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 | 4,339,926,707 | I_kwDOQb6kR88AAAABAq4Gsw | 73,141 | https://github.com/openclaw/openclaw/issues/73141 | https://api.github.com/repos/openclaw/openclaw/issues/73141 | Bug: Agent-initiated upgrade causes gateway restart cascade and session context loss | ## Description
When an agent attempts to run `openclaw upgrade` (which is excluded from `plugins.allow`), it triggers a config change that causes a gateway restart cascade. The repeated restarts eventually lead to session context loss — the main Telegram session context is not recovered after the final restart.
## Ti... | closed | completed | true | 1 | [] | [] | 2026-04-28T01:30:48Z | 2026-04-30T04:32:29Z | 2026-04-28T01:55:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | stayif | 7,867,389 | MDQ6VXNlcjc4NjczODk= | User | false |
openclaw/openclaw | 4,339,788,493 | I_kwDOQb6kR88AAAABAqvqzQ | 73,123 | https://github.com/openclaw/openclaw/issues/73123 | https://api.github.com/repos/openclaw/openclaw/issues/73123 | Bug: openclaw memory search times out when eager context-window warmup runs for the memory command | # Bug: `openclaw memory search` times out when eager context-window warmup runs for the `memory` command
## Summary
`openclaw memory search` can time out during Gemini query embedding generation even though:
- the Gemini API works directly;
- OpenClaw's Gemini embedding provider works when invoked directly;
- `Memor... | closed | completed | true | 0 | [] | [] | 2026-04-28T00:58:08Z | 2026-04-30T04:32:28Z | 2026-04-28T01:56:58Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | oalansilva | 126,212 | MDQ6VXNlcjEyNjIxMg== | User | false |
openclaw/openclaw | 3,915,522,728 | I_kwDOQb6kR87pYiKo | 12,539 | https://github.com/openclaw/openclaw/issues/12539 | https://api.github.com/repos/openclaw/openclaw/issues/12539 | [Security] Bundled Skills: Seven API Credentials Exposed in Plaintext | **Component**: Bundled Skills
**Category**: security
**Priority**: High
## Summary
Seven API credentials are exposed in plaintext through environment variables and configuration files. These include keys for Notion, OpenAI, Google, and other services. Credentials are accessible via `/proc/PID/environ` reading, coding... | closed | not_planned | true | 1 | [
"bug",
"security"
] | [] | 2026-02-09T10:01:35Z | 2026-04-30T04:32:27Z | 2026-04-28T01:57:50Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | fr33d3m0n | 127,989,854 | U_kgDOB6D4Xg | User | false |
openclaw/openclaw | 4,339,962,249 | I_kwDOQb6kR88AAAABAq6RiQ | 73,150 | https://github.com/openclaw/openclaw/issues/73150 | https://api.github.com/repos/openclaw/openclaw/issues/73150 | `memorySearch.provider` cannot target custom provider ids, limiting multi-Ollama / multi-GPU setups | ## Summary
`agents.defaults.memorySearch.provider` accepted generic provider kinds like `ollama` but rejected custom configured provider ids such as `ollama-5080`. This prevented routing memory embeddings/search to a specific local Ollama endpoint in a multi-GPU setup.
## Environment
- Historical OpenClaw setup, March... | closed | completed | true | 1 | [] | [] | 2026-04-28T01:40:33Z | 2026-04-30T04:32:26Z | 2026-04-28T02:11:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | oneandrewwang | 261,546,085 | U_kgDOD5bgZQ | User | false |
openclaw/openclaw | 4,339,955,805 | I_kwDOQb6kR88AAAABAq54XQ | 73,147 | https://github.com/openclaw/openclaw/issues/73147 | https://api.github.com/repos/openclaw/openclaw/issues/73147 | Custom local Ollama provider id fails auth resolution across agent lanes and falls back to paid model | ## Summary
A custom local Ollama provider (`ollama-gpu1`) was configured for a second GPU/Ollama endpoint. The provider was intended to route to a local model (`ollama-gpu1/qwen3:14b`) for heartbeat/light cron work. During gateway startup/diagnostics, OpenClaw treated the local provider as missing API auth and failed o... | closed | completed | true | 1 | [] | [] | 2026-04-28T01:38:44Z | 2026-04-30T04:32:26Z | 2026-04-28T02:11:46Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | oneandrewwang | 261,546,085 | U_kgDOD5bgZQ | User | false |
openclaw/openclaw | 4,340,091,873 | I_kwDOQb6kR88AAAABArCL4Q | 73,166 | https://github.com/openclaw/openclaw/issues/73166 | https://api.github.com/repos/openclaw/openclaw/issues/73166 | Cron scheduler silently stops firing after ~2.5 days of gateway uptime | # Cron scheduler silently stops firing after ~2.5 days of gateway uptime
## Summary
A cron job (schedule `0 9 * * *` with `tz: Asia/Shanghai`) silently failed to trigger after ~2.5 days of continuous gateway uptime. No error was logged; the job simply did not execute at its scheduled time. Manual `openclaw cron run <... | open | null | false | 1 | [] | [] | 2026-04-28T02:14:00Z | 2026-05-03T06:50:27Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | SkywingsWang | 87,686,296 | MDQ6VXNlcjg3Njg2Mjk2 | User | false |
openclaw/openclaw | 4,340,094,800 | I_kwDOQb6kR88AAAABArCXUA | 73,167 | https://github.com/openclaw/openclaw/issues/73167 | https://api.github.com/repos/openclaw/openclaw/issues/73167 | [Bug]: Microsoft Teams Asymmetrical communication | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Microsoft Teams inbound messages stopped reaching OpenClaw after upgrading from 2026.4.9 to 2026.4.25/2026.4.26.
Outbound Teams sending still works, and Discord still works both directions. Downgrading to 2026.4.9 restores ... | closed | completed | true | 5 | [
"bug",
"regression"
] | [
"BradGroux"
] | 2026-04-28T02:14:39Z | 2026-05-02T04:27:39Z | 2026-04-29T06:19:32Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | mikelavrik | 30,187,471 | MDQ6VXNlcjMwMTg3NDcx | User | false |
openclaw/openclaw | 4,068,646,797 | I_kwDOQb6kR87ygp-N | 44,625 | https://github.com/openclaw/openclaw/issues/44625 | https://api.github.com/repos/openclaw/openclaw/issues/44625 | [Bug]: Slack Bot scopes: auth.scopes: An API error occurred: unknown_method | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Version: OpenClaw 2026.3.11 (29dc654)
Query Slack channel capabilities failed
<img width="700" height="126" alt="Image" src="https://github.com/user-attachments/assets/1699ce07-1dc5-49b7-85b0-0019bb72f909" />
### Steps to reproduce
1. C... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T03:46:40Z | 2026-05-05T04:26:59Z | 2026-05-02T15:25:19Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | Qquanwei | 9,263,655 | MDQ6VXNlcjkyNjM2NTU= | User | false |
openclaw/openclaw | 4,146,653,748 | I_kwDOQb6kR873KOo0 | 55,334 | https://github.com/openclaw/openclaw/issues/55334 | https://api.github.com/repos/openclaw/openclaw/issues/55334 | [perf]: sessions.json unbounded growth causes gateway OOM — skillsSnapshot duplicated per session, no pruning of ephemeral sessions | ## Description
Gateway memory grows unbounded over time (~50-100 MB/min), eventually becoming unresponsive or OOM-killed. Root cause is `sessions.json` accumulating entries that are never pruned, with each entry containing a full `skillsSnapshot` duplicate.
## Environment
- OpenClaw: 2026.3.24
- OS: macOS 15.7.4 (al... | open | null | false | 7 | [] | [] | 2026-03-26T19:07:06Z | 2026-05-04T21:59:55Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | lisaage | 260,271,434 | U_kgDOD4NtSg | User | false |
openclaw/openclaw | 4,340,025,278 | I_kwDOQb6kR88AAAABAq-Hvg | 73,160 | https://github.com/openclaw/openclaw/issues/73160 | https://api.github.com/repos/openclaw/openclaw/issues/73160 | Bug: Session restart recovery fails silently on timeout, causing permanent context loss | ## Description
After a gateway restart, the main session context is permanently lost. The session restart recovery mechanism times out and silently gives up, creating a new session instead of recovering the old one. Transcript files exist on disk but the old session context is orphaned.
## Steps to Reproduce
1. Have... | open | null | false | 2 | [] | [] | 2026-04-28T01:57:36Z | 2026-05-03T06:50:23Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | stayif | 7,867,389 | MDQ6VXNlcjc4NjczODk= | User | false |
openclaw/openclaw | 4,182,458,607 | I_kwDOQb6kR875S0Dv | 58,567 | https://github.com/openclaw/openclaw/issues/58567 | https://api.github.com/repos/openclaw/openclaw/issues/58567 | Session ending with assistant message causes infinite prefill error loop with Opus 4.6 | ## Bug Description
When a session transcript ends with an `assistant` role message (e.g., due to OAuth prompt, tool timeout, or aborted response), subsequent attempts to resume that session fail with Anthropic API error because **Claude Opus 4.6 does not support assistant message prefill**.
This creates an infinite e... | closed | completed | true | 3 | [] | [] | 2026-03-31T21:22:32Z | 2026-04-30T04:32:25Z | 2026-04-28T02:25:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | BenHeee | 108,291,247 | U_kgDOBnRkrw | User | false |
openclaw/openclaw | 4,340,139,660 | I_kwDOQb6kR88AAAABArFGjA | 73,172 | https://github.com/openclaw/openclaw/issues/73172 | https://api.github.com/repos/openclaw/openclaw/issues/73172 | [Bug]: iMessage dmHistoryLimit injects blank messages — macOS writes outbound text to attributedBody, not text column | ## Summary
iMessage history injected via `dmHistoryLimit` is entirely blank for outbound (agent) messages. The gateway reads the `text` column from `chat.db`, but macOS writes outbound message content exclusively to the `attributedBody` column (NSAttributedString blob). This means agents start every new session with z... | open | null | false | 2 | [] | [] | 2026-04-28T02:25:56Z | 2026-05-03T11:25:33Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | Orionation | 260,579,663 | U_kgDOD4ghTw | User | false |
openclaw/openclaw | 4,339,873,902 | I_kwDOQb6kR88AAAABAq04bg | 73,134 | https://github.com/openclaw/openclaw/issues/73134 | https://api.github.com/repos/openclaw/openclaw/issues/73134 | [Bug]: Unhandled rejection from Telegram command handlers exits the gateway daemon via process.exit(1) | ## Bug type
Crash (process/app exits or hangs)
## Beta release blocker
No
## Summary
Telegram command handler errors that are not Grammy `HttpError` or recoverable network errors fall through both monitor handlers and trigger `process.exit(1)` in `infra/unhandled-rejections.ts`, terminating the gateway daemon mid-con... | closed | completed | true | 2 | [] | [] | 2026-04-28T01:17:41Z | 2026-04-30T04:32:23Z | 2026-04-28T02:28:18Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | ybgwon96 | 226,481,728 | U_kgDODX_WQA | User | false |
openclaw/openclaw | 4,339,924,390 | I_kwDOQb6kR88AAAABAq39pg | 73,140 | https://github.com/openclaw/openclaw/issues/73140 | https://api.github.com/repos/openclaw/openclaw/issues/73140 | [Bug]: Telegram grammy dependency is not resolved from global package on 2026.4.25; message send hangs | ## Bug type
Regression / runtime dependency resolution / channel delivery hang
## Summary
On `OpenClaw 2026.4.25 (aa36ee6)`, Telegram inbound updates were consumed but no reply was sent. Direct Telegram Bot API calls worked. The strongest local failure was that importing the OpenClaw Telegram send module from the glob... | closed | completed | true | 1 | [] | [] | 2026-04-28T01:30:17Z | 2026-04-30T04:32:23Z | 2026-04-28T02:30:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | oalansilva | 126,212 | MDQ6VXNlcjEyNjIxMg== | User | false |
openclaw/openclaw | 4,340,162,725 | I_kwDOQb6kR88AAAABArGgpQ | 73,177 | https://github.com/openclaw/openclaw/issues/73177 | https://api.github.com/repos/openclaw/openclaw/issues/73177 | test(cli): qr-cli.test.ts fails with ReferenceError due to vitest mock hoisting | ## Summary
`src/cli/qr-cli.test.ts` fails deterministically with a `ReferenceError: Cannot access '__vi_import_2__' before initialization` when the module import chain includes `../logging/subsystem.js`.
This is a pre-existing vitest mock hoisting bug that becomes exposed when any file in the same import graph adds a... | closed | completed | true | 2 | [] | [] | 2026-04-28T02:32:15Z | 2026-04-30T04:32:09Z | 2026-04-28T03:34:43Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | DoncicX | 13,766,184 | MDQ6VXNlcjEzNzY2MTg0 | User | false |
openclaw/openclaw | 4,340,175,542 | I_kwDOQb6kR88AAAABArHStg | 73,178 | https://github.com/openclaw/openclaw/issues/73178 | https://api.github.com/repos/openclaw/openclaw/issues/73178 | [Bug]: Gateway "full process restart" exits PID 1 with code 0 → Docker Swarm task stays Complete and never restarts (service stuck at 0/1) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When the gateway receives `SIGUSR1` and performs a "full process restart"
(triggered by a config change that requires restart), the entrypoint
process (PID 1 inside the container) **exits with code 0** after spawning
the rep... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-28T02:36:01Z | 2026-05-01T04:45:14Z | 2026-04-28T05:30:28Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | du-nguyen-IT007 | 107,523,720 | U_kgDOBmiuiA | User | false |
openclaw/openclaw | 4,340,149,221 | I_kwDOQb6kR88AAAABArFr5Q | 73,173 | https://github.com/openclaw/openclaw/issues/73173 | https://api.github.com/repos/openclaw/openclaw/issues/73173 | Bug Report: OpenClaw 4.24 — plugin-runtime-deps External Dependency Repair Incomplete | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
plugin-runtime-deps cannot handle packages referenced in dist but not in specs清单 (markdown-it,json5,dotenv,tslog,jiti)
### Steps to reproduce
1. Upgrade to OpenClaw 4.24\n2. Trigger web_search tool (depends on tavily)\n3. ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-28T02:28:36Z | 2026-04-30T04:32:22Z | 2026-04-28T02:36:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | kktttl | 266,642,194 | U_kgDOD-SjEg | User | false |
openclaw/openclaw | 4,340,182,127 | I_kwDOQb6kR88AAAABArHsbw | 73,180 | https://github.com/openclaw/openclaw/issues/73180 | https://api.github.com/repos/openclaw/openclaw/issues/73180 | sessions_spawn model parameter still ignored — all subagents run on parent's default model (2026.4.25) | ## Bug: `sessions_spawn` model override is completely ignored
### Summary
The `model` parameter in `sessions_spawn` is silently ignored. All subagents run on the parent session's default model (`agents.defaults.model.primary`), regardless of what model is specified.
This is a **different** bug from #17479 which was ... | closed | completed | true | 1 | [] | [] | 2026-04-28T02:37:59Z | 2026-04-30T04:32:13Z | 2026-04-28T03:25:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | danielzinhu99 | 98,975,395 | U_kgDOBeY-ow | User | false |
openclaw/openclaw | 4,337,241,402 | I_kwDOQb6kR88AAAABAoUNOg | 72,911 | https://github.com/openclaw/openclaw/issues/72911 | https://api.github.com/repos/openclaw/openclaw/issues/72911 | Gateway restart slow in 4.25 — hardcoded startup-grace and channel-connect-grace | ## Problem
After upgrading to OpenClaw `2026.4.25`, gateway restarts take significantly longer due to a **~27 second delay** between "gateway ready" and channel initialization.
### Timeline from logs
```
00:10:24 [gateway] ready (4 plugins; 4.6s)
00:10:24 [gateway] starting channels and sidecars...
00:10:50 [opencla... | closed | completed | true | 1 | [] | [] | 2026-04-27T16:22:45Z | 2026-04-30T04:32:21Z | 2026-04-28T02:38:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | liujuncg | 262,570,279 | U_kgDOD6aBJw | User | false |
openclaw/openclaw | 4,340,161,415 | I_kwDOQb6kR88AAAABArGbhw | 73,176 | https://github.com/openclaw/openclaw/issues/73176 | https://api.github.com/repos/openclaw/openclaw/issues/73176 | 2026.4.25: chokidar v5 default-import + plugin-loader regression breaks gateway (rollback to 2026.4.24 fixes) | ## Summary
OpenClaw 2026.4.25 fails to boot the embedded agent and most plugins due to a **chokidar v5 default-import incompatibility** in the bundled `dist/*.js` files. This causes Teams (and presumably Discord/Slack/etc.) to silently drop replies to inbound DMs, heartbeat lanes fail every cron tick, and only 3 of th... | closed | completed | true | 3 | [] | [] | 2026-04-28T02:31:53Z | 2026-04-30T04:29:51Z | 2026-04-28T03:03:26Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | w3i-William | 272,058,581 | U_kgDOEDdI1Q | User | false |
openclaw/openclaw | 4,340,195,896 | I_kwDOQb6kR88AAAABArIiOA | 73,181 | https://github.com/openclaw/openclaw/issues/73181 | https://api.github.com/repos/openclaw/openclaw/issues/73181 | [Bug]: After the OpenClaw 4.25 version update, the plugins.installs configuration item in the configuration file is missing. | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After the OpenClaw 4.25 version update, the plugins.installs configuration item in the configuration file is missing.
Running the command 'openclaw plugins update xxxx' will produce an error "no install record"
### Steps to... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-28T02:41:38Z | 2026-04-30T04:31:57Z | 2026-04-28T04:09:06Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | zym951223 | 51,111,325 | MDQ6VXNlcjUxMTExMzI1 | User | false |
openclaw/openclaw | 4,340,198,598 | I_kwDOQb6kR88AAAABArIsxg | 73,182 | https://github.com/openclaw/openclaw/issues/73182 | https://api.github.com/repos/openclaw/openclaw/issues/73182 | Reasoning default silently flipped to on for Claude models — doubles Anthropic spend, leaks thinking blocks to chat | ## Summary
An OpenClaw update around April 2026 silently flipped the default reasoning level for Claude models from `off` to `on`. The change has two compounding effects, both expensive:
1. **Extended-thinking is requested from the Anthropic API on every turn**, so users pay for thinking tokens with no behavioural ch... | open | null | false | 4 | [] | [] | 2026-04-28T02:42:25Z | 2026-05-03T06:50:30Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | w3i-William | 272,058,581 | U_kgDOEDdI1Q | User | false |
openclaw/openclaw | 4,339,840,783 | I_kwDOQb6kR88AAAABAqy3Dw | 73,129 | https://github.com/openclaw/openclaw/issues/73129 | https://api.github.com/repos/openclaw/openclaw/issues/73129 | [Bug] Config rewriter produces invalid JSON during install/upgrade — corrupted openclaw.json | ## Description
The OpenClaw config rewriter (which fires during \`npm install\` / version upgrade and during certain hot-reload events) can produce a syntactically invalid \`openclaw.json\`. The gateway then refuses to load the corrupted config and falls into auto-restore loops.
## Evidence from gateway.err.log
```
... | closed | completed | true | 1 | [] | [] | 2026-04-28T01:09:45Z | 2026-04-30T04:32:20Z | 2026-04-28T02:43:14Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,337,799,302 | I_kwDOQb6kR88AAAABAo2Qhg | 72,963 | https://github.com/openclaw/openclaw/issues/72963 | https://api.github.com/repos/openclaw/openclaw/issues/72963 | WebChat: duplicate message rendering after streaming completes | ## Description
When using the WebChat (Control UI), assistant messages appear **twice** after a response completes. The first copy is the streaming version, the second is the full message fetched via `chat.history`.
## Root Cause
The message dedup logic in the Control UI frontend fails because streaming messages hav... | closed | not_planned | true | 4 | [] | [] | 2026-04-27T17:51:44Z | 2026-05-02T04:28:29Z | 2026-04-29T11:08:43Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | lhtpluto | 132,045,861 | U_kgDOB97cJQ | User | false |
openclaw/openclaw | 4,337,936,981 | I_kwDOQb6kR88AAAABAo-qVQ | 72,974 | https://github.com/openclaw/openclaw/issues/72974 | https://api.github.com/repos/openclaw/openclaw/issues/72974 | Control UI sends duplicate chat.send requests causing double messages | ## Bug Description
**Control UI sends duplicate `chat.send` WebSocket requests**, causing every user message to appear twice in the session history, resulting in duplicate AI responses.
## Steps to Reproduce
1. Open Control UI in browser: `http://localhost:18789/`
2. Send any message (e.g., "你好")
3. Observe: message... | closed | completed | true | 3 | [] | [] | 2026-04-27T18:15:31Z | 2026-05-01T04:44:43Z | 2026-04-28T07:13:08Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Benjamin5281999 | 133,022,008 | U_kgDOB-3BOA | User | false |
openclaw/openclaw | 4,337,236,647 | I_kwDOQb6kR88AAAABAoT6pw | 72,909 | https://github.com/openclaw/openclaw/issues/72909 | https://api.github.com/repos/openclaw/openclaw/issues/72909 | [Bug]: 2026.4.25 stable breaks Intel macOS LaunchAgent + BlueBubbles after upgrade from 4.23; rollback restores service | ### Bug type
Regression (worked before, now fails)
### Summary
Upgrading a stable OpenClaw 2026.4.23 install to 2026.4.25 stable caused a macOS LaunchAgent-managed gateway to appear partially alive, but BlueBubbles/iMessage delivery became unreliable/unusable. CLI/channel probes were slow or timing out, the agent st... | closed | completed | true | 2 | [] | [] | 2026-04-27T16:22:06Z | 2026-04-30T04:32:19Z | 2026-04-28T02:48:54Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | SoumyaBanerjee2025 | 225,474,095 | U_kgDODXB2Lw | User | false |
openclaw/openclaw | 4,340,236,112 | I_kwDOQb6kR88AAAABArK_UA | 73,186 | https://github.com/openclaw/openclaw/issues/73186 | https://api.github.com/repos/openclaw/openclaw/issues/73186 | [Bug]: Thinking/reasoning content leaks into cron announce delivery for Matrix/Feishu | ## Summary
Thinking/reasoning content leaks into cron announce delivery for Matrix and Feishu channels in OpenClaw v2026.4.24. Duplicate delivery also occurs in some cases.
## Environment
- OpenClaw: `openclaw@2026.4.24` (npm global install)
- Runtime deps: `openclaw-2026.4.24-9805e7ca3f7e`
- Channels affected: Matr... | open | null | false | 1 | [] | [] | 2026-04-28T02:51:05Z | 2026-05-03T09:13:19Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | fryccerGit | 254,904,930 | U_kgDODzGKYg | User | false |
openclaw/openclaw | 4,340,249,362 | I_kwDOQb6kR88AAAABArLzEg | 73,187 | https://github.com/openclaw/openclaw/issues/73187 | https://api.github.com/repos/openclaw/openclaw/issues/73187 | Feature Request: Restore formatted gateway startup log output | ## Feature Request
**Current behavior:**
The gateway startup log output has been simplified to just 3 plain lines:
```
agent model: bailian/qwen3.6-plus
ready (5 plugins: browser, feishu, lossless-claw, memory-core, qqbot; 24.0s)
log file: C:\\Users\\Administrator\\.openclaw\\openclaw.log
```
**Previous behavior:**... | open | null | false | 1 | [] | [] | 2026-04-28T02:53:53Z | 2026-05-03T04:43:56Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | xupeijian | 270,532,249 | U_kgDOEB_-mQ | User | false |
openclaw/openclaw | 4,340,262,902 | I_kwDOQb6kR88AAAABArMn9g | 73,189 | https://github.com/openclaw/openclaw/issues/73189 | https://api.github.com/repos/openclaw/openclaw/issues/73189 | [Bug]: cron systemEvent payload.text lost in main-session handoff — agent receives bare heartbeat poll without payload (regression from 4.25 transient runtime context) | ## Summary
Cron jobs with `sessionTarget: "main"` + `payload.kind: "systemEvent"` fire on schedule but the `payload.text` is **not injected** into the session context. The agent only receives a bare `[OpenClaw heartbeat poll]` with no actual system event payload, resulting in "ghost runs" — the cron run records `statu... | closed | completed | true | 2 | [] | [] | 2026-04-28T02:57:00Z | 2026-05-03T04:36:01Z | 2026-04-30T16:17:30Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | WhiteGiverMa | 152,406,589 | U_kgDOCRWKPQ | User | false |
openclaw/openclaw | 4,337,770,989 | I_kwDOQb6kR88AAAABAo0h7Q | 72,960 | https://github.com/openclaw/openclaw/issues/72960 | https://api.github.com/repos/openclaw/openclaw/issues/72960 | Event loop blocked for ~97s during startup — startPluginServices causes 6min restart delay | ## Environment
- **OpenClaw version**: 2026.4.25 (aa36ee6)
- **Node.js**: v22.22.2
- **OS**: Debian 12 (bookworm), Linux 6.1.0-34-amd64, x86_64
- **Enabled plugins**: qqbot, telegram, memory-core, memory-wiki, active-memory, brave, deepseek, voyage
- **Agent model**: cpam/glm-5.1 (default), deepseek/deepseek-v4-pro
#... | closed | completed | true | 2 | [
"bug",
"gateway"
] | [] | 2026-04-27T17:47:14Z | 2026-04-30T04:32:18Z | 2026-04-28T02:57:36Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | rollingshmily | 62,282,098 | MDQ6VXNlcjYyMjgyMDk4 | User | false |
openclaw/openclaw | 4,340,232,392 | I_kwDOQb6kR88AAAABArKwyA | 73,185 | https://github.com/openclaw/openclaw/issues/73185 | https://api.github.com/repos/openclaw/openclaw/issues/73185 | [Bug]: `infer model run --local --prompt ""` reaches the provider; `--gateway` correctly rejects it (Claude shows "No text output returned"; DeepSeek silently bills) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
`pnpm openclaw infer model run --model <id> --prompt ""` (and `" "`, `$'\n'`) sends the empty/whitespace user turn to the provider with no client-side validation. Symptom is provider-dependent: against the de... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-04-28T02:50:08Z | 2026-05-01T04:44:40Z | 2026-04-28T07:33:59Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | iot2edge | 188,053,512 | U_kgDOCzV4CA | User | false |
openclaw/openclaw | 4,340,255,740 | I_kwDOQb6kR88AAAABArML_A | 73,188 | https://github.com/openclaw/openclaw/issues/73188 | https://api.github.com/repos/openclaw/openclaw/issues/73188 | Update 4.23 → 4.24+ causes gateway paralysis: empty runtime-deps package.json + missing chokidar cascades into full event loop starvation | # Bug Report: Update from 4.23 → 4.24+ causes gateway paralysis on power-user setups
## Environment
- **OpenClaw version:** 2026.4.23 (stable) → 2026.4.24 / 2026.4.25 (broken)
- **OS:** macOS 15.7.3 (Sequoia), Intel x86_64 (iMac Pro)
- **Node.js:** v25.9.0
- **npm:** 11.12.1
- **Install method:** `npm install -g openc... | closed | completed | true | 2 | [] | [] | 2026-04-28T02:55:14Z | 2026-04-30T04:32:15Z | 2026-04-28T03:13:59Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | leonlushgit | 256,653,063 | U_kgDOD0w3Bw | User | false |
openclaw/openclaw | 4,337,409,348 | I_kwDOQb6kR88AAAABAoedRA | 72,929 | https://github.com/openclaw/openclaw/issues/72929 | https://api.github.com/repos/openclaw/openclaw/issues/72929 | Gateway startup/status can self-DoS after restart and reports ready before channels are online | # OpenClaw 2026.4.25 gateway startup/status hardening ticket
Date: 2026-04-27
Version: OpenClaw 2026.4.25, commit aa36ee6
Host: macOS LaunchAgent gateway, loopback bind, port 18789
## Summary
After updating OpenClaw from 2026.4.24 to 2026.4.25 and restarting the gateway, Discord appeared offline. The gateway was not... | closed | completed | true | 2 | [] | [] | 2026-04-27T16:49:25Z | 2026-04-30T04:32:17Z | 2026-04-28T03:03:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Phluttr | 268,270,571 | U_kgDOD_176w | User | false |
openclaw/openclaw | 4,337,539,086 | I_kwDOQb6kR88AAAABAomYDg | 72,949 | https://github.com/openclaw/openclaw/issues/72949 | https://api.github.com/repos/openclaw/openclaw/issues/72949 | Plugin loaded twice per gateway run — hooks register at +2.0s but service.start() doesn't fire until +9.5s | ## Summary
> **In plain English:** the deep-observability plugin's hook-registration block is logged twice in a single gateway start — once at +2.0s under source `[gateway]`, then again at +9.5s under source `[plugins]`, with `service.start()` running only on the second load. This means there's a ~7-second window afte... | closed | not_planned | true | 1 | [] | [] | 2026-04-27T17:10:33Z | 2026-04-30T04:30:19Z | 2026-04-28T03:04:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mayank6136 | 16,468,779 | MDQ6VXNlcjE2NDY4Nzc5 | User | false |
openclaw/openclaw | 4,336,919,838 | I_kwDOQb6kR88AAAABAoAlHg | 72,879 | https://github.com/openclaw/openclaw/issues/72879 | https://api.github.com/repos/openclaw/openclaw/issues/72879 | [Bug] thought_signature 400 regression in 2026.4.25 – fix was supposedly in 2026.4.24 | ## Summary
The `thought_signature` Google Generative AI API 400 error has regressed in 2026.4.25. This was previously addressed and that issue was closed after the 2026.4.24 release, but the error is back.
## Error
```
Google Generative AI API error (400): Function call is missing a thought_signature in functionCall... | open | null | false | 4 | [] | [] | 2026-04-27T15:31:26Z | 2026-05-03T03:37:32Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | M00zyx | 135,591,382 | U_kgDOCBT11g | User | false |
openclaw/openclaw | 4,340,316,920 | I_kwDOQb6kR88AAAABArP6-A | 73,201 | https://github.com/openclaw/openclaw/issues/73201 | https://api.github.com/repos/openclaw/openclaw/issues/73201 | Duplicate plugin warning when bundled channel is configured | **Describe the bug**
When a bundled channel (e.g., `feishu`) is configured in `channels.*`, OpenClaw automatically enables the corresponding bundled plugin and adds it to `plugins.entries` at runtime. This causes a duplicate plugin ID warning:
```
plugins.entries.feishu: plugin feishu: duplicate plugin id detected; gl... | closed | completed | true | 1 | [] | [] | 2026-04-28T03:10:12Z | 2026-05-01T04:44:20Z | 2026-04-28T08:57:51Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | wangjy-openclaw | 264,916,084 | U_kgDOD8pMdA | User | false |
openclaw/openclaw | 4,340,336,485 | I_kwDOQb6kR88AAAABArRHZQ | 73,204 | https://github.com/openclaw/openclaw/issues/73204 | https://api.github.com/repos/openclaw/openclaw/issues/73204 | [Bug]: Session compaction blocks InteractionEventListener, causing silent message drops | ## Bug Description
When OpenClaw session compaction triggers on a long-running session, the `InteractionEventListener` on the `critical` lane times out with a 120-second `listener-timeout` error. **Incoming Discord messages are silently dropped** without any user-facing error or retry.
This is a regression from the e... | closed | completed | true | 2 | [] | [] | 2026-04-28T03:15:06Z | 2026-05-01T04:44:39Z | 2026-04-28T07:35:06Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | slideshow-dingo | 258,996,445 | U_kgDOD2_43Q | User | false |
openclaw/openclaw | 4,340,372,241 | I_kwDOQb6kR88AAAABArTTEQ | 73,211 | https://github.com/openclaw/openclaw/issues/73211 | https://api.github.com/repos/openclaw/openclaw/issues/73211 | [Bug]:Mattermost "draft preview" streaming feature is hardcoded and always enabled or disabled in different versions | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Mattermost 频道实现了一个"草稿预览"流式传输功能（由 PR [#47838](https://github.com/openclaw/openclaw/pull/47838) 引入于[2026.4.20](https://github.com/openclaw/openclaw/releases/tag/v2026.4.20)），允许 AI：
1. 通过 `POST /posts` 创建一个占位消息
2... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-28T03:25:08Z | 2026-05-03T09:49:16Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | lzn970721 | 62,925,906 | MDQ6VXNlcjYyOTI1OTA2 | User | false |
openclaw/openclaw | 4,340,411,896 | I_kwDOQb6kR88AAAABArVt-A | 73,222 | https://github.com/openclaw/openclaw/issues/73222 | https://api.github.com/repos/openclaw/openclaw/issues/73222 | [Feature Request] Control UI: Add session kind filter to session dropdown | ## Problem
The session dropdown in the Control UI header shows all sessions including subagent sessions and dreaming sessions, making the list very long and noisy.
## Expected Behavior
Control UI should allow filtering sessions by kind (e.g., `subagent`, `dreaming`, `cron`) in the session dropdown, similar to how th... | open | null | false | 1 | [] | [] | 2026-04-28T03:37:31Z | 2026-05-03T11:25:34Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | stevenliou | 9,360,792 | MDQ6VXNlcjkzNjA3OTI= | User | false |
openclaw/openclaw | 4,340,432,337 | I_kwDOQb6kR88AAAABArW90Q | 73,226 | https://github.com/openclaw/openclaw/issues/73226 | https://api.github.com/repos/openclaw/openclaw/issues/73226 | [Bug] dmScope: per-channel-peer does not work for webchat channel | **Environment:**
- OpenClaw v2026.4.26 (macOS, Node.js)
- Configured `session.dmScope: per-channel-peer`
- Channels: Feishu (websocket) + Web UI
---
**Bug Summary:**
Setting `session.dmScope` to `per-channel-peer` has no effect for the webchat channel. All webchat messages are routed to `agent:main:main` instead of ... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T03:44:20Z | 2026-05-04T04:36:55Z | 2026-05-01T09:57:35Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | OkeyDokeyClaw | 214,973,177 | U_kgDODNA6-Q | User | false |
openclaw/openclaw | 4,337,313,177 | I_kwDOQb6kR88AAAABAoYlmQ | 72,920 | https://github.com/openclaw/openclaw/issues/72920 | https://api.github.com/repos/openclaw/openclaw/issues/72920 | Gateway becomes unstable from bonjour plugin errors, causing Control UI disconnects (1006) | ## Summary
The local gateway appears to restart/crash due to the `bonjour` plugin, which causes Control UI websocket disconnects (`1006`) and makes the UI appear unstable.
## What happened
In Control UI, the interface disconnected with:
- `disconnected (1006): no reason`
At the same time, gateway logs showed resta... | closed | completed | true | 2 | [] | [
"vincentkoc"
] | 2026-04-27T16:33:47Z | 2026-04-30T04:32:07Z | 2026-04-28T03:45:02Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | drarturryzhov-boop | 241,340,210 | U_kgDODmKPMg | User | false |
openclaw/openclaw | 4,340,453,366 | I_kwDOQb6kR88AAAABArYP9g | 73,230 | https://github.com/openclaw/openclaw/issues/73230 | https://api.github.com/repos/openclaw/openclaw/issues/73230 | Cron delivery should support dynamic accountId resolution for channels with unstable session tokens (e.g. openclaw-weixin) | ## Problem
Channels like `openclaw-weixin` generate a new `accountId` on every re-login or Gateway restart (because WeChat session tokens are ephemeral). This means cron jobs with a static `accountId` in their delivery config silently break after any update, restart, or session refresh.
## Reproduction
1. Create a c... | open | null | false | 1 | [] | [] | 2026-04-28T03:50:06Z | 2026-05-03T09:49:18Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | sniu-git | 90,583,807 | MDQ6VXNlcjkwNTgzODA3 | User | false |
openclaw/openclaw | 4,340,393,891 | I_kwDOQb6kR88AAAABArUnow | 73,218 | https://github.com/openclaw/openclaw/issues/73218 | https://api.github.com/repos/openclaw/openclaw/issues/73218 | Bug: agents.defaults.models record-format allowlist stripped at gateway startup despite passing Zod validation (4.23/4.24) | ## Summary
Setting `agents.defaults.models` as a record (map keyed by `provider/model`, values are empty objects) per the documented schema results in the field being silently removed from the runtime config when the gateway starts. Reproducible 100% on OpenClaw 2026.4.23 and 2026.4.24.
## Reproduction
1. Write `/ro... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T03:31:25Z | 2026-04-30T04:32:02Z | 2026-04-28T03:52:19Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | JOY | 5,027,251 | MDQ6VXNlcjUwMjcyNTE= | User | false |
openclaw/openclaw | 4,340,421,301 | I_kwDOQb6kR88AAAABArWStQ | 73,224 | https://github.com/openclaw/openclaw/issues/73224 | https://api.github.com/repos/openclaw/openclaw/issues/73224 | [Bug] 2026.4.26: sharp not installed with media-understanding-core extension — image optimization fails | ## Bug Description
The `media-understanding-core` extension (new in 2026.4.26) declares `sharp` as a runtime dependency, but `sharp` is not installed when upgrading from 2026.4.25 via `npm update`. This causes image optimization to fail for all image-sending operations (message tool sending images to Telegram, etc.) w... | closed | completed | true | 1 | [] | [] | 2026-04-28T03:40:39Z | 2026-04-30T04:32:01Z | 2026-04-28T03:53:27Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | mlaihk | 25,972,362 | MDQ6VXNlcjI1OTcyMzYy | User | false |
openclaw/openclaw | 3,979,541,205 | I_kwDOQb6kR87tMvrV | 24,693 | https://github.com/openclaw/openclaw/issues/24693 | https://api.github.com/repos/openclaw/openclaw/issues/24693 | Security: Hook completion events leak cross-agent email content to default agent session | ## Summary
Hook completion system events are broadcast to the **default agent's main session** regardless of which agent processed the hook. This leaks email content (and potentially other hook payloads) across agent boundaries.
## Steps to Reproduce
1. Configure multi-agent setup with agents `exec` (default) and `d... | closed | completed | true | 2 | [
"security",
"stale"
] | [] | 2026-02-23T18:22:38Z | 2026-04-30T04:32:00Z | 2026-04-28T03:53:53Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | abersonFAC | 226,884,237 | U_kgDODYX6jQ | User | false |
openclaw/openclaw | 4,340,479,178 | I_kwDOQb6kR88AAAABArZ0yg | 73,234 | https://github.com/openclaw/openclaw/issues/73234 | https://api.github.com/repos/openclaw/openclaw/issues/73234 | WebChat/Control UI duplicates assistant replies when auto-TTS appends assistant-media transcript entries | ## Summary
When WebChat / Control UI is used with auto-TTS enabled (`messages.tts.auto` or persisted TTS prefs set to `always`), assistant text replies can appear twice in chat history/transcripts.
The duplicate entry is not a second model generation. It appears to be an extra `gateway-injected` / `assistant-media` t... | closed | completed | true | 2 | [] | [] | 2026-04-28T03:56:20Z | 2026-05-01T04:43:13Z | 2026-04-28T15:01:37Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | donghzhang8-lab | 272,378,712 | U_kgDOEDwrWA | User | false |
openclaw/openclaw | 4,340,486,669 | I_kwDOQb6kR88AAAABAraSDQ | 73,237 | https://github.com/openclaw/openclaw/issues/73237 | https://api.github.com/repos/openclaw/openclaw/issues/73237 | sessions_spawn rejects non-delivery channels (heartbeat, webhook, cron) | ## Bug
When an agent running on a non-delivery channel (e.g. `heartbeat`, `cron`, `webhook`) calls `sessions_spawn`, the server-side agent param validation rejects the request:
```
invalid agent params: unknown channel: heartbeat
```
This happens because `isGatewayMessageChannel()` only includes deliverable channel ... | closed | completed | true | 0 | [] | [] | 2026-04-28T03:58:04Z | 2026-05-01T04:45:04Z | 2026-04-28T06:32:25Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | fancymatt | 7,379,405 | MDQ6VXNlcjczNzk0MDU= | User | false |
openclaw/openclaw | 4,340,502,808 | I_kwDOQb6kR88AAAABArbRGA | 73,244 | https://github.com/openclaw/openclaw/issues/73244 | https://api.github.com/repos/openclaw/openclaw/issues/73244 | claude-live-session: bundled MCP tempDir cleaned up while persistent CLI subprocess still uses it (race) — plus misleading billing-error template | ## Bug 1 — Bundled MCP tempDir cleaned up while persistent CLI subprocess still references it
### Summary
In `claude-live-session` mode, the spawned `claude -p --input-format stream-json` subprocess persists across many turns (stored in the `liveSessions` Map and reused). Its argv permanently contains `--mcp-config /... | closed | completed | true | 0 | [] | [] | 2026-04-28T04:02:27Z | 2026-05-01T04:44:17Z | 2026-04-28T09:00:00Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Demented2984 | 224,897,327 | U_kgDODWepLw | User | false |
openclaw/openclaw | 4,340,523,246 | I_kwDOQb6kR88AAAABArcg7g | 73,248 | https://github.com/openclaw/openclaw/issues/73248 | https://api.github.com/repos/openclaw/openclaw/issues/73248 | [Feature]: Skills prompt injection consumes ~10K chars with 80+ skills, causing model selection noise and context window pressure | ### Summary
## Problem Description
With 80+ skills installed across multiple skill directories (`~/.openclaw/skills`, `~/.agents/skills`, `<workspace>/.agents/skills`, etc.), the XML-formatted skills list injected into the system prompt consumes approximately **7,000–10,000 characters**.
According to the documented ... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-04-28T04:08:47Z | 2026-05-04T04:36:07Z | 2026-05-01T14:51:34Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | ranlingfeng | 26,164,397 | MDQ6VXNlcjI2MTY0Mzk3 | User | false |
openclaw/openclaw | 4,340,370,101 | I_kwDOQb6kR88AAAABArTKtQ | 73,210 | https://github.com/openclaw/openclaw/issues/73210 | https://api.github.com/repos/openclaw/openclaw/issues/73210 | Auto-TTS: add voice-only delivery mode + improve system prompt for natural speech output | ## Feature Request
### Problem 1: No voice-only delivery mode
When TTS is enabled (`messages.tts.auto: "always"` / `/tts on`), OpenClaw sends **both** a voice message and the text reply on channels that support voice notes (Telegram, WhatsApp, Matrix, Feishu). There is no way to suppress the text and deliver voice-on... | open | null | false | 1 | [] | [] | 2026-04-28T03:24:34Z | 2026-05-03T04:43:57Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | altierac | 275,557,454 | U_kgDOEGysTg | User | false |
openclaw/openclaw | 4,340,531,126 | I_kwDOQb6kR88AAAABArc_tg | 73,250 | https://github.com/openclaw/openclaw/issues/73250 | https://api.github.com/repos/openclaw/openclaw/issues/73250 | WS handshake blocked during channel startup — WebUI inaccessible after gateway restart | ## Bug: WS handshake blocked during channel startup
### Summary
After upgrading to OpenClaw 2026.4.26, `openclaw gateway restart` leaves the WebUI (Control UI) inaccessible for approximately 60 seconds. The HTTP server starts in ~8s, but WebSocket connections are blocked until all channels (especially Feishu bots) fin... | closed | completed | true | 2 | [] | [] | 2026-04-28T04:11:22Z | 2026-05-01T04:46:55Z | 2026-04-28T04:31:34Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | ChrisCarlCao | 123,957,385 | U_kgDOB2NwiQ | User | false |
openclaw/openclaw | 4,340,321,679 | I_kwDOQb6kR88AAAABArQNjw | 73,202 | https://github.com/openclaw/openclaw/issues/73202 | https://api.github.com/repos/openclaw/openclaw/issues/73202 | ACPX gateway startup should not depend on live npx adapter resolution | ## Summary
OpenClaw ACPX became unstable when the ACP adapter command depended on live `npx` package resolution at gateway runtime. Pinning the ACP adapters locally and pointing OpenClaw at local wrapper scripts fixed gateway startup and allowed both Codex ACP and Claude ACP sessions to pass.
I am filing this as an u... | closed | completed | true | 1 | [] | [] | 2026-04-28T03:11:16Z | 2026-05-01T04:46:44Z | 2026-04-28T04:39:11Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | joerod26 | 277,153,564 | U_kgDOEIUHHA | User | false |
openclaw/openclaw | 4,340,630,525 | I_kwDOQb6kR88AAAABArjD_Q | 73,268 | https://github.com/openclaw/openclaw/issues/73268 | https://api.github.com/repos/openclaw/openclaw/issues/73268 | suppressToolErrors does not work for mutating tools (message/exec) | ## Problem
`agents.defaults.messages.suppressToolErrors` has no effect on mutating tools like `message`. Source: `pi-embedded-runner-BFOhIiuh.js:980` hardcodes that mutating tools always show warnings.
## Impact
Every failed `message` tool call pushes a ⚠️ warning to the user via the channel, even if the agent handles... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T04:40:33Z | 2026-05-03T04:36:37Z | 2026-04-30T14:58:03Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | 90yuwang | 279,619,049 | U_kgDOEKql6Q | User | false |
openclaw/openclaw | 4,340,630,723 | I_kwDOQb6kR88AAAABArjEww | 73,269 | https://github.com/openclaw/openclaw/issues/73269 | https://api.github.com/repos/openclaw/openclaw/issues/73269 | Assistant text output is auto-sent to channel even when tool calls follow | ## Problem
When an assistant response contains both text and tool calls, the text portion is automatically sent to the channel (e.g., Feishu) as an independent message. This causes 'narration leak' — internal thinking notes appear as standalone messages to the user.
## Impact
Users see fragmented messages like 'Let me... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T04:40:34Z | 2026-05-04T04:35:29Z | 2026-05-01T21:08:08Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | 90yuwang | 279,619,049 | U_kgDOEKql6Q | User | false |
openclaw/openclaw | 4,340,630,798 | I_kwDOQb6kR88AAAABArjFDg | 73,270 | https://github.com/openclaw/openclaw/issues/73270 | https://api.github.com/repos/openclaw/openclaw/issues/73270 | Feature request: cron payload kind=script for direct shell execution | ## Problem
All cron jobs currently use `kind=agentTurn` which spins up a full LLM agent session (~30-38k tokens of system prompt + skill injection). For cron jobs that simply execute a Python/Bash script and return NO_REPLY, this wastes significant tokens.
## Impact
288 runs/day of a 5-minute model-fallback check cron... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T04:40:35Z | 2026-05-03T04:36:36Z | 2026-04-30T14:58:09Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | 90yuwang | 279,619,049 | U_kgDOEKql6Q | User | false |
openclaw/openclaw | 4,340,630,855 | I_kwDOQb6kR88AAAABArjFRw | 73,271 | https://github.com/openclaw/openclaw/issues/73271 | https://api.github.com/repos/openclaw/openclaw/issues/73271 | Feature request: optional session state persistence across restarts | ## Problem
OpenClaw sessions are stateless — gateway restart clears all context. Agents must manually dump context to files before restart and manually read them after.
## Impact
Without manual dump, the agent loses all conversation history, in-progress tasks, and decisions made during the session.
## Current Workaro... | closed | completed | true | 1 | [] | [] | 2026-04-28T04:40:36Z | 2026-05-03T04:35:51Z | 2026-04-30T17:04:55Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | 90yuwang | 279,619,049 | U_kgDOEKql6Q | User | false |
openclaw/openclaw | 4,340,630,929 | I_kwDOQb6kR88AAAABArjFkQ | 73,272 | https://github.com/openclaw/openclaw/issues/73272 | https://api.github.com/repos/openclaw/openclaw/issues/73272 | Feature request: per-agent directive to disable [[reply_to_current]] tag | ## Problem
The `[[reply_to_current]]` tag is automatically processed by the runtime and results in replies that don't render correctly in some channels (e.g., Feishu DM doesn't show quote blocks). Agents use `message thread-reply` instead, but sometimes accidentally output `[[reply_to_current]]` in their text.
## Impa... | open | null | false | 1 | [] | [] | 2026-04-28T04:40:38Z | 2026-05-03T06:50:31Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | 90yuwang | 279,619,049 | U_kgDOEKql6Q | User | false |
openclaw/openclaw | 4,340,638,361 | I_kwDOQb6kR88AAAABArjimQ | 73,273 | https://github.com/openclaw/openclaw/issues/73273 | https://api.github.com/repos/openclaw/openclaw/issues/73273 | CLI hangs in package.json walk loop before signing connect.challenge nonce (v4.24 + v4.25) | ## Summary
After the gateway upgrades from 4.24 → 4.25, every `openclaw …` CLI invocation hangs for ~30s and self-times-out before completing the WS auth handshake. Downgrading to 4.24 does **not** fix it — local state appears to remain poisoned. Gateway, cron, plugins, and Telegram channels are otherwise healthy.
## ... | closed | completed | true | 1 | [] | [] | 2026-04-28T04:42:38Z | 2026-05-03T04:35:29Z | 2026-04-30T18:45:56Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | zackchiutw | 107,488,700 | U_kgDOBmglvA | User | false |
openclaw/openclaw | 4,069,488,146 | I_kwDOQb6kR87yj3YS | 44,833 | https://github.com/openclaw/openclaw/issues/44833 | https://api.github.com/repos/openclaw/openclaw/issues/44833 | [Bug]: Telegram group photo-only messages do not trigger image understanding | ### Bug type
Regression (worked before, now fails)
### Summary
Telegram DM photo-only messages are received and understood, but the same photo-only message in a Telegram group does not trigger image understanding unless a caption is added.
### Steps to reproduce
1. Configure OpenClaw with Telegram enabled and a gr... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-13T07:40:33Z | 2026-05-04T13:18:53Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | h12n0924 | 90,411,174 | MDQ6VXNlcjkwNDExMTc0 | User | false |
openclaw/openclaw | 4,069,543,880 | I_kwDOQb6kR87ykE_I | 44,845 | https://github.com/openclaw/openclaw/issues/44845 | https://api.github.com/repos/openclaw/openclaw/issues/44845 | [Bug]: Token usage shows 0/200k for Volcengine coding plan | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Problem: Session status always shows Context: 0/200k when using volcengine-plan/ark-code-latest(vocengine coding plan).
Root cause: OpenClaw expects usage.input_tokens / usage.output_tokens, but Volcengine Ark returns usage.prompt_tokens ... | open | null | false | 7 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T07:55:46Z | 2026-05-05T10:50:13Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | halocoding | 92,427,115 | U_kgDOBYJTaw | User | false |
openclaw/openclaw | 4,069,711,383 | I_kwDOQb6kR87ykt4X | 44,870 | https://github.com/openclaw/openclaw/issues/44870 | https://api.github.com/repos/openclaw/openclaw/issues/44870 | [Bug]: I have been unable to verify using the codex from the transfer station | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
openclaw version: 2026.3.13
When I encountered a 503 error while configuring a custom model
### Steps to reproduce
1.Configure model
The base_url I configured is https://example.com/v1
Endpoint selected OpenAI compatible
model Id is gpt... | open | null | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T08:32:12Z | 2026-05-04T00:05:45Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | 86jkuncle | 33,897,408 | MDQ6VXNlcjMzODk3NDA4 | User | false |
openclaw/openclaw | 4,069,781,884 | I_kwDOQb6kR87yk_F8 | 44,881 | https://github.com/openclaw/openclaw/issues/44881 | https://api.github.com/repos/openclaw/openclaw/issues/44881 | Gateway holds port after npm upgrade — new process crash-loops until openclaw gateway stop is run | ## Problem
When upgrading OpenClaw via `npm install -g openclaw@latest`, the old gateway process continues to hold the WebSocket port (`ws://127.0.0.1:18789`) even after the binary is replaced.
**Symptoms:**
- New gateway process fails to start: `Gateway failed to start: another gateway instance is already listening ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-13T08:47:28Z | 2026-05-01T04:44:56Z | 2026-04-28T06:45:39Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Mb29661 | 51,697,539 | MDQ6VXNlcjUxNjk3NTM5 | User | false |
openclaw/openclaw | 4,069,847,097 | I_kwDOQb6kR87ylPA5 | 44,896 | https://github.com/openclaw/openclaw/issues/44896 | https://api.github.com/repos/openclaw/openclaw/issues/44896 | Bug: Invalid combination of reasoning_effort and thinking type: low + disabled | ## Description
When using models with reasoning capabilities (e.g., bailian/glm-5, kimi-k2.5), the API returns an HTTP 400 error:
```
HTTP 400: Invalid combination of reasoning_effort and thinking type: low + disabled
```
## Root Cause
The issue is in the `createMoonshotThinkingWrapper` function in the compiled Jav... | open | null | false | 3 | [] | [] | 2026-03-13T09:01:50Z | 2026-05-04T13:18:56Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | newtontech | 45,420,372 | MDQ6VXNlcjQ1NDIwMzcy | User | false |
openclaw/openclaw | 4,069,895,085 | I_kwDOQb6kR87ylaut | 44,910 | https://github.com/openclaw/openclaw/issues/44910 | https://api.github.com/repos/openclaw/openclaw/issues/44910 | [BUG] OpenAI Codex errors leak into user chat | ## Problem
When OpenClaw spawns subagents or coding sessions, OpenAI Codex-related errors leak directly into the user's Telegram chat:
```
{"detail":"The 'gpt-5.3-codex' model is not supported when using Codex with a ChatGPT account."}
```
This happens even when:
- The user has never configured OpenAI or Codex
- The... | open | null | false | 5 | [] | [] | 2026-03-13T09:13:22Z | 2026-05-04T13:18:57Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | nox-openclaw | 267,769,742 | U_kgDOD_XXjg | User | false |
openclaw/openclaw | 4,069,964,059 | I_kwDOQb6kR87ylrkb | 44,922 | https://github.com/openclaw/openclaw/issues/44922 | https://api.github.com/repos/openclaw/openclaw/issues/44922 | [Bug]: Cron job with sessionTarget: "main" triggers both systemEvent and reminder despite delivery.mode: "none" | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Cron job with `sessionTarget: "main"` triggers both systemEvent and scheduled reminder despite `delivery.mode: "none"`.
### Steps to reproduce
1. Create a cron job with the following configuration:
- `sessionTarget: "main"`
- `paylo... | open | null | false | 6 | [
"bug",
"bug:behavior"
] | [] | 2026-03-13T09:28:32Z | 2026-05-04T00:11:53Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | GSL-R | 174,104,911 | U_kgDOCmChTw | User | false |
openclaw/openclaw | 4,069,969,196 | I_kwDOQb6kR87yls0s | 44,925 | https://github.com/openclaw/openclaw/issues/44925 | https://api.github.com/repos/openclaw/openclaw/issues/44925 | [Bug]: Subagent completion silently lost — no retry, no notification, no auto-restart on timeout | ## Environment
- **Version:** OpenClaw 2026.3.11
- **OS:** Ubuntu 24.04, Telegram forum bot
- **Channel:** Telegram (forum mode)
## Describe the bug
Subagent task orchestration has multiple failure modes where **results are silently lost**:
### Pattern 1: Completion announce fails (E31, E42, E45, E79)
Subagent compl... | open | null | false | 10 | [] | [] | 2026-03-13T09:29:37Z | 2026-05-04T13:18:59Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,069,979,012 | I_kwDOQb6kR87ylvOE | 44,930 | https://github.com/openclaw/openclaw/issues/44930 | https://api.github.com/repos/openclaw/openclaw/issues/44930 | [BUG] Telegram messages silently dropped due to update offset race condition | ## Problem
Messages sent via Telegram are sometimes silently dropped and never processed by OpenClaw. The user has to copy-paste the same message to get it through — the copy-paste works because it generates a new message ID that has not been skipped.
## Reproduction
1. Send a message while OpenClaw is actively proc... | open | null | false | 3 | [] | [] | 2026-03-13T09:31:58Z | 2026-05-04T00:11:54Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | nox-openclaw | 267,769,742 | U_kgDOD_XXjg | User | false |
openclaw/openclaw | 4,069,999,532 | I_kwDOQb6kR87yl0Os | 44,935 | https://github.com/openclaw/openclaw/issues/44935 | https://api.github.com/repos/openclaw/openclaw/issues/44935 | [Feature]: Auto-recover to primary model after fallback — with health probe, not blind switchback | ## Summary
When the primary model (e.g. Anthropic Claude) fails and falls back to a secondary model, there is **no safe mechanism to return to the primary** once it recovers. The current behavior is one of two extremes:
1. **Never returns** — user must manually switch back (our current situation; we disabled fallback... | open | null | false | 3 | [] | [] | 2026-03-13T09:36:59Z | 2026-05-04T13:19:00Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,070,032,326 | I_kwDOQb6kR87yl8PG | 44,941 | https://github.com/openclaw/openclaw/issues/44941 | https://api.github.com/repos/openclaw/openclaw/issues/44941 | [Feature]: Bidirectional fallback notifications — alert on switch AND on recovery, per-session/per-topic | ## Summary
When a model fallback happens, the user gets **zero notification**. When the primary model recovers, again **zero notification**. The user only discovers something changed when they notice degraded quality or manually check logs.
This should be a **built-in default behavior**, not something the user has to... | open | null | false | 2 | [] | [] | 2026-03-13T09:44:24Z | 2026-05-04T00:05:53Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,070,236,928 | I_kwDOQb6kR87ymuMA | 44,993 | https://github.com/openclaw/openclaw/issues/44993 | https://api.github.com/repos/openclaw/openclaw/issues/44993 | [Bug]: Heartbeat/Cron "Current time" timestamp is stale - not refreshing between runs | ### Bug type
Regression (worked before, now fails)
### Summary
Heartbeat/Cron messages inject a stale "Current time" timestamp that doesn't update between runs. Shows wrong time to agent.
### Steps to reproduce
1. Configure heartbeat or cron job that sends message to agent
2. Wait several hours (gateway runs for e... | open | null | false | 10 | [
"bug",
"regression"
] | [] | 2026-03-13T10:30:30Z | 2026-05-04T13:19:02Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | mclee1975 | 267,909,583 | U_kgDOD_f5zw | User | false |
openclaw/openclaw | 4,070,354,792 | I_kwDOQb6kR87ynK9o | 45,016 | https://github.com/openclaw/openclaw/issues/45016 | https://api.github.com/repos/openclaw/openclaw/issues/45016 | feat: Strip thinking blocks from history except last assistant message (token savings) | ## Problem
When extended thinking is enabled (or was previously enabled in a session), thinking/redacted_thinking blocks accumulate in conversation history. These are sent back to the API on every turn, consuming significant tokens with no benefit — the model doesn't need its prior reasoning to continue a conversation... | open | null | false | 2 | [] | [] | 2026-03-13T10:55:46Z | 2026-05-04T00:11:55Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | noliganda | 258,411,223 | U_kgDOD2cK1w | User | false |
openclaw/openclaw | 4,070,390,123 | I_kwDOQb6kR87ynTlr | 45,020 | https://github.com/openclaw/openclaw/issues/45020 | https://api.github.com/repos/openclaw/openclaw/issues/45020 | [Bug]: agent:main:main session page error:Unknown property ‘line-clamp’. Declaration dropped. | ### Bug type
Regression (worked before, now fails)
### Summary
After upgrade to 2026.3.12,dashboard chat page,agent:main:main session display a huge exclamation point;console in dev mode, display:Unknown property ‘line-clamp’. Declaration dropped.
My PC is a nvidia DGX SPark 4TB ,Ubuntu ,firefox 148.0.2。
<img width... | closed | completed | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-13T11:02:10Z | 2026-05-04T06:28:15Z | 2026-05-04T06:28:15Z | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | samskeny | 201,200,204 | U_kgDOC_4STA | User | false |
openclaw/openclaw | 4,070,432,973 | I_kwDOQb6kR87yneDN | 45,028 | https://github.com/openclaw/openclaw/issues/45028 | https://api.github.com/repos/openclaw/openclaw/issues/45028 | [Bug]: Control UI replies to Telegram-backed session are delivered to `webchat` instead of Telegram | ### Bug type
Regression (worked before, now fails)
### Summary
**Title:** Control UI replies to Telegram-backed session are delivered to `webchat` instead of Telegram
**Summary**
When opening or interacting with a Telegram direct session from the OpenClaw Control UI, replies are delivered to `webchat` rather than b... | open | null | false | 5 | [
"bug",
"regression"
] | [] | 2026-03-13T11:11:29Z | 2026-05-04T13:19:05Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | cleartangle | 253,114,784 | U_kgDODxY5oA | User | false |
openclaw/openclaw | 4,070,452,055 | I_kwDOQb6kR87ynitX | 45,031 | https://github.com/openclaw/openclaw/issues/45031 | https://api.github.com/repos/openclaw/openclaw/issues/45031 | Feature: Built-in security scanning for skill installation | ## Proposal
When users install skills via `clawhub install` or manually, there's currently no automated security check. Snyk's 2026 research found **36% of agent skills contain security flaws** across registries.
### Suggested integration
[AgentShield](https://github.com/elliotllliu/agent-shield) could run automatic... | open | null | false | 6 | [] | [] | 2026-03-13T11:15:21Z | 2026-05-04T13:19:07Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | elliotllliu | 55,885,132 | MDQ6VXNlcjU1ODg1MTMy | User | false |
openclaw/openclaw | 4,070,537,079 | I_kwDOQb6kR87yn3d3 | 45,041 | https://github.com/openclaw/openclaw/issues/45041 | https://api.github.com/repos/openclaw/openclaw/issues/45041 | [Feature] Output-level filter to strip leaked reasoning/meta-planning from assistant messages | ## Problem
In multi-agent group chat scenarios (multiple AI agents conversing in the same Discord channel), agents frequently leak internal reasoning/meta-planning into their actual assistant output. This creates noise and degrades the user experience.
### Examples of leaked content in assistant messages:
- `Reasonin... | open | null | false | 2 | [] | [] | 2026-03-13T11:32:07Z | 2026-05-04T00:11:56Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | 42opc | 118,930,848 | U_kgDOBxa9oA | User | false |
openclaw/openclaw | 4,070,592,516 | I_kwDOQb6kR87yoFAE | 45,050 | https://github.com/openclaw/openclaw/issues/45050 | https://api.github.com/repos/openclaw/openclaw/issues/45050 | Feishu leaks raw provider `server_error` payload to users instead of generic fallback, with no safe pre-output retry | ### Summary
When using the Feishu channel with `openai-codex/gpt-5.4`, a transient upstream provider failure can be delivered to the end user as a raw error string like:
```text
Codex error: {"type":"error","error":{"type":"server_error","code":"server_error","message":"An error occurred while processing your request... | open | null | false | 3 | [] | [] | 2026-03-13T11:43:42Z | 2026-05-04T00:06:04Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | zxfxpower | 4,416,703 | MDQ6VXNlcjQ0MTY3MDM= | User | false |
openclaw/openclaw | 4,070,888,950 | I_kwDOQb6kR87ypNX2 | 45,082 | https://github.com/openclaw/openclaw/issues/45082 | https://api.github.com/repos/openclaw/openclaw/issues/45082 | fix(mattermost): proactive message tool sends ignore thread context | ## Problem
When an agent uses the `message` tool to send a proactive message (without an explicit `replyTo`), the message lands in the Mattermost channel root — even when the active session is thread-scoped (`replyToMode: "all"`).
**Reproduction:**
1. Configure Mattermost with `replyToMode: "all"`
2. Start a conversa... | open | null | false | 3 | [] | [] | 2026-03-13T12:40:59Z | 2026-05-04T13:19:08Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | teconomix | 6,959,299 | MDQ6VXNlcjY5NTkyOTk= | User | false |
openclaw/openclaw | 4,070,964,803 | I_kwDOQb6kR87ypf5D | 45,102 | https://github.com/openclaw/openclaw/issues/45102 | https://api.github.com/repos/openclaw/openclaw/issues/45102 | [Bug] Gateway requires frequent restart when processing long tasks | ## Description
Gateway becomes unresponsive and requires frequent restarts when processing long-running tasks.
## Steps to reproduce
1. Start Gateway normally
2. Process several voice recognition requests
3. Gateway becomes slow and unresponsive
4. API calls start timing out
## Expected behavior
Gateway should handle... | open | null | false | 3 | [] | [] | 2026-03-13T12:56:18Z | 2026-05-04T13:19:10Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | 555comings | 150,606,742 | U_kgDOCPoTlg | User | false |
openclaw/openclaw | 4,071,174,521 | I_kwDOQb6kR87yqTF5 | 45,134 | https://github.com/openclaw/openclaw/issues/45134 | https://api.github.com/repos/openclaw/openclaw/issues/45134 | Intermediate assistant text leaks to Mattermost channels as unthreaded messages during multi-step tool calls | ## Bug Description
When an OpenClaw agent performs multi-step tool calls, any text written *between* tool calls appears as **standalone unthreaded messages** in the Mattermost channel — outside the reply thread. Only the final answer appears in the thread.
**Example:**
- Agent receives a question, writes \"Let me che... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-13T13:37:03Z | 2026-05-01T04:44:55Z | 2026-04-28T06:45:51Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | hnykda | 2,741,256 | MDQ6VXNlcjI3NDEyNTY= | User | false |
openclaw/openclaw | 4,071,307,572 | I_kwDOQb6kR87yqzk0 | 45,158 | https://github.com/openclaw/openclaw/issues/45158 | https://api.github.com/repos/openclaw/openclaw/issues/45158 | [Bug]: 飞书多 Agent 路由问题 - 所有消息都路由到同一个 agent | ### Bug type
Regression (worked before, now fails)
### Summary
## 问题概述
配置了 7 个飞书 agent，每个有独立的 binding，但所有消息都路由到同一个 agent。
## 环境信息
- OpenClaw 版本: 2026.3.12
- 操作系统: macOS 26.2
- 飞书插件: @openclaw/feishu@2026.3.7
## 实际行为
所有消息都路由到 agent:feishu-ou_0e1f866b530ddfa37146537a067b91f3:main
## 预期行为
每个用户的消息应该路由到对应的 agent
## 已尝试方案... | open | null | false | 4 | [
"bug",
"regression"
] | [] | 2026-03-13T14:02:14Z | 2026-05-04T00:06:05Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | jeffery85615-byte | 266,107,418 | U_kgDOD9x6Gg | User | false |
openclaw/openclaw | 4,071,442,028 | I_kwDOQb6kR87yrUZs | 45,179 | https://github.com/openclaw/openclaw/issues/45179 | https://api.github.com/repos/openclaw/openclaw/issues/45179 | Feature: temp artifact provenance + durable-path guardrails to prevent /tmp chaos | ## Summary
OpenClaw should make temp-audit artifacts safe by default so agents do not create ambiguous `/tmp` residue that later gets escalated as possibly-important work.
Right now, agents can clone repos, dump audit outputs, write prompt scaffolds, and stage review material in `/tmp` or `/private/tmp` with no prove... | open | null | false | 2 | [] | [] | 2026-03-13T14:24:57Z | 2026-05-04T13:19:11Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | sene1337 | 261,265,124 | U_kgDOD5KW5A | User | false |
openclaw/openclaw | 4,071,513,780 | I_kwDOQb6kR87yrl60 | 45,184 | https://github.com/openclaw/openclaw/issues/45184 | https://api.github.com/repos/openclaw/openclaw/issues/45184 | [Feature]: Skip dispatch when message mentions a different account's bot | Running a multi-account Matrix setup where different agents handle different rooms. The problem: when a user @mentions Bot A in a room where Bot B has `requireMention: false`, Bot B's handler still dispatches the message to the LLM — triggering a typing indicator, consuming API tokens, and generating a response that's ... | open | null | false | 3 | [] | [] | 2026-03-13T14:36:36Z | 2026-05-04T13:19:13Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | JustHereToHelp | 50,504,690 | MDQ6VXNlcjUwNTA0Njkw | User | false |
openclaw/openclaw | 4,071,636,169 | I_kwDOQb6kR87ysDzJ | 45,210 | https://github.com/openclaw/openclaw/issues/45210 | https://api.github.com/repos/openclaw/openclaw/issues/45210 | [Feature]: Per-session/topic pinning to prevent cleanup | ## Problem
Currently, `pruneAfter` and `maxEntries` are global settings. Users cannot protect specific sessions/topics from being cleaned up.
## Use Case
- Users may have important long-running sessions in specific Telegram topics (e.g., a topic for a major project)
- These sessions contain valuable context that sho... | closed | not_planned | true | 2 | [] | [] | 2026-03-13T14:57:58Z | 2026-05-03T04:38:25Z | 2026-04-30T05:18:40Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | YorrickBao | 18,000,481 | MDQ6VXNlcjE4MDAwNDgx | User | false |
openclaw/openclaw | 4,071,709,926 | I_kwDOQb6kR87ysVzm | 45,224 | https://github.com/openclaw/openclaw/issues/45224 | https://api.github.com/repos/openclaw/openclaw/issues/45224 | Unhandled Playwright assertion error in CRSession._onMessage crashes Gateway | ## Bug Description
An unhandled `Assertion error` from Playwright's Chrome DevTools Protocol (CDP) session management crashes the entire Gateway process. The error is not caught by OpenClaw, causing a full process exit that requires launchd to restart.
## Stack Trace
```
[ERROR] Unhandled promise rejection: Error: A... | open | null | false | 4 | [] | [] | 2026-03-13T15:11:01Z | 2026-05-04T13:19:15Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | yyxiaohui | 235,090,347 | U_kgDODgMxqw | User | false |
openclaw/openclaw | 4,072,029,751 | I_kwDOQb6kR87ytj43 | 45,268 | https://github.com/openclaw/openclaw/issues/45268 | https://api.github.com/repos/openclaw/openclaw/issues/45268 | status: context usage is misleading when totalTokensFresh=false | ## Summary
`/status` can show misleading context usage when session token totals are stale.
When `totalTokens` is missing and `totalTokensFresh=false`, the status output currently falls back to `inputTokens + outputTokens`, which can display values like `299k/200k (149%)` even though true context occupancy is unknown.... | closed | completed | true | 5 | [
"stale"
] | [] | 2026-03-13T16:09:44Z | 2026-05-01T04:40:23Z | 2026-04-29T04:08:18Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | xyooz | 101,877,493 | U_kgDOBhKG9Q | User | false |
openclaw/openclaw | 4,072,031,172 | I_kwDOQb6kR87ytkPE | 45,269 | https://github.com/openclaw/openclaw/issues/45269 | https://api.github.com/repos/openclaw/openclaw/issues/45269 | [Bug]: `apply_patch` is treated as an unknown/plugin-only tool in agent policy pipeline, so agent-routed runs cannot execute it | ### Bug type
Regression (worked before, now fails)
### Summary
## Summary
In OpenClaw 2026.3.12, `apply_patch` appears to exist as a built-in runtime tool, but the agent tool-policy pipeline strips it from allowlists before execution.
This causes repeated warnings like:
- `tools.profile (coding) allowlist contains ... | open | null | false | 7 | [
"bug",
"regression"
] | [] | 2026-03-13T16:09:56Z | 2026-05-04T00:12:02Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | Alfred-claw28 | 263,039,412 | U_kgDOD62ptA | User | false |
openclaw/openclaw | 4,072,060,654 | I_kwDOQb6kR87ytrbu | 45,271 | https://github.com/openclaw/openclaw/issues/45271 | https://api.github.com/repos/openclaw/openclaw/issues/45271 | [Bug]: Model does "tool calling narrations" since 2026.3.7 | ### Bug type
Regression (worked before, now fails)
### Summary
I just updated OpenClaw from 2026.3.2 to 2026.3.8. I found out that my agent now always sends text before calling a tool (e.g. `role":"assistant", content: [ {"type":"text","text":"Blabla"}, {"type":"toolCall",...} ]`).
This leads to much more messages ... | closed | completed | true | 10 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-13T16:15:25Z | 2026-05-01T04:42:10Z | 2026-04-28T19:17:29Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | MexHigh | 50,419,884 | MDQ6VXNlcjUwNDE5ODg0 | User | false |
openclaw/openclaw | 4,072,224,823 | I_kwDOQb6kR87yuTg3 | 45,301 | https://github.com/openclaw/openclaw/issues/45301 | https://api.github.com/repos/openclaw/openclaw/issues/45301 | [Bug]: The space_id parameter in Lark Knowledge Base tools has been tampered with. | ### Bug type
Regression (worked before, now fails)
### Summary
Problem Description: When calling the `feishu_create_doc` or `feishu_wiki_space_node` tools in the OpenClaw Lark plugin, the user-provided `space_id` (knowledge base ID) is automatically modified.
Reproduction Steps:
Call `feishu_wiki_space list` to get... | open | null | false | 3 | [
"bug",
"regression"
] | [] | 2026-03-13T16:45:40Z | 2026-05-04T00:06:18Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | wang1528186571 | 139,137,698 | U_kgDOCEsSog | User | false |
openclaw/openclaw | 4,340,651,752 | I_kwDOQb6kR88AAAABArkW6A | 73,274 | https://github.com/openclaw/openclaw/issues/73274 | https://api.github.com/repos/openclaw/openclaw/issues/73274 | Feature Request: Expose appendAssistantMessageToSessionTranscript in Plugin Runtime API | ## Problem
Plugins that hook `agent_end` (e.g. for cron output injection) need to persist messages into **other sessions' transcripts**. Currently, the only cross-session injection API available to plugins is `enqueueSystemEvent`, which is **in-memory only** and ephemeral.
This creates a fundamental limitation:
- `e... | open | null | false | 2 | [] | [] | 2026-04-28T04:46:33Z | 2026-05-03T11:25:36Z | null | CONTRIBUTOR | null | 20260503T120038Z | 2026-05-03T12:00:38Z | kagura-agent | 268,167,063 | U_kgDOD_vnlw | User | false |
openclaw/openclaw | 4,340,663,224 | I_kwDOQb6kR88AAAABArlDuA | 73,278 | https://github.com/openclaw/openclaw/issues/73278 | https://api.github.com/repos/openclaw/openclaw/issues/73278 | [Bug]: Discord autoThread mentions can produce duplicate parent-channel and thread replies | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Discord channel mentions with `autoThread: true` can produce two user-visible replies: one manual `message.send` reply in the parent channel and one normal final response in the created thread.
### Steps to re... | open | null | false | 1 | [] | [] | 2026-04-28T04:49:35Z | 2026-05-03T06:07:40Z | null | CONTRIBUTOR | null | 20260503T090142Z | 2026-05-03T09:01:42Z | solavrc | 145,330,217 | U_kgDOCKmQKQ | User | false |
openclaw/openclaw | 4,340,280,674 | I_kwDOQb6kR88AAAABArNtYg | 73,192 | https://github.com/openclaw/openclaw/issues/73192 | https://api.github.com/repos/openclaw/openclaw/issues/73192 | [Bug]: Update4.20->4.26，/new session 错误 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
创建新会话后，报错如下，多个Agent测试均存在。回退4.20并无该问题
### Steps to reproduce
1. 更新到2026.4.26
2. 直接/new
3. 就会报错
### Expected behavior
NOT_ENOUGH_INFO
### Actual behavior
Agent failed before reply: All models failed (4): minimax/MiniMax... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-28T03:01:30Z | 2026-05-01T04:46:19Z | 2026-04-28T04:54:27Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | mc275 | 17,558,568 | MDQ6VXNlcjE3NTU4NTY4 | User | false |
openclaw/openclaw | 4,081,507,639 | I_kwDOQb6kR87zRt03 | 48,123 | https://github.com/openclaw/openclaw/issues/48123 | https://api.github.com/repos/openclaw/openclaw/issues/48123 | Gateway silently drops non-image attachments in chat.send | ## Problem
`parseMessageWithAttachments()` in `src/gateway/chat-attachments.ts` silently drops any attachment that is not `image/*`:
```typescript
// chat-attachments.ts:123-126
if (sniffedMime && !isImageMime(sniffedMime)) {
log?.warn(`attachment ${label}: detected non-image (${sniffedMime}), dropping`);
continu... | closed | completed | true | 2 | [] | [] | 2026-03-16T10:08:26Z | 2026-05-01T04:46:18Z | 2026-04-28T04:55:02Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | samzong | 13,782,141 | MDQ6VXNlcjEzNzgyMTQx | User | false |
openclaw/openclaw | 4,340,695,876 | I_kwDOQb6kR88AAAABArnDRA | 73,284 | https://github.com/openclaw/openclaw/issues/73284 | https://api.github.com/repos/openclaw/openclaw/issues/73284 | fix(logging): subsystem console writes secrets verbatim at writeConsoleLine exit | ## Sub-issue of #64046
Follow-up to #67953 (file log redaction, now closed).
## Problem
`createSubsystemLogger()` writes console/terminal output without redaction.
File logging for subsystem loggers was fixed by #67953 via the `getChildLogger()` → `buildLogger()` file transport `redactSensitiveText()` gate. Console... | closed | completed | true | 0 | [] | [] | 2026-04-28T04:57:21Z | 2026-05-01T04:45:03Z | 2026-04-28T06:36:52Z | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Ziy1-Tan | 49,604,965 | MDQ6VXNlcjQ5NjA0OTY1 | User | false |
openclaw/openclaw | 4,340,559,555 | I_kwDOQb6kR88AAAABAreuww | 73,255 | https://github.com/openclaw/openclaw/issues/73255 | https://api.github.com/repos/openclaw/openclaw/issues/73255 | [Bug] Telegram deleteWebhook fails on startup: ENETUNREACH despite working curl |
## Bug Description
Telegram polling provider fails on every startup with "deleteWebhook failed: Network request for 'deleteWebhook' failed!" followed by retry loops. The network errors show `ETIMEDOUT` and `ENETUNREACH` codes, but `curl` from the same host reaches api.telegram.org fine.
## Steps to Reproduce
1. Sta... | closed | completed | true | 2 | [] | [] | 2026-04-28T04:20:43Z | 2026-05-01T04:46:03Z | 2026-04-28T05:04:06Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | ttomiczek | 265,786,127 | U_kgDOD9eTDw | User | false |
openclaw/openclaw | 4,338,237,591 | I_kwDOQb6kR88AAAABApRAlw | 72,996 | https://github.com/openclaw/openclaw/issues/72996 | https://api.github.com/repos/openclaw/openclaw/issues/72996 | Mid-flight pnpm upgrade leaves gateway dead with no launchd respawn (macOS); plus plaintext secrets in plist | # Mid-flight pnpm upgrade leaves gateway dead with no launchd respawn (macOS)
**Versions:** running 2026.4.24 → upgrade target 2026.4.25 (latest)
**Platform:** macOS 26.4.1 arm64, node 24.13.0, pnpm
**Service mode:** LaunchAgent (`ai.openclaw.gateway`, KeepAlive=true)
## Summary
When the npm-global package is replac... | closed | completed | true | 3 | [
"bug",
"app: macos",
"gateway",
"security"
] | [] | 2026-04-27T19:07:53Z | 2026-05-01T04:45:41Z | 2026-04-28T05:18:33Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Mathewb7 | 157,361,264 | U_kgDOCWEkcA | User | false |
openclaw/openclaw | 4,340,768,662 | I_kwDOQb6kR88AAAABArrflg | 73,293 | https://github.com/openclaw/openclaw/issues/73293 | https://api.github.com/repos/openclaw/openclaw/issues/73293 | sessions_spawn model parameter silently ignored — subagent always uses agents.defaults.model.primary | ## Bug Description
The `model` parameter in `sessions_spawn()` is accepted and saved to `runs.json` (with `modelApplied: true` in the response), but the child subagent session actually starts with `agents.defaults.model.primary` instead of the requested model.
## Reproduction
Three independent subagent spawns, all o... | closed | completed | true | 1 | [] | [] | 2026-04-28T05:17:02Z | 2026-05-01T04:45:34Z | 2026-04-28T05:21:15Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | lhtpluto | 132,045,861 | U_kgDOB97cJQ | User | false |
openclaw/openclaw | 4,083,723,637 | I_kwDOQb6kR87zaK11 | 48,388 | https://github.com/openclaw/openclaw/issues/48388 | https://api.github.com/repos/openclaw/openclaw/issues/48388 | [Bug]: Feishu file names with Chinese characters are garbled (UTF-8 encoding issue) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When receiving files via Feishu channel, file names containing Chinese characters are corrupted due to UTF-8 encoding being incorrectly interpreted as Latin-1/ISO-8859-1.
### Steps to reproduce
1. Configure Feishu channel in OpenClaw.
2. ... | closed | completed | true | 8 | [
"bug",
"bug:behavior",
"clawsweeper"
] | [] | 2026-03-16T17:09:56Z | 2026-05-01T04:45:32Z | 2026-04-28T05:23:21Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | liuhua1202 | 30,530,878 | MDQ6VXNlcjMwNTMwODc4 | User | false |
openclaw/openclaw | 4,191,439,804 | I_kwDOQb6kR8751Eu8 | 59,409 | https://github.com/openclaw/openclaw/issues/59409 | https://api.github.com/repos/openclaw/openclaw/issues/59409 | [Bug]: Feishu plugin: Chinese filenames in attachments display as garbled (Latin-1 encoding issue) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When receiving files with Chinese filenames via Feishu webhook, the filenames are displayed as garbled text (e.g., "六大组件全解析.md" shows as "å_å_ç_ä_å_è_æ.md"). Server locale is correctly set to UTF-8, but the fil... | closed | completed | true | 1 | [
"bug",
"bug:behavior",
"clawsweeper"
] | [] | 2026-04-02T04:16:36Z | 2026-05-01T04:45:31Z | 2026-04-28T05:23:30Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | cycle0 | 2,794,488 | MDQ6VXNlcjI3OTQ0ODg= | User | false |
openclaw/openclaw | 4,340,553,289 | I_kwDOQb6kR88AAAABAreWSQ | 73,253 | https://github.com/openclaw/openclaw/issues/73253 | https://api.github.com/repos/openclaw/openclaw/issues/73253 | [Bug]: image_generate MEDIA URL not delivered to Telegram — sendPhoto never called (v2026.4.26) | ## Bug Summary
When the `image_generate` tool generates an image, the `MEDIA:` URL appears correctly in the tool result text and `details.media.mediaUrls` is properly populated, but **Telegram receives only the text reply — no photo is sent**. The Telegram delivery layer calls `sendMessage` (text) but never calls `sen... | closed | completed | true | 1 | [] | [] | 2026-04-28T04:18:36Z | 2026-05-01T04:45:28Z | 2026-04-28T05:23:41Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | mlaihk | 25,972,362 | MDQ6VXNlcjI1OTcyMzYy | User | false |
openclaw/openclaw | 4,340,797,681 | I_kwDOQb6kR88AAAABArtQ8Q | 73,298 | https://github.com/openclaw/openclaw/issues/73298 | https://api.github.com/repos/openclaw/openclaw/issues/73298 | [Bug]: Severe slowdown, startup delay, and near-unusable responsiveness after upgrading from 2026.4.2 to 2026.4.26 on Codex setup | [Bug]: Severe slowdown, startup delay, and near-unusable responsiveness after upgrading from 2026.4.2 to 2026.4.26 on Codex setup
### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading this Debian host from OpenClaw 2026.4.2 to 2026.4.26 in a Codex-based setup, sta... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T05:24:22Z | 2026-05-04T04:36:50Z | 2026-05-01T10:58:21Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | ndj888 | 11,382,964 | MDQ6VXNlcjExMzgyOTY0 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.