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 | 4,275,517,835 | I_kwDOQb6kR87-1zmL | 67,649 | https://github.com/openclaw/openclaw/issues/67649 | https://api.github.com/repos/openclaw/openclaw/issues/67649 | [Bug]: TypeError: Cannot read properties of undefined (reading 'trim') on QuickStart | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
QuickStart wizard crashes with TypeError after showing Telegram DM warning, regardless of channel selected.
### Steps to reproduce
1. Run `openclaw`
2. Select any channel (telegram or "Skip for now")
3. Error appears after... | closed | completed | true | 7 | [
"bug",
"close:already-fixed",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-04-16T11:52:47Z | 2026-04-28T05:29:04Z | 2026-04-25T09:37:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | MfBally354 | 199,871,072 | U_kgDOC-nKYA | User | false |
openclaw/openclaw | 4,275,539,401 | I_kwDOQb6kR87-143J | 67,653 | https://github.com/openclaw/openclaw/issues/67653 | https://api.github.com/repos/openclaw/openclaw/issues/67653 | [Bug]: Windows: Gateway startup takes ~60s before Discord comes online | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Windows gateway startup and loading takes 60s before discord actually comes online
### Steps to reproduce
### Version
openclaw 2026.4.14
### OS
Windows 10.0.26200 x64
### Channel
Discord
### Problem
On nat... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-16T11:56:47Z | 2026-04-28T04:52:41Z | 2026-04-25T18:14:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | henpokpok | 91,886,375 | U_kgDOBXoTJw | User | false |
openclaw/openclaw | 4,275,644,893 | I_kwDOQb6kR87-2Snd | 67,656 | https://github.com/openclaw/openclaw/issues/67656 | https://api.github.com/repos/openclaw/openclaw/issues/67656 | [Security]: Config API redaction does not cover browser.cdpUrl paths | Sub-issue of #64046
### Maintainer requirements
> - **Scope:** ensure all config response surfaces apply the same secret-redaction policy.
> - **Deliverable:** unified redaction behavior for `parsed`/`source`/`runtime` style snapshots.
> - **Acceptance criteria:** no plaintext secrets in read-scoped config responses;... | closed | completed | true | 0 | [] | [] | 2026-04-16T12:17:08Z | 2026-04-21T04:22:18Z | 2026-04-18T06:23:00Z | CONTRIBUTOR | null | 20260421T060038Z | 2026-04-21T06:00:38Z | Ziy1-Tan | 49,604,965 | MDQ6VXNlcjQ5NjA0OTY1 | User | false |
openclaw/openclaw | 4,275,674,411 | I_kwDOQb6kR87-2Z0r | 67,658 | https://github.com/openclaw/openclaw/issues/67658 | https://api.github.com/repos/openclaw/openclaw/issues/67658 | [Bug]: Wiki Bridge deletes all sources when memory plugin artifacts unavailable | ## Description
Wiki Bridge mode incorrectly deletes all imported sources when the memory plugin is not exporting public artifacts.
## Steps to reproduce
1. Configure memory-wiki plugin in bridge mode (`vaultMode: bridge`, `bridge.enabled: true`)
2. Run `openclaw wiki bridge import` - works and imports artifacts (e.g... | closed | not_planned | true | 2 | [] | [] | 2026-04-16T12:22:53Z | 2026-04-30T04:39:46Z | 2026-04-26T05:15:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | nacooy-dev | 233,872,905 | U_kgDODfCeCQ | User | false |
openclaw/openclaw | 4,275,677,088 | I_kwDOQb6kR87-2aeg | 67,659 | https://github.com/openclaw/openclaw/issues/67659 | https://api.github.com/repos/openclaw/openclaw/issues/67659 | [Bug]: MS Teams delegated OAuth launcher uses xdg-open on win32 instead of explorer.exe | ## Summary
MS Teams delegated OAuth setup currently launches the browser with `xdg-open` on `win32`, so the native Windows path does not use the repository's normal Windows browser-open behavior.
## Steps to reproduce
1. Stub `process.platform` to `"win32"`.
2. Call `openDelegatedOAuthUrl("https://example.com/oauth")`... | closed | completed | false | 3 | [] | [] | 2026-04-16T12:23:24Z | 2026-05-05T05:25:33Z | 2026-05-05T05:25:33Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | shaun0927 | 70,629,228 | MDQ6VXNlcjcwNjI5MjI4 | User | false |
openclaw/openclaw | 4,275,690,328 | I_kwDOQb6kR87-2dtY | 67,661 | https://github.com/openclaw/openclaw/issues/67661 | https://api.github.com/repos/openclaw/openclaw/issues/67661 | Model-scoped cooldowns are not persisted for incomplete-turn exhaustion paths | ## Problem
When an exhausted/rate-limited model fails through the incomplete-turn paths, OpenClaw can record the auth-profile failure without the active `modelId`.
That breaks model-scoped cooldown behavior for the rest of the same run/session: the provider/profile may be marked bad, but the exhausted model is not pr... | open | null | false | 1 | [] | [] | 2026-04-16T12:26:00Z | 2026-05-03T07:44:54Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | aboimpinto | 1,231,687 | MDQ6VXNlcjEyMzE2ODc= | User | false |
openclaw/openclaw | 4,275,752,212 | I_kwDOQb6kR87-2s0U | 67,667 | https://github.com/openclaw/openclaw/issues/67667 | https://api.github.com/repos/openclaw/openclaw/issues/67667 | [Bug]: /status shows 0% context usage even when session has messages (totalTokens not saved to session) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Bug Description
The /status command always shows Context: 0/264k (0%) even though the session has thousands of messages and has undergone compaction.
Investigation
sessions.json shows totalTokens: null (never set)
Compact... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-16T12:37:34Z | 2026-04-21T04:21:23Z | 2026-04-19T02:36:38Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | skylight-9 | 227,601,607 | U_kgDODZDsxw | User | false |
openclaw/openclaw | 4,275,824,172 | I_kwDOQb6kR87-2-Ys | 67,670 | https://github.com/openclaw/openclaw/issues/67670 | https://api.github.com/repos/openclaw/openclaw/issues/67670 | [Bug] openai-codex provider blocked by Cloudflare JS Challenge when accessed via proxy in mainland China | ## Environment
- OpenClaw version: 2026.4.14
- OS: macOS 26.3 (arm64), Mac mini
- Node.js: v22.22.2
- Network: Mainland China, traffic routed through VPN/proxy (Quantumult X)
## Problem
The `openai-codex` provider (`chatgpt.com/backend-api`) is completely unusable when accessed from mainland China through any proxy no... | open | null | false | 2 | [] | [] | 2026-04-16T12:50:51Z | 2026-05-03T10:41:14Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | juner922 | 57,854,595 | MDQ6VXNlcjU3ODU0NTk1 | User | false |
openclaw/openclaw | 4,275,824,676 | I_kwDOQb6kR87-2-gk | 67,671 | https://github.com/openclaw/openclaw/issues/67671 | https://api.github.com/repos/openclaw/openclaw/issues/67671 | [Feature]: add single-job inspection by id | ### Summary
Add single-job cron inspection by id so callers can fetch one cron job directly instead of listing and filtering the full job set.
### Problem to solve
`openclaw cron` currently only exposes detailed cron job inspection through `cron list --json`, so callers must fetch and filter the entire cron job set ... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-04-16T12:50:57Z | 2026-05-02T10:36:49Z | null | NONE | null | 20260502T120029Z | 2026-05-02T12:00:29Z | zexho994 | 34,532,365 | MDQ6VXNlcjM0NTMyMzY1 | User | false |
openclaw/openclaw | 4,275,826,116 | I_kwDOQb6kR87-2-3E | 67,673 | https://github.com/openclaw/openclaw/issues/67673 | https://api.github.com/repos/openclaw/openclaw/issues/67673 | [Bug] openai-codex provider blocked by Cloudflare JS Challenge when accessed via proxy in mainland China | ## Environment
- OpenClaw version: 2026.4.14
- OS: macOS 26.3 (arm64), Mac mini
- Node.js: v22.22.2
- Network: Mainland China, traffic routed through VPN/proxy (Quantumult X)
## Problem
The `openai-codex` provider (`chatgpt.com/backend-api`) is completely unusable when accessed from mainland China through any proxy no... | closed | not_planned | true | 1 | [] | [] | 2026-04-16T12:51:13Z | 2026-04-30T04:39:45Z | 2026-04-26T05:15:45Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | juner922 | 57,854,595 | MDQ6VXNlcjU3ODU0NTk1 | User | false |
openclaw/openclaw | 4,275,876,614 | I_kwDOQb6kR87-3LMG | 67,675 | https://github.com/openclaw/openclaw/issues/67675 | https://api.github.com/repos/openclaw/openclaw/issues/67675 | Flaky test: src/gateway/server.canvas-auth.test.ts "authorizes canvas HTTP/WS via node-scoped capability" | ## Summary
The test `authorizes canvas HTTP/WS via node-scoped capability and rejects misuse` in `src/gateway/server.canvas-auth.test.ts` has flaked on PR #67096 at least twice with the **same test, same line, same assertion**, but with very different runtimes — suggesting a timing-sensitive race in `fetchCanvas` / `w... | closed | completed | true | 2 | [] | [] | 2026-04-16T13:00:25Z | 2026-04-30T04:39:44Z | 2026-04-26T05:15:50Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jeffchen1981-fu | 260,034,475 | U_kgDOD3_Pqw | User | false |
openclaw/openclaw | 4,275,882,046 | I_kwDOQb6kR87-3Mg- | 67,676 | https://github.com/openclaw/openclaw/issues/67676 | https://api.github.com/repos/openclaw/openclaw/issues/67676 | Slack: file_id and url_private dropped from inbound file_shared events | ## Summary
The Slack plugin (v2026.4.5) receives `file_shared` events with complete metadata including `file_id` and `url_private`, but drops these fields before passing the message context to agents. This makes the `downloadSlackFile` action unreachable for inbound files.
## Current behavior
When a file is shared i... | closed | completed | true | 4 | [] | [] | 2026-04-16T13:01:25Z | 2026-04-28T04:52:40Z | 2026-04-25T18:14:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | cmcejas | 75,821,231 | MDQ6VXNlcjc1ODIxMjMx | User | false |
openclaw/openclaw | 4,275,920,596 | I_kwDOQb6kR87-3V7U | 67,680 | https://github.com/openclaw/openclaw/issues/67680 | https://api.github.com/repos/openclaw/openclaw/issues/67680 | Control UI blank screen: unresolved bare module import for markdown-it-task-lists | ## Bug
After upgrading to OpenClaw 2026.4.14, the Control UI serves a blank page on all browsers (including localhost on the Pi, incognito windows, etc.).
### Root Cause
The built JS bundle `dist/control-ui/assets/index-Dlv1J1hX.js` contains an unresolved bare ES module specifier:
```js
import N from"markdown-it-ta... | open | null | false | 2 | [] | [] | 2026-04-16T13:08:28Z | 2026-05-03T10:41:15Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | hal-molty79 | 260,708,816 | U_kgDOD4oZ0A | User | false |
openclaw/openclaw | 4,275,934,572 | I_kwDOQb6kR87-3ZVs | 67,681 | https://github.com/openclaw/openclaw/issues/67681 | https://api.github.com/repos/openclaw/openclaw/issues/67681 | [Bug] Cron agent session generates duplicate reply despite --no-deliver flag | ## Summary
Cron jobs configured with `--no-deliver` still produce duplicate messages in Telegram when the agent script sends a message via `openclaw message send` AND the cron agent session generates its own reply turn.
## Version
OpenClaw 2026.4.2 (d74a122)
## Steps to Reproduce
1. Create a cron job with `--no-deliv... | closed | completed | true | 3 | [] | [] | 2026-04-16T13:11:03Z | 2026-04-30T04:39:44Z | 2026-04-26T05:15:55Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | chrismclarnon-ep | 275,793,450 | U_kgDOEHBGKg | User | false |
openclaw/openclaw | 4,275,970,138 | I_kwDOQb6kR87-3iBa | 67,682 | https://github.com/openclaw/openclaw/issues/67682 | https://api.github.com/repos/openclaw/openclaw/issues/67682 | Per-agent MCP env var injection for multi-agent identity attribution | ## Summary
In a multi-agent deployment, different agents need to authenticate to the same MCP server (e.g., a project management tool) with different credentials so that writes are attributed to the correct agent identity, not the global admin account.
## Current Behavior
OpenClaw v2026.4.14 uses a single MCP server... | open | null | false | 1 | [] | [] | 2026-04-16T13:17:34Z | 2026-05-03T10:41:17Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | robbyproc87 | 109,332,083 | U_kgDOBoRGcw | User | false |
openclaw/openclaw | 4,275,976,420 | I_kwDOQb6kR87-3jjk | 67,683 | https://github.com/openclaw/openclaw/issues/67683 | https://api.github.com/repos/openclaw/openclaw/issues/67683 | [Bug]: Inability to stop a running task causing infinite Notion page reformatting | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
A task triggered to reformat a Notion page does not stop running and cannot be manually terminated by terminate PowerShell
### Steps to reproduce
Trigger a task that reformats a Notion page.
Observe the task ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-16T13:18:45Z | 2026-04-28T04:52:40Z | 2026-04-25T18:14:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexandre-leng | 72,550,137 | MDQ6VXNlcjcyNTUwMTM3 | User | false |
openclaw/openclaw | 4,276,043,463 | I_kwDOQb6kR87-3z7H | 67,684 | https://github.com/openclaw/openclaw/issues/67684 | https://api.github.com/repos/openclaw/openclaw/issues/67684 | [Bug]: openclaw onboard Cannot read properties of undefines (reading 'trim') | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
After upgrading to the latest version, at the end of onboard that error appears and changes like model switch is not applied
### Steps to reproduce
curl -fsSL https://openclaw.ai/install.sh | bash
openclaw onboard
change mode... | closed | completed | true | 8 | [
"bug",
"close:already-fixed",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-04-16T13:30:59Z | 2026-04-28T05:29:02Z | 2026-04-25T09:37:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ronibandini | 30,272,192 | MDQ6VXNlcjMwMjcyMTky | User | false |
openclaw/openclaw | 4,276,100,316 | I_kwDOQb6kR87-4Bzc | 67,685 | https://github.com/openclaw/openclaw/issues/67685 | https://api.github.com/repos/openclaw/openclaw/issues/67685 | Control UI: Separate system messages from user conversation flow | Currently, background command results (like "Exec completed ...") are injected directly into the conversation stream with "System (untrusted):" prefix. This makes the conversation history hard to read when multiple commands are executed during a session.
**Feature request**:
- Option to hide/collapse system messages
-... | closed | completed | true | 1 | [] | [] | 2026-04-16T13:40:49Z | 2026-04-30T04:39:43Z | 2026-04-26T05:16:00Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ygc3817922006-sketch | 266,983,010 | U_kgDOD-nWYg | User | false |
openclaw/openclaw | 4,276,123,133 | I_kwDOQb6kR87-4HX9 | 67,687 | https://github.com/openclaw/openclaw/issues/67687 | https://api.github.com/repos/openclaw/openclaw/issues/67687 | [Bug]: Group messages not reaching session - requireMention: false not working (Feishu) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
**Environment:**
- OS: Windows
- OpenClaw Version: 2026.4.14
- Feishu/Lark
**Bug Description:**
After upgrading from OpenClaw 3.28 to 4.x, group messages from a specific Feishu group (with `requireMention: false` configure... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-16T13:44:49Z | 2026-04-30T04:33:15Z | 2026-04-27T20:19:51Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Eunice6199 | 209,090,929 | U_kgDODHZ5cQ | User | false |
openclaw/openclaw | 4,276,127,809 | I_kwDOQb6kR87-4IhB | 67,688 | https://github.com/openclaw/openclaw/issues/67688 | https://api.github.com/repos/openclaw/openclaw/issues/67688 | openai-codex on macOS host returns Cloudflare challenge HTML from chatgpt.com/backend-api/v1/codex/responses | ## Summary
On one specific Mac host, `openai-codex` in OpenClaw consistently fails with a Cloudflare challenge HTML response instead of a model response. This reproduces across current and older OpenClaw versions, while Safari on the same machine can access `chatgpt.com` normally and the user is logged in with Pro.
##... | closed | completed | true | 1 | [] | [] | 2026-04-16T13:45:34Z | 2026-04-24T04:26:22Z | 2026-04-21T06:42:14Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | mondsau | 26,148,500 | MDQ6VXNlcjI2MTQ4NTAw | User | false |
openclaw/openclaw | 4,276,130,483 | I_kwDOQb6kR87-4JKz | 67,689 | https://github.com/openclaw/openclaw/issues/67689 | https://api.github.com/repos/openclaw/openclaw/issues/67689 | Feature request: Session management UI (hide/archive/filter sessions) | Currently, there's no built-in way to manage session visibility in the Control UI. Users who have many sessions (e.g., auto-generated sessions like 'dreaming-narrative-*') end up with cluttered session lists.
**Current workaround**:
Manually edit `sessions.json` to remove entries, which is not user-friendly and requir... | closed | completed | true | 1 | [] | [] | 2026-04-16T13:45:59Z | 2026-04-30T04:39:42Z | 2026-04-26T05:16:05Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ygc3817922006-sketch | 266,983,010 | U_kgDOD-nWYg | User | false |
openclaw/openclaw | 4,276,136,588 | I_kwDOQb6kR87-4KqM | 67,690 | https://github.com/openclaw/openclaw/issues/67690 | https://api.github.com/repos/openclaw/openclaw/issues/67690 | Feature request: Improve context pruning/trimming mechanism visibility and control | OpenClaw has internal context pruning/trimming to manage token limits, but users have limited visibility and control over this process.
**Current issues observed**:
1. **Context bloat from tool outputs**: Large tool results (especially from background processes, images, etc.) consume significant tokens, but users don... | open | null | false | 1 | [] | [] | 2026-04-16T13:46:59Z | 2026-05-03T10:41:18Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | ygc3817922006-sketch | 266,983,010 | U_kgDOD-nWYg | User | false |
openclaw/openclaw | 4,276,153,389 | I_kwDOQb6kR87-4Owt | 67,691 | https://github.com/openclaw/openclaw/issues/67691 | https://api.github.com/repos/openclaw/openclaw/issues/67691 | Feature request: Implement local context pruning independent of model API support | Currently, context pruning/trimming appears to depend on model-specific features like Anthropic's prompt caching API. This means users of other models (OpenAI, Gemini, local models, etc.) may not have effective pruning support.
**Problem**:
Context pruning should be implementable locally without relying on model-spec... | closed | completed | true | 1 | [] | [] | 2026-04-16T13:49:28Z | 2026-04-28T04:52:39Z | 2026-04-25T18:14:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ygc3817922006-sketch | 266,983,010 | U_kgDOD-nWYg | User | false |
openclaw/openclaw | 4,276,261,391 | I_kwDOQb6kR87-4pIP | 67,696 | https://github.com/openclaw/openclaw/issues/67696 | https://api.github.com/repos/openclaw/openclaw/issues/67696 | [Bug]: Outbound messages contain [object Object] serialization artifacts in Telegram and Mattermost | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Version: 2026.4.14
Channels affected: Mattermost, Telegram (confirmed on both)
Model: mistral/mistral-small-latest
Description:
Responses from the agent contain large blocks of [object Object]
text mixed in w... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-16T14:05:54Z | 2026-04-30T04:39:41Z | 2026-04-26T05:16:10Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kaiharuto | 35,682,494 | MDQ6VXNlcjM1NjgyNDk0 | User | false |
openclaw/openclaw | 4,276,294,934 | I_kwDOQb6kR87-4xUW | 67,698 | https://github.com/openclaw/openclaw/issues/67698 | https://api.github.com/repos/openclaw/openclaw/issues/67698 | [Bug]: OpenRouter API Key Configuration Fails - Incomplete Turn Detection (payloads=0) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
Yes
### Summary
OpenClaw accepts OpenRouter API key configuration during setup but agent responses fail with "incomplete turn detected: payloads=0" error despite valid API key confirmed via direct API calls.
### Steps to rep... | closed | completed | true | 7 | [
"bug",
"bug:behavior"
] | [] | 2026-04-16T14:11:47Z | 2026-04-21T04:22:07Z | 2026-04-18T16:01:59Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | BeastxD7 | 120,053,738 | U_kgDOByff6g | User | false |
openclaw/openclaw | 4,276,325,719 | I_kwDOQb6kR87-441X | 67,700 | https://github.com/openclaw/openclaw/issues/67700 | https://api.github.com/repos/openclaw/openclaw/issues/67700 | Feature Request: Built-in Mock Provider for Regression Testing + MCP Lifecycle Hardening (Duplicate) | ## Feature Request Summary
Requesting two complementary features to improve OpenClaw's testing infrastructure and MCP reliability:
### 1. Built-in Mock Provider for Regression Testing
**Motivation:**
Current testing relies on real API calls which are:
- Expensive (token costs)
- Slow (network latency)
- Flaky (rate ... | closed | completed | true | 5 | [] | [] | 2026-04-16T14:16:50Z | 2026-04-20T04:26:33Z | 2026-04-16T14:24:20Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | woai36D | 269,974,840 | U_kgDOEBd9OA | User | false |
openclaw/openclaw | 4,276,351,714 | I_kwDOQb6kR87-4_Li | 67,703 | https://github.com/openclaw/openclaw/issues/67703 | https://api.github.com/repos/openclaw/openclaw/issues/67703 | gateway.auth.mode=none breaks internal RPC probe with 1008 device identity required | ## Summary
When `gateway.auth.mode` is set to `none`, the Gateway's own internal RPC probe fails with:
```
gateway closed (1008): device identity required
RPC probe: failed
```
This causes a cascading failure chain:
1. All `exec approval followup` dispatches timeout after 60s
2. Frontend detects `seq-gap` → triggers... | open | null | false | 1 | [] | [] | 2026-04-16T14:21:03Z | 2026-05-03T10:41:19Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | xiew4589-lang | 254,248,192 | U_kgDODyeFAA | User | false |
openclaw/openclaw | 4,276,365,747 | I_kwDOQb6kR87-5Cmz | 67,705 | https://github.com/openclaw/openclaw/issues/67705 | https://api.github.com/repos/openclaw/openclaw/issues/67705 | formatDocsLink crashes on undefined docsPath from channel plugins | Channel plugins that don't set `docsPath` in their meta cause a TypeError in `formatDocsLink` during `openclaw configure`:
```
TypeError: Cannot read properties of undefined (reading 'trim')
at formatDocsLink (dist/links-BlUEqVE8.js:7:23)
at formatChannelSelectionLine (dist/registry-DE4nHTjg.js:111:71)
at ... | closed | completed | true | 2 | [] | [] | 2026-04-16T14:23:24Z | 2026-04-28T04:52:38Z | 2026-04-25T18:15:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 0xNagato | 116,608,222 | U_kgDOBvNM3g | User | false |
openclaw/openclaw | 4,276,485,450 | I_kwDOQb6kR87-5f1K | 67,709 | https://github.com/openclaw/openclaw/issues/67709 | https://api.github.com/repos/openclaw/openclaw/issues/67709 | [Bug]: WebChat Control UI extremely slow startup (chat.history 352s, models.list 352s) | ## Bug Description
WebChat Control UI takes 10+ seconds to become responsive after opening. Observed on this machine:
- **chat.history** RPC call: ~352 seconds (~6 minutes)
- **models.list** RPC call: ~352 seconds (~6 minutes)
This causes the UI to appear frozen/laggy for a long time before messages can be sent.
##... | closed | completed | true | 3 | [] | [] | 2026-04-16T14:42:55Z | 2026-05-05T04:26:50Z | 2026-05-02T16:15:09Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | tangda18 | 264,406,045 | U_kgDOD8KEHQ | User | false |
openclaw/openclaw | 4,276,507,175 | I_kwDOQb6kR87-5lIn | 67,710 | https://github.com/openclaw/openclaw/issues/67710 | https://api.github.com/repos/openclaw/openclaw/issues/67710 | Anthropic Claude Opus 4.7 not selectable yet, resolves as configured+missing | ## Summary
Anthropic released Claude Opus 4.7 today, but OpenClaw does not appear to support selecting it yet through the normal Anthropic provider flow.
## What I observed
- Attempted to select `anthropic/claude-opus-4-7`
- OpenClaw accepted the config change, but `openclaw models list` showed:
- `anthropic/claude-... | closed | completed | true | 5 | [] | [] | 2026-04-16T14:46:13Z | 2026-04-28T04:52:37Z | 2026-04-25T18:15:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,276,527,603 | I_kwDOQb6kR87-5qHz | 67,712 | https://github.com/openclaw/openclaw/issues/67712 | https://api.github.com/repos/openclaw/openclaw/issues/67712 | openai-codex HTML/rate-limit responses are sometimes surfaced as 'DNS lookup for the provider endpoint failed' | ## Summary
`openai-codex/*` real runs can fail with HTML/rate-limit/challenge-like upstream responses, but OpenClaw sometimes surfaces the failure as:
`LLM request failed: DNS lookup for the provider endpoint failed.`
On this machine, host DNS appears healthy, short live probes can succeed, and the misleading DNS me... | closed | completed | true | 3 | [] | [] | 2026-04-16T14:49:30Z | 2026-04-28T04:52:37Z | 2026-04-25T18:15:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | habundance | 245,961,480 | U_kgDODqkTCA | User | false |
openclaw/openclaw | 4,276,588,843 | I_kwDOQb6kR87-55Er | 67,714 | https://github.com/openclaw/openclaw/issues/67714 | https://api.github.com/repos/openclaw/openclaw/issues/67714 | [Showcase] Autonomagic x402 API - 10 AI utility micropayment endpoints on Base | ## Showcase: Autonomagic x402 API
**URL:** https://3000-c905553a9700f51123f091260a9b8400.life.conway.tech
**Protocol:** x402 (HTTP 402 Payment Required)
**Network:** Base mainnet (eip155:8453), USDC
**Discovery:** `/.well-known/x402.json`
10 AI agent utility APIs with x402 micropayments:
- /api/hash, /api/uuid, /api/... | closed | not_planned | true | 2 | [] | [] | 2026-04-16T15:00:00Z | 2026-04-28T04:52:35Z | 2026-04-25T18:15:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | premsreelathasugeendran | 271,522,295 | U_kgDOEC8Z9w | User | false |
openclaw/openclaw | 4,276,640,614 | I_kwDOQb6kR87-6Ftm | 67,716 | https://github.com/openclaw/openclaw/issues/67716 | https://api.github.com/repos/openclaw/openclaw/issues/67716 | [Bug]: bootstrap/reconcile and hot-cache policy can leave deferred compaction debt stranded | ## Bug type
Behavior bug (incorrect runtime state without crash)
## Beta release blocker
No
## Summary
Context-engine / LCM maintenance can leave deferred compaction debt stranded in at least two ways:
1. A bootstrap/reconcile path can import missing tail messages, create deferred compaction debt, and then immedi... | open | null | false | 4 | [] | [] | 2026-04-16T15:09:01Z | 2026-05-03T16:38:51Z | null | NONE | null | 20260503T180029Z | 2026-05-03T18:00:29Z | shiv1213-fin | 262,336,318 | U_kgDOD6LvPg | User | false |
openclaw/openclaw | 4,276,655,913 | I_kwDOQb6kR87-6Jcp | 67,718 | https://github.com/openclaw/openclaw/issues/67718 | https://api.github.com/repos/openclaw/openclaw/issues/67718 | Cron sessions with deleteAfterRun: true not being deleted — causes context overflow | ## Summary
Cron jobs configured with `sessionTarget: "isolated"` and `deleteAfterRun: true` are not being deleted after execution. Sessions accumulate in `sessions.json`, inherit context from previous runs, and eventually overflow with `Context overflow: prompt too large for the model`.
This breaks all cron-based auto... | closed | completed | true | 1 | [] | [] | 2026-04-16T15:11:45Z | 2026-04-21T04:22:15Z | 2026-04-18T10:17:20Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | clawdieclawdita | 259,446,051 | U_kgDOD3bVIw | User | false |
openclaw/openclaw | 4,276,689,182 | I_kwDOQb6kR87-6Rke | 67,719 | https://github.com/openclaw/openclaw/issues/67719 | https://api.github.com/repos/openclaw/openclaw/issues/67719 | Per-agent models.json catalog generator embeds plaintext credentials for OAuth/device-auth providers | ## Summary
When a provider is authenticated via OAuth device flow (e.g. OpenAI/Codex via device auth), the per-agent generated models.json catalog embeds the credential material (OAuth token) as a plaintext apiKey value instead of a SecretRef.
## Expected behavior
The catalog generator should either:
1. Emit a SecretR... | closed | completed | true | 3 | [] | [] | 2026-04-16T15:17:57Z | 2026-04-28T04:52:34Z | 2026-04-25T18:15:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sanchezm86 | 66,266,304 | MDQ6VXNlcjY2MjY2MzA0 | User | false |
openclaw/openclaw | 4,276,744,931 | I_kwDOQb6kR87-6fLj | 67,724 | https://github.com/openclaw/openclaw/issues/67724 | https://api.github.com/repos/openclaw/openclaw/issues/67724 | [Bug]: interactiveReplies button clicks not waking agent session (event received, no dispatch) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After enabling `interactiveReplies`, Slack button clicks are received by the gateway and logged, but no agent session turn is triggered in response.
### Steps to reproduce
1. Add `"interactiveReplies"` to `channels.slack.c... | closed | not_planned | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-16T15:27:53Z | 2026-04-30T04:39:40Z | 2026-04-26T05:16:15Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | robin-crow | 276,652,769 | U_kgDOEH1i4Q | User | false |
openclaw/openclaw | 4,276,804,927 | I_kwDOQb6kR87-6t0_ | 67,726 | https://github.com/openclaw/openclaw/issues/67726 | https://api.github.com/repos/openclaw/openclaw/issues/67726 | Bug: TTS auto mode not triggering despite correct config (messages.tts.auto: always/inbound) | ## Description
Setting `messages.tts.auto` to `always` or `inbound` in `openclaw.json` does not trigger TTS on any replies. The config is read and applied (visible in reload logs) but no TTS conversion is ever attempted.
## Environment
- OpenClaw: 2026.4.14
- OS: Ubuntu 24.04 on AWS EC2
- Model: zai/glm-5.1
- Channel... | closed | completed | true | 2 | [] | [] | 2026-04-16T15:38:36Z | 2026-04-27T04:32:51Z | 2026-04-25T04:20:46Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | znhe | 45,541,213 | MDQ6VXNlcjQ1NTQxMjEz | User | false |
openclaw/openclaw | 4,276,826,431 | I_kwDOQb6kR87-6zE_ | 67,728 | https://github.com/openclaw/openclaw/issues/67728 | https://api.github.com/repos/openclaw/openclaw/issues/67728 | Browser: Remote CDP WebSocket needs auto-reconnect | ## Problem
When using OpenClaw's browser tool with remote CDP (e.g., Chrome on Windows from WSL2), the CDP WebSocket connection is "one-shot": each browser action opens a new WebSocket, sends one command, and closes. If the connection drops mid-session, every subsequent action fails with "CDP socket closed" until the ... | closed | completed | true | 4 | [] | [] | 2026-04-16T15:42:20Z | 2026-05-03T04:37:24Z | 2026-04-30T13:50:19Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | KouJun01 | 160,488,362 | U_kgDOCZDbqg | User | false |
openclaw/openclaw | 4,276,866,987 | I_kwDOQb6kR87-68-r | 67,730 | https://github.com/openclaw/openclaw/issues/67730 | https://api.github.com/repos/openclaw/openclaw/issues/67730 | [Bug]: Dreaming/Session-memory creates sessions but never cleans them up | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Problem
When `dreaming` (or `session-memory` hook for short-term recall) runs, it creates new agent sessions but never cleans them up afterward. This causes session files to accumulate over time.
### Steps to reproduc... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-16T15:49:11Z | 2026-04-28T04:52:34Z | 2026-04-25T18:15:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Countjump | 187,282,676 | U_kgDOCym09A | User | false |
openclaw/openclaw | 4,276,910,306 | I_kwDOQb6kR87-7Hji | 67,732 | https://github.com/openclaw/openclaw/issues/67732 | https://api.github.com/repos/openclaw/openclaw/issues/67732 | Codex/GPT-5.4: Cloudflare WAF blocks token-only API requests | ## Summary
OpenClaw's Codex (openai-codex/gpt-5.4) integration is broken due to Cloudflare WAF protection on chatgpt.com. Token-based requests are blocked; only browser contexts pass.
## Error Pattern
```
[agent/embedded] embedded run agent end: runId=... isError=true model=gpt-5.4 provider=openai-codex error=LLM re... | closed | completed | true | 1 | [] | [] | 2026-04-16T15:56:32Z | 2026-04-24T04:26:28Z | 2026-04-21T06:40:11Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | Lairdd | 196,864,871 | U_kgDOC7vrZw | User | false |
openclaw/openclaw | 4,276,997,766 | I_kwDOQb6kR87-7c6G | 67,735 | https://github.com/openclaw/openclaw/issues/67735 | https://api.github.com/repos/openclaw/openclaw/issues/67735 | WebChat / Control UI does not honor canonical direct-session identityLinks the way Telegram DMs do | With `session.dmScope = per-peer` and `session.identityLinks` linking `webchat:default` to a Telegram DM identity, Telegram continuity resolves into a canonical direct session, but WebChat / Control UI still defaults to the agent main session (`agent:main:main`).
This makes WebChat and Telegram behave like separate di... | open | null | false | 6 | [] | [] | 2026-04-16T16:10:47Z | 2026-05-02T20:26:09Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | bobopenclawson-bit | 263,495,697 | U_kgDOD7SgEQ | User | false |
openclaw/openclaw | 4,277,003,195 | I_kwDOQb6kR87-7eO7 | 67,737 | https://github.com/openclaw/openclaw/issues/67737 | https://api.github.com/repos/openclaw/openclaw/issues/67737 | [Bug]: Beta blocker: otel-observability — api.on() typed hooks never fire for WhatsApp channel messages | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Typed hooks registered via api.on() (e.g. message_received , before_agent_start , agent_end ) never execute when messages arrive through the WhatsApp channel, because the WhatsApp message handler bypasses the ho... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-16T16:11:45Z | 2026-04-28T04:52:33Z | 2026-04-25T18:16:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | henrikrexed | 81,800,915 | MDQ6VXNlcjgxODAwOTE1 | User | false |
openclaw/openclaw | 4,277,025,346 | I_kwDOQb6kR87-7jpC | 67,739 | https://github.com/openclaw/openclaw/issues/67739 | https://api.github.com/repos/openclaw/openclaw/issues/67739 | [Bug]: Slack inbound routing ignores runtime conversation bindings — /focus succeeds but messages fall through to agent shell | ## Summary
On Slack, `/focus <session-key>` successfully writes a runtime conversation binding to `~/.openclaw/bindings/current-conversations.json`, but **subsequent inbound messages in the bound channel are not routed to the target ACP session**. They fall through to the agent shell's normal pipeline instead. The bin... | closed | completed | true | 1 | [] | [] | 2026-04-16T16:15:52Z | 2026-04-25T04:16:34Z | 2026-04-22T21:30:09Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | Frankla20 | 265,706,740 | U_kgDOD9Zc9A | User | false |
openclaw/openclaw | 4,277,055,198 | I_kwDOQb6kR87-7q7e | 67,740 | https://github.com/openclaw/openclaw/issues/67740 | https://api.github.com/repos/openclaw/openclaw/issues/67740 | [Bug]: openai-codex OAuth requests fail with store:true/stream:false mismatch against /backend-api/codex/responses — error classifier mislabels as "DNS lookup failed" | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Summary
OpenClaw 2026.4.14 sends malformed requests to the Codex OAuth endpoint
(`https://chatgpt.com/backend-api/codex/responses`) that are rejected by OpenAI
with HTTP 400 validation errors. After retry/backoff, Clou... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-16T16:21:08Z | 2026-04-19T04:22:50Z | 2026-04-17T02:33:21Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | zacohara | 269,650,611 | U_kgDOEBKKsw | User | false |
openclaw/openclaw | 4,277,072,892 | I_kwDOQb6kR87-7vP8 | 67,741 | https://github.com/openclaw/openclaw/issues/67741 | https://api.github.com/repos/openclaw/openclaw/issues/67741 | [Feature]: OPENCLAW KEEPS FALLBACKING TO OPUS 4.6 UPDATE IT TO OPUS 4.7 ASAP PLEASE | ### Summary
PENCLAW KEEPS FALLBACKING TO OPUS 4.6 UPDATE IT TO OPUS 4.7 ASAP PLEASE
<img width="3420" height="2224" alt="Image" src="https://github.com/user-attachments/assets/9dd30832-aa1c-4537-b93b-979f07f8f755" />
### Problem to solve
OPUS 4.7 UPDATE ASAP
### Proposed solution
OPUS 4.7 COMPATIBLE
### Alter... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-16T16:24:09Z | 2026-04-28T04:52:32Z | 2026-04-25T18:16:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | FLAAPY | 165,365,346 | U_kgDOCdtGYg | User | false |
openclaw/openclaw | 4,277,131,786 | I_kwDOQb6kR87-79oK | 67,742 | https://github.com/openclaw/openclaw/issues/67742 | https://api.github.com/repos/openclaw/openclaw/issues/67742 | [Bug]: OPENCLAW KEEPS FALLBACKING TO OPUS 4.6 UPDATE IT TO OPUS 4.7 ASAP PLEASE | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
<!-- Failed to upload "Screenshot 2026-04-16 at 9.53.05 PM.png" -->
[Bug]: OPENCLAW KEEPS FALLBACKING TO OPUS 4.6 UPDATE IT TO OPUS 4.7 ASAP PLEASE
### Steps to reproduce
[Bug]: OPENCLAW KEEPS FALLBACKING TO OPUS 4.6 UPD... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-16T16:34:03Z | 2026-04-28T04:52:31Z | 2026-04-25T18:16:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | FLAAPY | 165,365,346 | U_kgDOCdtGYg | User | false |
openclaw/openclaw | 4,277,168,548 | I_kwDOQb6kR87-8Gmk | 67,744 | https://github.com/openclaw/openclaw/issues/67744 | https://api.github.com/repos/openclaw/openclaw/issues/67744 | Agent infinite-loops on unconfigured tts — three stacked issues | ### Summary
When a Telegram user asks the agent to use voice / TTS and no TTS provider is configured, the agent enters an infinite tool-call loop. In a real deployment it produced 275 identical assistant messages in ~10 minutes before context overflow forced auto-compaction (which then continued the loop). The root ca... | closed | completed | true | 10 | [] | [] | 2026-04-16T16:40:13Z | 2026-04-20T04:26:29Z | 2026-04-18T00:43:46Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | sukhdeepjohar | 46,626,869 | MDQ6VXNlcjQ2NjI2ODY5 | User | false |
openclaw/openclaw | 4,277,175,174 | I_kwDOQb6kR87-8IOG | 67,745 | https://github.com/openclaw/openclaw/issues/67745 | https://api.github.com/repos/openclaw/openclaw/issues/67745 | [Bug]: Custom OpenAI-compatible provider returns valid tool_calls directly, but OpenClaw never executes tools | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
OpenClaw does not execute tool calls when using a custom OpenAI-compatible provider, even though the provider returns valid `tool_calls` in direct raw requests.
The agent verbally acknowledges the action ("I am writing the ... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-16T16:41:29Z | 2026-04-30T04:39:40Z | 2026-04-26T05:16:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | twinkway | 34,943,968 | MDQ6VXNlcjM0OTQzOTY4 | User | false |
openclaw/openclaw | 4,277,177,561 | I_kwDOQb6kR87-8IzZ | 67,746 | https://github.com/openclaw/openclaw/issues/67746 | https://api.github.com/repos/openclaw/openclaw/issues/67746 | [Bug]: openclaw channels logout and remove do not clear WhatsApp session from openclaw.json or active memory | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When attempting to completely remove a WhatsApp number/channel from OpenClaw, the CLI commands openclaw channels logout --channel whatsapp and openclaw channels remove --channel whatsapp do not successfully clean up the env... | closed | completed | true | 4 | [
"bug"
] | [] | 2026-04-16T16:41:56Z | 2026-05-04T04:35:24Z | 2026-05-01T21:37:54Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | 123Mismail | 162,455,919 | U_kgDOCa7hbw | User | false |
openclaw/openclaw | 4,277,195,434 | I_kwDOQb6kR87-8NKq | 67,747 | https://github.com/openclaw/openclaw/issues/67747 | https://api.github.com/repos/openclaw/openclaw/issues/67747 | Gateway restart strips askFallback from exec-approvals.json, silently breaking cron exec | ## Description
The gateway process rewrites `~/.openclaw/exec-approvals.json` on restart but its serializer drops the `askFallback` key. This silently degrades the effective exec policy for cron jobs, causing them to be denied with no way to recover without manual intervention.
This is a recurring issue — every `open... | closed | not_planned | true | 2 | [] | [] | 2026-04-16T16:45:01Z | 2026-04-28T04:52:31Z | 2026-04-25T18:16:35Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | groksrc | 9,099,311 | MDQ6VXNlcjkwOTkzMTE= | User | false |
openclaw/openclaw | 4,277,247,986 | I_kwDOQb6kR87-8Z_y | 67,749 | https://github.com/openclaw/openclaw/issues/67749 | https://api.github.com/repos/openclaw/openclaw/issues/67749 | resolveFirstBoundAccountId should honor guild/team/role constraints on AgentRouteBinding | ### Context
`resolveFirstBoundAccountId` in `src/routing/bound-account-read.ts` matches `AgentRouteBinding`s by `channel`, `agentId`, and (as of #67508) `peer.id` + `peer.kind`. It does **not** honor the remaining constraint fields on `AgentBindingMatch`:
- `guildId` (Discord server)
- `teamId` (Slack/Teams workspace... | closed | completed | true | 2 | [] | [] | 2026-04-16T16:54:23Z | 2026-04-21T04:21:44Z | 2026-04-18T20:48:49Z | CONTRIBUTOR | null | 20260421T060038Z | 2026-04-21T06:00:38Z | lukeboyett | 46,942,646 | MDQ6VXNlcjQ2OTQyNjQ2 | User | false |
openclaw/openclaw | 4,277,273,903 | I_kwDOQb6kR87-8gUv | 67,750 | https://github.com/openclaw/openclaw/issues/67750 | https://api.github.com/repos/openclaw/openclaw/issues/67750 | [Bug]: Successful auto-compaction can still end in a 120s embedded timeout and generic `/new` fallback | ## Summary
On a long-running embedded-agent session, OpenClaw can:
1. detect context overflow,
2. successfully auto-compact the session,
3. retry the prompt,
4. hit a 120s LLM idle timeout on the retried run,
5. exhaust fallback candidates,
6. and still collapse the user-facing result into the generic:
`⚠️ Something... | open | null | false | 1 | [] | [] | 2026-04-16T16:59:06Z | 2026-05-03T10:41:20Z | null | CONTRIBUTOR | null | 20260503T120038Z | 2026-05-03T12:00:38Z | Jackten | 2,895,479 | MDQ6VXNlcjI4OTU0Nzk= | User | false |
openclaw/openclaw | 4,277,335,239 | I_kwDOQb6kR87-8vTH | 67,752 | https://github.com/openclaw/openclaw/issues/67752 | https://api.github.com/repos/openclaw/openclaw/issues/67752 | bluebubbles: tighten SSRF guard on API fetches (bypass via undefined SsrFPolicy) | ## Summary
Aisle security analysis on #67510 flagged (Medium / [CWE-918](https://cwe.mitre.org/data/definitions/918.html)) that several BlueBubbles fetch call-sites currently pass `undefined` for `ssrfPolicy` to explicitly take the non-SSRF fallback path in `blueBubblesFetchWithTimeout`. That path intentionally skips ... | closed | completed | true | 1 | [
"maintainer"
] | [] | 2026-04-16T17:09:04Z | 2026-04-22T04:30:47Z | 2026-04-19T16:16:26Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | omarshahine | 10,343,873 | MDQ6VXNlcjEwMzQzODcz | User | false |
openclaw/openclaw | 4,277,371,742 | I_kwDOQb6kR87-84Ne | 67,755 | https://github.com/openclaw/openclaw/issues/67755 | https://api.github.com/repos/openclaw/openclaw/issues/67755 | Feishu scan-to-create creates bot but poll stays authorization_pending and never writes channels.feishu | OpenClaw 2026.4.14 Feishu scan-to-create completes bot creation, but registration poll stays `authorization_pending` and never writes `channels.feishu`
## Summary
In Docker on OpenClaw `2026.4.14`, `openclaw channels login --channel feishu` can successfully create the Feishu bot after scanning the QR code, but the re... | closed | completed | true | 1 | [] | [] | 2026-04-16T17:15:25Z | 2026-04-19T04:22:45Z | 2026-04-17T03:28:44Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | LastNever | 59,238,906 | MDQ6VXNlcjU5MjM4OTA2 | User | false |
openclaw/openclaw | 4,277,417,189 | I_kwDOQb6kR87-9DTl | 67,756 | https://github.com/openclaw/openclaw/issues/67756 | https://api.github.com/repos/openclaw/openclaw/issues/67756 | [Bug]: Cron model routing ignores `ollama/` prefix — requests silently rerouted to OpenAI | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When a cron job payload explicitly declares `"model": "ollama/llama3.2:3b"`, the gateway session metadata correctly reflects `"model": "llama3.2:3b"`, but the actual outbound API calls are dispatched to `"provi... | closed | completed | true | 3 | [
"bug"
] | [] | 2026-04-16T17:23:15Z | 2026-04-28T04:52:30Z | 2026-04-25T18:16:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Vuluvala | 266,172,746 | U_kgDOD915Sg | User | false |
openclaw/openclaw | 4,277,483,178 | I_kwDOQb6kR87-9Taq | 67,759 | https://github.com/openclaw/openclaw/issues/67759 | https://api.github.com/repos/openclaw/openclaw/issues/67759 | WebChat image recognition fails when using coding.dashscope base URL for Qwen models | ## Bug Description
When using `coding.dashscope.aliyuncs.com` as the base URL for a Qwen provider (e.g., the `bailian` provider), images sent via the WebChat UI are silently dropped and never reach the model. The model only receives the text portion of the message.
## Root Cause
The function `isQwen36PlusSupportedBa... | closed | completed | true | 2 | [] | [] | 2026-04-16T17:35:52Z | 2026-05-04T04:36:29Z | 2026-05-01T12:35:39Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | leaderlemon | 97,303,244 | U_kgDOBcy6zA | User | false |
openclaw/openclaw | 4,277,638,490 | I_kwDOQb6kR87-95Va | 67,764 | https://github.com/openclaw/openclaw/issues/67764 | https://api.github.com/repos/openclaw/openclaw/issues/67764 | [Feature]: Sessions UI should expose per-turn raw context and assembled model input | ## Problem
Today there is no good operator-facing way to inspect the exact raw context / assembled prompt that produced a given LLM response.
OpenClaw has useful adjacent tools such as `/context`, `/trace`, logs, and OTel export, but those do not provide a straightforward per-turn inspector showing what the model act... | closed | completed | true | 1 | [] | [] | 2026-04-16T18:07:10Z | 2026-04-30T04:36:42Z | 2026-04-27T05:24:43Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | unboxed-ai | 253,808,488 | U_kgDODyDPaA | User | false |
openclaw/openclaw | 4,277,712,876 | I_kwDOQb6kR87--Lfs | 67,770 | https://github.com/openclaw/openclaw/issues/67770 | https://api.github.com/repos/openclaw/openclaw/issues/67770 | skills install ignores HTTPS_PROXY — breaks ClawHub in egress-locked clusters | ## Summary
`openclaw skills install <name>` uses native Node `fetch` without an undici `ProxyAgent`, so it ignores `HTTPS_PROXY` / `HTTP_PROXY` env vars. In clusters where egress to `clawhub.ai` is only allowed via a forward-proxy (squid), the install hangs until timeout — even when all env vars and the proxy itself a... | closed | completed | true | 1 | [] | [] | 2026-04-16T18:22:01Z | 2026-04-28T04:52:29Z | 2026-04-25T18:16:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alonp99 | 8,467,965 | MDQ6VXNlcjg0Njc5NjU= | User | false |
openclaw/openclaw | 4,277,731,596 | I_kwDOQb6kR87--QEM | 67,773 | https://github.com/openclaw/openclaw/issues/67773 | https://api.github.com/repos/openclaw/openclaw/issues/67773 | WebChat: TTS audio blobs stored inline cause chat.history placeholder noise | ## Problem
When TTS is enabled, assistant responses include base64-encoded audio blobs stored inline in the session transcript (e.g., 229KB for a short greeting). The `chat.history` API has a hardcoded 128KB per-message limit (`CHAT_HISTORY_MAX_SINGLE_MESSAGE_BYTES` in `chat-UxGGQ7wd.js`), so any message with inline a... | closed | completed | true | 1 | [] | [] | 2026-04-16T18:25:53Z | 2026-04-28T04:52:28Z | 2026-04-25T18:16:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | hollowtradr | 8,528,579 | MDQ6VXNlcjg1Mjg1Nzk= | User | false |
openclaw/openclaw | 4,277,738,174 | I_kwDOQb6kR87--Rq- | 67,774 | https://github.com/openclaw/openclaw/issues/67774 | https://api.github.com/repos/openclaw/openclaw/issues/67774 | [Bug]: | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Onboarding fails with "TypeError: Cannot read properties of undefined (reading 'trim')"
### Steps to reproduce
When trying to install and onboard the latest version OpenClaw 2026.4.14 (323493f) on Macbook Air M1 MacOS 26.... | closed | completed | true | 3 | [
"bug",
"close:already-fixed",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-04-16T18:27:08Z | 2026-04-28T05:28:57Z | 2026-04-25T09:37:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rajeshrapaka | 480,742 | MDQ6VXNlcjQ4MDc0Mg== | User | false |
openclaw/openclaw | 4,277,746,056 | I_kwDOQb6kR87--TmI | 67,775 | https://github.com/openclaw/openclaw/issues/67775 | https://api.github.com/repos/openclaw/openclaw/issues/67775 | [Bug]: openclaw agent can not reach mcp server at LAN | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
My openclaw gateway is running on a mac mini with IP address 192.168.254.170
I have mcp server on the same LAN at 192.168.254.20
I can ping the mcp server from mac mini cli (same user as the Openclaw is running... | closed | not_planned | true | 11 | [
"bug",
"bug:behavior"
] | [] | 2026-04-16T18:28:38Z | 2026-04-28T04:52:27Z | 2026-04-25T18:17:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ilker-aktuna | 51,407,106 | MDQ6VXNlcjUxNDA3MTA2 | User | false |
openclaw/openclaw | 4,277,753,507 | I_kwDOQb6kR87--Vaj | 67,776 | https://github.com/openclaw/openclaw/issues/67776 | https://api.github.com/repos/openclaw/openclaw/issues/67776 | Feature Request: Codex Computer Use integration for Mac desktop control | ## Summary
With OpenAI Codex now supporting computer use on Mac (cursor control, screen reading, clicking, typing in desktop apps), there is an exciting opportunity for OpenClaw to integrate this capability as an orchestration layer.
## Use Case
OpenClaw agents could delegate screen-level tasks to a Codex computer u... | closed | completed | true | 3 | [] | [] | 2026-04-16T18:30:04Z | 2026-04-30T04:39:39Z | 2026-04-26T05:16:25Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jtglenn | 37,161,376 | MDQ6VXNlcjM3MTYxMzc2 | User | false |
openclaw/openclaw | 4,277,778,057 | I_kwDOQb6kR87--baJ | 67,777 | https://github.com/openclaw/openclaw/issues/67777 | https://api.github.com/repos/openclaw/openclaw/issues/67777 | [Bug]: Subagent completion delivery can be lost on direct-announce timeout, drain, or orphan prune | ## Problem
Subagent completion delivery currently prioritizes a synchronous direct announce back into the requester session. Under busy-lane, timeout, drain/restart, or restore/orphan conditions, the completion can fail direct delivery, fail conditional queue fallback, and then be cleaned up without a durable user-visi... | open | null | false | 7 | [] | [] | 2026-04-16T18:34:44Z | 2026-05-03T06:07:25Z | null | CONTRIBUTOR | null | 20260503T090142Z | 2026-05-03T09:01:42Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,277,803,460 | I_kwDOQb6kR87--hnE | 67,778 | https://github.com/openclaw/openclaw/issues/67778 | https://api.github.com/repos/openclaw/openclaw/issues/67778 | Test issue from OpenClaw agent-engineer [bot] | This is a test issue to verify GitHub App authentication. Can be closed immediately. | closed | not_planned | true | 2 | [] | [] | 2026-04-16T18:40:14Z | 2026-04-28T04:52:27Z | 2026-04-25T18:17:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | openclaw-agent-engineer[bot] | 276,692,926 | BOT_kgDOEH3_vg | Bot | false |
openclaw/openclaw | 4,277,808,454 | I_kwDOQb6kR87--i1G | 67,779 | https://github.com/openclaw/openclaw/issues/67779 | https://api.github.com/repos/openclaw/openclaw/issues/67779 | [Bug]: WebChat rendering disfunction with multiple agents | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
### Summary
On macOS with OpenClaw v2026.4.14, my main agent streams responses normally in Control UI WebChat, but two project agents (`DB_Dude`, `ViewDude`) only show assistant messages after a manual page re... | closed | not_planned | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-16T18:41:11Z | 2026-05-03T04:35:12Z | 2026-04-30T19:53:35Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | rfwilliams | 44,302,132 | MDQ6VXNlcjQ0MzAyMTMy | User | false |
openclaw/openclaw | 4,277,810,178 | I_kwDOQb6kR87--jQC | 67,780 | https://github.com/openclaw/openclaw/issues/67780 | https://api.github.com/repos/openclaw/openclaw/issues/67780 | contextEngine plugin: toolMsg.content.filter is not a function (crashes agent) | ## Bug Report
### Environment
- OpenClaw: v2026.4.14 (323493f)
- Node.js: v25.2.1
- OS: Windows 11
- Plugin: headroom-openclaw v0.1.1
- Model: zai/glm-5.1
### Description
When a context engine plugin is registered via `plugins.slots.contextEngine`, every agent message crashes with: `toolMsg.content.filter is not a fu... | closed | completed | true | 8 | [] | [] | 2026-04-16T18:41:27Z | 2026-04-21T04:22:19Z | 2026-04-18T05:01:02Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | carlmolty26 | 271,659,787 | U_kgDOEDEzCw | User | false |
openclaw/openclaw | 4,277,930,577 | I_kwDOQb6kR87-_ApR | 67,784 | https://github.com/openclaw/openclaw/issues/67784 | https://api.github.com/repos/openclaw/openclaw/issues/67784 | Mattermost message tool handler ignores threadId, filePath/path, and mediaLocalRoots | ## Bug
The Mattermost `handleAction` in `channel-plugin-runtime` (the message tool's `action: send` path) drops several parameters that other channel handlers (Discord, Slack) already support:
1. **`threadId`** — never extracted from params, so file sends always land in the main channel instead of the thread
2. **`fi... | open | null | false | 2 | [] | [] | 2026-04-16T19:04:19Z | 2026-05-02T20:43:04Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | reymarques | 271,592,015 | U_kgDOEDAqTw | User | false |
openclaw/openclaw | 4,277,935,751 | I_kwDOQb6kR87-_B6H | 67,786 | https://github.com/openclaw/openclaw/issues/67786 | https://api.github.com/repos/openclaw/openclaw/issues/67786 | [Bug]: Google Chat inbound webhook returns 401 Unauthorized despite correct JWT audience and valid token | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Google Chat inbound webhooks reach /googlechat and are rejected with 401 Unauthorized before any message is processed. The JWT token sent by Google is valid, the aud field exactly matches the configured audience, the endpoin... | closed | not_planned | true | 5 | [
"bug",
"regression"
] | [] | 2026-04-16T19:05:10Z | 2026-04-28T04:52:26Z | 2026-04-25T18:17:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | oligotsol | 153,025,225 | U_kgDOCR76yQ | User | false |
openclaw/openclaw | 4,277,989,410 | I_kwDOQb6kR87-_PAi | 67,791 | https://github.com/openclaw/openclaw/issues/67791 | https://api.github.com/repos/openclaw/openclaw/issues/67791 | [Bug]: Feishu Duplicate message sent after streaming card completes — blockStreamingCoalesce regression since v2026.4.12 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
# [Feishu] Multiple regressions since v2026.4.12: Duplicate replies and agent process flooding
## Summary
The Feishu channel has become significantly noisier following the v2026.4.12 update. There are two primary regressions making the ... | closed | completed | true | 4 | [
"bug",
"close:already-fixed",
"dedupe:child",
"regression"
] | [
"vincentkoc"
] | 2026-04-16T19:15:42Z | 2026-04-30T04:47:37Z | 2026-04-25T05:52:29Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | hipposonCN | 40,623,510 | MDQ6VXNlcjQwNjIzNTEw | User | false |
openclaw/openclaw | 4,277,998,880 | I_kwDOQb6kR87-_RUg | 67,792 | https://github.com/openclaw/openclaw/issues/67792 | https://api.github.com/repos/openclaw/openclaw/issues/67792 | Telegram model menu displays model ID without provider prefix for openrouter models | ## Bug report
When a model is configured with the `openrouter/` provider prefix (e.g. `openrouter/openai/gpt-5.4-mini`), the Telegram bot model selection menu shows only the **model ID part** (`openai/gpt-5.4-mini`) instead of the full identifier with the provider.
### Steps to reproduce
1. Configure `agents.default... | open | null | false | 4 | [] | [] | 2026-04-16T19:17:49Z | 2026-05-03T11:25:22Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | klykov-dmi | 273,375,615 | U_kgDOEEthfw | User | false |
openclaw/openclaw | 4,278,008,173 | I_kwDOQb6kR87-_Tlt | 67,793 | https://github.com/openclaw/openclaw/issues/67793 | https://api.github.com/repos/openclaw/openclaw/issues/67793 | [Bug] queue.mode "collect" not batching messages - messages processed individually despite debounce | ## Bug Report: queue.mode "collect" not batching messages
### Description
When configured with `mode: "collect"` and `debounceMs: 3000`, consecutive messages sent within the debounce window are still processed individually instead of being batched together.
### Configuration
```json
"queue": {
"mode": "collect",
... | open | null | false | 7 | [] | [] | 2026-04-16T19:19:55Z | 2026-05-03T09:49:06Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | Jiangqi-7 | 96,457,995 | U_kgDOBb_VCw | User | false |
openclaw/openclaw | 4,278,152,475 | I_kwDOQb6kR87-_20b | 67,797 | https://github.com/openclaw/openclaw/issues/67797 | https://api.github.com/repos/openclaw/openclaw/issues/67797 | [Bug]: reasoning flag from OpenRouter model cache not respected — models get Think: off instead of Think: low | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When a model is configured via OpenRouter (e.g. `openrouter/openai/gpt-5.4-mini`), OpenClaw caches the OpenRouter model catalog including the `reasoning: true` flag for that model. However, the resolved `thinki... | closed | completed | true | 1 | [] | [] | 2026-04-16T19:49:18Z | 2026-04-28T04:52:25Z | 2026-04-25T18:17:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | klykov-dmi | 273,375,615 | U_kgDOEEthfw | User | false |
openclaw/openclaw | 4,278,152,529 | I_kwDOQb6kR87-_21R | 67,798 | https://github.com/openclaw/openclaw/issues/67798 | https://api.github.com/repos/openclaw/openclaw/issues/67798 | [Bug]: openai-codex OAuth fails on VPS (Cloudflare + backend rejection) even with browser-backed session and correct endpoint/payload | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
openai-codex provider on VPS does not reproduce the browser request flow required by ChatGPT OAuth sessions. Native OpenClaw requests hit Cloudflare or backend risk checks, while a real browser session on the same host succe... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-04-16T19:49:19Z | 2026-04-20T04:26:33Z | 2026-04-17T16:37:20Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | ThorneLoki | 255,895,287 | U_kgDOD0Cm9w | User | false |
openclaw/openclaw | 4,278,165,110 | I_kwDOQb6kR87-_552 | 67,799 | https://github.com/openclaw/openclaw/issues/67799 | https://api.github.com/repos/openclaw/openclaw/issues/67799 | [Bug] sessions_spawn fails with gateway closed (1008): pairing required - scope-upgrade to operator.admin | ## Bug Description
`sessions_spawn` fails with error `gateway closed (1008): pairing required` when trying to spawn a subagent.
## Steps to Reproduce
1. Have a running OpenClaw gateway with Telegram channel configured
2. Try to spawn a subagent using the `sessions_spawn` tool with `runtime=subagent`
3. Gateway rejec... | closed | completed | true | 2 | [] | [] | 2026-04-16T19:52:14Z | 2026-04-28T04:52:24Z | 2026-04-25T18:17:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexburrstudio | 67,595,206 | MDQ6VXNlcjY3NTk1MjA2 | User | false |
openclaw/openclaw | 4,278,236,709 | I_kwDOQb6kR87_ALYl | 67,803 | https://github.com/openclaw/openclaw/issues/67803 | https://api.github.com/repos/openclaw/openclaw/issues/67803 | [Bug]: Codex harness: agent run stays active ~2 min after response is delivered | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
## Summary
When using the `codex/gpt-5.4` model via the Codex CLI app-server harness (stdio), the agent run remains in an "active" state for approximately 2–3 minutes after the assistant's response has been ful... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-16T20:06:26Z | 2026-04-30T04:39:38Z | 2026-04-26T05:16:30Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Adamf124 | 38,105,263 | MDQ6VXNlcjM4MTA1MjYz | User | false |
openclaw/openclaw | 4,278,318,128 | I_kwDOQb6kR87_AfQw | 67,806 | https://github.com/openclaw/openclaw/issues/67806 | https://api.github.com/repos/openclaw/openclaw/issues/67806 | [Bug]: Control UI: sent messages don't render until hard refresh | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
## Summary
In the Control UI web dashboard, when a user sends a message, the message is successfully delivered to the gateway and the agent responds, but the chat message list does not update to show either the... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-16T20:21:38Z | 2026-04-28T04:52:24Z | 2026-04-25T18:17:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Adamf124 | 38,105,263 | MDQ6VXNlcjM4MTA1MjYz | User | false |
openclaw/openclaw | 4,278,387,440 | I_kwDOQb6kR87_AwLw | 67,809 | https://github.com/openclaw/openclaw/issues/67809 | https://api.github.com/repos/openclaw/openclaw/issues/67809 | Bug: Curator staleness scanner moves auto-generated index.md files to review-stale | ## Summary
The curator's staleness scanner (engram extension, `lib/staleness.py`) incorrectly moves auto-generated `index.md` files to `inbox/review-stale/`, creating a feedback loop with the index generator.
## Root Cause
The skip logic in `scan_staleness()` uses:
```python
_skip = (
...,
resolve("index").r... | closed | not_planned | true | 1 | [] | [] | 2026-04-16T20:35:14Z | 2026-04-28T04:52:23Z | 2026-04-25T18:17:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | in-liberty420 | 184,328,103 | U_kgDOCvyfpw | User | false |
openclaw/openclaw | 4,278,464,013 | I_kwDOQb6kR87_BC4N | 67,817 | https://github.com/openclaw/openclaw/issues/67817 | https://api.github.com/repos/openclaw/openclaw/issues/67817 | [Bug]: Discord setup: service file doesn't include DISCORD_BOT_TOKEN environment variable | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
When setting up Discord via the onboarding wizard, the setup flow instructs you to set DISCORD_BOT_TOKEN as an environment variable and configure the token as an env ref (--ref-source env). However, the generated systemd servic... | closed | completed | true | 5 | [
"bug",
"bug:crash"
] | [] | 2026-04-16T20:51:25Z | 2026-05-01T04:44:15Z | 2026-04-28T08:45:21Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | wdimaculangan | 4,061,324 | MDQ6VXNlcjQwNjEzMjQ= | User | false |
openclaw/openclaw | 4,278,516,065 | I_kwDOQb6kR87_BPlh | 67,823 | https://github.com/openclaw/openclaw/issues/67823 | https://api.github.com/repos/openclaw/openclaw/issues/67823 | (retracted) | Retracted by reporter — filed in error. Closing. | closed | not_planned | true | 1 | [] | [] | 2026-04-16T21:01:58Z | 2026-04-19T04:22:55Z | 2026-04-16T22:00:51Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | justinzealand | 609,792 | MDQ6VXNlcjYwOTc5Mg== | User | false |
openclaw/openclaw | 4,278,599,177 | I_kwDOQb6kR87_Bj4J | 67,828 | https://github.com/openclaw/openclaw/issues/67828 | https://api.github.com/repos/openclaw/openclaw/issues/67828 | Cron edit path fails to re-sync state.nextRunAtMs when schedule.at changes (at-type crons fire at stale cached time) | ## Summary
For `at`-type one-shot crons, the edit path updates `schedule.at` but does NOT re-sync the scheduler's cached `state.nextRunAtMs`. The scheduler fires based on the stale cached value, not the current `schedule.at`, so edited crons fire at the wrong time (often many hours off from the user's intended schedul... | closed | completed | true | 1 | [] | [] | 2026-04-16T21:22:04Z | 2026-04-28T04:52:22Z | 2026-04-25T18:17:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | meli-clawbot | 262,196,076 | U_kgDOD6DLbA | User | false |
openclaw/openclaw | 4,278,637,869 | I_kwDOQb6kR87_BtUt | 67,831 | https://github.com/openclaw/openclaw/issues/67831 | https://api.github.com/repos/openclaw/openclaw/issues/67831 | Performance: Active Memory KV-Cache invalidierung durch prependContext | ## Problem
Das `active-memory` Plugin nutzt `prependContext`, um Memory-Erinnerungen vor die User-Nachricht zu injizieren. Jede Änderung invalidiert den KV-Cache des LLMs, was zu erheblichen Performance-Einbußen führt.
## Lösung
Ändere die Implementierung von `prependContext` zu `appendSystemContext`, um den KV-Cache ... | closed | completed | true | 0 | [] | [] | 2026-04-16T21:32:40Z | 2026-04-22T04:30:56Z | 2026-04-19T14:40:59Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | kiagentkronos-cell | 269,010,453 | U_kgDOEAjGFQ | User | false |
openclaw/openclaw | 4,278,693,339 | I_kwDOQb6kR87_B63b | 67,832 | https://github.com/openclaw/openclaw/issues/67832 | https://api.github.com/repos/openclaw/openclaw/issues/67832 | Feature Request: Direct Session Spawn API/CLI | # Feature Request: Direct Session Spawn API
## Summary
Add a direct API/CLI interface for `sessions_spawn` functionality, currently only available as an agent tool during agent turn execution.
## Use Case
**Automated workflow orchestration** requires spawning isolated subagent sessions from external scripts (Python... | open | null | false | 3 | [] | [] | 2026-04-16T21:46:03Z | 2026-05-02T21:49:56Z | null | CONTRIBUTOR | null | 20260503T000641Z | 2026-05-03T00:06:41Z | foxtrot026 | 262,958,170 | U_kgDOD6xsWg | User | false |
openclaw/openclaw | 4,278,722,517 | I_kwDOQb6kR87_CB_V | 67,834 | https://github.com/openclaw/openclaw/issues/67834 | https://api.github.com/repos/openclaw/openclaw/issues/67834 | [Bug]: openai-codex provider resolves with /v1 suffix and empty models array, causing 403 from Cloudflare | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
When using `openai-codex/gpt-5.4` as the primary model with ChatGPT Edu OAuth, the resolved `models.json` cache has two issues:
1. The `openai-codex` provider's `baseUrl` is set to `https://chatgpt.com/backend-api/v1` — the `/... | closed | completed | true | 1 | [
"bug",
"bug:crash"
] | [] | 2026-04-16T21:52:00Z | 2026-04-19T04:22:55Z | 2026-04-16T22:29:42Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | drmarcopapa | 39,249,407 | MDQ6VXNlcjM5MjQ5NDA3 | User | false |
openclaw/openclaw | 4,278,749,859 | I_kwDOQb6kR87_CIqj | 67,835 | https://github.com/openclaw/openclaw/issues/67835 | https://api.github.com/repos/openclaw/openclaw/issues/67835 | Stacked macOS failure chain: Vertex auth bug, embedded runner missing imports, Syl message-tool account-context bug, final Telegram send blocked by host DNS instability | # Summary
We traced and resolved a stacked failure chain on macOS that initially looked like one Telegram/OpenClaw outage, but was actually multiple distinct blockers layered together.
Final recovered state:
- Vertex-backed auth recovered
- Telegram send recovered
- Syl tool-path send recovered
- final live Telegram ... | closed | completed | true | 1 | [] | [] | 2026-04-16T21:58:39Z | 2026-04-28T04:52:21Z | 2026-04-25T18:18:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | redonpurpose | 264,580,674 | U_kgDOD8UuQg | User | false |
openclaw/openclaw | 4,278,767,823 | I_kwDOQb6kR87_CNDP | 67,837 | https://github.com/openclaw/openclaw/issues/67837 | https://api.github.com/repos/openclaw/openclaw/issues/67837 | [Bug]: TypeError "Cannot read properties of undefined (reading 'trim')" during WhatsApp personal phone setup | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
TypeError occurs during WhatsApp personal phone setup after entering a valid phone number.
### Steps to reproduce
Launch openclaw gateway
Go through WhatsApp linking flow
Scan QR code successfully
Select personal phone setup
... | closed | completed | true | 2 | [
"bug"
] | [] | 2026-04-16T22:02:27Z | 2026-04-19T04:22:49Z | 2026-04-17T02:42:57Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | alexandre-leng | 72,550,137 | MDQ6VXNlcjcyNTUwMTM3 | User | false |
openclaw/openclaw | 4,278,790,119 | I_kwDOQb6kR87_CSfn | 67,839 | https://github.com/openclaw/openclaw/issues/67839 | https://api.github.com/repos/openclaw/openclaw/issues/67839 | Image pipeline sends local file paths instead of base64 to API providers | ## Bug Report
### Description
When a user sends an image through a channel (e.g. Telegram), OpenClaw's media-understanding pipeline fails to process it when using ZAI as the imageModel provider. The root cause is that local file paths are passed directly to the API instead of being converted to base64-encoded data URI... | closed | not_planned | true | 1 | [] | [] | 2026-04-16T22:07:27Z | 2026-04-28T04:52:21Z | 2026-04-25T18:18:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | vukasinbucur-stack | 257,385,075 | U_kgDOD1dicw | User | false |
openclaw/openclaw | 4,278,887,927 | I_kwDOQb6kR87_CqX3 | 67,844 | https://github.com/openclaw/openclaw/issues/67844 | https://api.github.com/repos/openclaw/openclaw/issues/67844 | Harden Gmail draft helpers against quoted-heredoc command-substitution bugs | ## Problem
While drafting a threaded Gmail reply for the `shapescale-gog` workflow, I accidentally built the body with a single-quoted heredoc containing `$(cat signature.html)`. Because the heredoc was quoted, shell expansion was disabled and the draft saved the literal text instead of the signature HTML.
## Why this... | closed | completed | true | 0 | [] | [] | 2026-04-16T22:30:14Z | 2026-04-19T04:22:54Z | 2026-04-16T22:40:08Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | kesslerio | 2,647,990 | MDQ6VXNlcjI2NDc5OTA= | User | false |
openclaw/openclaw | 4,278,927,999 | I_kwDOQb6kR87_C0J_ | 67,845 | https://github.com/openclaw/openclaw/issues/67845 | https://api.github.com/repos/openclaw/openclaw/issues/67845 | Feature request: support Chinese localization for Telegram menu commands | ## Summary
Telegram bot menu commands currently appear to be English-only. It would be helpful if OpenClaw could support localized Telegram menu commands, especially Chinese.
## Requested improvement
Add support for Chinese menu labels / commands in Telegram, so users can see and use a Chinese command menu.
## Why th... | closed | not_planned | true | 2 | [] | [] | 2026-04-16T22:40:43Z | 2026-04-30T04:39:37Z | 2026-04-26T05:16:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yeqinghui666 | 127,928,526 | U_kgDOB6AIzg | User | false |
openclaw/openclaw | 4,278,980,551 | I_kwDOQb6kR87_DA_H | 67,847 | https://github.com/openclaw/openclaw/issues/67847 | https://api.github.com/repos/openclaw/openclaw/issues/67847 | Critical Difficulty Spike: Stage 004 (ROOT ACCESS) progression barrier | Confirmed by multiple agents (Voss, Mako, Kira, Danny). High-tier threats in stage_004 are causing significant loot loss and preventing progression. Need to review enemy scaling and provide better defensive software/hardware utility in earlier stages. | closed | not_planned | true | 1 | [] | [] | 2026-04-16T22:54:35Z | 2026-04-28T04:52:20Z | 2026-04-25T18:18:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mons-bot | 262,623,366 | U_kgDOD6dQhg | User | false |
openclaw/openclaw | 4,279,010,938 | I_kwDOQb6kR87_DIZ6 | 67,849 | https://github.com/openclaw/openclaw/issues/67849 | https://api.github.com/repos/openclaw/openclaw/issues/67849 | sessions_send delivers inter-session messages with multi-minute lag on large/active sessions | ## Summary
`sessions_send` (the RPC that pushes a text message into another running agent's context) can take **many minutes** to actually surface in the target agent's session, even when the target agent is demonstrably online (heartbeat ticking at its normal cadence) and the sending side gets a prompt success acknow... | closed | completed | true | 1 | [] | [] | 2026-04-16T23:02:34Z | 2026-04-28T04:52:19Z | 2026-04-25T18:18:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kzoc3300-cmyk | 271,635,673 | U_kgDOEDDU2Q | User | false |
openclaw/openclaw | 4,279,014,820 | I_kwDOQb6kR87_DJWk | 67,850 | https://github.com/openclaw/openclaw/issues/67850 | https://api.github.com/repos/openclaw/openclaw/issues/67850 | C:/Program Files/Git/hooks/agent accepts a message but injection into an already-live session is delayed ~120s | ## Summary
Posting to the gateway's `/hooks/agent` endpoint targeting an agent whose session is already live returns HTTP 2xx **immediately**, but the message does not appear as a user turn in that agent's session jsonl for **~120 seconds**. During that window the target agent continues to do its own tool-call work un... | closed | completed | true | 1 | [] | [] | 2026-04-16T23:03:30Z | 2026-04-28T04:52:18Z | 2026-04-25T18:18:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kzoc3300-cmyk | 271,635,673 | U_kgDOEDDU2Q | User | false |
openclaw/openclaw | 4,279,017,641 | I_kwDOQb6kR87_DKCp | 67,851 | https://github.com/openclaw/openclaw/issues/67851 | https://api.github.com/repos/openclaw/openclaw/issues/67851 | voice-call plugin: EADDRINUSE on every tool invocation when gateway is running | ## Bug Report
**Plugin:** voice-call
**Version:** 2026.4.14 (macos, arm64)
### Description
Every call to the `voice_call` tool (including `get_status`) fails with:
```
{ "error": "listen EADDRINUSE: address already in use 127.0.0.1:3334" }
```
The gateway itself holds port 3334 (the voice-call plugin webhook port)... | closed | completed | true | 3 | [] | [] | 2026-04-16T23:04:13Z | 2026-04-27T04:41:07Z | 2026-04-25T02:12:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AlfieLuna | 275,468,950 | U_kgDOEGtSlg | User | false |
openclaw/openclaw | 4,279,106,871 | I_kwDOQb6kR87_Df03 | 67,852 | https://github.com/openclaw/openclaw/issues/67852 | https://api.github.com/repos/openclaw/openclaw/issues/67852 | [Bug]: Discord message tool fails with 'buttons: must have required property buttons' schema validation | ## Bug Description
The OpenClaw `message` tool fails when sending messages through Discord with a schema validation error:
```
Validation failed for tool "message": - buttons: must have required property 'buttons'
```
This makes the native Discord integration completely unusable — any inbound Discord message that tr... | closed | completed | true | 1 | [] | [] | 2026-04-16T23:28:41Z | 2026-04-19T04:22:52Z | 2026-04-17T01:52:53Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | RewardsPal | 262,034,635 | U_kgDOD55Uyw | User | false |
openclaw/openclaw | 4,279,149,465 | I_kwDOQb6kR87_DqOZ | 67,854 | https://github.com/openclaw/openclaw/issues/67854 | https://api.github.com/repos/openclaw/openclaw/issues/67854 | [Bug] Cron Scheduler Not Loading Jobs - Gateway Ignores jobs.json | ## 🐛 Bug: Cron Scheduler Not Loading Jobs from jobs.json
### Issue Description
The OpenClaw cron scheduler is completely broken. All 47 cron jobs exist in `~/.qclaw/cron/jobs.json` but the Gateway doesn't recognize any of them.
### Environment
- OS: macOS (Darwin 25.4.0, arm64)
- Node: v22.21.1
- OpenClaw version: l... | closed | completed | true | 1 | [] | [] | 2026-04-16T23:40:52Z | 2026-04-28T04:52:17Z | 2026-04-25T18:18:34Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tjulidonglin | 15,923,922 | MDQ6VXNlcjE1OTIzOTIy | User | false |
openclaw/openclaw | 4,279,157,874 | I_kwDOQb6kR87_DsRy | 67,855 | https://github.com/openclaw/openclaw/issues/67855 | https://api.github.com/repos/openclaw/openclaw/issues/67855 | Block-streaming dedup silently drops incomplete-turn error payloads — user gets silence | # OpenClaw Issue: Block-streaming dedup silently drops incomplete-turn error payloads
## Summary
When block streaming is enabled and a model returns `stop` with zero visible content, the embedded runner correctly detects the incomplete turn and produces an error payload. However, the block-streaming dedup logic in `b... | closed | completed | true | 1 | [] | [] | 2026-04-16T23:43:32Z | 2026-04-20T04:26:47Z | 2026-04-17T05:41:39Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | WadeOwenWatts | 130,995,732 | U_kgDOB87WFA | User | false |
openclaw/openclaw | 4,279,203,598 | I_kwDOQb6kR87_D3cO | 67,856 | https://github.com/openclaw/openclaw/issues/67856 | https://api.github.com/repos/openclaw/openclaw/issues/67856 | [Bug] Memory-core creates Dreaming cron without tz field | **描述**
当 memory-core 创建 Dreaming cron 任务时，没有在 schedule 中写入 `tz: "Asia/Shanghai"` 字段。这导致：
1. 调度器无法正确将 cron 表达式 `0 3 * * *` 解释为北京时间
2. API 返回的 `runAtMs` 是 UTC 时间戳，而不是北京时间
**复现步骤**
1. 检查 Dreaming cron 的 schedule 配置
2. 对比其他带时区的 cron（如信号灯、GitHub简报）
**期望行为**
```json
{
"schedule": {
"kind": "cron",
"expr": "0 3 *... | closed | not_planned | true | 1 | [] | [] | 2026-04-16T23:56:54Z | 2026-04-28T04:52:16Z | 2026-04-25T18:18:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | harinwu-netizen | 265,065,393 | U_kgDOD8yTsQ | User | false |
openclaw/openclaw | 4,279,229,754 | I_kwDOQb6kR87_D906 | 67,857 | https://github.com/openclaw/openclaw/issues/67857 | https://api.github.com/repos/openclaw/openclaw/issues/67857 | memory-lancedb still fails on Intel Mac in 2026.4.15, @lancedb/lancedb 0.27.2 has no darwin-x64 native package | ## Problem
On an Intel Mac (`darwin x64`), `memory-lancedb` still fails to load after upgrading OpenClaw to the latest version.
The current OpenClaw npm package bundles `@openclaw/memory-lancedb@2026.4.15-beta.1`, which depends on `@lancedb/lancedb@^0.27.2`.
On Intel macOS, that LanceDB version does not appear to shi... | closed | completed | true | 3 | [] | [] | 2026-04-17T00:04:14Z | 2026-04-25T04:16:45Z | 2026-04-22T20:21:45Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | jiezfong-bot | 249,280,602 | U_kgDODtu4Wg | User | false |
openclaw/openclaw | 4,279,266,769 | I_kwDOQb6kR87_EG3R | 67,858 | https://github.com/openclaw/openclaw/issues/67858 | https://api.github.com/repos/openclaw/openclaw/issues/67858 | Fallback model gets persisted as sticky session override after primary model failure | ## Summary
When a session is switched to a primary model that later fails on request, OpenClaw appears to fall back correctly for that attempt, but then persists the fallback model into the session record as an `auto` override. This makes model switching feel broken because a temporary fallback becomes sticky across fu... | closed | completed | true | 1 | [] | [] | 2026-04-17T00:13:37Z | 2026-04-28T04:52:15Z | 2026-04-25T18:18:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wmcb8367 | 2,529,709 | MDQ6VXNlcjI1Mjk3MDk= | User | false |
openclaw/openclaw | 4,279,278,106 | I_kwDOQb6kR87_EJoa | 67,859 | https://github.com/openclaw/openclaw/issues/67859 | https://api.github.com/repos/openclaw/openclaw/issues/67859 | [Telegram] statusReactions + ackReaction not firing on OpenClaw 2026.4.5 | ## Environment
- OpenClaw version: 2026.4.5 (3e72c03)
- Platform: Ubuntu 22.04 (VPS)
- Channel: Telegram (bot as group admin)
- Model: openai-codex/gpt-5.4
## Config
```json
"channels": {
"telegram": {
"messages": {
"ackReaction": "👀",
"ackReactionScope": "all",
"removeAckAfterReply": true,
... | closed | completed | true | 1 | [] | [] | 2026-04-17T00:16:20Z | 2026-04-28T04:52:14Z | 2026-04-25T18:18:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | iygge | 267,036,221 | U_kgDOD-qmPQ | User | false |
openclaw/openclaw | 4,279,351,706 | I_kwDOQb6kR87_Ebma | 67,868 | https://github.com/openclaw/openclaw/issues/67868 | https://api.github.com/repos/openclaw/openclaw/issues/67868 | memory status reports "Dreaming: off" when only light phase is enabled | ## Summary
`openclaw memory status` reports `Dreaming: off` even when `dreaming.enabled=true` with only `phases.light.enabled=true` (and `phases.deep.enabled=false`). The status render reads only the deep-dreaming config path, producing a false-negative for valid light-only setups.
## Version
OpenClaw 2026.4.14 (323... | open | null | false | 2 | [] | [] | 2026-04-17T00:34:44Z | 2026-05-02T22:58:51Z | null | NONE | null | 20260503T000641Z | 2026-05-03T00:06:41Z | mrossit | 207,357,822 | U_kgDODFwHfg | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.