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 | 4,004,127,565 | I_kwDOQb6kR87uqiNN | 29,552 | https://github.com/openclaw/openclaw/issues/29552 | https://api.github.com/repos/openclaw/openclaw/issues/29552 | [Bug]: Version mismatch | ### Summary
global version mismatch with the current version, it's happening since 2026.2.25
### Steps to reproduce
make sure all your modules and components are upto the current version before pushing it to origin main
### Expected behavior
gatway succesfuly pairs
### Actual behavior
gatway pair faulier
### O... | closed | completed | false | 11 | [
"bug"
] | [] | 2026-02-28T07:31:22Z | 2026-05-04T03:25:00Z | 2026-05-03T19:11:47Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | Bekiman1 | 101,992,829 | U_kgDOBhRJfQ | User | false |
openclaw/openclaw | 4,038,317,895 | I_kwDOQb6kR87ws9dH | 38,826 | https://github.com/openclaw/openclaw/issues/38826 | https://api.github.com/repos/openclaw/openclaw/issues/38826 | TTS voice reply leaks NO_REPLY / internal tokens to Telegram chat | ## Bug Description
When the agent uses the ts tool to send a voice message on Telegram, the follow-up NO_REPLY silent token (and sometimes No response generated. Please try again.) leaks through to the user as a visible text message, instead of being suppressed.
## Steps to Reproduce
1. User requests the agent to r... | closed | completed | true | 5 | [
"stale"
] | [] | 2026-03-07T11:56:48Z | 2026-04-30T04:44:53Z | 2026-04-25T07:00:20Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Ma-tiezhu | 265,853,321 | U_kgDOD9iZiQ | User | false |
openclaw/openclaw | 4,014,465,023 | I_kwDOQb6kR87vR9__ | 32,473 | https://github.com/openclaw/openclaw/issues/32473 | https://api.github.com/repos/openclaw/openclaw/issues/32473 | [Bug]: control ui requires device identity (use HTTPS or localhost secure context) | ### Bug type
Regression (worked before, now fails)
### Summary
I'm using a Hostinger VPS and Docker, and after configuring the Brave key, I'm receiving the reported error.
control ui requires device identity (use HTTPS or localhost secure context)
I can't find how to solve this, can anyone help me?
### Steps to r... | open | null | false | 15 | [
"bug",
"regression"
] | [] | 2026-03-03T03:24:41Z | 2026-05-05T03:11:02Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | RafaelLee | 681,629 | MDQ6VXNlcjY4MTYyOQ== | User | false |
openclaw/openclaw | 4,021,050,703 | I_kwDOQb6kR87vrF1P | 34,252 | https://github.com/openclaw/openclaw/issues/34252 | https://api.github.com/repos/openclaw/openclaw/issues/34252 | [Feature]: Allow multiple Ollama Providers | ### Summary
I am using 2 instances of Ollama on different Ports. I is using a GPU (mostly used by Home Assistant) and the Dumb CPU one but a ton of RAM is dedicated to OpenClaw. I would like to selectively allocate the GPU Instance to OpenClaw MemorySearch (nomic-embed-text) and HeartBeat (same Model HA uses), the r... | closed | completed | true | 6 | [
"enhancement"
] | [] | 2026-03-04T09:41:10Z | 2026-04-30T04:34:44Z | 2026-04-27T10:57:55Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | AlbyZ | 15,170,053 | MDQ6VXNlcjE1MTcwMDUz | User | false |
openclaw/openclaw | 4,189,496,037 | I_kwDOQb6kR875tqLl | 59,228 | https://github.com/openclaw/openclaw/issues/59228 | https://api.github.com/repos/openclaw/openclaw/issues/59228 | [Bug]: command-dispatch skill routing strips arguments from slash commands | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Slash commands with arguments (e.g., `/command 115`) do not pass arguments to the skill dispatch script. Only the base command name is received — arguments are stripped.
### Steps to reproduce
1. Configure a skill with `co... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-04-01T19:52:15Z | 2026-04-29T04:40:02Z | 2026-04-26T08:57:55Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | leobrigassi | 171,322,326 | U_kgDOCjYr1g | User | false |
openclaw/openclaw | 4,111,739,826 | I_kwDOQb6kR871FCuy | 51,487 | https://github.com/openclaw/openclaw/issues/51487 | https://api.github.com/repos/openclaw/openclaw/issues/51487 | Feature Request: Support paste image directly into webchat input box | ## Feature Request
**Problem:**
Currently in the webchat interface, users cannot paste images directly into the input box. When users try to paste a screenshot (Ctrl+V), nothing happens. The only way to share images is:
1. Use the file upload button
2. Send the local file path
3. Use a different channel that support... | closed | completed | true | 3 | [] | [] | 2026-03-21T07:41:07Z | 2026-04-30T04:51:34Z | 2026-04-25T05:27:46Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | stat11 | 99,306,077 | U_kgDOBetKXQ | User | false |
openclaw/openclaw | 4,040,967,916 | I_kwDOQb6kR87w3Ebs | 39,690 | https://github.com/openclaw/openclaw/issues/39690 | https://api.github.com/repos/openclaw/openclaw/issues/39690 | Bug: Multimodal (Vision) Messages Ignored for Custom Ollama Cloud Models | # Bug: Multimodal (Vision) Messages Ignored for Custom Ollama Cloud Models
## Summary
OpenClaw silently strips images from messages when using custom Ollama Cloud models, even when the model is explicitly configured with input: ["text", "image"] in the provider config.
## Root Cause Found
In dist/auth-profiles-CNyDTs... | closed | completed | true | 5 | [] | [] | 2026-03-08T09:49:09Z | 2026-04-30T04:37:30Z | 2026-04-27T04:50:28Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | Skrblik | 139,452,656 | U_kgDOCE_g8A | User | false |
openclaw/openclaw | 4,191,408,468 | I_kwDOQb6kR87509FU | 59,405 | https://github.com/openclaw/openclaw/issues/59405 | https://api.github.com/repos/openclaw/openclaw/issues/59405 | [Bug]: Telegram channel runtime drops exec/grep tool output for openai/gpt-5.1-codex | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Version: 2026.3.13
Model: openai/gpt-5.1-codex (API key auth)
Channel: Telegram
Symptom: exec, grep, find tools fire but return no output to agent
context in Telegram sessions. Same queries work corre... | open | null | false | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-02T04:08:13Z | 2026-05-02T04:19:26Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | motnaj | 63,837,645 | MDQ6VXNlcjYzODM3NjQ1 | User | false |
openclaw/openclaw | 4,166,025,051 | I_kwDOQb6kR874UH9b | 57,326 | https://github.com/openclaw/openclaw/issues/57326 | https://api.github.com/repos/openclaw/openclaw/issues/57326 | CLI-backed helper paths still bypass CLI dispatch on latest main | ## Summary
Latest `main` already contains the original CLI backend ownership / `isCliProvider()` fix that prompted this issue.
What is still broken on latest `main` is a smaller remaining surface: some helper paths still bypass `runCliAgent()` for CLI-backed models and go through the embedded/API path instead.
Affect... | open | null | false | 9 | [] | [] | 2026-03-29T23:07:25Z | 2026-05-05T11:51:46Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | nicko-ai | 39,348,559 | MDQ6VXNlcjM5MzQ4NTU5 | User | false |
openclaw/openclaw | 4,218,907,008 | I_kwDOQb6kR877d2mA | 62,579 | https://github.com/openclaw/openclaw/issues/62579 | https://api.github.com/repos/openclaw/openclaw/issues/62579 | [Windows] QMD collection name mismatch: creates memory-root-main but queries memory-alt-main | ### Summary
On Windows 11 native install (OpenClaw 2026.4.5), enabling `memory.backend: "qmd"` causes all `memory_search` calls to fail with `Collection not found: memory-alt-main` and fall back to builtin FTS.
OpenClaw creates QMD collections named `memory-root-main` and `memory-dir-main` on boot, but queries for `m... | closed | completed | true | 2 | [] | [] | 2026-04-07T15:50:40Z | 2026-04-28T05:05:50Z | 2026-04-25T14:40:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Hag-Fish | 271,590,354 | U_kgDOEDAj0g | User | false |
openclaw/openclaw | 4,164,228,294 | I_kwDOQb6kR874NRTG | 57,055 | https://github.com/openclaw/openclaw/issues/57055 | https://api.github.com/repos/openclaw/openclaw/issues/57055 | [Bug]: Sandbox browser ignores global browser.ssrfPolicy configuration, causing navigation to be blocked | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When using sandbox browser with browser.ssrfPolicy.dangerouslyAllowPrivateNetwork: true configured, browser navigation still fails with "Navigation blocked: strict browser SSRF policy requires Playwright-backed... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-29T13:10:32Z | 2026-04-27T04:39:47Z | 2026-04-25T02:38:03Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | linjiaqin12138 | 38,090,888 | MDQ6VXNlcjM4MDkwODg4 | User | false |
openclaw/openclaw | 4,026,059,705 | I_kwDOQb6kR87v-Mu5 | 35,560 | https://github.com/openclaw/openclaw/issues/35560 | https://api.github.com/repos/openclaw/openclaw/issues/35560 | Doctor injects orphan top-level keys into multi-account telegram config on every CLI command | ## Bug Description
When using a multi-account Telegram configuration (e.g., `channels.telegram.accounts.default`), the Doctor automatically injects orphan top-level keys (`enabled`, `dmPolicy`, `groupPolicy`, `streaming`) into `channels.telegram` on virtually every CLI command invocation (`gateway start`, `gateway sto... | closed | completed | true | 3 | [
"stale"
] | [] | 2026-03-05T06:11:46Z | 2026-04-28T04:43:50Z | 2026-04-26T01:24:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kilhyeonjun | 41,348,539 | MDQ6VXNlcjQxMzQ4NTM5 | User | false |
openclaw/openclaw | 4,221,188,773 | I_kwDOQb6kR877mjql | 62,761 | https://github.com/openclaw/openclaw/issues/62761 | https://api.github.com/repos/openclaw/openclaw/issues/62761 | iMessage channel replays old messages after update restart — no cursor persistence | ## Summary
After running `openclaw update` (updated to 2026.4.5), the iMessage channel replayed a large batch of already-processed messages, causing an outgoing message loop.
## Version
- OpenClaw: 2026.4.5 (3e72c03)
- imsg: 0.5.0
- macOS: 25.3.1 arm64
## Steps to Reproduce
1. Have iMessage channel configured and ... | open | null | false | 3 | [] | [] | 2026-04-07T23:40:17Z | 2026-05-02T11:16:35Z | null | NONE | null | 20260502T120029Z | 2026-05-02T12:00:29Z | brianroccoforte1 | 227,307,261 | U_kgDODYxu_Q | User | false |
openclaw/openclaw | 4,131,734,765 | I_kwDOQb6kR872RUTt | 54,131 | https://github.com/openclaw/openclaw/issues/54131 | https://api.github.com/repos/openclaw/openclaw/issues/54131 | Bug: WhatsApp outbound media (MEDIA: token) fails silently — workaround: openclaw message send --media | ## Summary
When an agent includes a `MEDIA:` path in its reply text, the file is **never delivered** to WhatsApp. The outbound log always shows `mediaUrl=null`, `mediaKind=null`. No error is logged.
## Environment
- OpenClaw: `2026.3.23-2`
- macOS: Darwin 25.3.0 arm64 (Mac mini M4)
- Gateway mode: LaunchDaemon (boot... | closed | completed | true | 7 | [] | [] | 2026-03-25T01:24:36Z | 2026-04-28T05:27:15Z | 2026-04-25T09:54:46Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | felpsanchez | 268,795,311 | U_kgDOEAV9rw | User | false |
openclaw/openclaw | 3,932,837,605 | I_kwDOQb6kR87qalbl | 14,785 | https://github.com/openclaw/openclaw/issues/14785 | https://api.github.com/repos/openclaw/openclaw/issues/14785 | Reduce tool schema token overhead (~3,500 tok/session) | ## Problem
Every session loads the full JSON schemas for all available tools into context. Currently this costs **~3,500 tokens (13,972 chars)** — a fixed tax on every single session regardless of what the user is doing.
The biggest offenders by parameter count:
| Tool | Approx. Share | Notes |
|------|-------------... | open | null | false | 6 | [
"enhancement"
] | [] | 2026-02-12T16:45:07Z | 2026-05-05T02:13:11Z | null | CONTRIBUTOR | null | 20260505T030055Z | 2026-05-05T03:00:55Z | seans-openclawbot | 259,728,142 | U_kgDOD3sjDg | User | false |
openclaw/openclaw | 4,237,451,504 | I_kwDOQb6kR878kmDw | 64,175 | https://github.com/openclaw/openclaw/issues/64175 | https://api.github.com/repos/openclaw/openclaw/issues/64175 | OpenRouter Perplexity Sonar Deep Research falls back in OpenClaw agent/runtime because the model is treated as tool-capable | lobster-biscuit
### Summary
`openrouter/perplexity/sonar-deep-research` is configured correctly and is available in the OpenClaw model catalog, but it does not run successfully through OpenClaw agent/runtime execution paths.
Instead, OpenClaw attempts to use it through a tool-capable execution surface and the model ... | open | null | false | 2 | [] | [] | 2026-04-10T07:39:31Z | 2026-05-03T12:31:08Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | Catfish-75 | 275,010,223 | U_kgDOEGRSrw | User | false |
openclaw/openclaw | 4,235,538,860 | I_kwDOQb6kR878dTGs | 63,974 | https://github.com/openclaw/openclaw/issues/63974 | https://api.github.com/repos/openclaw/openclaw/issues/63974 | [Bug]: Discord replies can become unreliable when a live session overflows and a deferred restart-required config change lands during active user wait | # [Bug]: Discord replies can become unreliable when a live session overflows and a deferred restart-required config change lands during active user wait
## Bug type
Behavior bug (incorrect output/state without crash)
## Beta release blocker
No
## Summary
During a live Discord conversation, replies became unreliab... | closed | completed | true | 3 | [] | [] | 2026-04-09T23:43:49Z | 2026-04-28T05:02:17Z | 2026-04-25T15:54:01Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | slideshow-dingo | 258,996,445 | U_kgDOD2_43Q | User | false |
openclaw/openclaw | 4,235,535,335 | I_kwDOQb6kR878dSPn | 63,973 | https://github.com/openclaw/openclaw/issues/63973 | https://api.github.com/repos/openclaw/openclaw/issues/63973 | Discord ACP thread-bound session can fail with thread_binding_invalid due to binding manager init / channel-id normalization issues | ## Summary
`Discord` 上通过 ACP 创建 **thread-bound persistent session** 时，`sessions_spawn(runtime="acp", agentId="codex", thread=true, mode="session")` 会稳定失败，报：
- `thread_binding_invalid`
- `Session binding adapter failed to bind target conversation`
我们在 `2026.4.8` 上复现并本地排查后，发现问题不只是权限或 Discord 配置，更像是 **Discord thread bi... | closed | completed | true | 4 | [] | [] | 2026-04-09T23:42:39Z | 2026-04-25T04:16:51Z | 2026-04-22T20:12:03Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | delinecnlin | 135,121,608 | U_kgDOCA3KyA | User | false |
openclaw/openclaw | 4,236,505,031 | I_kwDOQb6kR878g-_H | 64,071 | https://github.com/openclaw/openclaw/issues/64071 | https://api.github.com/repos/openclaw/openclaw/issues/64071 | Completed subagent sessions reconnect and re-announce after gateway restart | ## Description
After a gateway restart, all previously completed subagent sessions (mode="run") reconnect to the gateway and re-send their announce messages. This causes:
1. **72+ zombie WebSocket connections** from completed subagents that should have been cleaned up
2. **Stale announce messages** delivered to the p... | closed | completed | true | 2 | [] | [] | 2026-04-10T04:13:18Z | 2026-04-28T05:02:01Z | 2026-04-25T15:56:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Zhengyang-Liu | 14,222,421 | MDQ6VXNlcjE0MjIyNDIx | User | false |
openclaw/openclaw | 4,235,736,519 | I_kwDOQb6kR878eDXH | 63,991 | https://github.com/openclaw/openclaw/issues/63991 | https://api.github.com/repos/openclaw/openclaw/issues/63991 | acpx persistent sessions: sessionDir is null, prompt cannot find session | ## Bug
`acpx gemini` persistent sessions register in the index but cannot be resumed. The `sessionDir` field is always `null` in the session JSON, causing `prompt` to fail with "No acpx session found" even though `status` and `sessions` show the session exists.
## Steps to Reproduce
1. Create a persistent session:
... | closed | not_planned | true | 2 | [] | [] | 2026-04-10T00:45:02Z | 2026-04-28T05:02:15Z | 2026-04-25T15:54:14Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | DranboFieldston | 258,684,645 | U_kgDOD2s25Q | User | false |
openclaw/openclaw | 4,235,730,694 | I_kwDOQb6kR878eB8G | 63,990 | https://github.com/openclaw/openclaw/issues/63990 | https://api.github.com/repos/openclaw/openclaw/issues/63990 | Feature: Multi-index embedding memory with model-aware failover (no mixed vector spaces) | ## Summary
Introduce first-class multi-index embedding support in OpenClaw memory to allow resilient provider/model failover without corrupting vector semantics.
Today memory is effectively single-embedding-model per agent (e.g. `openai/text-embedding-3-small`). For production reliability, we need dual-index operation... | open | null | false | 5 | [] | [] | 2026-04-10T00:43:16Z | 2026-05-03T02:14:50Z | null | NONE | null | 20260503T030030Z | 2026-05-03T03:00:30Z | DIZ-admin | 188,769,789 | U_kgDOC0Bl_Q | User | false |
openclaw/openclaw | 4,235,826,124 | I_kwDOQb6kR878eZPM | 63,998 | https://github.com/openclaw/openclaw/issues/63998 | https://api.github.com/repos/openclaw/openclaw/issues/63998 | Session transcript doomloop: crash-restart cycle inflates transcript until gateway OOMs | ## Summary
Gateway enters an unrecoverable crash loop when a session transcript grows large enough to overflow context on reload. Each restart appends bootstrap entries, growing the transcript further, creating a death spiral.
## Reproduction
1. A session accumulates large payloads (e.g., 16× 197KB media attachments... | open | null | false | 3 | [] | [] | 2026-04-10T01:09:43Z | 2026-05-03T11:49:26Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | liu51115 | 31,350,941 | MDQ6VXNlcjMxMzUwOTQx | User | false |
openclaw/openclaw | 4,235,818,281 | I_kwDOQb6kR878eXUp | 63,997 | https://github.com/openclaw/openclaw/issues/63997 | https://api.github.com/repos/openclaw/openclaw/issues/63997 | [Bug]: | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
This was submitted by my bot who insists this is a bug. Discord is now broken. Here are the details:
"Your version: 4.9
The exact error: "discord missing register/activate export"
Note that the plugin has registerFull() bu... | closed | not_planned | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-10T01:07:38Z | 2026-04-28T05:02:13Z | 2026-04-25T15:54:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nicknero | 3,698,945 | MDQ6VXNlcjM2OTg5NDU= | User | false |
openclaw/openclaw | 4,235,800,013 | I_kwDOQb6kR878eS3N | 63,996 | https://github.com/openclaw/openclaw/issues/63996 | https://api.github.com/repos/openclaw/openclaw/issues/63996 | Circuit Breaker cooldown probe does not fire for 'timeout' reason — no auto-recovery after network outage | ## Summary
When the primary model provider becomes unreachable due to a network outage (e.g., VPN disconnect), OpenClaw correctly fails over to the fallback model. However, after the network recovers, the system **never automatically switches back** to the primary model. The user must manually restart the gateway or s... | closed | completed | true | 3 | [] | [] | 2026-04-10T01:02:34Z | 2026-04-28T05:02:14Z | 2026-04-25T15:54:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fuyizheng3120 | 268,386,865 | U_kgDOD_9CMQ | User | false |
openclaw/openclaw | 4,235,854,543 | I_kwDOQb6kR878egLP | 64,004 | https://github.com/openclaw/openclaw/issues/64004 | https://api.github.com/repos/openclaw/openclaw/issues/64004 | [Bug]: Control UI remains slow although `sessions.list` returns quickly | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The Control UI remains slow to update (about 40 seconds to load) after opening Sessions even when observed `sessions.list` requests complete in about 20ms to 70ms.
### Steps to reproduce
1. Open the Control U... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T01:16:28Z | 2026-05-05T04:27:54Z | 2026-05-02T09:33:38Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | WaMaSeDu | 265,785,875 | U_kgDOD9eSEw | User | false |
openclaw/openclaw | 4,235,837,767 | I_kwDOQb6kR878ecFH | 64,001 | https://github.com/openclaw/openclaw/issues/64001 | https://api.github.com/repos/openclaw/openclaw/issues/64001 | Bug: memory.qmd binary unavailable (ENOENT) even with absolute path in memory.qmd.command | ## Summary
On Linux with OpenClaw v2026.4.8, the QMD memory backend fails to spawn the `qmd` binary at gateway startup, logging:
```
[memory] qmd binary unavailable (/home/openclaw/.npm-global/bin/qmd); falling back to builtin: spawn /home/openclaw/.npm-global/bin/qmd ENOENT
```
This occurs **even when `memory.qmd.c... | closed | completed | true | 2 | [] | [] | 2026-04-10T01:12:31Z | 2026-04-30T04:42:36Z | 2026-04-26T04:43:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | fayenix | 273,238,055 | U_kgDOEElIJw | User | false |
openclaw/openclaw | 4,235,956,650 | I_kwDOQb6kR878e5Gq | 64,012 | https://github.com/openclaw/openclaw/issues/64012 | https://api.github.com/repos/openclaw/openclaw/issues/64012 | ComfyUI provider config rejected by schema validation (models.providers.comfy) | ## Description
The documented ComfyUI provider configuration under `models.providers.comfy` is rejected by the JSON schema validator at gateway startup. The docs at https://docs.openclaw.ai/providers/comfy show this config:
```json
{
"models": {
"providers": {
"comfy": {
"mode": "local",
"... | closed | completed | true | 2 | [] | [] | 2026-04-10T01:40:37Z | 2026-04-27T04:59:34Z | 2026-04-24T20:32:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | halburrows | 30,581,470 | MDQ6VXNlcjMwNTgxNDcw | User | false |
openclaw/openclaw | 4,235,931,183 | I_kwDOQb6kR878ey4v | 64,009 | https://github.com/openclaw/openclaw/issues/64009 | https://api.github.com/repos/openclaw/openclaw/issues/64009 | [Bug]: Slack socket-mode connection becomes stale, misses ping/pong, and restarts repeatedly | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The Slack socket-mode connection repeatedly misses ping/pong deadlines, enters a stale-socket state, restarts, and sometimes fails to send messages because no active connection is available.
### Steps to repro... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-10T01:34:43Z | 2026-04-28T05:02:12Z | 2026-04-25T15:54:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | WaMaSeDu | 265,785,875 | U_kgDOD9eSEw | User | false |
openclaw/openclaw | 4,236,418,948 | I_kwDOQb6kR878gp-E | 64,065 | https://github.com/openclaw/openclaw/issues/64065 | https://api.github.com/repos/openclaw/openclaw/issues/64065 | [Bug]: Three async interleaving races in cron, command-queue, and plugin-binding subsystems | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Formal concurrency interleaving analysis (predictive race detection via RAPID) identified three async races across `src/process/command-queue.ts`, `src/plugins/hook-runner-global.ts`, and `src/plugins/conversat... | open | null | false | 2 | [] | [] | 2026-04-10T03:53:08Z | 2026-05-02T18:31:20Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | lihaokun | 48,173,523 | MDQ6VXNlcjQ4MTczNTIz | User | false |
openclaw/openclaw | 4,236,294,988 | I_kwDOQb6kR878gLtM | 64,055 | https://github.com/openclaw/openclaw/issues/64055 | https://api.github.com/repos/openclaw/openclaw/issues/64055 | [Bug]: openclaw gateway restart on arm64 platform always failed by port 18789 is used | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
rk3568 chip arm64
ubuntu 24.04
use nvm install node 24
after modify config file
openclaw gateway restart will lost channel connection
check by 'openclaw doctor' , the port 18789 is used
### Steps to reproduce
modify confi... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-10T03:18:53Z | 2026-04-28T05:02:04Z | 2026-04-25T15:56:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ghmole | 7,156,846 | MDQ6VXNlcjcxNTY4NDY= | User | false |
openclaw/openclaw | 4,236,279,435 | I_kwDOQb6kR878gH6L | 64,047 | https://github.com/openclaw/openclaw/issues/64047 | https://api.github.com/repos/openclaw/openclaw/issues/64047 | [Bug] Feishu /new command delivered=false (2026.4.9, DM, 100% reproducible) | ## Bug Description
When triggering /new from Feishu DM, the command is detected but delivered=false — the message never reaches the agent and no reply is sent.
**Environment:**
- OpenClaw version: 2026.4.9
- Channel: Feishu (WebSocket mode)
- Agent: xiaohongshu (non-main agent)
- Feishu DM
## Steps to reproduce
1. ... | closed | completed | true | 4 | [] | [] | 2026-04-10T03:14:01Z | 2026-04-28T05:02:08Z | 2026-04-25T15:55:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tomidada1002-droid | 266,193,111 | U_kgDOD93I1w | User | false |
openclaw/openclaw | 4,236,016,269 | I_kwDOQb6kR878fHqN | 64,016 | https://github.com/openclaw/openclaw/issues/64016 | https://api.github.com/repos/openclaw/openclaw/issues/64016 | BUG: heartbeat fires duplicate runs when external wake events (openclaw agent CLI) arrive during scheduled heartbeat | ## Bug type
Behavior bug (duplicate/extra heartbeat runs beyond configured interval)
## Summary
When heartbeat is configured with isolatedSession and every=60m, extra heartbeat runs occur within the same hour window. These duplicate heartbeats are triggered by external wake events (e.g. openclaw agent CLI calls) tha... | closed | completed | true | 2 | [] | [] | 2026-04-10T01:53:39Z | 2026-05-05T04:27:08Z | 2026-05-02T13:52:19Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | foreverxdord | 44,799,061 | MDQ6VXNlcjQ0Nzk5MDYx | User | false |
openclaw/openclaw | 4,236,015,839 | I_kwDOQb6kR878fHjf | 64,015 | https://github.com/openclaw/openclaw/issues/64015 | https://api.github.com/repos/openclaw/openclaw/issues/64015 | doctor/status false positives on 2026.4.9: plugin allowlist, disabled memory-core, Discord runtime, SecretRef, and claude-cli drift | ## Summary
After updating to `OpenClaw 2026.4.9 (0512059)` and completing the local remediation pass, the gateway is healthy and the boot/heartbeat regressions are fixed, but the diagnostics surface still contains contradictions that appear upstream:
1. `openclaw doctor --deep` warns `plugins.allow is empty` even tho... | closed | not_planned | true | 2 | [] | [] | 2026-04-10T01:53:32Z | 2026-05-03T04:34:58Z | 2026-04-30T21:30:29Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | RotoDude93 | 236,109,455 | U_kgDODhK-jw | User | false |
openclaw/openclaw | 4,236,190,183 | I_kwDOQb6kR878fyHn | 64,034 | https://github.com/openclaw/openclaw/issues/64034 | https://api.github.com/repos/openclaw/openclaw/issues/64034 | [Bug]: 2026.4.9 multi-channel degradation on Linux: Discord direct-session overflow + slash lag + Telegram 'reading length' lane crashes | ## Summary
After upgrading to **OpenClaw 2026.4.9 (0512059)** on Linux, the assistant shows multi-channel degradation that matches user-visible symptoms across **Discord, Telegram, and webchat**:
- slash commands become noticeably lagged
- direct conversations degrade after a few turns and can effectively stop respon... | closed | not_planned | true | 2 | [] | [] | 2026-04-10T02:46:18Z | 2026-04-30T04:42:35Z | 2026-04-26T04:43:21Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | cygnostik | 5,272,292 | MDQ6VXNlcjUyNzIyOTI= | User | false |
openclaw/openclaw | 4,236,154,444 | I_kwDOQb6kR878fpZM | 64,030 | https://github.com/openclaw/openclaw/issues/64030 | https://api.github.com/repos/openclaw/openclaw/issues/64030 | cron: encode sessionKey when deriving jobId to support channel-native conversation IDs | ## Problem
`runtime.channel.routing.buildAgentSessionKey(...)` passes the channel-provided id segment through verbatim. For DingTalk, that segment is the platform's native `conversationId`, which routinely contains `/`. A real example:
```
agent:main:dingtalk:group:cid3tmd4xb19xjfk/wogxwy2a==
```
This sessionKey is ... | open | null | false | 3 | [] | [] | 2026-04-10T02:35:27Z | 2026-05-03T12:31:03Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | BrilliantWang | 120,538,413 | U_kgDOBy9FLQ | User | false |
openclaw/openclaw | 4,236,106,615 | I_kwDOQb6kR878fdt3 | 64,025 | https://github.com/openclaw/openclaw/issues/64025 | https://api.github.com/repos/openclaw/openclaw/issues/64025 | [Bug]: ClawHub CLI cannot search/inspect/install plugins — skills and plugins are disconnected | ## Problem
ClawHub website has a dedicated **Plugins** tab (`https://clawhub.ai/plugins`) with many published plugins, but the **ClawHub CLI (`clawhub` v0.9.0) cannot find any of them**.
## Steps to Reproduce
### 1. Plugin exists on the website
- Visit: `https://clawhub.ai/plugins/episodic-claw` → **200 OK**, page l... | closed | completed | true | 5 | [] | [] | 2026-04-10T02:19:59Z | 2026-05-01T04:45:03Z | 2026-04-28T06:38:17Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | downwind7clawd-ctrl | 258,506,560 | U_kgDOD2h_QA | User | false |
openclaw/openclaw | 4,236,093,397 | I_kwDOQb6kR878fafV | 64,024 | https://github.com/openclaw/openclaw/issues/64024 | https://api.github.com/repos/openclaw/openclaw/issues/64024 | Discord channel repeats previous response after LLM idle timeout | ## Summary
When the LLM fails to respond within the 60-second idle timeout, the Discord gateway re-processes the user's message and returns the **previous cached response** instead of generating a new one. This results in the bot repeating an unrelated answer to a completely different question.
## Steps to Reproduce
... | closed | completed | true | 3 | [] | [] | 2026-04-10T02:15:34Z | 2026-04-28T05:02:11Z | 2026-04-25T15:54:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yokmama | 312,475 | MDQ6VXNlcjMxMjQ3NQ== | User | false |
openclaw/openclaw | 4,236,300,269 | I_kwDOQb6kR878gM_t | 64,056 | https://github.com/openclaw/openclaw/issues/64056 | https://api.github.com/repos/openclaw/openclaw/issues/64056 | Telegram: subagent completion announce and exec approvals bypass message chunking | ## Summary
When a subagent completes and its result is announced back to the parent session on Telegram, the message is sent raw via `sendMessage` without going through the outbound chunking pipeline. This causes `400: Bad Request: message is too long` errors when the subagent response exceeds Telegram's 4096-characte... | closed | completed | true | 2 | [] | [] | 2026-04-10T03:20:24Z | 2026-04-28T05:02:03Z | 2026-04-25T15:56:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | xyqshi | 1,771,471 | MDQ6VXNlcjE3NzE0NzE= | User | false |
openclaw/openclaw | 4,236,285,338 | I_kwDOQb6kR878gJWa | 64,051 | https://github.com/openclaw/openclaw/issues/64051 | https://api.github.com/repos/openclaw/openclaw/issues/64051 | terminated errors can prematurely finalize chat UI during retry/fallback, causing ghost output | ## English
### Summary
When an underlying LLM request fails with a recoverable error such as `terminated` or `timeout`, OpenClaw may prematurely treat that failure as a terminal lifecycle error for the current chat run.
As a result, the frontend chat UI is finalized too early, while retry or model fallback may still ... | closed | completed | true | 2 | [] | [] | 2026-04-10T03:15:53Z | 2026-04-28T05:02:06Z | 2026-04-25T15:55:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ChenYongJie666 | 33,951,512 | MDQ6VXNlcjMzOTUxNTEy | User | false |
openclaw/openclaw | 4,236,219,610 | I_kwDOQb6kR878f5Ta | 64,038 | https://github.com/openclaw/openclaw/issues/64038 | https://api.github.com/repos/openclaw/openclaw/issues/64038 | Bug: Telegram sendMessage retried without idempotency key causes duplicate messages | ## Summary
Two related issues cause duplicate messages to be sent to users on Telegram:
**Bug 1 (confirmed): No idempotency key on sendMessage retries**
When a `sendMessage` call fails at the network level *after* Telegram has already delivered the message, OpenClaw retries the send without an idempotency key. This ... | closed | completed | true | 2 | [] | [] | 2026-04-10T02:55:42Z | 2026-04-28T05:02:09Z | 2026-04-25T15:55:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | BillBOT33 | 271,105,074 | U_kgDOECi8Mg | User | false |
openclaw/openclaw | 4,074,648,313 | I_kwDOQb6kR87y3jL5 | 45,740 | https://github.com/openclaw/openclaw/issues/45740 | https://api.github.com/repos/openclaw/openclaw/issues/45740 | gh-issues skill: untrusted issue body injected directly into sub-agent prompt | ## Problem
The `gh-issues` skill injects raw GitHub issue bodies and review comment bodies directly into sub-agent prompts without any sanitization or isolation.
**Locations:**
- `skills/gh-issues/SKILL.md` line 369: `Body: {body}` — issue body injected into fix sub-agent prompt
- `skills/gh-issues/SKILL.md` line 739... | open | null | false | 12 | [] | [] | 2026-03-14T04:33:03Z | 2026-05-04T00:12:59Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | zients | 53,077,736 | MDQ6VXNlcjUzMDc3NzM2 | User | false |
openclaw/openclaw | 4,206,122,551 | I_kwDOQb6kR876tFY3 | 61,103 | https://github.com/openclaw/openclaw/issues/61103 | https://api.github.com/repos/openclaw/openclaw/issues/61103 | [Bug]: Control UI screenshot/image upload broken after upgrade to 2026.4.2 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Bug Description
After upgrading to OpenClaw 2026.4.2, images sent via the Control UI (webchat) no longer reach the agent. The sender sees the image thumbnail in the chat, but the agent receives only the text portion — no... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-04-05T00:53:59Z | 2026-04-28T05:09:39Z | 2026-04-25T13:54:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | liangxiaoyan307-lab | 273,660,401 | U_kgDOEE-58Q | User | false |
openclaw/openclaw | 4,202,308,509 | I_kwDOQb6kR876eiOd | 60,550 | https://github.com/openclaw/openclaw/issues/60550 | https://api.github.com/repos/openclaw/openclaw/issues/60550 | Read tool path validation misses filePath and file aliases | ## Summary
The warning check for the `read` tool's path parameter in `pi-embedded-BYdcxQ5A.js` only validates against `path` and `file_path`, but `CLAUDE_PARAM_GROUPS` and `CLAUDE_PARAM_ALIASES` define four accepted aliases: `path`, `file_path`, `filePath`, and `file`.
This causes spurious `read tool called without p... | closed | not_planned | true | 2 | [] | [] | 2026-04-03T21:44:16Z | 2026-04-29T04:39:33Z | 2026-04-26T09:18:43Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | inire | 12,611,289 | MDQ6VXNlcjEyNjExMjg5 | User | false |
openclaw/openclaw | 4,202,002,288 | I_kwDOQb6kR876dXdw | 60,504 | https://github.com/openclaw/openclaw/issues/60504 | https://api.github.com/repos/openclaw/openclaw/issues/60504 | MC Skills UI: gateway path label shows wrong profile directory | ## Bug Report
**Summary:** In Mission Control Skills section, the gateway path label incorrectly shows ~/.openclaw/skills (gateway) even when running under a non-default profile.
**Steps to Reproduce:**
1. Run OpenClaw with a non-default profile (e.g. --profile new)
2. Open Mission Control and navigate to Skills
3. O... | closed | completed | true | 2 | [] | [] | 2026-04-03T19:58:33Z | 2026-04-28T05:10:55Z | 2026-04-25T13:39:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yannismatsis | 262,978,308 | U_kgDOD6y7BA | User | false |
openclaw/openclaw | 4,195,094,355 | I_kwDOQb6kR876DA9T | 59,737 | https://github.com/openclaw/openclaw/issues/59737 | https://api.github.com/repos/openclaw/openclaw/issues/59737 | GitHub: Create two PR templates (small fixes vs substantial changes) | ## Problem
The current PR template (`pull_request_template.md`) is comprehensive but overwhelming for small fixes. It has 30+ sections including:
- Summary (4 bullet points required)
- Change Type (5 checkboxes)
- Scope (8 checkboxes)
- Root Cause / Regression History
- Regression Test Plan
- User-visible Changes
- D... | open | null | false | 2 | [] | [] | 2026-04-02T14:44:53Z | 2026-05-02T07:49:35Z | null | NONE | null | 20260502T090030Z | 2026-05-02T09:00:30Z | dboone323 | 65,318,437 | MDQ6VXNlcjY1MzE4NDM3 | User | false |
openclaw/openclaw | 4,201,758,740 | I_kwDOQb6kR876ccAU | 60,479 | https://github.com/openclaw/openclaw/issues/60479 | https://api.github.com/repos/openclaw/openclaw/issues/60479 | Fix gemini-cli detection and project discovery issues | When using the npm-installed `@google/gemini-cli`, OpenClaw fails to detect it because it expects an `oauth2.js` file that is not present in the bundled version. Additionally, project discovery fails with a `400 Bad Request` due to a mismatch in metadata and headers compared to what `gemini-cli` uses. | closed | completed | true | 1 | [] | [
"vincentkoc"
] | 2026-04-03T18:44:37Z | 2026-04-14T14:49:26Z | 2026-04-05T07:41:44Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | wzfmini01 | 273,381,436 | U_kgDOEEt4PA | User | false |
openclaw/openclaw | 4,200,214,368 | I_kwDOQb6kR876Wi9g | 60,262 | https://github.com/openclaw/openclaw/issues/60262 | https://api.github.com/repos/openclaw/openclaw/issues/60262 | [Bug]: Cron jobs fire but don't wake agent sessions | # OpenClaw Bug Report: Cron Jobs Not Waking Sessions
**Date:** April 3, 2026
**OpenClaw Version:** 2026.4.1
**Environment:** macOS, Telegram channel
---
## Summary
Cron jobs are firing successfully (showing `status: "ok"` and `deliveryStatus: "not-requested"`) but **not creating active sessions** for the agent ... | closed | completed | true | 6 | [] | [] | 2026-04-03T11:48:39Z | 2026-04-26T04:27:52Z | 2026-04-23T06:15:44Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | dylanpassey-hub | 263,126,731 | U_kgDOD67-yw | User | false |
openclaw/openclaw | 4,009,062,767 | I_kwDOQb6kR87u9XFv | 31,231 | https://github.com/openclaw/openclaw/issues/31231 | https://api.github.com/repos/openclaw/openclaw/issues/31231 | Update 2026.2.26: Matrix plugin broken — missing dependency + duplicate plugin conflict | ## Summary
The `2026.2.26` update broke Matrix connectivity for users who had previously installed the Matrix plugin locally. Two separate issues compound each other.
---
## Issue 1: Bundled Matrix extension missing `@vector-im/matrix-bot-sdk`
The newly bundled `extensions/matrix` plugin requires `@vector-im/matrix... | closed | not_planned | true | 4 | [] | [] | 2026-03-02T03:27:08Z | 2026-04-14T14:49:17Z | 2026-03-19T13:29:43Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | nibblesnbits | 6,042,917 | MDQ6VXNlcjYwNDI5MTc= | User | false |
openclaw/openclaw | 3,953,877,224 | I_kwDOQb6kR87rq2Do | 19,312 | https://github.com/openclaw/openclaw/issues/19312 | https://api.github.com/repos/openclaw/openclaw/issues/19312 | [Bug] v2026.2.15 Bundled Plugin Ecosystem Failure - 81% Plugins Affected by workspace:* Dependencies | **Version:** 2026.2.15
**Environment:** Ubuntu LTS, Node.js
**Install Method:** npm install -g openclaw
**Discovered:** 2026-02-17
**Analysis Method:** Systematic sub-agent audit of all 36 bundled plugins
---
## Executive Summary
**CRITICAL:** Systematic build pipeline failure affects 81% of OpenClaw's bundl... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-02-17T17:28:31Z | 2026-04-14T14:49:17Z | 2026-03-22T04:08:57Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | derbronko | 2,217,509 | MDQ6VXNlcjIyMTc1MDk= | User | false |
openclaw/openclaw | 3,987,725,338 | I_kwDOQb6kR87tr9wa | 26,250 | https://github.com/openclaw/openclaw/issues/26250 | https://api.github.com/repos/openclaw/openclaw/issues/26250 | SSH transport security check rejects connection before tunnel is established | ### Summary
When configuring a node to connect to a remote gateway using transport: "ssh", the security check fires on the raw sshTarget IP address before the SSH tunnel is actually established. This means the connection is always rejected with a SECURITY ERROR, making the SSH transport mode unusable.
### Steps to re... | open | null | false | 4 | [
"bug",
"security"
] | [] | 2026-02-25T07:31:57Z | 2026-04-26T13:59:35Z | null | NONE | null | 20260426T150015Z | 2026-04-26T15:00:15Z | happynostrich | 153,582,260 | U_kgDOCSd6tA | User | false |
openclaw/openclaw | 3,943,819,934 | I_kwDOQb6kR87rEeqe | 17,069 | https://github.com/openclaw/openclaw/issues/17069 | https://api.github.com/repos/openclaw/openclaw/issues/17069 | loginctl enable-linger required for SSH-based OpenClaw installations | When running OpenClaw Gateway on a Raspberry Pi (or other remote machine) accessed via SSH, the gateway service dies every time the SSH session closes.
**Root cause:** systemd-logind's default `KillUserProcesses=yes` kills all user processes (including systemd user services) when a user logs out.
**Current workaround... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-02-15T11:06:15Z | 2026-04-14T14:49:15Z | 2026-03-27T04:15:03Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | axschr73 | 41,828,883 | MDQ6VXNlcjQxODI4ODgz | User | false |
openclaw/openclaw | 3,922,850,172 | I_kwDOQb6kR87p0fF8 | 13,552 | https://github.com/openclaw/openclaw/issues/13552 | https://api.github.com/repos/openclaw/openclaw/issues/13552 | [Bug]: macOS companion app fails to pass gateway auth token over SSH tunnel — no UI field, no auto-read from remote config | ## Summary
The macOS companion app cannot connect to a remote gateway over SSH tunnel when the gateway has `auth.mode: "token"` configured. The app establishes the SSH tunnel and discovers the gateway, but fails with:
```
Gateway error: gateway connect: connect to gateway @ ws://127.0.0.1:18789: unauthorized: gateway... | closed | completed | true | 16 | [
"bug"
] | [] | 2026-02-10T18:02:19Z | 2026-04-14T14:49:15Z | 2026-03-08T19:28:18Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | wgj | 463,564 | MDQ6VXNlcjQ2MzU2NA== | User | false |
openclaw/openclaw | 4,014,348,359 | I_kwDOQb6kR87vRhhH | 32,439 | https://github.com/openclaw/openclaw/issues/32439 | https://api.github.com/repos/openclaw/openclaw/issues/32439 | [Bug]: Can't install on my mac. | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
i just run the install command, it install faild.
and this is my installation environment:
· Active Node.js: v24.14.0 (/Users/mine/.nvm/versions/node/v24.14.0/bin/node)
· Active npm: 11.9.0 (/Users/mine/.nvm/versions/node/v24.14.0/bin/npm)... | closed | completed | true | 7 | [
"bug",
"bug:behavior"
] | [] | 2026-03-03T02:36:15Z | 2026-04-14T14:49:07Z | 2026-03-03T03:42:40Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Minessential | 34,987,057 | MDQ6VXNlcjM0OTg3MDU3 | User | false |
openclaw/openclaw | 4,058,652,440 | I_kwDOQb6kR87x6h8Y | 43,260 | https://github.com/openclaw/openclaw/issues/43260 | https://api.github.com/repos/openclaw/openclaw/issues/43260 | Support `model` field in SKILL.md frontmatter for per-skill model routing | ## Problem
Model selection is available at the agent level (`agents.list[].model.primary`) and default level (`agents.defaults.model.primary`), but not at the skill level. This means all skills invoked by an agent run on the same model, regardless of task complexity.
In practice, agents often run on a cost-efficient ... | open | null | false | 8 | [] | [] | 2026-03-11T14:41:10Z | 2026-05-04T00:42:56Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | AlethiaQuizForge | 254,334,588 | U_kgDODyjWfA | User | false |
openclaw/openclaw | 4,156,587,559 | I_kwDOQb6kR873wH4n | 56,086 | https://github.com/openclaw/openclaw/issues/56086 | https://api.github.com/repos/openclaw/openclaw/issues/56086 | Discord health-monitor: stale-socket triggers full gateway restart instead of surgical reconnect | ## Description
The health-monitor's `stale-socket` detection for the Discord provider triggers a **full gateway restart** instead of reconnecting only the Discord WebSocket. This causes significant downtime for users with active Discord integrations.
## Impact
- **22+ full gateway restarts per day** on our setup
- 1... | closed | completed | true | 5 | [] | [] | 2026-03-28T00:08:53Z | 2026-04-28T05:22:28Z | 2026-04-25T11:25:09Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sahan-silva | 16,109,981 | MDQ6VXNlcjE2MTA5OTgx | User | false |
openclaw/openclaw | 4,049,244,109 | I_kwDOQb6kR87xWo_N | 41,716 | https://github.com/openclaw/openclaw/issues/41716 | https://api.github.com/repos/openclaw/openclaw/issues/41716 | [Feature]: nodes tool + node command surface: add file_push/file_pull (gateway ↔ node file transfer) | ### Summary
Add bidirectional file transfer between gateway and paired nodes via new `file.read` / `file.write` node commands and corresponding `file_push` / `file_pull` actions on the `nodes` agent tool.
### Problem to solve
`system.run` lets agents execute commands on remote node hosts, but agent workflows routine... | closed | completed | true | 6 | [] | [] | 2026-03-10T04:01:10Z | 2026-05-04T04:35:43Z | 2026-05-01T20:11:24Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | chaosreload | 6,723,037 | MDQ6VXNlcjY3MjMwMzc= | User | false |
openclaw/openclaw | 4,210,049,383 | I_kwDOQb6kR8768EFn | 61,724 | https://github.com/openclaw/openclaw/issues/61724 | https://api.github.com/repos/openclaw/openclaw/issues/61724 | sessions_spawn(runtime="subagent") fails with "streamTo is only supported for runtime=acp" | ## Bug Description
\`sessions_spawn\` with \`runtime: "subagent"\` (the default runtime) immediately throws:
```
streamTo is only supported for runtime=acp; got runtime=subagent
```
This happens even when \`streamTo\` is **not** explicitly passed in the call. An internal ACP function \`implicitStreamToParent\` auto-i... | closed | completed | true | 7 | [
"clawsweeper"
] | [] | 2026-04-06T06:39:01Z | 2026-04-30T04:33:41Z | 2026-04-27T16:43:17Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | flyawayzack | 79,414,127 | MDQ6VXNlcjc5NDE0MTI3 | User | false |
openclaw/openclaw | 4,207,827,273 | I_kwDOQb6kR876zllJ | 61,390 | https://github.com/openclaw/openclaw/issues/61390 | https://api.github.com/repos/openclaw/openclaw/issues/61390 | cli-backend: "No conversation found" not detected as session_expired — stale session retry loop | ## Summary
`isCliSessionExpiredErrorMessage()` in `pi-embedded-helpers` does not match Claude CLI's `"No conversation found with session ID: ..."` error. The existing recovery logic (clear stale session + retry fresh) never triggers, causing an infinite retry loop with the same dead session ID across all fallback mode... | closed | not_planned | true | 4 | [] | [] | 2026-04-05T15:19:56Z | 2026-04-28T04:33:38Z | 2026-04-26T04:27:25Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | hoh-dev-bot | 219,382,756 | U_kgDODROD5A | User | false |
openclaw/openclaw | 4,208,009,436 | I_kwDOQb6kR8760SDc | 61,421 | https://github.com/openclaw/openclaw/issues/61421 | https://api.github.com/repos/openclaw/openclaw/issues/61421 | Docs mention Dreaming config/commands, but 2026.4.2 local build does not expose them | ## Summary
The public OpenClaw docs describe experimental Dreaming support for `memory-core`, including config at `plugins.entries.memory-core.config.dreaming.mode` and CLI/chat surfaces like `openclaw memory promote` / `/dreaming`.
However, on a local install of OpenClaw `2026.4.2`, the installed CLI/docs/plugin sch... | closed | completed | true | 9 | [] | [] | 2026-04-05T16:15:03Z | 2026-04-28T05:08:52Z | 2026-04-25T14:04:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | henry-at-weddi | 273,816,125 | U_kgDOEFIaPQ | User | false |
openclaw/openclaw | 4,075,464,813 | I_kwDOQb6kR87y6qht | 46,002 | https://github.com/openclaw/openclaw/issues/46002 | https://api.github.com/repos/openclaw/openclaw/issues/46002 | [Bug]: Feishu blockStreaming does not deliver intermediate text before tool calls | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When using the Feishu channel, text written by the assistant before tool calls in the same response is not delivered to the user until after all tool calls complete. Intermediate text between tool invocations never reaches the user.
### St... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-03-14T10:41:56Z | 2026-04-29T04:38:39Z | 2026-04-26T09:44:50Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | abstract-sum | 121,620,652 | U_kgDOBz_IrA | User | false |
openclaw/openclaw | 4,197,313,055 | I_kwDOQb6kR876Leof | 59,933 | https://github.com/openclaw/openclaw/issues/59933 | https://api.github.com/repos/openclaw/openclaw/issues/59933 | [Bug]: sessions_list does not surface group sessions for the same agent | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`sessions_list` returns only DM sessions even when active group sessions exist with messages and responses. This causes agents to believe group messages are not being received, leading to misdiagnosis during setup.
This is distinct from #5... | open | null | false | 2 | [] | [] | 2026-04-02T22:38:03Z | 2026-05-02T08:59:58Z | null | NONE | null | 20260502T090030Z | 2026-05-02T09:00:30Z | fattchris | 2,224,082 | MDQ6VXNlcjIyMjQwODI= | User | false |
openclaw/openclaw | 4,135,417,959 | I_kwDOQb6kR872fXhn | 54,484 | https://github.com/openclaw/openclaw/issues/54484 | https://api.github.com/repos/openclaw/openclaw/issues/54484 | Telegram inbound image messages should expose local media paths | ## Problem
Telegram inbound image messages are still surfaced to the agent as a generic `[Image]` placeholder, even when the media has already been downloaded to a local file path.
Example observed path:
`/home/ubuntu/.openclaw/media/inbound/file_372---ab72cd4a-5f5c-46c7-9909-43e1c30675cd.jpg`
## Expected behavior
Wh... | closed | completed | true | 2 | [] | [] | 2026-03-25T13:24:35Z | 2026-04-28T04:34:54Z | 2026-04-26T04:13:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | bulutarkan | 199,141,809 | U_kgDOC96psQ | User | false |
openclaw/openclaw | 4,136,018,826 | I_kwDOQb6kR872hqOK | 54,519 | https://github.com/openclaw/openclaw/issues/54519 | https://api.github.com/repos/openclaw/openclaw/issues/54519 | [Feature]: read-only channel mode (inbound only, no outbound) | ### Summary
Per-account read-only channel mode — block all outbound message delivery at the gateway level while still receiving inbound messages.
### Problem to solve
There's no config option to make a channel receive-only. When a channel is connected, the agent can both receive and send. Boundary rules in system pr... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-03-25T14:48:53Z | 2026-05-04T21:41:52Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | liu51115 | 31,350,941 | MDQ6VXNlcjMxMzUwOTQx | User | false |
openclaw/openclaw | 4,060,463,710 | I_kwDOQb6kR87yBcJe | 43,447 | https://github.com/openclaw/openclaw/issues/43447 | https://api.github.com/repos/openclaw/openclaw/issues/43447 | Kimi/Moonshot 'Rate Limit' error masks insufficient funds, causes UI lockout | # Bug Report: Kimi/Moonshot "Rate Limit" Error Masking Insufficient Funds
## Environment
- **OpenClaw Version:** 2026.3.8
- **OS:** macOS
- **Plugin:** kimi-claw (auto-loaded)
- **Provider:** Moonshot (Kimi)
## Problem Summary
When Kimi/Moonshot API account balance reaches zero, the API returns a **429 Rate Limit Rea... | open | null | false | 5 | [] | [] | 2026-03-11T20:20:04Z | 2026-05-04T12:05:06Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | superstizzle | 210,292,019 | U_kgDODIjNMw | User | false |
openclaw/openclaw | 4,008,360,820 | I_kwDOQb6kR87u6rt0 | 30,998 | https://github.com/openclaw/openclaw/issues/30998 | https://api.github.com/repos/openclaw/openclaw/issues/30998 | Feature: Tool output compression middleware (context optimization) | ## Problem
Every tool call dumps raw output into the context window. Large outputs (exec logs, browser snapshots, file reads, web fetches) consume context quickly, reducing effective session length.
Projects like [claude-context-mode](https://github.com/mksglu/claude-context-mode) demonstrate that 98% context reducti... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-03-01T21:51:11Z | 2026-04-27T05:03:38Z | 2026-04-24T19:10:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | decomontenegro | 15,161,416 | MDQ6VXNlcjE1MTYxNDE2 | User | false |
openclaw/openclaw | 4,062,606,421 | I_kwDOQb6kR87yJnRV | 43,767 | https://github.com/openclaw/openclaw/issues/43767 | https://api.github.com/repos/openclaw/openclaw/issues/43767 | [Bug]: Heartbeat ignores lightContext: true, loads full agent context + unbounded session history | ### Bug type
Regression (worked before, now fails)
### Summary
# Heartbeat ignores lightContext: true, loads full agent context + unbounded session history
### Steps to reproduce
## Bug Description
The heartbeat feature loads the **full agent context and accumulated conversation history** on every heartbeat tic... | closed | completed | true | 6 | [
"bug",
"regression"
] | [] | 2026-03-12T06:53:03Z | 2026-04-27T04:50:16Z | 2026-04-24T22:22:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | dwdupont | 43,862,194 | MDQ6VXNlcjQzODYyMTk0 | User | false |
openclaw/openclaw | 4,163,713,149 | I_kwDOQb6kR874LTh9 | 56,941 | https://github.com/openclaw/openclaw/issues/56941 | https://api.github.com/repos/openclaw/openclaw/issues/56941 | [Bug]: Heartbeat isolatedSession: true does not create fresh sessions | ## `isolatedSession: true` does not actually isolate heartbeat sessions
Related: #51542 (locked)
### Environment
- OpenClaw **2026.3.28** (f9b1079)
- Linux 6.17.0-1009-aws (x64), Node 24.14.0
### Config
```json
{
"heartbeat": {
"every": "30m",
"target": "last",
"directPolicy": "allow",
"isolatedS... | closed | completed | true | 2 | [] | [] | 2026-03-29T10:29:28Z | 2026-04-30T04:51:43Z | 2026-04-25T05:17:13Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | yubingjiaocn | 9,165,347 | MDQ6VXNlcjkxNjUzNDc= | User | false |
openclaw/openclaw | 4,041,411,111 | I_kwDOQb6kR87w4won | 39,945 | https://github.com/openclaw/openclaw/issues/39945 | https://api.github.com/repos/openclaw/openclaw/issues/39945 | Feature: Progressive Disclosure for Skills (lazy-loading) | ## Problem
Currently, all installed skills are injected into the system prompt at init time. With 83+ skills installed, this causes truncation (currently 41 of 83 skills are visible in the prompt) and wastes context tokens on skills that may never be triggered in a given session.
## Proposed Solution: Progressive Dis... | closed | completed | true | 3 | [] | [] | 2026-03-08T14:34:21Z | 2026-04-27T04:58:53Z | 2026-04-24T20:37:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | EtroxTaran | 24,548,577 | MDQ6VXNlcjI0NTQ4NTc3 | User | false |
openclaw/openclaw | 4,070,945,327 | I_kwDOQb6kR87ypbIv | 45,096 | https://github.com/openclaw/openclaw/issues/45096 | https://api.github.com/repos/openclaw/openclaw/issues/45096 | [Bug]: v3.12 diagnostics-otel plugin only reports system metrics, no LLM call traces/spans (input/output missing) to langfuse | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
### Environment
- OpenClaw version: v3.12 (2026-03-12 latest release)
- Deployment: Docker container
- OTel backend: Self-hosted Langfuse v3.155.1 OSS
- OTel protocol: http/protobuf
### Problem Description
The `diagnostics-otel` plugin onl... | closed | completed | true | 8 | [
"bug",
"bug:behavior"
] | [
"vincentkoc"
] | 2026-03-13T12:51:55Z | 2026-04-28T04:32:35Z | 2026-04-25T20:00:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zhangyongjun10 | 125,647,109 | U_kgDOB305BQ | User | false |
openclaw/openclaw | 4,023,129,569 | I_kwDOQb6kR87vzBXh | 34,749 | https://github.com/openclaw/openclaw/issues/34749 | https://api.github.com/repos/openclaw/openclaw/issues/34749 | BlueBubbles: image attachments blocked by SSRF guard (localhost URL) | ## Problem
BlueBubbles runs locally and serves attachment downloads at `http://127.0.0.1:1234/api/v1/attachment/*/download`. The gateway's SSRF guard blocks these as private/internal IPs, so **all inbound images from iMessage are silently dropped**.
## Evidence
```
[security] blocked URL fetch (url-fetch) target=http:... | closed | completed | true | 2 | [
"stale"
] | [] | 2026-03-04T17:14:03Z | 2026-04-22T04:30:55Z | 2026-04-19T15:43:33Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | auxclawdbot | 262,904,753 | U_kgDOD6ubsQ | User | false |
openclaw/openclaw | 4,023,125,924 | I_kwDOQb6kR87vzAek | 34,748 | https://github.com/openclaw/openclaw/issues/34748 | https://api.github.com/repos/openclaw/openclaw/issues/34748 | Optimization audit: performance, reliability, maintainability, security, and UX improvements | ## Summary
Comprehensive optimization audit across:
1) Performance 2) Readability/Maintainability 3) Security 4) Error handling/Logging 5) Test coverage 6) Documentation 7) Dependency management 8) Configuration management 9) Architecture 10) UX.
## Highest-priority findings
- P0 Sync I/O on request path:
- src/gate... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T17:13:12Z | 2026-04-23T04:23:15Z | 2026-04-20T04:29:44Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | evan966890 | 236,738,035 | U_kgDODhxV8w | User | false |
openclaw/openclaw | 4,023,117,305 | I_kwDOQb6kR87vy-X5 | 34,744 | https://github.com/openclaw/openclaw/issues/34744 | https://api.github.com/repos/openclaw/openclaw/issues/34744 | Cron job scheduled for 9:10 AM ET fired 2.5 hours late | # OpenClaw Cron Job Timing Bug Report
**Summary:** Cron job scheduled for 9:10 AM ET fired at 11:45 AM ET (2.5 hours late). A second job scheduled for 12:00 PM ET fired on time.
## Job Details
### Late Job (MacBook Reminder)
- **Scheduled:** 2026-03-04T09:10:00-05:00 (9:10 AM ET)
- **Actual execution:** ~11:45 AM ET... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-04T17:11:14Z | 2026-04-23T04:23:16Z | 2026-04-20T04:29:43Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | rowbotik | 4,590,280 | MDQ6VXNlcjQ1OTAyODA= | User | false |
openclaw/openclaw | 4,022,994,066 | I_kwDOQb6kR87vygSS | 34,710 | https://github.com/openclaw/openclaw/issues/34710 | https://api.github.com/repos/openclaw/openclaw/issues/34710 | Memory sync fails with "attempt to write a readonly database" | ## Bug Report
**Error:**
```
memory sync failed (session-delta): Error: attempt to write a readonly database
```
**Frequency:** Multiple times per hour
**Impact:** Memory persistence may fail
## Logs
- 16:01:04 - memory sync failed (watch)
- 16:19:56 - memory sync failed (session-delta)
- 16:34:58 - memory sync fail... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T16:46:09Z | 2026-04-23T04:23:17Z | 2026-04-20T04:29:37Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | oggixx | 247,152,204 | U_kgDODrs-TA | User | false |
openclaw/openclaw | 4,022,925,116 | I_kwDOQb6kR87vyPc8 | 34,693 | https://github.com/openclaw/openclaw/issues/34693 | https://api.github.com/repos/openclaw/openclaw/issues/34693 | Feature: Customizable Slack Assistant status messages | ## Summary
Add config options to customize the Slack Assistant API status messages (the "Finding answers..." text that appears while the bot is thinking).
## Background
Slack's Agent/Assistant framework uses `assistant.threads.setStatus` to show status while processing. Currently Clawdbot uses a hardcoded default. Use... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T16:32:25Z | 2026-04-23T04:23:17Z | 2026-04-20T04:29:33Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | pbc-os | 259,675,171 | U_kgDOD3pUIw | User | false |
openclaw/openclaw | 4,022,894,921 | I_kwDOQb6kR87vyIFJ | 34,683 | https://github.com/openclaw/openclaw/issues/34683 | https://api.github.com/repos/openclaw/openclaw/issues/34683 | feat: expose staleEventThresholdMs in config (gateway.healthMonitor.timing) | ## Problem
The gateway health monitor's `DEFAULT_STALE_EVENT_THRESHOLD_MS` is hardcoded at 30 minutes. During quiet periods (e.g., nighttime), channels with no inbound messages get flagged as `stale-socket` and restarted unnecessarily.
On my setup, this caused **152 restarts in one day** on Telegram — all false posit... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T16:26:47Z | 2026-04-23T04:23:18Z | 2026-04-20T04:29:31Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | Timememo8210 | 125,830,046 | U_kgDOB4ADng | User | false |
openclaw/openclaw | 4,022,891,601 | I_kwDOQb6kR87vyHRR | 34,681 | https://github.com/openclaw/openclaw/issues/34681 | https://api.github.com/repos/openclaw/openclaw/issues/34681 | macOS app crashes on WebSocket ping when connection drops simultaneously | **Version:** 2026.3.2 (2026030290)
**OS:** macOS 26.3 (25D125), ARM64
## Crash summary
Thread 9 (`com.apple.NSURLSession-delegate`) crashes with `EXC_BREAKPOINT` / `_assertionFailure` in `WebSocketTaskBox.sendPing()`.
## Crashing stack trace
```
0 libswiftCore.dylib _assertionFailure(_:_:file:line:flags... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T16:26:08Z | 2026-04-23T04:23:19Z | 2026-04-20T04:29:29Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | CrackerFormula | 55,393,829 | MDQ6VXNlcjU1MzkzODI5 | User | false |
openclaw/openclaw | 4,022,863,780 | I_kwDOQb6kR87vyAek | 34,665 | https://github.com/openclaw/openclaw/issues/34665 | https://api.github.com/repos/openclaw/openclaw/issues/34665 | [Bug] Feishu: Agent replies fail with Invalid user_id (discord ID used instead of feishu ou_ ID) | ## Bug Description
When using Feishu channel, agent successfully generates replies but fails to send them back to Feishu. Error: `Invalid ids: [discord:1089077850088415312]` - system incorrectly uses Discord user ID instead of Feishu user ID (ou_xxx) for message delivery.
## Steps to Reproduce
1. Configure OpenClaw w... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T16:20:28Z | 2026-04-23T04:23:20Z | 2026-04-20T04:29:25Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | michaelxmchn | 124,868,816 | U_kgDOB3FY0A | User | false |
openclaw/openclaw | 4,022,827,313 | I_kwDOQb6kR87vx3kx | 34,661 | https://github.com/openclaw/openclaw/issues/34661 | https://api.github.com/repos/openclaw/openclaw/issues/34661 | Channel delivery only includes last text segment after tool calls, discarding pre-tool-call content | ## Description
When an agent response contains text before and after tool calls, channel delivery (e.g. Telegram) only delivers the **last text segment** after the final tool call. Text written before tool calls is silently discarded. The TUI correctly shows the full transcript including all text segments.
## Reprodu... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T16:13:06Z | 2026-04-23T04:23:20Z | 2026-04-20T04:29:23Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | adamzelene | 83,459,429 | MDQ6VXNlcjgzNDU5NDI5 | User | false |
openclaw/openclaw | 4,022,825,644 | I_kwDOQb6kR87vx3Ks | 34,659 | https://github.com/openclaw/openclaw/issues/34659 | https://api.github.com/repos/openclaw/openclaw/issues/34659 | [Bug]: openclaw tui doesn't show reply directly | ### Bug type
Regression (worked before, now fails)
### Summary
After the upgrade to 3.2
I open openclaw tui and send the message, but the answer doesn't show directly.
I need to ctrl+c exit the tui and reopen it, the answer will show.
### Steps to reproduce
After the upgrade to 3.2
I open openclaw tui and send the... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T16:12:44Z | 2026-04-23T04:23:21Z | 2026-04-20T04:29:22Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | prophetyy18 | 39,017,272 | MDQ6VXNlcjM5MDE3Mjcy | User | false |
openclaw/openclaw | 4,022,823,545 | I_kwDOQb6kR87vx2p5 | 34,657 | https://github.com/openclaw/openclaw/issues/34657 | https://api.github.com/repos/openclaw/openclaw/issues/34657 | Canvas: capability URLs not generated for nodes connecting through reverse proxy | ## Problem
When the OpenClaw gateway runs behind a reverse proxy (e.g., Traefik) in a Docker setup, the Canvas panel on macOS nodes fails to authenticate with the gateway's canvas/A2UI HTTP endpoints.
The Mac app auto-navigates to `https://<domain>/__openclaw__/a2ui/?platform=macos` on Canvas open, but the request ar... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T16:12:20Z | 2026-04-23T04:23:22Z | 2026-04-20T04:29:20Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | jaylane | 1,765,190 | MDQ6VXNlcjE3NjUxOTA= | User | false |
openclaw/openclaw | 4,022,822,750 | I_kwDOQb6kR87vx2de | 34,656 | https://github.com/openclaw/openclaw/issues/34656 | https://api.github.com/repos/openclaw/openclaw/issues/34656 | [Bug]: HTTP 401 authentication_error: invalid api key on every message send (dashboard & Telegram) after config change / gateway restart | ### Bug type
Regression (worked before, now fails)
### Summary
After updating API keys in models.json and restarting the gateway, both the dashboard and Telegram bot start throwing HTTP 401 authentication_error: invalid api key every time I send any message.
Dashboard: Page loads normally after refresh, but sending... | closed | not_planned | true | 8 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T16:12:12Z | 2026-04-23T04:23:22Z | 2026-04-20T04:29:18Z | NONE | null | 20260423T060020Z | 2026-04-23T06:00:20Z | Zacisnewuser | 259,759,796 | U_kgDOD3uetA | User | false |
openclaw/openclaw | 4,022,810,158 | I_kwDOQb6kR87vxzYu | 34,654 | https://github.com/openclaw/openclaw/issues/34654 | https://api.github.com/repos/openclaw/openclaw/issues/34654 | Auth profile order filter uses wrong field name for api_key credentials | ## Bug Description
`resolveAuthProfileOrder()` in `model-auth` filters out `api_key` type profiles using `cred.key`, but profiles persisted via onboarding/CLI store the API key in `cred.apiKey`. This causes API key profiles to silently drop from the auth order, breaking model failover.
## Reproduction
1. Configure a... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T16:09:43Z | 2026-04-22T04:29:29Z | 2026-04-20T04:29:16Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | hamhoagie | 467,533 | MDQ6VXNlcjQ2NzUzMw== | User | false |
openclaw/openclaw | 4,022,803,592 | I_kwDOQb6kR87vxxyI | 34,652 | https://github.com/openclaw/openclaw/issues/34652 | https://api.github.com/repos/openclaw/openclaw/issues/34652 | [Bug]: Cron定时任务不执行 - nextWakeAtMs始终为null | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
## 问题描述
配置定时任务后，cron服务的nextWakeAtMs始终为null，定时任务无法自动执行。
## 日志
- heartbeat已启动: 2026-03-04T15:44:03.899Z [heartbeat] started
- 但cron服务的nextWakeAtMs始终为null
## 已尝试
1. 重启Gateway - 无效
2. 发送wake事件 - 无效
## 预期
定时任务应该按照配置的时间自动执行
## 实际
任务never触发
#... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-04T16:08:21Z | 2026-04-22T04:29:29Z | 2026-04-20T04:29:14Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Openclaw-projectdavid | 265,304,618 | U_kgDOD9A6Kg | User | false |
openclaw/openclaw | 4,022,802,864 | I_kwDOQb6kR87vxxmw | 34,651 | https://github.com/openclaw/openclaw/issues/34651 | https://api.github.com/repos/openclaw/openclaw/issues/34651 | [Bug] o3/reasoning models fail with Codex OAuth due to auto-injected reasoning.summary requiring org verification | ## Problem
Using OpenClaw 2026.3.2 with Codex OAuth (ChatGPT Pro subscription), attempting to use o3 or gpt-5.2 with `reasoning.effort` fails with:
```
Error: "Your organization must be verified to generate reasoning summaries"
Error param: reasoning.summary
Error code: unsupported_value
```
Pro plan users cannot ve... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T16:08:11Z | 2026-04-22T04:29:30Z | 2026-04-20T04:29:12Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | alicebotalex | 260,494,797 | U_kgDOD4bVzQ | User | false |
openclaw/openclaw | 4,022,745,984 | I_kwDOQb6kR87vxjuA | 34,645 | https://github.com/openclaw/openclaw/issues/34645 | https://api.github.com/repos/openclaw/openclaw/issues/34645 | Feature: per-phase model routing (different models for conversation vs tool execution) | ## Feature Request: Per-tool or per-phase model routing
### Motivation
We use Kimi K2.5 as our primary model because it excels at understanding user intent, conversation, and task routing. However, it has a critical tool-call loop bug ([#34574](https://github.com/openclaw/openclaw/issues/34574), [MoonshotAI/Kimi-K2.5... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T15:56:45Z | 2026-04-22T04:29:31Z | 2026-04-20T04:29:10Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | gucasbrg | 3,182,658 | MDQ6VXNlcjMxODI2NTg= | User | false |
openclaw/openclaw | 4,022,736,728 | I_kwDOQb6kR87vxhdY | 34,642 | https://github.com/openclaw/openclaw/issues/34642 | https://api.github.com/repos/openclaw/openclaw/issues/34642 | sessions.patch rejects unique provider-scoped model shorthand (e.g. nvidia/kimi) | ## Summary
When a user patches session model with a provider-scoped shorthand like `nvidia/kimi`, gateway may reject it as `model not allowed` even when there is exactly one allowlisted model under that provider (for example `nvidia/moonshotai/kimi-k2.5`).
This creates avoidable failures in session model switching path... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T15:54:54Z | 2026-04-22T04:29:31Z | 2026-04-20T04:29:09Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | AytuncYildizli | 47,717,026 | MDQ6VXNlcjQ3NzE3MDI2 | User | false |
openclaw/openclaw | 4,022,706,072 | I_kwDOQb6kR87vxZ-Y | 34,636 | https://github.com/openclaw/openclaw/issues/34636 | https://api.github.com/repos/openclaw/openclaw/issues/34636 | openclaw doctor: False positives on session files flagged as "orphaned" and "stale" | **Environment:**
- OpenClaw: (version at time of filing)
- Node: v25.5.0 (NVM-managed)
- Session count: 45 active session files
- Cleanup status: 10 `.deleted` markers (cleanup completed 2026-03-03)
**Issue:**
`openclaw doctor` reports "orphaned transcript files" and "stale session lock" warnings when the system state... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T15:49:22Z | 2026-04-22T04:29:32Z | 2026-04-20T04:29:07Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | subimage | 12,508 | MDQ6VXNlcjEyNTA4 | User | false |
openclaw/openclaw | 4,022,697,666 | I_kwDOQb6kR87vxX7C | 34,633 | https://github.com/openclaw/openclaw/issues/34633 | https://api.github.com/repos/openclaw/openclaw/issues/34633 | Gateway --bind lan only listens on IPv4, breaks Railway/IPv6 proxies | ## Summary
When using `--bind lan`, the gateway binds to `0.0.0.0` (IPv4 only). Railway's proxy connects over IPv6, resulting in a 502 "Application failed to respond" error.
## Steps to Reproduce
1. Deploy OpenClaw to Railway using the Dockerfile
2. Use `railway.toml` with `startCommand = "node openclaw.mjs gateway ... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T15:48:01Z | 2026-04-22T04:29:33Z | 2026-04-20T04:29:05Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | brandoncollins7 | 36,413,482 | MDQ6VXNlcjM2NDEzNDgy | User | false |
openclaw/openclaw | 4,022,691,997 | I_kwDOQb6kR87vxWid | 34,632 | https://github.com/openclaw/openclaw/issues/34632 | https://api.github.com/repos/openclaw/openclaw/issues/34632 | model_context_window_exceeded stop reason not handled — auto-compaction never triggers | ## Bug
When using OpenAI-compatible providers (e.g. ZhipuAI/glm-5), the model can return `finish_reason: model_context_window_exceeded` when the context window is exceeded.
Openclaw (via `@mariozechner/pi-ai`) does not recognize this stop reason:
- `mapStopReason()` in `openai-completions.js` throws `Error("Unhandled... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T15:46:56Z | 2026-04-22T04:29:34Z | 2026-04-20T04:29:04Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | aliyaredpilled | 148,153,466 | U_kgDOCNSkeg | User | false |
openclaw/openclaw | 4,022,683,666 | I_kwDOQb6kR87vxUgS | 34,631 | https://github.com/openclaw/openclaw/issues/34631 | https://api.github.com/repos/openclaw/openclaw/issues/34631 | [Bug]: Title: LINE webhook route intermittently returns 404 until gateway restart (2026.3.2) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Title: LINE webhook route intermittently returns 404 until gateway restart (2026.3.2)
Summary: On macOS LaunchAgent gateway (loopback:18789), LINE channel shows running/configured, but GET /line/webhook intermittently returns 404. LINE web... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-04T15:45:23Z | 2026-04-22T04:29:34Z | 2026-04-20T04:29:02Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | a0983186600-a11y | 261,878,009 | U_kgDOD5vw-Q | User | false |
openclaw/openclaw | 4,022,677,172 | I_kwDOQb6kR87vxS60 | 34,630 | https://github.com/openclaw/openclaw/issues/34630 | https://api.github.com/repos/openclaw/openclaw/issues/34630 | [Bug] macOS Talk Mode toggle crashes with Swift exclusivity violation in TalkOverlayView | ## Environment
- **OpenClaw App Version:** 2026.3.2 (build 2026030290)
- **macOS Version:** 26.2 (25C56)
- **Hardware:** Mac16,10 (Apple Silicon)
## Summary
Clicking the Talk Mode toggle in the macOS menu bar app causes an immediate crash due to a Swift exclusivity violation in `TalkOverlayView.body.getter`. The app t... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T15:44:10Z | 2026-04-22T04:29:35Z | 2026-04-20T04:29:00Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Co5mos | 16,522,225 | MDQ6VXNlcjE2NTIyMjI1 | User | false |
openclaw/openclaw | 4,022,648,088 | I_kwDOQb6kR87vxL0Y | 34,624 | https://github.com/openclaw/openclaw/issues/34624 | https://api.github.com/repos/openclaw/openclaw/issues/34624 | Feature: Global circuit breaker for model failover chains | ## Problem
When a model provider experiences an outage (e.g., Anthropic returning `overloaded_error` / HTTP 529), every active session independently discovers the failure through its own retry cycle. Each session burns 1-2+ minutes per attempt before timing out, and if multiple sessions are active, the system spends 1... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T15:38:30Z | 2026-04-22T04:29:36Z | 2026-04-20T04:28:58Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | ryandecook7 | 139,201,002 | U_kgDOCEwJ6g | User | false |
openclaw/openclaw | 4,022,638,874 | I_kwDOQb6kR87vxJka | 34,621 | https://github.com/openclaw/openclaw/issues/34621 | https://api.github.com/repos/openclaw/openclaw/issues/34621 | openclaw doctor: Node warning conflates version with installation method | **Environment:**
- OpenClaw: version unknown
- Node: v25.5.0 (NVM-managed)
- OS: Ubuntu 24.04
**Issue:**
warns "install System Node 22+ and migrate from NVM-managed Node" when running Node **v25.5.0** via NVM.
**Problem:**
The warning conveys two pieces of information that get conflated:
1. "Install System Node 22+"... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T15:36:41Z | 2026-04-22T04:29:37Z | 2026-04-20T04:28:56Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | subimage | 12,508 | MDQ6VXNlcjEyNTA4 | User | false |
openclaw/openclaw | 4,022,612,368 | I_kwDOQb6kR87vxDGQ | 34,613 | https://github.com/openclaw/openclaw/issues/34613 | https://api.github.com/repos/openclaw/openclaw/issues/34613 | Feature Request: session.inject — Contract-compliant session message injection (enqueue-only) | ## Purpose
PR Factory's Long-Term Compatibility Contract §11 requires that external signals (CI failures, deploy failures, retry exhaustion, etc.) be injected into the agent session as structured messages — not as overrides, not as orchestration logic.
We need a Gateway WebSocket RPC method that supports this pattern... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T15:31:22Z | 2026-04-22T04:29:38Z | 2026-04-20T04:28:55Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | lloydarmbrust | 191,840 | MDQ6VXNlcjE5MTg0MA== | User | false |
openclaw/openclaw | 4,022,609,896 | I_kwDOQb6kR87vxCfo | 34,611 | https://github.com/openclaw/openclaw/issues/34611 | https://api.github.com/repos/openclaw/openclaw/issues/34611 | 🚨 Critical: Gateway does not load config, defaults to auth none | ## Problem
After Gateway process starts, it does NOT load `openclaw.json` configuration, resulting in `auth: none` being used instead of the configured `auth: token` mode.
This is a DAY0 security vulnerability - Gateway runs without authentication even when explicitly configured with token.
## Impact
- ❌ Gateway ac... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T15:30:53Z | 2026-04-22T04:29:38Z | 2026-04-20T04:28:53Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | 2233admin | 57,929,895 | MDQ6VXNlcjU3OTI5ODk1 | User | false |
openclaw/openclaw | 4,022,605,318 | I_kwDOQb6kR87vxBYG | 34,610 | https://github.com/openclaw/openclaw/issues/34610 | https://api.github.com/repos/openclaw/openclaw/issues/34610 | [Bug]: Gateway更新后仍是旧版本 + Agent不通知重启完成 + 浏览器标签不关闭导致内存泄漏 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Agent 在执行需要重启 Gateway 的操作后存在三个问题，影响用户体验和系统稳定性。
### Steps to reproduce
问题 1: Gateway 更新后重启仍是旧版本
1. 通过对话让 Agent 执行 `npm update -g openclaw`
2. Agent 执行 `systemctl --user restart openclaw-gateway.service`
3. 检查版本发现仍是旧版本
问题 2: Agent 不主动通知重启完... | closed | not_planned | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-04T15:29:57Z | 2026-04-22T04:29:39Z | 2026-04-20T04:28:51Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | himiko776-v-6 | 262,384,915 | U_kgDOD6OtEw | User | false |
openclaw/openclaw | 4,022,591,808 | I_kwDOQb6kR87vw-FA | 34,609 | https://github.com/openclaw/openclaw/issues/34609 | https://api.github.com/repos/openclaw/openclaw/issues/34609 | [Bug]: Markdown→mrkdwn pipeline misinterprets Slack mrkdwn bold (*text*) as Markdown italic | **Bug type**
Behavior bug
**Summary**
When an LLM outputs Slack mrkdwn formatting (e.g., `*text*` for bold), the Markdown→Slack mrkdwn conversion pipeline misinterprets it as Markdown italic and converts it to `_text_`, reversing the intended semantics.
**Steps to reproduce**
1. Configure a Slack channel agent wit... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T15:27:15Z | 2026-04-22T04:29:40Z | 2026-04-20T04:28:49Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | carrotRakko | 24,588,751 | MDQ6VXNlcjI0NTg4NzUx | User | false |
openclaw/openclaw | 4,022,553,082 | I_kwDOQb6kR87vw0n6 | 34,594 | https://github.com/openclaw/openclaw/issues/34594 | https://api.github.com/repos/openclaw/openclaw/issues/34594 | Feature: support idmap (UID remapping) on docker.binds and workspaceAccess mounts | ## Problem
When running OpenClaw with rootless Podman on Linux, files rsync'd into the agent workspace from the host are not writable by the sandboxed agent process.
**Root cause:** Rootless Podman maps the host user (e.g. `openclaw`, UID 996) to UID 0 (root) inside the container. The sandbox process runs as a differ... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T15:19:48Z | 2026-04-22T04:29:40Z | 2026-04-20T04:28:47Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | goern | 260,331 | MDQ6VXNlcjI2MDMzMQ== | User | false |
openclaw/openclaw | 4,022,550,968 | I_kwDOQb6kR87vw0G4 | 34,592 | https://github.com/openclaw/openclaw/issues/34592 | https://api.github.com/repos/openclaw/openclaw/issues/34592 | [Bug]: Discord channel connection failure causes gateway infinite restart loop | ## Summary
When the Discord API is unreachable (e.g., due to network restrictions or no proxy/TUN mode enabled), the gateway process crashes into an infinite restart loop due to unhandled promise rejections in `GatewayPlugin.registerClient`. This kills the entire gateway, affecting all channels (including Telegram whi... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T15:19:21Z | 2026-04-22T04:29:41Z | 2026-04-20T04:28:45Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | yjfkpyu | 19,964,829 | MDQ6VXNlcjE5OTY0ODI5 | User | false |
openclaw/openclaw | 4,022,503,747 | I_kwDOQb6kR87vwolD | 34,582 | https://github.com/openclaw/openclaw/issues/34582 | https://api.github.com/repos/openclaw/openclaw/issues/34582 | [Bug]: Messages Only Show Up on Whatsapp when type on Terminal | ### Bug type
Regression (worked before, now fails)
### Summary
# OpenClaw TUI Bug Report
## Issue: TUI messages not reaching agent / replies going to wrong channel
### Environment
- OpenClaw Version: 2026.3.2
- OS: Linux (VPS)
- TUI running on: VS Code terminal (Ubuntu)
### Channels Configured
- WhatsApp (enab... | closed | not_planned | true | 6 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T15:09:53Z | 2026-04-22T04:29:42Z | 2026-04-20T04:28:43Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | jzpod27-art | 265,569,709 | U_kgDOD9RFrQ | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.