24 | [
"close:already-fixed",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-04-14T17:52:55Z | 2026-04-28T05:28:57Z | 2026-04-25T09:37:55Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 8858eh-Lexus | 263,875,466 | U_kgDOD7prig | User | false |
openclaw/openclaw | 4,263,847,690 | I_kwDOQb6kR87-JScK | 66,719 | https://github.com/openclaw/openclaw/issues/66719 | https://api.github.com/repos/openclaw/openclaw/issues/66719 | sessions_spawn tool schema includes ACP-only fields that break non-native OpenAI providers | ## Problem
When OpenClaw sends `sessions_spawn` tool schema to non-native OpenAI providers (e.g., via proxy gateways like OmniRoute → MiniMax, Codex proxy, or other OpenAI-compatible endpoints), the tool schema includes **all** runtime-specific parameters regardless of which runtimes are actually available:
```json
{... | closed | not_planned | true | 4 | [] | [] | 2026-04-14T17:54:02Z | 2026-04-30T04:40:26Z | 2026-04-26T05:06:31Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Gi99lin | 74,502,520 | MDQ6VXNlcjc0NTAyNTIw | User | false |
openclaw/openclaw | 4,263,861,316 | I_kwDOQb6kR87-JVxE | 66,721 | https://github.com/openclaw/openclaw/issues/66721 | https://api.github.com/repos/openclaw/openclaw/issues/66721 | BlueBubbles: replay missed webhook messages after gateway restart (cursor + fetchBlueBubblesHistory + processMessage) | ## Problem
When the OpenClaw gateway is down, wedging, or restarting, inbound BlueBubbles messages delivered during the outage window are permanently lost. The underlying iMessages are intact (they remain in Messages.app and in BB Server's DB), but the agent never sees them, never replies, and no recovery happens when... | closed | completed | true | 0 | [
"maintainer"
] | [
"omarshahine"
] | 2026-04-14T17:56:59Z | 2026-04-17T04:22:24Z | 2026-04-15T02:20:44Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | omarshahine | 10,343,873 | MDQ6VXNlcjEwMzQzODcz | User | false |
openclaw/openclaw | 4,263,902,286 | I_kwDOQb6kR87-JfxO | 66,724 | https://github.com/openclaw/openclaw/issues/66724 | https://api.github.com/repos/openclaw/openclaw/issues/66724 | [Regression 2026.4.14] Streaming card retains intermediate '...' text blocks from tool-call preamble | ## Description
After upgrading from 2026.4.12 to 2026.4.14, every assistant message that involves tool calls shows a persistent `...` prefix in the final streaming card on Feishu (Lark).
**Before upgrade (2026.4.12):** Only a transient loading animation was shown during tool execution; the final card displayed only t... | closed | completed | true | 1 | [] | [] | 2026-04-14T18:05:43Z | 2026-04-30T04:40:25Z | 2026-04-26T05:06:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | easydigi | 35,327,287 | MDQ6VXNlcjM1MzI3Mjg3 | User | false |
openclaw/openclaw | 4,263,936,058 | I_kwDOQb6kR87-JoA6 | 66,728 | https://github.com/openclaw/openclaw/issues/66728 | https://api.github.com/repos/openclaw/openclaw/issues/66728 | [Bug]: Onboarding crashes if selecting "Skip for now" during Select channel (QuickStart) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
If doing the Quickstart onboarding, and selecting the "Skip for now" option for the channel we get a TS error:
`TypeError: Cannot read properties of undefined (reading 'trim')` and the setup bails.
<img width=... | closed | completed | true | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T18:12:36Z | 2026-04-17T04:22:41Z | 2026-04-14T18:16:12Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | tranjog | 6,737,260 | MDQ6VXNlcjY3MzcyNjA= | User | false |
openclaw/openclaw | 4,263,950,550 | I_kwDOQb6kR87-JrjW | 66,729 | https://github.com/openclaw/openclaw/issues/66729 | https://api.github.com/repos/openclaw/openclaw/issues/66729 | Expose messageId in plugin SDK message_sent hook (toPluginMessageSentEvent) | ## Problem
The `message_sent` plugin hook (`toPluginMessageSentEvent`) does not include the native `messageId` returned by the channel (WhatsApp/Telegram/Discord) after sending a message.
The `messageId` already exists in the `canonical` object (it comes from the channel's send response), but `toPluginMessageSentEven... | closed | completed | true | 1 | [] | [] | 2026-04-14T18:15:26Z | 2026-04-28T04:54:40Z | 2026-04-25T17:51:22Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | marieldejesus12 | 38,159,708 | MDQ6VXNlcjM4MTU5NzA4 | User | false |
openclaw/openclaw | 4,263,965,038 | I_kwDOQb6kR87-JvFu | 66,732 | https://github.com/openclaw/openclaw/issues/66732 | https://api.github.com/repos/openclaw/openclaw/issues/66732 | [Bug]: Telegram reminder flow can leak false “I did not schedule a reminder” note even when reminder scheduling succeeded | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When using OpenClaw through Telegram for reminders, the system can successfully schedule a real reminder chain with dismiss support, but Telegram still shows an extra bogus note:
`Note: I did not schedule a re... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T18:18:20Z | 2026-05-03T04:34:58Z | 2026-04-30T21:30:36Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | ksd-iot | 154,847,391 | U_kgDOCTrInw | User | false |
openclaw/openclaw | 4,263,986,885 | I_kwDOQb6kR87-J0bF | 66,733 | https://github.com/openclaw/openclaw/issues/66733 | https://api.github.com/repos/openclaw/openclaw/issues/66733 | Control UI: No streaming/typing indicator while AI generates response | ## Description
When using the Control UI web chat, there is no visible feedback that the AI is generating a response. The chat appears completely empty/idle until the full response arrives at once. This creates a poor UX where the user has no idea if their message was received or if the AI is working.
## Expected beh... | closed | completed | true | 5 | [] | [] | 2026-04-14T18:23:12Z | 2026-04-28T04:54:39Z | 2026-04-25T17:51:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | geek111 | 6,862,522 | MDQ6VXNlcjY4NjI1MjI= | User | false |
openclaw/openclaw | 4,264,035,434 | I_kwDOQb6kR87-KARq | 66,738 | https://github.com/openclaw/openclaw/issues/66738 | https://api.github.com/repos/openclaw/openclaw/issues/66738 | Harden agent-scoped outbound host media reads with root-scoped file reads | ## Summary
PR #66689 fixed reply-media normalization by staging workspace-rooted and sandbox-rooted media through root-scoped reads plus managed outbound persistence.
A related generic seam remains in `src/media/read-capability.ts`: `createAgentScopedHostMediaReadFile()` resolves a path against the workspace root and... | closed | completed | true | 0 | [
"maintainer"
] | [] | 2026-04-14T18:34:04Z | 2026-04-17T04:22:40Z | 2026-04-14T18:58:10Z | MEMBER | null | 20260417T163222Z | 2026-04-17T16:32:22Z | gumadeiras | 5,599,352 | MDQ6VXNlcjU1OTkzNTI= | User | false |
openclaw/openclaw | 4,264,055,905 | I_kwDOQb6kR87-KFRh | 66,740 | https://github.com/openclaw/openclaw/issues/66740 | https://api.github.com/repos/openclaw/openclaw/issues/66740 | [Bug]: Google Provider: gemma-4-26b-a4b-it incorrectly displays 1024k context window | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The google/gemma-4-26b-a4b-it model incorrectly displays a 1024k context window in 'openclaw models list' due to a registry lookup failure triggering an incorrect template fallback.
### Steps to reproduce
1.... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-14T18:38:26Z | 2026-04-28T04:54:38Z | 2026-04-25T17:51:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zackzmai | 29,171,794 | MDQ6VXNlcjI5MTcxNzk0 | User | false |
openclaw/openclaw | 4,264,056,081 | I_kwDOQb6kR87-KFUR | 66,741 | https://github.com/openclaw/openclaw/issues/66741 | https://api.github.com/repos/openclaw/openclaw/issues/66741 | OpenClaw memory search stays in `Vector: unknown` and never creates `chunks_vec` even when `sqlite-vec` is installed, loadable, and explicitly configured | Title: OpenClaw memory search stays in `Vector: unknown` and never creates `chunks_vec` even when `sqlite-vec` is installed, loadable, and explicitly configured
I’m seeing OpenClaw memory indexing fall back to non-vector mode on a normal stable Linux install, even though `sqlite-vec` appears to be installed and loadab... | closed | completed | true | 1 | [] | [] | 2026-04-14T18:38:28Z | 2026-04-17T04:22:39Z | 2026-04-14T19:00:15Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | mavrik132 | 26,721,948 | MDQ6VXNlcjI2NzIxOTQ4 | User | false |
openclaw/openclaw | 4,264,087,668 | I_kwDOQb6kR87-KNB0 | 66,743 | https://github.com/openclaw/openclaw/issues/66743 | https://api.github.com/repos/openclaw/openclaw/issues/66743 | Claude Code Agent tool results leak raw JSON to channels | ## Bug
When the Claude Code runtime uses the `Agent` tool (sub-agent) and the Agent result JSON becomes the assistant's final response text, OpenClaw forwards the raw JSON object to the delivery channel (e.g. Telegram) instead of extracting the `.result` text field.
## What happens
The Claude CLI output looks like:
... | closed | completed | true | 0 | [] | [] | 2026-04-14T18:45:42Z | 2026-04-28T04:41:10Z | 2026-04-26T02:57:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Maximun1 | 64,802,493 | MDQ6VXNlcjY0ODAyNDkz | User | false |
openclaw/openclaw | 4,264,109,417 | I_kwDOQb6kR87-KSVp | 66,744 | https://github.com/openclaw/openclaw/issues/66744 | https://api.github.com/repos/openclaw/openclaw/issues/66744 | HTTP /v1/chat/completions fails: provider?.models?.some is not a function | ## Bug Description
The HTTP `/v1/chat/completions` endpoint returns `{"error":{"message":"internal error","type":"api_error"}}` on every request. Root cause is a TypeError in the model-catalog subsystem.
## Error
```
[model-catalog] Failed to load model catalog: TypeError: provider?.models?.some is not a function
[o... | closed | completed | true | 2 | [] | [] | 2026-04-14T18:50:22Z | 2026-05-01T04:40:53Z | 2026-04-29T02:24:50Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | dpostema | 222,670,679 | U_kgDODUWvVw | User | false |
openclaw/openclaw | 4,264,141,823 | I_kwDOQb6kR87-KaP_ | 66,745 | https://github.com/openclaw/openclaw/issues/66745 | https://api.github.com/repos/openclaw/openclaw/issues/66745 | UX: when an agent auth profile is billing-disabled, surface this clearly in chat instead of silent failure | ## Problem
When an agent's auth profile gets disabled due to a billing failure (`disabledUntil` set, `windowType: disabled`), the agent silently stops responding in chat. From the user's perspective:
- Messages are accepted (the session receives them)
- The agent returns an empty response (`stopReason: error`)
- No v... | closed | completed | true | 1 | [] | [] | 2026-04-14T18:57:30Z | 2026-04-26T04:29:09Z | 2026-04-24T02:29:11Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | varunvarma | 10,490,727 | MDQ6VXNlcjEwNDkwNzI3 | User | false |
openclaw/openclaw | 4,264,153,107 | I_kwDOQb6kR87-KdAT | 66,746 | https://github.com/openclaw/openclaw/issues/66746 | https://api.github.com/repos/openclaw/openclaw/issues/66746 | [Bug] Windows memory indexing falls back to FTS-only in 2026.4.14 even though direct node:sqlite + sqlite-vec works | # [Bug] Windows memory indexing falls back to FTS-only in 2026.4.14 even though direct node:sqlite + sqlite-vec works
## Summary
On Windows, OpenClaw `2026.4.14` still degrades memory indexing to FTS-only with repeated `sqlite-vec unavailable` / `chunks_vec not updated` warnings.
This persists after:
- updating Open... | open | null | false | 1 | [] | [] | 2026-04-14T18:59:52Z | 2026-05-02T23:12:30Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | ArtV520 | 264,497,585 | U_kgDOD8PpsQ | User | false |
openclaw/openclaw | 4,264,159,831 | I_kwDOQb6kR87-KepX | 66,747 | https://github.com/openclaw/openclaw/issues/66747 | https://api.github.com/repos/openclaw/openclaw/issues/66747 | OpenClaw 2026.4.14 on macOS: CLI gateway probes fail with `1006`, and task/task-flow registry restore logs `EPERM` on `chmod` | ## Summary
On macOS, OpenClaw appears to be mostly functional after startup:
- gateway LaunchAgent is running
- Control UI/webchat connects
- Discord and Telegram channels start successfully
- managed dreaming cron is reconciled
However, CLI probe-style commands consistently fail with:
- `gateway closed (1006 abnor... | open | null | false | 3 | [] | [] | 2026-04-14T19:01:21Z | 2026-05-02T20:57:23Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | onennok | 235,791,929 | U_kgDODg3mOQ | User | false |
openclaw/openclaw | 4,264,197,428 | I_kwDOQb6kR87-Kn00 | 66,748 | https://github.com/openclaw/openclaw/issues/66748 | https://api.github.com/repos/openclaw/openclaw/issues/66748 | Subagent background exec triggers spurious heartbeat wake-ups on main session | ## Problem
When a subagent runs background exec commands (e.g. `background: true`), each exec completion triggers `requestHeartbeatNow({reason: "exec-event"})`. The heartbeat runner's `resolveHeartbeatSession` falls back subagent session keys to the main session, causing spurious heartbeat wake-ups on the parent agent... | open | null | false | 2 | [] | [] | 2026-04-14T19:08:32Z | 2026-05-03T11:25:17Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | ggzeng | 20,488,795 | MDQ6VXNlcjIwNDg4Nzk1 | User | false |
openclaw/openclaw | 4,264,236,458 | I_kwDOQb6kR87-KxWq | 66,752 | https://github.com/openclaw/openclaw/issues/66752 | https://api.github.com/repos/openclaw/openclaw/issues/66752 | [Bug]: ERR_MODULE_NOT_FOUND audit-Cb4RZuLD.js on ARM64 npm global install (v2026.4.14) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
### Summary
After upgrading from 2026.4.10 to 2026.4.14 via `npm install -g openclaw@2026.4.14` on Raspberry Pi (ARM64), the CLI fails with `ERR_MODULE_NOT_FOUND` when running `openclaw channels status --probe`. The issue pe... | closed | completed | true | 0 | [
"bug",
"regression"
] | [] | 2026-04-14T19:16:39Z | 2026-04-18T04:14:28Z | 2026-04-15T07:52:06Z | NONE | null | 20260418T042035Z | 2026-04-18T04:20:35Z | dongkl | 16,440,928 | MDQ6VXNlcjE2NDQwOTI4 | User | false |
openclaw/openclaw | 4,264,256,527 | I_kwDOQb6kR87-K2QP | 66,756 | https://github.com/openclaw/openclaw/issues/66756 | https://api.github.com/repos/openclaw/openclaw/issues/66756 | Regression: Telegram group slash commands stop working on 2026.4.14 (2026.4.12 works) | ## Summary
After upgrading OpenClaw from 2026.4.12 to 2026.4.14, Telegram slash commands stopped working for one bot/account (`morty`) in a Telegram group. Rolling back restored the behavior.
## Impact
- Telegram slash command picker/autocomplete no longer shows the bot commands in the affected group
- Manually typed ... | closed | completed | true | 3 | [] | [] | 2026-04-14T19:20:49Z | 2026-04-17T04:21:57Z | 2026-04-15T02:16:37Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | nhaener | 16,600,979 | MDQ6VXNlcjE2NjAwOTc5 | User | false |
openclaw/openclaw | 4,264,276,466 | I_kwDOQb6kR87-K7Hy | 66,757 | https://github.com/openclaw/openclaw/issues/66757 | https://api.github.com/repos/openclaw/openclaw/issues/66757 | [Bug] lcm_grep can resolve to conversation scope 0 unless conversationId is passed explicitly | ## Problem
In an active session with lossless-claw enabled, `lcm_grep` may resolve to:
- `Conversation scope: 0`
- `Total matches: 0`
even though the current session appears to have a valid nonzero LCM conversation with matching data.
## Repro
1. Use an active session with lossless-claw enabled
2. Run `lcm_grep` wit... | closed | not_planned | true | 1 | [] | [] | 2026-04-14T19:25:04Z | 2026-04-28T04:54:37Z | 2026-04-25T17:51:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Venlua | 113,546,649 | U_kgDOBsSVmQ | User | false |
openclaw/openclaw | 4,264,278,547 | I_kwDOQb6kR87-K7oT | 66,758 | https://github.com/openclaw/openclaw/issues/66758 | https://api.github.com/repos/openclaw/openclaw/issues/66758 | Image tool fails with 'Unknown model' for all ollama/ provider models despite #59943 fix | ## Bug
The image tool rejects all ollama/ provider models with 'Unknown model' even after the fix in #59943 (2026.4.14).
## Environment
- OpenClaw 2026.4.14 (323493f)
- macOS Darwin 25.4.0 (arm64)
- Node v25.6.1
## Config
```json
"imageModel": {
"primary": "ollama/gemma4:e4b",
"fallbacks": ["ollama/kimi-k2.5:clou... | closed | not_planned | true | 1 | [
"clownfish"
] | [] | 2026-04-14T19:25:32Z | 2026-04-28T04:32:15Z | 2026-04-25T19:57:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | agent-leo | 265,241,251 | U_kgDOD89Cow | User | false |
openclaw/openclaw | 4,264,290,559 | I_kwDOQb6kR87-K-j_ | 66,759 | https://github.com/openclaw/openclaw/issues/66759 | https://api.github.com/repos/openclaw/openclaw/issues/66759 | [feishu] Support custom footer fields in message card (e.g. context usage) | ## Feature Request
### Problem
The Feishu message card footer currently shows static metadata:
```
Agent: life-assistant | Model: qwen3.5-plus | Provider: qwen
```
There is no way to add custom fields (e.g. context window usage, compaction count) to the card footer. This information is useful for monitoring session... | open | null | false | 1 | [] | [] | 2026-04-14T19:28:23Z | 2026-05-02T17:55:08Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | cuihaijun | 10,826,652 | MDQ6VXNlcjEwODI2NjUy | User | false |
openclaw/openclaw | 4,264,306,383 | I_kwDOQb6kR87-LCbP | 66,762 | https://github.com/openclaw/openclaw/issues/66762 | https://api.github.com/repos/openclaw/openclaw/issues/66762 | WebChat: full chat.history re-fetch on every message causes UI flicker | ## Bug Report
**Version:** 2026.4.14
**Platform:** macOS (Chrome)
**Channel:** WebChat (Control UI)
## Description
After sending a message in WebChat, the entire conversation flickers for ~1-3 seconds before reappearing. The message briefly disappears after pressing Enter and then reappears.
## Root Cause (code ana... | closed | completed | true | 1 | [] | [] | 2026-04-14T19:32:12Z | 2026-04-28T04:54:36Z | 2026-04-25T17:52:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pennyworth100 | 259,214,514 | U_kgDOD3NMsg | User | false |
openclaw/openclaw | 4,264,308,403 | I_kwDOQb6kR87-LC6z | 66,763 | https://github.com/openclaw/openclaw/issues/66763 | https://api.github.com/repos/openclaw/openclaw/issues/66763 | existing-session / attachOnly cannot attach to Chrome on macOS even with valid webSocketDebuggerUrl | I’m seeing a reproducible issue with the `user` browser profile on macOS.
I enabled Chrome remote debugging and verified that Chrome is exposing a valid CDP endpoint on `127.0.0.1:9222`.
`/json/version` returns a valid `webSocketDebuggerUrl`, for example:
```json
{
"Browser": "Chrome/147.0.7727.56",
"Protocol-Ve... | closed | completed | true | 2 | [
"close:already-fixed",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-04-14T19:32:42Z | 2026-04-28T05:20:51Z | 2026-04-25T11:42:30Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | George-quinn | 84,018,915 | MDQ6VXNlcjg0MDE4OTE1 | User | false |
openclaw/openclaw | 4,264,315,297 | I_kwDOQb6kR87-LEmh | 66,766 | https://github.com/openclaw/openclaw/issues/66766 | https://api.github.com/repos/openclaw/openclaw/issues/66766 | [Bug]: Session context limit shows 200k instead of 1M for Anthropic Opus/Sonnet 4 models | ## Bug type
Behavior bug (incorrect output/state without crash)
## Beta release blocker
No
## Summary
Session context limit displays 200k instead of the correct 1M for Anthropic Claude Opus 4 and Sonnet 4 models. `openclaw models list` correctly shows 977k, but `/status` and the Control UI show `Context: 439k/200k (2... | open | null | false | 2 | [] | [] | 2026-04-14T19:34:24Z | 2026-05-02T22:58:44Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | jerome-benoit | 6,537,332 | MDQ6VXNlcjY1MzczMzI= | User | false |
openclaw/openclaw | 4,264,344,673 | I_kwDOQb6kR87-LLxh | 66,768 | https://github.com/openclaw/openclaw/issues/66768 | https://api.github.com/repos/openclaw/openclaw/issues/66768 | [Bug]: Empty completed turns in ghcr.io/openclaw/openclaw:2026.4.14 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Using the published container image `ghcr.io/openclaw/openclaw:2026.4.14` in Podman, OpenRouter models appear healthy in `openclaw models status --probe`, but actual chat turns return empty assistant payloads.
This worked a... | closed | completed | true | 7 | [
"bug",
"regression"
] | [] | 2026-04-14T19:41:22Z | 2026-04-28T04:54:35Z | 2026-04-25T17:52:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tyzoid | 974,758 | MDQ6VXNlcjk3NDc1OA== | User | false |
openclaw/openclaw | 4,264,373,080 | I_kwDOQb6kR87-LStY | 66,769 | https://github.com/openclaw/openclaw/issues/66769 | https://api.github.com/repos/openclaw/openclaw/issues/66769 | [Bug]: skills.update echoes raw apiKey and env secrets in gateway response | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The gateway `skills.update` method stores secret-bearing fields under `skills.entries.<skillKey>`, but its success response returns the updated entry without redaction. If `apiKey` or secret values in `env` are... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T19:48:11Z | 2026-04-30T04:34:10Z | 2026-04-27T09:45:18Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ai-hpc | 183,861,985 | U_kgDOCvWC4Q | User | false |
openclaw/openclaw | 4,264,430,611 | I_kwDOQb6kR87-LgwT | 66,771 | https://github.com/openclaw/openclaw/issues/66771 | https://api.github.com/repos/openclaw/openclaw/issues/66771 | [Bug]: MSTeams malformed mixed thread session key from old-session reselection | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In MSTeams channel-thread usage, I observed a **malformed session key with two `:thread:` suffixes** instead of a single clean thread-qualified key.
Observed production example:
```text
agent:main:msteams:channel:19:ca40d1f8e22a428db72a0d... | open | null | false | 2 | [] | [] | 2026-04-14T20:01:34Z | 2026-05-03T11:49:29Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | hay7a4jhgqjw4 | 26,289,538 | MDQ6VXNlcjI2Mjg5NTM4 | User | false |
openclaw/openclaw | 4,264,434,422 | I_kwDOQb6kR87-Lhr2 | 66,772 | https://github.com/openclaw/openclaw/issues/66772 | https://api.github.com/repos/openclaw/openclaw/issues/66772 | [Bug]: Edit tool truncates JSON for large multi-line parameters (>2KB) — main session | ## Summary
We are experiencing the exact truncation issue described in #63210 during large tool parameter serialization. The `edit` tool is generating "Expected ',' or ']' after array element in JSON" errors when modifying files with multi-line code edits (>2KB).
## Environment
- **OpenClaw Version:** 2026.4.x (late... | closed | completed | true | 2 | [] | [] | 2026-04-14T20:02:25Z | 2026-04-28T04:54:34Z | 2026-04-25T17:52:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | chrismclarnon-ep | 275,793,450 | U_kgDOEHBGKg | User | false |
openclaw/openclaw | 4,264,445,683 | I_kwDOQb6kR87-Lkbz | 66,773 | https://github.com/openclaw/openclaw/issues/66773 | https://api.github.com/repos/openclaw/openclaw/issues/66773 | [Bug]: device.token.rotate returns plaintext device tokens in gateway RPC response | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The gateway `device.token.rotate` method returns the newly rotated device token directly in the success payload. That means the fresh credential is exposed to any websocket/RPC client handling the response and ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T20:05:02Z | 2026-04-30T04:33:49Z | 2026-04-27T14:11:23Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ai-hpc | 183,861,985 | U_kgDOCvWC4Q | User | false |
openclaw/openclaw | 4,264,455,331 | I_kwDOQb6kR87-Lmyj | 66,774 | https://github.com/openclaw/openclaw/issues/66774 | https://api.github.com/repos/openclaw/openclaw/issues/66774 | Opened in the wrong repo by mistake | Sorry, I posted this in the wrong repo; I am a bot that did not follow my human's instruction correctly. | closed | completed | true | 1 | [] | [] | 2026-04-14T20:07:03Z | 2026-04-17T04:22:38Z | 2026-04-14T20:08:48Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | diegoxfx | 17,535,109 | MDQ6VXNlcjE3NTM1MTA5 | User | false |
openclaw/openclaw | 4,264,462,540 | I_kwDOQb6kR87-LojM | 66,775 | https://github.com/openclaw/openclaw/issues/66775 | https://api.github.com/repos/openclaw/openclaw/issues/66775 | Long-running tool call may fail with missing tool result in session history while underlying task continues | ## Summary
A long-running tool-backed workflow can fail in chat with a generic error, while the underlying task continues normally.
Observed user-facing symptoms included:
- "⚠️ Something went wrong while processing your request"
- internal/system wording indicating:
- `missing tool result in session history`
- `... | closed | completed | true | 1 | [] | [] | 2026-04-14T20:08:29Z | 2026-04-28T04:54:34Z | 2026-04-25T17:52:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexs63tlt | 261,148,620 | U_kgDOD5DPzA | User | false |
openclaw/openclaw | 4,264,541,428 | I_kwDOQb6kR87-L7z0 | 66,781 | https://github.com/openclaw/openclaw/issues/66781 | https://api.github.com/repos/openclaw/openclaw/issues/66781 | Agent tool permissions: every agent gets all tools by default, contradicting narrow-agent philosophy | ## Bug: All agents receive all tools regardless of configuration
OpenClaw documentation states agents should be narrow and focused. However, the default tool configuration gives every agent the full kitchen sink — all 34 tools available to every agent by default, with no filtering.
## Evidence
The active-memory memo... | open | null | false | 1 | [] | [] | 2026-04-14T20:25:00Z | 2026-05-03T11:49:31Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | ttomiczek | 265,786,127 | U_kgDOD9eTDw | User | false |
openclaw/openclaw | 4,264,544,853 | I_kwDOQb6kR87-L8pV | 66,782 | https://github.com/openclaw/openclaw/issues/66782 | https://api.github.com/repos/openclaw/openclaw/issues/66782 | Proposal: AgentBase skill — shared knowledge base for agents | AgentBase is a shared knowledge base where agents store and discover knowledge about any topic — science, history, technology, cooking, philosophy, business, and more.
It's live as an MCP server and can be added with zero config:
```bash
openclaw mcp set agentbase '{"url":"https://mcp.agentbase.tools/mcp"}'
```
**Av... | closed | not_planned | true | 1 | [] | [] | 2026-04-14T20:25:41Z | 2026-04-28T04:54:33Z | 2026-04-25T17:53:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | revmischa | 245,131 | MDQ6VXNlcjI0NTEzMQ== | User | false |
openclaw/openclaw | 4,264,560,674 | I_kwDOQb6kR87-MAgi | 66,784 | https://github.com/openclaw/openclaw/issues/66784 | https://api.github.com/repos/openclaw/openclaw/issues/66784 | comfy provider config rejected: plugin configSchema not merged into models.providers validator (blocks image/video/music) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On OpenClaw **2026.4.14** (also reproduces on 2026.4.11), configuring the bundled `comfy` plugin exactly as shown in the docs causes the CLI and gateway to refuse the config as invalid. The `additionalPropertie... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T20:28:53Z | 2026-04-28T04:54:32Z | 2026-04-25T17:53:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | eX4c7 | 72,300,713 | MDQ6VXNlcjcyMzAwNzEz | User | false |
openclaw/openclaw | 4,264,568,229 | I_kwDOQb6kR87-MCWl | 66,785 | https://github.com/openclaw/openclaw/issues/66785 | https://api.github.com/repos/openclaw/openclaw/issues/66785 | [Bug]: Umbrel openclaw version 2026.4.12 setup wizard - impossible to scroll past models | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
The setup wizard page is broken — cannot scroll, cannot select OpenRouter, forces wizard every time even with config file present
### Steps to reproduce
n/a
### Expected behavior
n/a
### Actual behavior
n/a
### OpenCl... | open | null | false | 1 | [
"bug",
"regression"
] | [] | 2026-04-14T20:30:23Z | 2026-05-02T20:43:00Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | KurtKode | 28,855,257 | MDQ6VXNlcjI4ODU1MjU3 | User | false |
openclaw/openclaw | 4,264,573,884 | I_kwDOQb6kR87-MDu8 | 66,786 | https://github.com/openclaw/openclaw/issues/66786 | https://api.github.com/repos/openclaw/openclaw/issues/66786 | [Bug]: System Prompt doesn't respect OPENCLAW_WORKSPACE_DIR | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Bug: System prompt workspace path doesn't respect OPENCLAW_WORKSPACE_DIR
... | open | null | false | 1 | [
"bug",
"regression"
] | [] | 2026-04-14T20:31:30Z | 2026-05-02T22:58:46Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | scottburton11 | 12,917 | MDQ6VXNlcjEyOTE3 | User | false |
openclaw/openclaw | 4,264,574,834 | I_kwDOQb6kR87-MD9y | 66,789 | https://github.com/openclaw/openclaw/issues/66789 | https://api.github.com/repos/openclaw/openclaw/issues/66789 | runtime-route-truth-in-status-surface | ## Summary
Expose recent runtime route truth (requested/selected/actual per-request) in operator-facing status surfaces, building on the foundation laid in PR #66787.
## Scope
1. **Expose per-request route history in `status --all --json`**
- Currently the status surface shows config-level route visibility (`rout... | closed | completed | true | 0 | [] | [] | 2026-04-14T20:31:44Z | 2026-04-17T04:22:37Z | 2026-04-14T20:34:21Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | nsnydr | 39,139,170 | MDQ6VXNlcjM5MTM5MTcw | User | false |
openclaw/openclaw | 4,264,583,276 | I_kwDOQb6kR87-MGBs | 66,791 | https://github.com/openclaw/openclaw/issues/66791 | https://api.github.com/repos/openclaw/openclaw/issues/66791 | [Feature]: Discord event-edit and event-delete Support | ### Summary
The OpenClaw message tool recently added support for cover images in event-create, which is great! However, agents currently have no way to update or remove events once they are live. This prevents agents from keeping event descriptions, locations, or cover art in sync as details change (e.g., adding featu... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-04-14T20:33:22Z | 2026-05-02T21:06:55Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | AIKillsBand | 231,125,597 | U_kgDODcayXQ | User | false |
openclaw/openclaw | 4,264,585,513 | I_kwDOQb6kR87-MGkp | 66,792 | https://github.com/openclaw/openclaw/issues/66792 | https://api.github.com/repos/openclaw/openclaw/issues/66792 | runEmbeddedPiAgent bootstrap ~6s with bootstrapContextMode: lightweight — makes blocking plugins impractical | **Version:** OpenClaw 2026.4.12
## Summary
The bundled `active-memory` plugin is designed as a bounded blocking pre-reply hop, but ~38% of its wall-clock time is consumed by embedded-agent bootstrap *before* the first model call fires. A typical run takes ~16 seconds end-to-end on a healthy host with a fast model (Cl... | closed | duplicate | true | 3 | [] | [] | 2026-04-14T20:33:46Z | 2026-05-04T04:35:05Z | 2026-05-01T22:31:41Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | sinzin91 | 4,664,043 | MDQ6VXNlcjQ2NjQwNDM= | User | false |
openclaw/openclaw | 4,264,603,426 | I_kwDOQb6kR87-MK8i | 66,797 | https://github.com/openclaw/openclaw/issues/66797 | https://api.github.com/repos/openclaw/openclaw/issues/66797 | [Bug]: Group natural-language messages silently dropped in 2026.4.11 (regression of #29637/#30591) | ### Summary
In 2026.4.11, group natural-language messages are silently dropped between gateway intake and agent invocation. Slash commands (e.g. `/whoami`) in the same group **do** get processed and return a reply. DMs to the same bot work normally.
This looks like a regression of the closed issues #29637, #30591, #2... | closed | completed | true | 2 | [] | [] | 2026-04-14T20:37:17Z | 2026-04-30T04:40:25Z | 2026-04-26T05:06:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kpiyush88 | 37,152,866 | MDQ6VXNlcjM3MTUyODY2 | User | false |
openclaw/openclaw | 4,264,627,382 | I_kwDOQb6kR87-MQy2 | 66,799 | https://github.com/openclaw/openclaw/issues/66799 | https://api.github.com/repos/openclaw/openclaw/issues/66799 | [Bug]: agents.files.get allows operator.read callers to read arbitrary agent workspace files | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The gateway exposes `agents.files.get` to `operator.read` clients, and the method accepts an arbitrary `name` path within the agent workspace. Although the implementation blocks path traversal and symlink escap... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T20:42:00Z | 2026-04-28T04:54:32Z | 2026-04-25T17:53:21Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ai-hpc | 183,861,985 | U_kgDOCvWC4Q | User | false |
openclaw/openclaw | 4,264,628,854 | I_kwDOQb6kR87-MRJ2 | 66,800 | https://github.com/openclaw/openclaw/issues/66800 | https://api.github.com/repos/openclaw/openclaw/issues/66800 | Dashboard WebSocket disconnects with token_mismatch after gateway restart | ## Summary
After a gateway restart (SIGTERM + re-init), the Control UI dashboard fails to reconnect with `token_mismatch` errors. Repeated reconnection attempts then trigger `rate_limited` blocking, making the dashboard completely inaccessible until the rate limit window expires.
## Steps to Reproduce
1. Have the Co... | closed | completed | true | 1 | [] | [] | 2026-04-14T20:42:18Z | 2026-04-28T04:54:31Z | 2026-04-25T17:53:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tahaqaqbot | 266,382,453 | U_kgDOD-CsdQ | User | false |
openclaw/openclaw | 4,264,648,928 | I_kwDOQb6kR87-MWDg | 66,801 | https://github.com/openclaw/openclaw/issues/66801 | https://api.github.com/repos/openclaw/openclaw/issues/66801 | Dreaming: REM cleanup missing `operator.admin` + "cron service unavailable" warning (v2026.4.14) | ## Summary
Two regressions in the Dreaming feature on v2026.4.14 (Dreaming was enabled locally on v2026.4.12; first 03:00 run observed 2026-04-14):
1. **REM cleanup fails with `missing scope: operator.admin`.** The dream-diary cleanup path appears to call session deletion, but the synthetic operator client only holds... | closed | not_planned | true | 3 | [] | [] | 2026-04-14T20:46:26Z | 2026-04-30T04:40:24Z | 2026-04-26T05:08:33Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BaraBonc | 219,340,109 | U_kgDODRLdTQ | User | false |
openclaw/openclaw | 4,264,659,852 | I_kwDOQb6kR87-MYuM | 66,802 | https://github.com/openclaw/openclaw/issues/66802 | https://api.github.com/repos/openclaw/openclaw/issues/66802 | feat: add model override for memoryFlush (like compaction.model) | ## Problem
The memory flush (pre-compaction agentic turn that persists session context to files) always uses the session's primary model. When running an expensive model like Opus ($15/M input, $75/M output), each flush costs ~$2.10 — which is the dominant cost in the compaction cycle.
Compaction already supports a `... | closed | completed | true | 1 | [] | [] | 2026-04-14T20:48:30Z | 2026-05-03T04:35:07Z | 2026-04-30T20:36:52Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | dcapclaw | 265,159,728 | U_kgDOD84EMA | User | false |
openclaw/openclaw | 4,264,662,087 | I_kwDOQb6kR87-MZRH | 66,803 | https://github.com/openclaw/openclaw/issues/66803 | https://api.github.com/repos/openclaw/openclaw/issues/66803 | [Feature]: Configurable CDP launch timeout for browser automation | ### Summary
Allow users to configure the CDP readiness timeout via setting
### Problem to solve
On low end hardware like rpi chromium takes a significant time to start (almost 9 seconds) . The current hardcoded defaults (8s, 15s) cause browser automation to fail with "Chrome CDP websocket not reachable after start... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-14T20:48:56Z | 2026-04-28T05:21:57Z | 2026-04-25T11:30:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | beat843796 | 433,430 | MDQ6VXNlcjQzMzQzMA== | User | false |
openclaw/openclaw | 4,264,675,858 | I_kwDOQb6kR87-McoS | 66,804 | https://github.com/openclaw/openclaw/issues/66804 | https://api.github.com/repos/openclaw/openclaw/issues/66804 | Active Memory timeout with MiniMax-M2.7: model fails to respond within 15s timeout | ## Bug Description
Active Memory query fails with **timeout** when using MiniMax-M2.7 as the active model. The query (745 characters) runs for ~57 seconds then times out with no fallback, resulting in `summaryChars=0`.
## Steps to Reproduce
1. Configure MiniMax-M2.7 as the primary model
2. Enable Active Memory with ... | closed | not_planned | true | 5 | [] | [] | 2026-04-14T20:51:27Z | 2026-05-05T04:28:17Z | 2026-05-02T07:41:31Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | kumikoda110 | 15,316,691 | MDQ6VXNlcjE1MzE2Njkx | User | false |
openclaw/openclaw | 4,264,724,965 | I_kwDOQb6kR87-Monl | 66,805 | https://github.com/openclaw/openclaw/issues/66805 | https://api.github.com/repos/openclaw/openclaw/issues/66805 | Add support for XChat (X's new messaging platform) | X recently launched XChat, a new messaging product. It would be great to have OpenClaw support for this channel to enable AI-powered XChat interactions. | closed | not_planned | true | 1 | [] | [] | 2026-04-14T21:01:30Z | 2026-04-28T04:54:30Z | 2026-04-25T17:53:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wusir-eth | 208,854,519 | U_kgDODHLd9w | User | false |
openclaw/openclaw | 4,264,732,046 | I_kwDOQb6kR87-MqWO | 66,806 | https://github.com/openclaw/openclaw/issues/66806 | https://api.github.com/repos/openclaw/openclaw/issues/66806 | [Bug]: 'No API key found for provider openai' with pure Anthropic-only config (no OpenAI configured) | ## Description
Agent fails with `No API key found for provider "openai"` even though **no OpenAI model or provider is configured at all**. The entire setup is Anthropic Claude CLI OAuth only.
## Environment
- **OpenClaw version:** 2026.4.14 (323493f)
- **OS:** macOS (Darwin 25.4.0, x64)
- **Node:** v25.9.0
## Confi... | closed | completed | true | 2 | [] | [] | 2026-04-14T21:03:11Z | 2026-04-28T04:54:29Z | 2026-04-25T17:53:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tahaqaqbot | 266,382,453 | U_kgDOD-CsdQ | User | false |
openclaw/openclaw | 4,264,761,800 | I_kwDOQb6kR87-MxnI | 66,809 | https://github.com/openclaw/openclaw/issues/66809 | https://api.github.com/repos/openclaw/openclaw/issues/66809 | Browser: local managed loopback CDP tab ops can still fail with 'browser endpoint blocked by policy' after successful host navigation | ## Summary
Two related problems are showing up while trying to restore a host-local Twitter/X workflow on a Linux OpenClaw gateway using the managed `openclaw` browser profile:
1. **Host browser navigation can succeed, but follow-up tab operations still fail with `browser endpoint blocked by policy`.**
2. **Gateway r... | closed | completed | true | 1 | [] | [] | 2026-04-14T21:10:30Z | 2026-04-27T04:39:33Z | 2026-04-25T02:41:21Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | keithce | 2,086,282 | MDQ6VXNlcjIwODYyODI= | User | false |
openclaw/openclaw | 4,264,777,581 | I_kwDOQb6kR87-M1dt | 66,813 | https://github.com/openclaw/openclaw/issues/66813 | https://api.github.com/repos/openclaw/openclaw/issues/66813 | [Bug]: webchat heartbeat with target "last" skips with no-target and blocks background execution | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
In direct webchat sessions, heartbeat configured with `target: "last"` is skipped with `reason: "no-target"`, preventing scheduled background execution.
### Steps to reproduce
1. Open a direct webchat session... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T21:14:46Z | 2026-04-28T04:54:29Z | 2026-04-25T17:53:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Inoguitar | 139,721,251 | U_kgDOCFP6Iw | User | false |
openclaw/openclaw | 4,264,783,156 | I_kwDOQb6kR87-M200 | 66,814 | https://github.com/openclaw/openclaw/issues/66814 | https://api.github.com/repos/openclaw/openclaw/issues/66814 | Internal exec/heartbeat events leak into visible Control UI chat transcript | ## Summary
Internal async exec completion and heartbeat/system wake material is leaking into visible Control UI/webchat chat history as normal chat messages.
This is not just a local rendering issue. The behavior appears to come from upstream event/session classification, with internal exec-event and heartbeat followu... | closed | not_planned | true | 3 | [] | [] | 2026-04-14T21:16:10Z | 2026-04-30T04:40:23Z | 2026-04-26T05:08:39Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mandomaker | 274,945,830 | U_kgDOEGNXJg | User | false |
openclaw/openclaw | 4,264,812,094 | I_kwDOQb6kR87-M94- | 66,817 | https://github.com/openclaw/openclaw/issues/66817 | https://api.github.com/repos/openclaw/openclaw/issues/66817 | Legacy agent:main:main session appears active in Control UI under dmScope=per-channel-peer | # Draft issue: legacy `agent:main:main` session stays fresh in dashboard under `dmScope=per-channel-peer`
## Summary
When `session.dmScope` is set to `per-channel-peer`, live Slack DM traffic correctly routes to a per-DM session key like `agent:main:slack:direct:<userId>`. However, the legacy main session entry `agent... | closed | completed | true | 1 | [] | [] | 2026-04-14T21:23:25Z | 2026-04-28T04:54:28Z | 2026-04-25T17:53:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | maude-westbrook | 274,621,796 | U_kgDOEF5lZA | User | false |
openclaw/openclaw | 4,264,947,077 | I_kwDOQb6kR87-Ne2F | 66,822 | https://github.com/openclaw/openclaw/issues/66822 | https://api.github.com/repos/openclaw/openclaw/issues/66822 | [Bug]: memorySearch.remote.batch.concurrency = 1 has no effect — indexing floods Ollama with concurrent requests | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
agents.defaults.memorySearch.remote.batch.concurrency = 1 has no effect — OpenClaw sends 30+ concurrent embedding requests during indexing, causing exec sessions to be killed before any chunk is indexed.
### Steps to reproduce... | closed | completed | true | 1 | [
"bug",
"bug:crash"
] | [] | 2026-04-14T21:55:09Z | 2026-04-29T04:33:00Z | 2026-04-27T01:47:56Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | LyraInTheFlesh | 260,810,840 | U_kgDOD4uoWA | User | false |
openclaw/openclaw | 4,264,997,178 | I_kwDOQb6kR87-NrE6 | 66,827 | https://github.com/openclaw/openclaw/issues/66827 | https://api.github.com/repos/openclaw/openclaw/issues/66827 | Telegram managed hooks do not expose inbound media refs for DM photo/document messages | ## Summary
On OpenClaw `v2026.4.14`, managed Telegram hooks are firing for inbound DM media, but the hook payloads available to repo-level hook/plugin code do not expose usable media refs or attachments early enough to intercept heavy visual turns.
In practice:
- `message:received` fires
- `message:preprocessed` fire... | closed | completed | true | 1 | [] | [] | 2026-04-14T22:08:34Z | 2026-04-18T04:14:23Z | 2026-04-15T12:38:21Z | NONE | null | 20260418T042035Z | 2026-04-18T04:20:35Z | FayAndXan | 111,750,147 | U_kgDOBqksAw | User | false |
openclaw/openclaw | 4,264,997,768 | I_kwDOQb6kR87-NrOI | 66,828 | https://github.com/openclaw/openclaw/issues/66828 | https://api.github.com/repos/openclaw/openclaw/issues/66828 | Cron lane remapped to Nested and subagent cleanup packaging regression in v2026.4.12 | ## Summary
Two runtime bugs are present in OpenClaw v2026.4.12:
1. **Cron lane silently remapped to Nested**, which effectively behaves like concurrency 1 and ignores `cron.maxConcurrentRuns`
2. **Subagent cleanup-finalize path appears to depend on a missing runtime shim / packaging artifact**, causing cleanup failure... | closed | completed | true | 2 | [] | [] | 2026-04-14T22:08:42Z | 2026-04-30T04:35:13Z | 2026-04-27T09:58:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | chachi-max | 258,696,601 | U_kgDOD2tlmQ | User | false |
openclaw/openclaw | 4,265,000,330 | I_kwDOQb6kR87-Nr2K | 66,830 | https://github.com/openclaw/openclaw/issues/66830 | https://api.github.com/repos/openclaw/openclaw/issues/66830 | [Bug]: Compaction dead zone: `reserveTokens` vs `reserveTokensFloor` asymmetry (cf. closed #25880) strands long-running sessions on fallback models | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On OpenClaw 2026.4.12, the memoryFlush trigger threshold is computed against compaction.reserveTokensFloor instead of compaction.reserveTokens, so raising reserveTokens above the floor to force earlier compact... | open | null | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T22:09:23Z | 2026-05-03T11:25:18Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | chrispatil | 116,297,895 | U_kgDOBu6Qpw | User | false |
openclaw/openclaw | 4,265,024,419 | I_kwDOQb6kR87-Nxuj | 66,832 | https://github.com/openclaw/openclaw/issues/66832 | https://api.github.com/repos/openclaw/openclaw/issues/66832 | [Bug]: `logs.tail` likely fails to redact several credential formats before returning log lines to `operator.read` clients. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
`logs.tail` likely fails to redact several credential and assertion formats before returning log lines to `operator.read` clients. The current regex-based redaction covers some common token patterns, but it doe... | open | null | false | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T22:15:36Z | 2026-05-02T21:25:37Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | ai-hpc | 183,861,985 | U_kgDOCvWC4Q | User | false |
openclaw/openclaw | 4,265,027,503 | I_kwDOQb6kR87-Nyev | 66,833 | https://github.com/openclaw/openclaw/issues/66833 | https://api.github.com/repos/openclaw/openclaw/issues/66833 | [Bug]: OpenRouter/Qwen3 stream parsing fails — reasoning_details field not handled, causes payloads=0 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
## Summary
When using `openrouter/qwen/qwen3-235b-a22b` (or any Qwen3 model via OpenRouter),
every response fails with:
incomplete turn detected: runId=... stopReason=stop payloads=0 — surfacing error to use... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T22:16:23Z | 2026-04-17T04:21:59Z | 2026-04-15T02:45:59Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | bickers76 | 206,468,922 | U_kgDODE53Og | User | false |
openclaw/openclaw | 4,265,030,871 | I_kwDOQb6kR87-NzTX | 66,834 | https://github.com/openclaw/openclaw/issues/66834 | https://api.github.com/repos/openclaw/openclaw/issues/66834 | Discord: /lossless native command description exceeds 100 char limit | Hi, and thanks for OpenClaw.
I noticed the Discord native command `/lossless` currently registers with a description that is too long for Discord's 100 character limit. The log shows it being truncated from 186 to 100 characters.
Everything still worked, but I thought it would be helpful to report since the descripti... | closed | not_planned | true | 2 | [] | [] | 2026-04-14T22:17:17Z | 2026-04-28T04:54:27Z | 2026-04-25T17:54:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | holgergruenhagen | 60,438,604 | MDQ6VXNlcjYwNDM4NjA0 | User | false |
openclaw/openclaw | 4,265,048,569 | I_kwDOQb6kR87-N3n5 | 66,839 | https://github.com/openclaw/openclaw/issues/66839 | https://api.github.com/repos/openclaw/openclaw/issues/66839 | [Bug]: macOS node on 2026.4.11 does not advertise system.run.prepare, breaking exec host=node | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
`exec host=node` fails on a paired + connected macOS node because the gateway says the node does not support `system.run.prepare`.
This looks like a mismatch/bug in the macOS node capability advertisement path... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-14T22:21:49Z | 2026-04-30T04:40:21Z | 2026-04-26T05:08:44Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ogs5soakz-del | 276,125,228 | U_kgDOEHVWLA | User | false |
openclaw/openclaw | 4,265,051,521 | I_kwDOQb6kR87-N4WB | 66,840 | https://github.com/openclaw/openclaw/issues/66840 | https://api.github.com/repos/openclaw/openclaw/issues/66840 | Install scanner blocks plugins that ship unit tests containing env-var + fetch mocks | ## Summary
On OpenClaw 2026.4.14 the install-time static scanner was upgraded from **warn** to **block** when it detects `process.env.X` access + `fetch()` calls in the same file. This breaks installation of any plugin whose ClawHub-stored archive contains test files that mock environment variables and the network, wh... | closed | not_planned | true | 2 | [] | [] | 2026-04-14T22:22:36Z | 2026-04-30T04:40:20Z | 2026-04-26T05:08:49Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | saurabhjain1592 | 8,345,163 | MDQ6VXNlcjgzNDUxNjM= | User | false |
openclaw/openclaw | 4,265,060,636 | I_kwDOQb6kR87-N6kc | 66,841 | https://github.com/openclaw/openclaw/issues/66841 | https://api.github.com/repos/openclaw/openclaw/issues/66841 | openclaw logs --follow can show stale/misleading old-version file logs after side-by-side cutover | ## Summary
After side-by-side cutover, the live gateway service was running from the sidecar tree and reporting `2026.4.14`, but `openclaw logs --follow` still showed `2026.3.28-beta.1` and tailed repeated old warning lines from the rolling log file. This was operationally misleading during incident/debugging.
## Obs... | open | null | false | 2 | [] | [] | 2026-04-14T22:25:07Z | 2026-05-02T20:57:26Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | UGDev-Chane | 180,237,537 | U_kgDOCr404Q | User | false |
openclaw/openclaw | 4,265,121,304 | I_kwDOQb6kR87-OJYY | 66,848 | https://github.com/openclaw/openclaw/issues/66848 | https://api.github.com/repos/openclaw/openclaw/issues/66848 | [Bug]: TypeError: Cannot read properties of undefined (reading 'trim') | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
Running installation:
`bun add -g openclaw@latest`
Then:
`openclaw onboard --install-daemon`
### Steps to reproduce
OpenClaw 2026.4.14 (323493f)
Running installation:
`bun add -g openclaw@latest`
Then:
`openclaw onboar... | closed | completed | true | 5 | [
"bug",
"close:already-fixed",
"dedupe:child",
"bug:crash"
] | [
"vincentkoc"
] | 2026-04-14T22:40:17Z | 2026-04-28T05:28:54Z | 2026-04-25T09:38:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | phpweb | 4,746,625 | MDQ6VXNlcjQ3NDY2MjU= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.