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 | 4,289,190,220 | I_kwDOQb6kR87_p9lM | 68,696 | https://github.com/openclaw/openclaw/issues/68696 | https://api.github.com/repos/openclaw/openclaw/issues/68696 | [Bug]: | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Telegram session unexpectedly switches from GPT-5.4 to Claude Sonnet after async system events / gateway restart
### Steps to reproduce
I am seeing a repeatable issue where my Telegram direct session unexpect... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T21:10:33Z | 2026-04-28T04:50:28Z | 2026-04-25T18:40:54Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | migo-09 | 277,340,732 | U_kgDOEIfiPA | User | false |
openclaw/openclaw | 4,073,238,042 | I_kwDOQb6kR87yyK4a | 45,438 | https://github.com/openclaw/openclaw/issues/45438 | https://api.github.com/repos/openclaw/openclaw/issues/45438 | perf: structuredClone in session store cache causes native memory leak (~1GB/min) | ## Problem
The gateway leaks ~1GB/min of native memory due to `structuredClone` usage in the session store cache. RSS grows to 4-5GB in minutes while V8 heap stays at ~1.2GB. The gap is native memory from structuredClone's serialization buffer that V8 GC cannot reclaim.
## Evidence
- Heap sampling profile: **242MB all... | closed | completed | true | 9 | [] | [] | 2026-03-13T20:21:41Z | 2026-05-05T04:27:15Z | 2026-05-02T13:24:15Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | markus-lassfolk | 3,661,143 | MDQ6VXNlcjM2NjExNDM= | User | false |
openclaw/openclaw | 4,289,177,218 | I_kwDOQb6kR87_p6aC | 68,695 | https://github.com/openclaw/openclaw/issues/68695 | https://api.github.com/repos/openclaw/openclaw/issues/68695 | Feature request: expose attached files on `before_dispatch` plugin hook event | Closing by author — this was submitted prematurely without my full review. Apologies for the noise. I'll re-open or re-submit if/when the design is ready. | closed | not_planned | true | 0 | [] | [] | 2026-04-18T21:04:18Z | 2026-04-21T04:21:43Z | 2026-04-18T21:11:48Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | atmkei | 262,648,314 | U_kgDOD6ex-g | User | false |
openclaw/openclaw | 3,944,468,904 | I_kwDOQb6kR87rG9Go | 17,311 | https://github.com/openclaw/openclaw/issues/17311 | https://api.github.com/repos/openclaw/openclaw/issues/17311 | SecretsProvider: env, keyring, 1Password providers (building on #16663) | ## Context
PR #16663 by @amor71 delivers GCP, AWS, Azure, and Vault secret providers with CLI tooling, per-agent IAM isolation, stale-while-revalidate caching, migration scripts, and 130+ tests.
Our PR #11539 had a `SecretsProvider` interface and implementations for env vars and OS keyring. We closed it and are build... | closed | completed | true | 10 | [] | [
"vincentkoc"
] | 2026-02-15T16:52:31Z | 2026-05-02T04:28:13Z | 2026-04-29T13:09:50Z | CONTRIBUTOR | null | 20260502T060142Z | 2026-05-02T06:01:42Z | akoscz | 1,360,047 | MDQ6VXNlcjEzNjAwNDc= | User | false |
openclaw/openclaw | 4,289,193,806 | I_kwDOQb6kR87_p-dO | 68,697 | https://github.com/openclaw/openclaw/issues/68697 | https://api.github.com/repos/openclaw/openclaw/issues/68697 | Pluggable STT Providers for voice-call Plugin | ### Summary
Add a pluggable STT provider interface to the voice-call plugin, mirroring the existing TTS provider pattern (api.registerSpeechProvider).
### Problem to solve
The voice-call plugin's streaming STT is hardcoded to openai-realtime in three places: the zod schema enum, the initializeMediaStreaming() method... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-04-18T21:12:01Z | 2026-04-25T04:15:49Z | 2026-04-23T03:30:25Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | agenticbrian | 275,851,811 | U_kgDOEHEqIw | User | false |
openclaw/openclaw | 4,096,022,553 | I_kwDOQb6kR870JFgZ | 49,888 | https://github.com/openclaw/openclaw/issues/49888 | https://api.github.com/repos/openclaw/openclaw/issues/49888 | [Bug]: Persisted oversized toolResult entries can poison shared main sessions (lag/token blowup/OOM risk) | ### Bug type
Performance/reliability bug
### Summary
Large tool results are persisted to session transcripts by default, which can poison shared human-facing sessions (`session.dmScope = "main"`) and cause severe lag, token blowups, and eventual instability/OOM under tool-heavy workflows.
### Steps to reproduce
1.... | closed | completed | true | 7 | [] | [] | 2026-03-18T15:38:48Z | 2026-04-30T04:56:15Z | 2026-04-25T04:32:59Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | sene1337 | 261,265,124 | U_kgDOD5KW5A | User | false |
openclaw/openclaw | 4,289,319,387 | I_kwDOQb6kR87_qdHb | 68,703 | https://github.com/openclaw/openclaw/issues/68703 | https://api.github.com/repos/openclaw/openclaw/issues/68703 | [Bug]: Discord guild-admin actions execute without requester authorization checks | ## Severity Assessment
### CVSS Assessment
| Metric | v3.1 | v4.0 |
|--------|------|------|
| **Score** | 7.1 / 10.0 | 7.1 / 10.0 |
| **Severity** | High | High |
| **Vector** | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:L | CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:H/VA:L/SC:N/SI:N/SA:N |
| **Calculator** | [CVSS v... | open | null | false | 1 | [] | [] | 2026-04-18T22:13:37Z | 2026-05-02T19:50:23Z | null | CONTRIBUTOR | null | 20260502T210030Z | 2026-05-02T21:00:30Z | coygeek | 65,363,919 | MDQ6VXNlcjY1MzYzOTE5 | User | false |
openclaw/openclaw | 4,289,319,395 | I_kwDOQb6kR87_qdHj | 68,704 | https://github.com/openclaw/openclaw/issues/68704 | https://api.github.com/repos/openclaw/openclaw/issues/68704 | [Bug]: Gemma 4 on OpenAI-compat transports — `reasoning_content` is re-sent in conversation history, violating Gemma's documented contract | ### Bug type
Behavior bug (incorrect outbound payload)
### Beta release blocker
No
### Summary
When using Gemma 4 models (google/gemma-4-26b-a4b, google/gemma-4-31b, etc.) via an OpenAI-compatible endpoint such as LM Studio, OpenClaw re-sends the model's prior-turn `reasoning_content` on subsequent turns. Google's... | closed | completed | true | 2 | [] | [] | 2026-04-18T22:13:38Z | 2026-04-30T04:36:04Z | 2026-04-27T07:26:57Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | chip-snomo | 269,966,955 | U_kgDOEBdeaw | User | false |
openclaw/openclaw | 4,289,321,075 | I_kwDOQb6kR87_qdhz | 68,707 | https://github.com/openclaw/openclaw/issues/68707 | https://api.github.com/repos/openclaw/openclaw/issues/68707 | [Bug]: openai-completions provider never sends stream_options.include_usage: true, causing context token tracking to always show 0% | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Root cause: resolveIncludeUsageForStreaming() in openai-http-CBh7eNgq.js requires stream_options.include_usage === true in the inbound payload, but OpenClaw never injects this when proxying to openai-completion... | closed | completed | true | 0 | [
"bug"
] | [] | 2026-04-18T22:14:36Z | 2026-04-21T04:21:26Z | 2026-04-19T02:14:50Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | arifakyol | 9,477,480 | MDQ6VXNlcjk0Nzc0ODA= | User | false |
openclaw/openclaw | 4,289,320,361 | I_kwDOQb6kR87_qdWp | 68,706 | https://github.com/openclaw/openclaw/issues/68706 | https://api.github.com/repos/openclaw/openclaw/issues/68706 | [Bug]: Fallback model can stick to session after auto fallback, causing later runs to start from persisted override | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
Automatic fallback model selection appears to persist into session state and can affect later turns as if it were a durable session model override.
I observed a Telegram session that seemed to switch away from the configured p... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-04-18T22:14:08Z | 2026-04-24T04:26:20Z | 2026-04-21T06:59:30Z | NONE | null | 20260424T060021Z | 2026-04-24T06:00:21Z | kh-mitya | 264,714,009 | U_kgDOD8c3GQ | User | false |
openclaw/openclaw | 4,208,990,615 | I_kwDOQb6kR8764BmX | 61,589 | https://github.com/openclaw/openclaw/issues/61589 | https://api.github.com/repos/openclaw/openclaw/issues/61589 | Browser start reports CDP unreachable even while managed Chromium is live and /json/version responds | ### Summary
On this OpenClaw 2026.4.2 Linux host, the official managed browser lane for profile `openclaw` is degraded even though Chromium/CDP itself is healthy.
### Environment
- OpenClaw: 2026.4.2
- Host: Linux VPS
- Browser config (effective baseline during investigation):
- `browser.enabled=true`
- `browser.d... | closed | completed | true | 2 | [] | [] | 2026-04-06T01:10:24Z | 2026-04-21T04:21:41Z | 2026-04-18T22:19:32Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | tadeogutierrez | 16,327,260 | MDQ6VXNlcjE2MzI3MjYw | User | false |
openclaw/openclaw | 4,289,338,978 | I_kwDOQb6kR87_qh5i | 68,709 | https://github.com/openclaw/openclaw/issues/68709 | https://api.github.com/repos/openclaw/openclaw/issues/68709 | [Bug]: `openclaw.json` written with mode 0664 during gateway hot-save on Linux — chmod 0o600 not applied, exposes API keys group-readable | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On Linux systemd-managed deployments, `openclaw.json` ends up with mode `0664` (group-writable, world-readable) after the gateway performs a runtime hot-save (e.g., `plugins.entries.<provider>.config.*` or `age... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-18T22:25:04Z | 2026-04-21T04:21:25Z | 2026-04-19T02:20:47Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | rgerrans | 18,623,772 | MDQ6VXNlcjE4NjIzNzcy | User | false |
openclaw/openclaw | 4,289,351,550 | I_kwDOQb6kR87_qk9- | 68,712 | https://github.com/openclaw/openclaw/issues/68712 | https://api.github.com/repos/openclaw/openclaw/issues/68712 | Bug: Realtime voice bridge fails on outbound calls — status callback race overwrites <Connect><Stream> TwiML | ## Bug: Realtime voice bridge fails on outbound calls — status callback race condition overwrites `<Connect><Stream>` TwiML
### Description
When making outbound calls with `realtime.enabled = true`, the OpenAI Realtime voice bridge consistently fails with `"WebSocket was closed before the connection was established"`... | closed | not_planned | true | 0 | [] | [] | 2026-04-18T22:33:05Z | 2026-04-21T04:21:39Z | 2026-04-18T22:34:48Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | primecoder-legonhill | 265,703,021 | U_kgDOD9ZObQ | User | false |
openclaw/openclaw | 4,228,432,250 | I_kwDOQb6kR878CMF6 | 63,431 | https://github.com/openclaw/openclaw/issues/63431 | https://api.github.com/repos/openclaw/openclaw/issues/63431 | Nested workspace (workspace-local/local) is accepted, initialized, and recreated after deletion | +
## 🐛 Bug: Nested workspace () is accepted, initialized, and re-created after deletion
### Summary
OpenClaw can accept and operate on a **non-canonical nested workspace path** (e.g. ).
Once activated, the runtime **bootstraps it as a valid workspace**, writes state and memory inside it, and **recreates it after m... | closed | not_planned | true | 6 | [] | [] | 2026-04-08T23:46:37Z | 2026-04-28T05:03:38Z | 2026-04-25T15:38:05Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Jmarzab | 196,876,602 | U_kgDOC7wZOg | User | false |
openclaw/openclaw | 4,289,407,807 | I_kwDOQb6kR87_qys_ | 68,719 | https://github.com/openclaw/openclaw/issues/68719 | https://api.github.com/repos/openclaw/openclaw/issues/68719 | feat(bluebubbles): inbound audio enricher — pre-dispatch transcription via /api/v1/message/audio-transcript | ## Problem
When an iMessage voice note arrives via BlueBubbles, the agent receives an empty string or `🎤 [Audio]` and responds apologetically. BlueBubbles v1.14.0+ already exposes `GET /api/v1/message/audio-transcript/:guid` (Apple's on-device transcription, no ASR cost). OpenClaw has no built-in mechanism to fetch t... | closed | not_planned | true | 2 | [
"channel: bluebubbles"
] | [] | 2026-04-18T23:01:47Z | 2026-05-04T04:37:20Z | 2026-05-01T06:47:45Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | markthebest12 | 6,416,874 | MDQ6VXNlcjY0MTY4NzQ= | User | false |
openclaw/openclaw | 4,289,409,915 | I_kwDOQb6kR87_qzN7 | 68,720 | https://github.com/openclaw/openclaw/issues/68720 | https://api.github.com/repos/openclaw/openclaw/issues/68720 | 2 local models on 2 local server ? | ### Summary
I've tried setting up 2 local server and it's all most impossible , so instead of one local and one cloud they should be a easy way to set up a second local sever .
### Problem to solve
if you want all local you will set up 2 models , 1 quick 4b model for chat and a second 20-40b model for coding etc , ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-18T23:02:57Z | 2026-04-28T04:50:27Z | 2026-04-25T18:41:01Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | peter247 | 3,247,383 | MDQ6VXNlcjMyNDczODM= | User | false |
openclaw/openclaw | 4,289,355,118 | I_kwDOQb6kR87_ql1u | 68,713 | https://github.com/openclaw/openclaw/issues/68713 | https://api.github.com/repos/openclaw/openclaw/issues/68713 | Bug: Realtime voice bridge fails on outbound calls — status callback race overwrites <Connect><Stream> TwiML | ## Bug: Realtime voice bridge fails on outbound calls — status callback race condition overwrites `<Connect><Stream>` TwiML
### Description
When making outbound calls with `realtime.enabled = true`, the OpenAI Realtime voice bridge consistently fails with `"WebSocket was closed before the connection was established"`... | closed | completed | true | 3 | [] | [] | 2026-04-18T22:35:31Z | 2026-04-27T04:55:25Z | 2026-04-24T21:35:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | legonhilltech-jpg | 272,396,537 | U_kgDOEDxw-Q | User | false |
openclaw/openclaw | 4,175,671,187 | I_kwDOQb6kR87446-T | 58,128 | https://github.com/openclaw/openclaw/issues/58128 | https://api.github.com/repos/openclaw/openclaw/issues/58128 | [Bug]: AWS Bedrock converse API schema change breaks non-Anthropic models (Nova Pro affected) | ## Summary
The AWS Bedrock `converse` API no longer accepts the `type` field in message content blocks. This breaks non-Anthropic Bedrock models (Amazon Nova Pro confirmed affected). **Anthropic/Claude models on Bedrock continue to work.**
## Error
```
Parameter validation failed:
Invalid number of parameters set fo... | closed | not_planned | true | 3 | [] | [] | 2026-03-31T04:29:42Z | 2026-04-28T05:16:47Z | 2026-04-25T12:30:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rixcorp-oc | 265,470,397 | U_kgDOD9LBvQ | User | false |
openclaw/openclaw | 4,289,462,910 | I_kwDOQb6kR87_rAJ- | 68,723 | https://github.com/openclaw/openclaw/issues/68723 | https://api.github.com/repos/openclaw/openclaw/issues/68723 | memory-core: cron-fired dreaming writes raw memory snippets to DREAMS.md because subagent runtime is request-scoped | ## Summary
When `memory-core.dreaming` fires via its managed cron (systemEvent `__openclaw_memory_core_short_term_promotion_dream__`), the narrative generator at `extensions/memory-core/src/dreaming-narrative.ts` falls back because the plugin runtime's subagent API is request-scoped and unavailable inside a system-eve... | closed | completed | true | 0 | [] | [] | 2026-04-18T23:23:27Z | 2026-04-21T04:21:36Z | 2026-04-18T23:29:42Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | LightDriverCS | 255,745,086 | U_kgDODz5cPg | User | false |
openclaw/openclaw | 4,289,500,766 | I_kwDOQb6kR87_rJZe | 68,727 | https://github.com/openclaw/openclaw/issues/68727 | https://api.github.com/repos/openclaw/openclaw/issues/68727 | Audio auto transcription can prefer local Whisper over API provider and break Groq multipart uploads | ### Summary
Telegram/audio transcription can unexpectedly use a local Whisper CLI even when an API provider is configured/available, and Groq OpenAI-compatible audio transcription can fail when OpenClaw passes a proxy-wrapped fetch into the audio provider.
### Local reproduction
Environment:
- macOS gateway
- `tool... | closed | completed | true | 1 | [] | [] | 2026-04-18T23:47:21Z | 2026-04-26T04:30:31Z | 2026-04-23T04:32:06Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | dyz2102 | 21,371,785 | MDQ6VXNlcjIxMzcxNzg1 | User | false |
openclaw/openclaw | 4,289,503,191 | I_kwDOQb6kR87_rJ_X | 68,728 | https://github.com/openclaw/openclaw/issues/68728 | https://api.github.com/repos/openclaw/openclaw/issues/68728 | [Bug]: Gemma 4 models should be detected as reasoning models (Ollama think mode) | Gemma 4 models (e.g. `gemma4`, `gemma4:12b`, `gemma4:27b`, `gemma4:31b`) support Ollama native thinking mode via the `think` parameter and may emit answer text in the `thinking`/`reasoning` fields.
OpenClaw currently infers whether a model is a “reasoning/thinking” model via `isReasoningModelHeuristic`, but upstream `... | closed | completed | true | 1 | [] | [] | 2026-04-18T23:49:14Z | 2026-04-29T04:33:17Z | 2026-04-27T00:33:13Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | frankhli843 | 51,806,469 | MDQ6VXNlcjUxODA2NDY5 | User | false |
openclaw/openclaw | 4,289,506,166 | I_kwDOQb6kR87_rKt2 | 68,731 | https://github.com/openclaw/openclaw/issues/68731 | https://api.github.com/repos/openclaw/openclaw/issues/68731 | [Feature]: Broaden BlueBubbles contact-name enrichment from groups to 1:1 DMs | ### Summary
Extend `enrichBlueBubblesParticipantsWithContactNames` (currently invoked only on the group-chat path) to also resolve 1:1 inbound sender names from the agent user's local Contacts.app.
### Problem to solve
Today, `monitor-processing.ts` only calls `enrichBlueBubblesParticipantsWithContactNames` when `isGr... | closed | not_planned | true | 4 | [
"channel: bluebubbles"
] | [
"omarshahine"
] | 2026-04-18T23:50:42Z | 2026-05-04T04:37:20Z | 2026-05-01T06:49:46Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | markthebest12 | 6,416,874 | MDQ6VXNlcjY0MTY4NzQ= | User | false |
openclaw/openclaw | 4,289,536,760 | I_kwDOQb6kR87_rSL4 | 68,735 | https://github.com/openclaw/openclaw/issues/68735 | https://api.github.com/repos/openclaw/openclaw/issues/68735 | [Bug]: LLM request failed: provider rejected the request schema or tool payload. | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.14 to 2026.4.15, configured use model github-copilot/gpt-5-mini, worked at first message, failed in later messages, the error is : LLM request failed: provider rejected the request schema or tool ... | closed | completed | true | 28 | [
"bug",
"regression"
] | [] | 2026-04-19T00:09:23Z | 2026-04-29T04:29:17Z | 2026-04-26T14:55:48Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | yokarno | 23,699,567 | MDQ6VXNlcjIzNjk5NTY3 | User | false |
openclaw/openclaw | 4,190,461,591 | I_kwDOQb6kR875xV6X | 59,287 | https://github.com/openclaw/openclaw/issues/59287 | https://api.github.com/repos/openclaw/openclaw/issues/59287 | [Bug]: openclaw health --json reports telegram.running=false while probe succeeds and status --deep shows Telegram OK | ## Summary
`openclaw health --json` reports Telegram `running=false` even while the Telegram probe succeeds and `openclaw status --deep` shows Telegram `OK`.
On the same system and same runtime state:
- `openclaw status --deep` shows Telegram healthy/OK
- `openclaw health --json` shows:
- `configured: true`
- `r... | open | null | false | 5 | [] | [] | 2026-04-01T23:20:33Z | 2026-04-30T23:22:47Z | null | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | smonett | 50,761,481 | MDQ6VXNlcjUwNzYxNDgx | User | false |
openclaw/openclaw | 4,080,189,663 | I_kwDOQb6kR87zMsDf | 47,899 | https://github.com/openclaw/openclaw/issues/47899 | https://api.github.com/repos/openclaw/openclaw/issues/47899 | Bedrock provider: baseUrl not passed as endpoint to BedrockRuntimeClient | ## Summary
The `amazon-bedrock` provider in OpenClaw does not pass `models.json`'s `baseUrl` configuration to the AWS SDK's `BedrockRuntimeClient` as the `endpoint` parameter. This prevents users from routing Bedrock requests through a custom proxy (e.g., for cost monitoring, rate limiting, or logging).
## Current be... | closed | completed | true | 5 | [] | [] | 2026-03-16T04:41:25Z | 2026-04-22T04:31:12Z | 2026-04-19T06:47:11Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | L7-admin | 261,360,786 | U_kgDOD5QMkg | User | false |
openclaw/openclaw | 4,289,512,140 | I_kwDOQb6kR87_rMLM | 68,732 | https://github.com/openclaw/openclaw/issues/68732 | https://api.github.com/repos/openclaw/openclaw/issues/68732 | Plugin update writes trigger false gateway restarts via runtime-config paths (firecrawl webFetch + gateway.auth.token) | # OpenClaw core bug: plugin update persists runtime-only `plugins.entries.firecrawl.config.webFetch` into source config
## Summary
`updatePluginsAfterCoreUpdate()` uses `params.configSnapshot.config` (runtime config) as the base config for plugin update persistence. When that runtime config includes normalized/materia... | closed | completed | true | 4 | [] | [] | 2026-04-18T23:53:31Z | 2026-04-25T04:16:10Z | 2026-04-22T23:45:08Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | fchaudhryspear | 198,639,243 | U_kgDOC9b-iw | User | false |
openclaw/openclaw | 4,226,370,872 | I_kwDOQb6kR8776U04 | 63,274 | https://github.com/openclaw/openclaw/issues/63274 | https://api.github.com/repos/openclaw/openclaw/issues/63274 | [Bug]: Cron Telegram delivery broken on v2026.4.8 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
<html><head></head><body><h1>[Bug]: Cron Telegram delivery broken on v2026.4.8 — isolated+announce false positive, main+system-event hangs</h1>
<h2>Summary</h2>
<p>No working pattern exists for delivering cron-based reminder... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-08T17:24:51Z | 2026-04-28T05:03:59Z | 2026-04-25T15:34:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sudoAPWH | 186,760,306 | U_kgDOCyG8cg | User | false |
openclaw/openclaw | 4,289,578,480 | I_kwDOQb6kR87_rcXw | 68,738 | https://github.com/openclaw/openclaw/issues/68738 | https://api.github.com/repos/openclaw/openclaw/issues/68738 | readPageSummaries: no concurrency limit on fs.readFile, triggers EDEADLK on iCloud/FileProvider-backed vaults | ## Problem
`readPageSummaries` in `dist/cli-mPaKxVzJ.js` (line 586) unbounded-parallel `fs.readFile`s every markdown page in the wiki vault via `Promise.all(filePaths.map(readFile))`. For vaults backed by iCloud Drive / macOS `FileProvider` (e.g. Obsidian-in-iCloud), this kernel-level parallel read burst can deadlock ... | open | null | false | 1 | [] | [] | 2026-04-19T00:37:58Z | 2026-05-02T19:50:25Z | null | NONE | null | 20260502T210030Z | 2026-05-02T21:00:30Z | richardmqq | 14,178,267 | MDQ6VXNlcjE0MTc4MjY3 | User | false |
openclaw/openclaw | 4,049,156,383 | I_kwDOQb6kR87xWTkf | 41,685 | https://github.com/openclaw/openclaw/issues/41685 | https://api.github.com/repos/openclaw/openclaw/issues/41685 | [Bug]: Control UI displays sessions from unregistered agents | ## Summary
When deleting an agent, the agent directory is not fully cleaned up. This causes the Control UI to still display sessions from deleted agents, creating user confusion.
## How I discovered this
I was trying to clean up my agents and ran `openclaw agents delete taizi --force`. After deleting, I opened the C... | open | null | false | 5 | [] | [] | 2026-03-10T03:31:43Z | 2026-05-04T12:44:53Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Amateur0x1 | 157,032,610 | U_kgDOCVwgog | User | false |
openclaw/openclaw | 4,289,591,978 | I_kwDOQb6kR87_rfqq | 68,739 | https://github.com/openclaw/openclaw/issues/68739 | https://api.github.com/repos/openclaw/openclaw/issues/68739 | Telegram native approvals still start with execApprovals.enabled=false, then loop on pairing required (scope-upgrade) | ## Summary
On OpenClaw 2026.4.15, Telegram native approvals still attempt to start even when `channels.telegram.execApprovals.enabled` is explicitly set to `false`. The native approval handler then loops on `GatewayClientRequestError: pairing required`, with websocket logs showing `reason: "scope-upgrade"`.
In the sam... | closed | completed | true | 2 | [] | [] | 2026-04-19T00:47:07Z | 2026-04-28T04:50:26Z | 2026-04-25T18:41:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | stuart-pebble | 175,665,393 | U_kgDOCnhw8Q | User | false |
openclaw/openclaw | 4,289,605,629 | I_kwDOQb6kR87_ri_9 | 68,743 | https://github.com/openclaw/openclaw/issues/68743 | https://api.github.com/repos/openclaw/openclaw/issues/68743 | [Bug]: Dreaming cron job times out at 10 minutes even though manual trigger completes in ~2-4 minutes | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
reaming cron job times out at 10 minutes even though manual trigger completes in 2-4 minutes
### Steps to reproduce
1. Leave the Memory Dreaming Promotion cron job enabled with schedule '0 3 * * *'
2. Observe the job at th... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-19T00:56:56Z | 2026-04-28T04:50:25Z | 2026-04-25T18:41:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | whnnick | 229,894,359 | U_kgDODbPo1w | User | false |
openclaw/openclaw | 4,289,622,454 | I_kwDOQb6kR87_rnG2 | 68,751 | https://github.com/openclaw/openclaw/issues/68751 | https://api.github.com/repos/openclaw/openclaw/issues/68751 | session-memory: raw prior-session turns replay as current input on /reset (autonomous re-execution) | ### Summary
Bundled hook `session-memory` persists raw user/assistant turns from prior sessions to `<workspace>/memory/YYYY-MM-DD-{slug}.md` without untrusted-data markers, so on the next `/new` or `/reset` those prior user commands re-enter the agent's context as current input and the agent re-executes them.
### Ste... | open | null | false | 4 | [] | [] | 2026-04-19T01:08:08Z | 2026-05-02T14:22:13Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | cygnostik | 5,272,292 | MDQ6VXNlcjUyNzIyOTI= | User | false |
openclaw/openclaw | 4,289,622,628 | I_kwDOQb6kR87_rnJk | 68,752 | https://github.com/openclaw/openclaw/issues/68752 | https://api.github.com/repos/openclaw/openclaw/issues/68752 | doctor: 'Multiple state directories' check is noisy on multi-user hosts | ## Problem
On a host that legitimately runs OpenClaw as multiple Unix users (e.g. separate user accounts for separate agents on a single Mac Mini), \`openclaw doctor\` walking \`/Users/*/.openclaw\` reports every sibling user's state directory as a potential split-history concern:
\`\`\`
◇ State integrity
│ - Multi... | closed | not_planned | true | 1 | [] | [] | 2026-04-19T01:08:14Z | 2026-04-30T04:39:11Z | 2026-04-26T05:22:25Z | CONTRIBUTOR | null | 20260430T060018Z | 2026-04-30T06:00:18Z | markthebest12 | 6,416,874 | MDQ6VXNlcjY0MTY4NzQ= | User | false |
openclaw/openclaw | 4,289,643,122 | I_kwDOQb6kR87_rsJy | 68,757 | https://github.com/openclaw/openclaw/issues/68757 | https://api.github.com/repos/openclaw/openclaw/issues/68757 | [CLOSED - CREATED IN ERROR] | I apologize, this issue was created in error. Please ignore and delete if possible. | closed | not_planned | true | 0 | [] | [] | 2026-04-19T01:22:59Z | 2026-04-21T04:21:17Z | 2026-04-19T01:24:02Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | arhang | 4,433,757 | MDQ6VXNlcjQ0MzM3NTc= | User | false |
openclaw/openclaw | 4,099,820,924 | I_kwDOQb6kR870Xk18 | 50,295 | https://github.com/openclaw/openclaw/issues/50295 | https://api.github.com/repos/openclaw/openclaw/issues/50295 | Feature Request: Add tools.exec.obfuscationCheck config option to disable obfuscation detection | ## Problem
The exec tool includes a hardcoded obfuscation detection layer (`src/infra/exec-obfuscation-detect.ts`) that flags commands exceeding 10,000 characters (`MAX_COMMAND_CHARS = 1e4`) or matching heuristic patterns (hex escapes, octal sequences, variable expansion, etc.). When triggered, it forces an approval f... | closed | completed | true | 7 | [] | [] | 2026-03-19T07:13:55Z | 2026-04-30T04:55:18Z | 2026-04-25T04:41:40Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | joesinvestments | 257,641,472 | U_kgDOD1tMAA | User | false |
openclaw/openclaw | 4,289,654,852 | I_kwDOQb6kR87_rvBE | 68,760 | https://github.com/openclaw/openclaw/issues/68760 | https://api.github.com/repos/openclaw/openclaw/issues/68760 | [Bug]: Cron delivery channel reverts to 'last' after job execution in 4.15 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
After upgrading to 4.15, cron job delivery channel reverts to last after each execution, even when explicitly set to telegram. First observed after upgrading to 4.15 — unable to confirm if this is a regression ... | closed | completed | true | 0 | [
"bug",
"bug:behavior"
] | [] | 2026-04-19T01:29:43Z | 2026-04-22T04:31:13Z | 2026-04-19T06:39:18Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | fenglanhua | 257,674,093 | U_kgDOD1vLbQ | User | false |
openclaw/openclaw | 4,289,655,041 | I_kwDOQb6kR87_rvEB | 68,761 | https://github.com/openclaw/openclaw/issues/68761 | https://api.github.com/repos/openclaw/openclaw/issues/68761 | Severe regression from 2026.3.13 to 2026.4.15: web actions degraded, image grabbing worse, tasks often go silent | ## Summary
I upgraded from `2026.3.13` to `2026.4.15`, and the regression is very obvious.
After the update, the assistant became dramatically worse. This is not a subtle change, it feels like a major capability regression and is now barely usable for normal work.
## Regressions observed
Compared with `2026.3.13`, se... | open | null | false | 1 | [] | [] | 2026-04-19T01:29:54Z | 2026-05-03T04:59:43Z | null | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | royrenwb | 18,721,231 | MDQ6VXNlcjE4NzIxMjMx | User | false |
openclaw/openclaw | 4,188,541,879 | I_kwDOQb6kR875qBO3 | 59,165 | https://github.com/openclaw/openclaw/issues/59165 | https://api.github.com/repos/openclaw/openclaw/issues/59165 | RFC: Credential Provider Plugin | # RFC: Credential Provider Plugin
| Field | Value |
|--------------|------------------------------------------------|
| **Status** | Draft |
| **Authors** | [@loveyana](https://github.com/loveyana) |
| **Created** | 2026... | open | null | false | 2 | [] | [] | 2026-04-01T17:12:48Z | 2026-04-30T23:22:38Z | null | CONTRIBUTOR | null | 20260501T113057Z | 2026-05-01T11:30:57Z | loveyana | 87,516,818 | MDQ6VXNlcjg3NTE2ODE4 | User | false |
openclaw/openclaw | 4,289,697,987 | I_kwDOQb6kR87_r5jD | 68,767 | https://github.com/openclaw/openclaw/issues/68767 | https://api.github.com/repos/openclaw/openclaw/issues/68767 | sessions_spawn bloats subagent context even with lightContext:true | ## Bug Description
When spawning subagents via `sessions_spawn` with `lightContext: true`, the subagent still receives 11-32k tokens of context instead of the expected 1-2k tokens. This causes performance degradation for models with smaller context windows (e.g., Qwen 3.6 Plus).
## Environment
- **OpenClaw version:*... | closed | completed | true | 1 | [] | [] | 2026-04-19T02:02:11Z | 2026-04-28T04:50:23Z | 2026-04-25T18:41:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | syuvers | 180,635,274 | U_kgDOCsRGig | User | false |
openclaw/openclaw | 4,289,693,476 | I_kwDOQb6kR87_r4ck | 68,766 | https://github.com/openclaw/openclaw/issues/68766 | https://api.github.com/repos/openclaw/openclaw/issues/68766 | [Bug]: Feishu WebSocket connection does not recover after transient token refresh failure | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When the Feishu `tenant_access_token` refresh fails due to a transient timeout (e.g., `open.feishu.cn` responds slowly during off-peak hours), the Feishu WebSocket connection drops and does not automatically re... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-04-19T01:58:53Z | 2026-04-30T04:32:44Z | 2026-04-28T00:32:38Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | jw8957 | 4,081,007 | MDQ6VXNlcjQwODEwMDc= | User | false |
openclaw/openclaw | 4,289,699,283 | I_kwDOQb6kR87_r53T | 68,768 | https://github.com/openclaw/openclaw/issues/68768 | https://api.github.com/repos/openclaw/openclaw/issues/68768 | Docs: contextTokens placement in agents.defaults.model doesn't match schema | ## Problem
The configuration reference at `/gateway/configuration-reference` documents `contextTokens` as a field inside `agents.defaults.model`:
```json5
{
agents: {
defaults: {
model: {
primary: "anthropic/claude-opus-4-6",
contextTokens: 200000
}
}
}
}
```
The gateway rejec... | closed | not_planned | true | 1 | [
"invalid"
] | [] | 2026-04-19T02:03:07Z | 2026-04-21T04:21:23Z | 2026-04-19T03:21:59Z | NONE | null | 20260421T060038Z | 2026-04-21T06:00:38Z | ajfonthemove | 263,406,761 | U_kgDOD7NEqQ | User | false |
openclaw/openclaw | 4,289,699,373 | I_kwDOQb6kR87_r54t | 68,769 | https://github.com/openclaw/openclaw/issues/68769 | https://api.github.com/repos/openclaw/openclaw/issues/68769 | TTS `parseTtsDirectives` is markdown-blind: `[[tts:xxx]]` inside code spans / code blocks triggers auto TTS in `tagged` mode | ## Summary
`parseTtsDirectives` in `provider-error-utils-CyJAWFR1.js` uses the regex `/\[\[tts:([^\]]+)\]\]/gi`, which matches **any literal `[[tts:xxx]]` occurrence** regardless of surrounding markdown context. An assistant writing *about* TTS directives — for example in a troubleshooting reply, a `| TTS 链路诊断 | 查 \`[... | closed | completed | true | 1 | [] | [] | 2026-04-19T02:03:11Z | 2026-04-27T04:37:22Z | 2026-04-25T03:13:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | richardmqq | 14,178,267 | MDQ6VXNlcjE0MTc4MjY3 | User | false |
openclaw/openclaw | 4,289,699,511 | I_kwDOQb6kR87_r563 | 68,770 | https://github.com/openclaw/openclaw/issues/68770 | https://api.github.com/repos/openclaw/openclaw/issues/68770 | Telegram plugin: outbound media path (sendVoice / sendAudio / sendDocument / ...) emits no success log line | ## Summary
The Telegram plugin writes a `telegram sendMessage ok chat=... message=...` log line for every text send, but the media path (`sendVoice`, `sendAudio`, `sendDocument`, `sendVideo`, etc.) has **no equivalent log line**. This breaks observability for TTS-backed voice notes and any other media-bearing replies:... | open | null | false | 2 | [] | [] | 2026-04-19T02:03:16Z | 2026-05-02T12:37:07Z | null | NONE | null | 20260502T150030Z | 2026-05-02T15:00:30Z | richardmqq | 14,178,267 | MDQ6VXNlcjE0MTc4MjY3 | User | false |
openclaw/openclaw | 4,166,214,455 | I_kwDOQb6kR874U2M3 | 57,360 | https://github.com/openclaw/openclaw/issues/57360 | https://api.github.com/repos/openclaw/openclaw/issues/57360 | Feature: Expose requestMetadata config for Bedrock cost allocation tagging | ## Summary
The `@mariozechner/pi-ai` library (bundled in OpenClaw v2026.3.28, pi-ai v0.63.1) added `requestMetadata` support to `BedrockOptions`, which forwards key-value pairs to the AWS Bedrock Converse API for [split cost allocation](https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_ConverseStream... | open | null | false | 1 | [] | [] | 2026-03-30T00:28:43Z | 2026-05-03T09:12:29Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | wjonaskr | 137,480,581 | U_kgDOCDHJhQ | User | false |
openclaw/openclaw | 4,289,708,614 | I_kwDOQb6kR87_r8JG | 68,771 | https://github.com/openclaw/openclaw/issues/68771 | https://api.github.com/repos/openclaw/openclaw/issues/68771 | [Bug] openclaw-weixin: MEDIA sends duplicate file content and renames files to UUID | ## Bug Description
When sending multiple files via `MEDIA:` tag through the openclaw-weixin channel, two issues occur:
1. **All MEDIA lines send the same file content** - regardless of different file paths specified, only the first file's content is delivered for all attachments
2. **Filenames become UUIDs** - e.g. `... | closed | not_planned | true | 1 | [] | [] | 2026-04-19T02:09:29Z | 2026-04-28T04:50:22Z | 2026-04-25T18:41:34Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | wonggp6 | 106,688,846 | U_kgDOBlvxTg | User | false |
openclaw/openclaw | 4,289,712,603 | I_kwDOQb6kR87_r9Hb | 68,772 | https://github.com/openclaw/openclaw/issues/68772 | https://api.github.com/repos/openclaw/openclaw/issues/68772 | Ajv 2020-12 schema compatibility: playwright-mcp tools fail to initialize | ## Bug Description
When using `@playwright/mcp` v1.60+ as an MCP server in OpenClaw, tool registration fails with:
```
Error: no schema with key or ref "https://json-schema.org/draft/2020-12/schema"
```
## Environment
- **OpenClaw version:** 2026.4.15 (041266a)
- **Node.js:** v22.22.0
- **@playwright/mcp:** v1.60.0... | closed | completed | true | 1 | [] | [] | 2026-04-19T02:12:19Z | 2026-04-27T04:37:09Z | 2026-04-25T03:15:27Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | liemnhoang | 58,699,349 | MDQ6VXNlcjU4Njk5MzQ5 | User | false |
openclaw/openclaw | 4,289,732,408 | I_kwDOQb6kR87_sB84 | 68,776 | https://github.com/openclaw/openclaw/issues/68776 | https://api.github.com/repos/openclaw/openclaw/issues/68776 | Discord inbound silently hangs in `ensureConfiguredBindingRouteReady` for `type: "acp"` bindings | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On the first Discord inbound that resolves to a `type: "acp"` configured binding, the gateway silently suspends inside `await conversationRuntime.ensureConfiguredBindingRouteReady(...)` in the Discord preflight... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-19T02:26:25Z | 2026-04-28T04:43:33Z | 2026-04-26T01:37:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | legacycapitalholdings | 273,457,464 | U_kgDOEEyhOA | User | false |
openclaw/openclaw | 4,289,735,255 | I_kwDOQb6kR87_sCpX | 68,777 | https://github.com/openclaw/openclaw/issues/68777 | https://api.github.com/repos/openclaw/openclaw/issues/68777 | Control UI assets missing from npm installs — Dashboard blank | ## Problem
Users report that after installation, the Dashboard is blank. Errors point to missing `dist/control-ui` assets.
## Root Cause Analysis
Source code tracing reveals:
1. **`prepack` script** (`scripts/openclaw-prepack.ts:114-115`) correctly runs `pnpm build && pnpm ui:build`
2. **Main build script** (`scrip... | closed | completed | true | 1 | [] | [] | 2026-04-19T02:28:36Z | 2026-04-28T04:50:22Z | 2026-04-25T18:41:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JIRBOY | 49,422,297 | MDQ6VXNlcjQ5NDIyMjk3 | User | false |
openclaw/openclaw | 4,289,735,532 | I_kwDOQb6kR87_sCts | 68,778 | https://github.com/openclaw/openclaw/issues/68778 | https://api.github.com/repos/openclaw/openclaw/issues/68778 | WhatsApp plugin depends on RC library and requires postinstall hotfix — unstable on clean install | ## Problem
After upgrading or clean installing, the WhatsApp channel is unavailable. `openclaw plugins list` shows the WhatsApp plugin as broken or not loaded.
## Root Cause Analysis
Three compounding risks found in source code:
1. **RC dependency**: `extensions/whatsapp/package.json` depends on `@whiskeysockets/ba... | closed | completed | true | 1 | [] | [] | 2026-04-19T02:28:50Z | 2026-04-28T04:50:21Z | 2026-04-25T18:41:47Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JIRBOY | 49,422,297 | MDQ6VXNlcjQ5NDIyMjk3 | User | false |
openclaw/openclaw | 4,289,735,832 | I_kwDOQb6kR87_sCyY | 68,779 | https://github.com/openclaw/openclaw/issues/68779 | https://api.github.com/repos/openclaw/openclaw/issues/68779 | Browser existing-session mode has overly tight timeouts — multi-step automation is unstable | ## Problem
When using `browser --browser-profile user` for browser automation:
- Operations drop mid-flow with `gateway closed (1000)`
- Handshake timeouts are frequent
- Multi-step flows ("open → click → navigate") fail repeatedly
## Root Cause Analysis
1. **Aggressive timeout values** (`extensions/browser/src/brow... | open | null | false | 2 | [] | [] | 2026-04-19T02:29:05Z | 2026-05-03T07:50:08Z | null | CONTRIBUTOR | null | 20260503T090142Z | 2026-05-03T09:01:42Z | JIRBOY | 49,422,297 | MDQ6VXNlcjQ5NDIyMjk3 | User | false |
openclaw/openclaw | 4,289,736,231 | I_kwDOQb6kR87_sC4n | 68,780 | https://github.com/openclaw/openclaw/issues/68780 | https://api.github.com/repos/openclaw/openclaw/issues/68780 | Plugin allowlist empty warning is unactionable and alarming for first-time users | ## Problem
After a fresh install or upgrade, users see:
```
[plugins] plugins.allow is empty; discovered non-bundled plugins may auto-load: openclaw-lark, openclaw-weixin
[plugins] openclaw-lark: loaded without install/load-path provenance
```
These warnings are unactionable for new users — they have not configured ... | open | null | false | 2 | [] | [] | 2026-04-19T02:29:23Z | 2026-05-02T20:57:33Z | null | CONTRIBUTOR | null | 20260502T210030Z | 2026-05-02T21:00:30Z | JIRBOY | 49,422,297 | MDQ6VXNlcjQ5NDIyMjk3 | User | false |
openclaw/openclaw | 4,289,736,808 | I_kwDOQb6kR87_sDBo | 68,781 | https://github.com/openclaw/openclaw/issues/68781 | https://api.github.com/repos/openclaw/openclaw/issues/68781 | Gateway startup extremely slow (70-100s), causing TUI timeouts and spontaneous restarts | ## Problem
Users report three interrelated issues:
1. **Extremely slow startup**: `openclaw gateway start` takes 70-100+ seconds before the gateway is usable, causing the TUI to timeout
2. **TUI connection timeouts**:
```
history failed: Error: gateway request timeout for chat.history
sessions list failed: E... | closed | completed | true | 1 | [] | [] | 2026-04-19T02:29:50Z | 2026-04-28T04:50:20Z | 2026-04-25T18:41:54Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | JIRBOY | 49,422,297 | MDQ6VXNlcjQ5NDIyMjk3 | User | false |
openclaw/openclaw | 4,289,737,301 | I_kwDOQb6kR87_sDJV | 68,782 | https://github.com/openclaw/openclaw/issues/68782 | https://api.github.com/repos/openclaw/openclaw/issues/68782 | Add selective installation for plugins, skills, and channels — reduce install size and startup overhead | ## User Pain Point
OpenClaw currently uses an **all-or-nothing installation** strategy: every channel plugin (WhatsApp, Telegram, Discord, WeChat, Feishu, etc.), skill module, and API interface is installed regardless of whether the user needs them.
### Specific Impacts
1. **Bloated install size**: Unneeded plugins ... | open | null | false | 1 | [] | [] | 2026-04-19T02:30:15Z | 2026-05-02T20:26:15Z | null | CONTRIBUTOR | null | 20260502T210030Z | 2026-05-02T21:00:30Z | JIRBOY | 49,422,297 | MDQ6VXNlcjQ5NDIyMjk3 | User | false |
openclaw/openclaw | 4,289,751,872 | I_kwDOQb6kR87_sGtA | 68,784 | https://github.com/openclaw/openclaw/issues/68784 | https://api.github.com/repos/openclaw/openclaw/issues/68784 | [Bug]: native Codex app-server does not surface app approval/input prompts, and Codex Apps GitHub writes fail as "user rejected MCP tool call" | TL;DR: in a native Codex app-server session, Codex Apps GitHub reads worked, but GitHub write actions failed almost immediately as `user rejected MCP tool call` even though no approval/input prompt was shown.
## Bug type
Behavior bug (incorrect output/state without crash)
## Beta release blocker
No
## Summary
In a n... | closed | completed | true | 3 | [] | [] | 2026-04-19T02:40:04Z | 2026-04-25T04:15:54Z | 2026-04-23T01:30:50Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | kesslerio | 2,647,990 | MDQ6VXNlcjI2NDc5OTA= | User | false |
openclaw/openclaw | 4,289,777,074 | I_kwDOQb6kR87_sM2y | 68,786 | https://github.com/openclaw/openclaw/issues/68786 | https://api.github.com/repos/openclaw/openclaw/issues/68786 | Security audit false positive: plugins.allow_phantom_entries flags bundled stock plugins as phantom in CLI context | # Security audit false positive: `plugins.allow_phantom_entries` flags bundled stock plugins as phantom in CLI context
## Summary
`openclaw security audit --deep` can incorrectly warn that `plugins.allow` contains phantom entries even when those plugin IDs are valid bundled stock plugins and are visibly loaded in `ope... | closed | duplicate | true | 3 | [
"duplicate",
"dedupe:child"
] | [] | 2026-04-19T02:57:24Z | 2026-04-28T04:37:11Z | 2026-04-26T03:49:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fchaudhryspear | 198,639,243 | U_kgDOC9b-iw | User | false |
openclaw/openclaw | 4,289,781,709 | I_kwDOQb6kR87_sN_N | 68,790 | https://github.com/openclaw/openclaw/issues/68790 | https://api.github.com/repos/openclaw/openclaw/issues/68790 | Slack: internal replyToId can override valid thread anchor and leak replies to channel root | ## Summary
Slack replies can leak to the channel root when `[[reply_to_current]]` resolves to an internal OpenClaw message id instead of a real Slack `thread_ts`, and the Slack outbound path gives that `replyToId` priority over the known thread anchor.
This appears in inter-session / subagent-completion flows where t... | closed | completed | true | 2 | [] | [] | 2026-04-19T03:00:39Z | 2026-04-27T04:47:41Z | 2026-04-24T23:38:37Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jongjinkim61-cpu | 276,873,620 | U_kgDOEIDBlA | User | false |
openclaw/openclaw | 4,289,779,783 | I_kwDOQb6kR87_sNhH | 68,789 | https://github.com/openclaw/openclaw/issues/68789 | https://api.github.com/repos/openclaw/openclaw/issues/68789 | [Bug]: Slack WebClient path intermittently fails with `EAI_AGAIN slack.com` even when host DNS and Node HTTPS are healthy | ## Summary
On `OpenClaw 2026.4.14`, Slack delivery can fail intermittently with:
```text
Error: A request error occurred: getaddrinfo EAI_AGAIN slack.com
```
In this case the gateway **does not crash-loop**. The process stays up, but Slack sends fail and Socket Mode also shows repeated pong timeouts / stale-socket r... | closed | completed | true | 3 | [] | [] | 2026-04-19T02:59:21Z | 2026-05-04T04:34:00Z | 2026-05-02T04:16:14Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | sonnyb9 | 43,247,590 | MDQ6VXNlcjQzMjQ3NTkw | User | false |
openclaw/openclaw | 4,094,782,144 | I_kwDOQb6kR870EWrA | 49,791 | https://github.com/openclaw/openclaw/issues/49791 | https://api.github.com/repos/openclaw/openclaw/issues/49791 | Feishu group chat: @Bot /think medium is normalized correctly but returns no reply, while the same command works in p2p | ## Summary
In Feishu, `/think medium` works in p2p/direct chat, but fails in group chat even when the bot is explicitly mentioned.
The group-chat message appears to be normalized correctly to `/think medium`, so this does not look like the old mention-normalization issue alone. The same input works in p2p and returns... | open | null | false | 5 | [] | [] | 2026-03-18T12:21:38Z | 2026-05-04T20:14:15Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | TopMan-123 | 175,288,232 | U_kgDOCnKvqA | User | false |
openclaw/openclaw | 4,289,787,526 | I_kwDOQb6kR87_sPaG | 68,791 | https://github.com/openclaw/openclaw/issues/68791 | https://api.github.com/repos/openclaw/openclaw/issues/68791 | Native Windows regression on 2026.4.15: Scheduled Task Gateway unhealthy, agent turns hang / return incomplete-turn payloads=0, Telegram replies Agent couldn't generate a response | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
I’m seeing a native Windows issue after updating from 2026.4.14 to 2026.4.15.
On 2026.4.15, the Gateway Scheduled Task does not stay healthy, and even when I run the Gateway manually in the foreground, agent turns either ha... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-19T03:04:37Z | 2026-04-30T04:38:23Z | 2026-04-26T05:59:41Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | limpredator | 114,434,805 | U_kgDOBtIi9Q | User | false |
openclaw/openclaw | 4,289,794,463 | I_kwDOQb6kR87_sRGf | 68,793 | https://github.com/openclaw/openclaw/issues/68793 | https://api.github.com/repos/openclaw/openclaw/issues/68793 | qqbot: reduce default log verbosity; avoid logging full chat payloads at info | ## Summary
The bundled QQ Bot channel (`qqbot`) emits very chatty **info-level** logs during normal operation. Several log lines include **full inbound message text** and large **WebSocket dispatch JSON payloads** (for example `C2C_MESSAGE_CREATE` with `content`, full `d={...}`). This makes gateway logs hard to read an... | open | null | false | 1 | [] | [] | 2026-04-19T03:09:12Z | 2026-05-03T11:25:26Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | llzzww316 | 11,289,981 | MDQ6VXNlcjExMjg5OTgx | User | false |
openclaw/openclaw | 4,289,779,045 | I_kwDOQb6kR87_sNVl | 68,788 | https://github.com/openclaw/openclaw/issues/68788 | https://api.github.com/repos/openclaw/openclaw/issues/68788 | [Bug]: Windows: spawn claude ENOENT "claude" missing from cmdCommands shim allowlist | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
On Windows, OpenClaw fails to spawn the Claude CLI with `Error: spawn claude ENOENT`. Every model in the fallback chain fails. Root cause: `"claude"` is missing from the `cmdCommands` allowlist used by `resolve... | closed | not_planned | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-04-19T02:58:49Z | 2026-04-30T04:39:10Z | 2026-04-26T05:22:30Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | HaseebDev | 49,374,529 | MDQ6VXNlcjQ5Mzc0NTI5 | User | false |
openclaw/openclaw | 4,289,804,316 | I_kwDOQb6kR87_sTgc | 68,794 | https://github.com/openclaw/openclaw/issues/68794 | https://api.github.com/repos/openclaw/openclaw/issues/68794 | Control UI: context token counter inflates to >4× window (claude-cli) — cacheRead summed as fresh input | ## Summary
The Control UI context/token indicator reports values far exceeding the model's context window (observed: ~800k shown against a 200k Opus 4.7 window) after a handful of cache-hit turns on the `claude-cli` provider path.
## Version
- OpenClaw `2026.4.19-beta.1` (4862d34)
- Primary model: `claude-cli/claude-o... | closed | completed | true | 1 | [] | [] | 2026-04-19T03:15:15Z | 2026-04-28T04:50:20Z | 2026-04-25T18:42:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexprime1889-prog | 229,182,760 | U_kgDODakNKA | User | false |
openclaw/openclaw | 4,289,805,028 | I_kwDOQb6kR87_sTrk | 68,795 | https://github.com/openclaw/openclaw/issues/68795 | https://api.github.com/repos/openclaw/openclaw/issues/68795 | Control UI webchat loses all chat history on page reload — no history hydration | ## Summary
Reloading the Control UI webchat page (`/chat?session=<key>`) wipes the entire visible conversation. The server-side transcript is intact on disk; the UI simply never asks for it on load.
## Version
- OpenClaw `2026.4.19-beta.1` (4862d34)
- Host: macOS 15.x, Node 25.4.0, Chrome 140
## Evidence
After `locat... | closed | completed | true | 1 | [] | [] | 2026-04-19T03:15:40Z | 2026-04-28T04:50:19Z | 2026-04-25T18:42:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexprime1889-prog | 229,182,760 | U_kgDODakNKA | User | false |
openclaw/openclaw | 4,289,810,240 | I_kwDOQb6kR87_sU9A | 68,796 | https://github.com/openclaw/openclaw/issues/68796 | https://api.github.com/repos/openclaw/openclaw/issues/68796 | [Feature/Bug]: Expose undici connect.timeout for Ollama provider + make fallback decision consistent on reason=timeout | ## Summary
Two related asks affecting remote-Ollama setups:
1. **Feature**: Expose the undici `connect.timeout` (TCP connect) so operators running Ollama on a different host (LAN, tailnet) can tolerate slow first-byte responses. Today it is hardcoded to undici's default 10 s and is not reachable from `openclaw.json`.... | closed | completed | true | 1 | [] | [] | 2026-04-19T03:19:05Z | 2026-04-29T04:32:05Z | 2026-04-27T03:58:15Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | Juankcba | 66,220,814 | MDQ6VXNlcjY2MjIwODE0 | User | false |
openclaw/openclaw | 3,979,352,813 | I_kwDOQb6kR87tMBrt | 24,662 | https://github.com/openclaw/openclaw/issues/24662 | https://api.github.com/repos/openclaw/openclaw/issues/24662 | [Bug] Webchat image paste not sent to Agent | **Description**
When pasting an image into the webchat input box, the image appears locally in the browser but is not sent to the Agent. The Agent never receives the image content.
**Environment**
- OpenClaw: 2026.2.22-2
- macOS: 26.3
- Browser: Chrome (webchat at http://127.0.0.1:18789/chat)
**Steps to reproduce**
1... | closed | completed | true | 12 | [] | [] | 2026-02-23T17:35:24Z | 2026-04-27T05:04:15Z | 2026-04-24T19:07:27Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Tan-Kayon | 258,035,658 | U_kgDOD2FPyg | User | false |
openclaw/openclaw | 4,059,026,031 | I_kwDOQb6kR87x79Jv | 43,305 | https://github.com/openclaw/openclaw/issues/43305 | https://api.github.com/repos/openclaw/openclaw/issues/43305 | [Bug]: openai-responses fails with "Invalid input[n].id: string too long" after tool calls / session startup | ### Bug type
Regression (worked before, now fails)
### Summary
When OpenClaw is configured with api: "openai-responses" against an OpenAI-compatible proxy, some sessions fail immediately after tool calls or session startup with HTTP 400: Invalid 'input[n].id': string too long.... Switching the same provider to api: ... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-03-11T15:45:20Z | 2026-04-27T04:51:05Z | 2026-04-24T22:19:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | adirgan | 9,556,026 | MDQ6VXNlcjk1NTYwMjY= | User | false |
openclaw/openclaw | 4,173,487,305 | I_kwDOQb6kR874wlzJ | 57,898 | https://github.com/openclaw/openclaw/issues/57898 | https://api.github.com/repos/openclaw/openclaw/issues/57898 | [Bug]: ARE YOU GOING TO FRIGGING KEEP BREAKING THIS THING OR WE CAN GET A STABLE VERSION??????? | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
v2026.3.28
Few weeks ago I complained that the latest update broke tool usage and the fix was to modify some idiotic setting to "profile": full or something.
The next version fixed it. Now it's broken again..... Are you st... | closed | completed | true | 16 | [
"bug",
"regression"
] | [] | 2026-03-30T19:27:02Z | 2026-04-28T04:58:40Z | 2026-04-25T17:07:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | paciox | 6,546,735 | MDQ6VXNlcjY1NDY3MzU= | User | false |
openclaw/openclaw | 4,289,838,083 | I_kwDOQb6kR87_sbwD | 68,805 | https://github.com/openclaw/openclaw/issues/68805 | https://api.github.com/repos/openclaw/openclaw/issues/68805 | [Bug] openclaw-weixin: outbound push reports success but never delivers to personal WeChat accounts | **Plugin:** @tencent/openclaw-weixin
**Environment:** macOS 25.4.0, OpenClaw latest, Node v22.16.0
**Account Type:** Personal WeChat (not 企业微信), bot_type=3
**baseUrl:** https://ilinkai.weixin.qq.com
**Account ID:** e97a6e609973-im-bot
**User ID:** o9cq806I1a0W_IZksCrTaeiCf6gk@im.wechat
---
## Bug Description
Outboun... | closed | not_planned | true | 3 | [] | [] | 2026-04-19T03:38:46Z | 2026-04-28T04:50:17Z | 2026-04-25T18:42:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Sanosika | 118,650,972 | U_kgDOBxJ4XA | User | false |
openclaw/openclaw | 4,289,817,211 | I_kwDOQb6kR87_sWp7 | 68,799 | https://github.com/openclaw/openclaw/issues/68799 | https://api.github.com/repos/openclaw/openclaw/issues/68799 | 对话持久化模块(walProtocol.ts)未被运行时调用 | ## 问题描述
workspace目录下写好了对话持久化模块 `src/walProtocol.ts`，包含WAL协议和`appendConversationLog()`函数，但OpenClaw运行时没有调用这些函数，导致对话无法实时写入`conversations/`目录。
## 复现步骤
1. 在workspace编写并保存了walProtocol.ts和conversationPersistence.ts
2. 对话结束后，检查`.openclaw/conversations/`目录为空
3. 对话内容未持久化
## 预期行为
对话应该实时写入`.openclaw/conversations/YYYY-MM-DD.md`
... | closed | not_planned | true | 1 | [] | [] | 2026-04-19T03:23:59Z | 2026-04-28T04:50:18Z | 2026-04-25T18:42:20Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Pasaidon | 46,835,765 | MDQ6VXNlcjQ2ODM1NzY1 | User | false |
openclaw/openclaw | 4,289,857,299 | I_kwDOQb6kR87_sgcT | 68,812 | https://github.com/openclaw/openclaw/issues/68812 | https://api.github.com/repos/openclaw/openclaw/issues/68812 | Make memory embedding retry/concurrency parameters configurable | ## Problem
Providers like Gemini have strict per-minute (RPM) rate limits on embedding requests. Currently, three memory embedding parameters are **hardcoded constants** in the bundled manager JS:
- `EMBEDDING_INDEX_CONCURRENCY` (default: 4) — indexing concurrency
- `EMBEDDING_RETRY_BASE_DELAY_MS` (default: 500) — ba... | open | null | false | 1 | [] | [] | 2026-04-19T03:51:12Z | 2026-05-02T15:23:01Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | lubao515 | 14,576,356 | MDQ6VXNlcjE0NTc2MzU2 | User | false |
openclaw/openclaw | 4,289,856,243 | I_kwDOQb6kR87_sgLz | 68,811 | https://github.com/openclaw/openclaw/issues/68811 | https://api.github.com/repos/openclaw/openclaw/issues/68811 | [Feature]: Real-time console/activity stream in webchat while awaiting response | ## Problem
When a user sends a message in the OpenClaw webchat UI, there is no visual feedback while the agent is processing the request. The chat appears completely frozen with no spinner, no progress indicator, nothing until the full response arrives.
This creates a poor UX because:
- Users cannot tell if the syst... | closed | completed | true | 1 | [] | [] | 2026-04-19T03:50:36Z | 2026-04-28T04:50:15Z | 2026-04-25T18:42:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | geek111 | 6,862,522 | MDQ6VXNlcjY4NjI1MjI= | User | false |
openclaw/openclaw | 4,289,813,221 | I_kwDOQb6kR87_sVrl | 68,797 | https://github.com/openclaw/openclaw/issues/68797 | https://api.github.com/repos/openclaw/openclaw/issues/68797 | [Bug]: BUG_REPORT_subagent_context_leakage.md | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The OpenClaw orchestration layer failed to sanitize the `Subagent Context` block before injecting it into a new agent session. This resulted in the agent receiving a "ghost task" (creating a weather-image skill... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-19T03:21:10Z | 2026-04-30T04:39:09Z | 2026-04-26T05:22:35Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | cgiakmoz | 40,509,550 | MDQ6VXNlcjQwNTA5NTUw | User | false |
openclaw/openclaw | 4,289,852,010 | I_kwDOQb6kR87_sfJq | 68,809 | https://github.com/openclaw/openclaw/issues/68809 | https://api.github.com/repos/openclaw/openclaw/issues/68809 | MCP child process leak: disposeSession() doesn't kill process groups | ## Bug
openclaw-gateway leaks MCP child processes. Over 2+ days of uptime, my instance accumulated ~840 orphan processes (~5 GB RAM). After a restart, it hit 28 orphans in just 31 minutes.
## Root Cause
`disposeSession()` in `model-context-tokens-z5hvDVkk.js` (line ~1963) calls:
```js
session.process.kill();
```
T... | closed | completed | true | 1 | [] | [] | 2026-04-19T03:48:21Z | 2026-04-25T04:16:41Z | 2026-04-22T21:00:13Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | wiggdevin | 202,901,685 | U_kgDODBgItQ | User | false |
openclaw/openclaw | 4,289,826,229 | I_kwDOQb6kR87_sY21 | 68,804 | https://github.com/openclaw/openclaw/issues/68804 | https://api.github.com/repos/openclaw/openclaw/issues/68804 | [Bug]: LINE channel status shows false warning despite valid token configured | ## Bug Description
LINE channel shows warning: "LINE channel access token not configured" even though the token is correctly configured and fully functional.
## Steps to Reproduce
1. Configure LINE channel with channelAccessToken and channelSecret in config
2. Run `openclaw status`
3. Observe LINE shows WARN despite... | closed | not_planned | true | 1 | [] | [] | 2026-04-19T03:30:49Z | 2026-04-28T04:50:17Z | 2026-04-25T18:42:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kimuraproart-a11y | 267,546,251 | U_kgDOD_Juiw | User | false |
openclaw/openclaw | 4,289,854,283 | I_kwDOQb6kR87_sftL | 68,810 | https://github.com/openclaw/openclaw/issues/68810 | https://api.github.com/repos/openclaw/openclaw/issues/68810 | [Bug]: Windows QClaw gateway exit code 1, WebSocket message delivery failure, crash dumps not uploaded | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
QClaw gateway on Windows 11 exits with code 1 on all CLI commands. WeChat channel message delivery intermittently fails with "This operation was aborted". 5 Crashpad crash dumps (33MB each) have not been uploaded to the crash s... | closed | not_planned | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-04-19T03:49:35Z | 2026-04-28T04:50:16Z | 2026-04-25T18:42:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 15285701683 | 235,316,189 | U_kgDODgaj3Q | User | false |
openclaw/openclaw | 4,289,886,847 | I_kwDOQb6kR87_snp_ | 68,815 | https://github.com/openclaw/openclaw/issues/68815 | https://api.github.com/repos/openclaw/openclaw/issues/68815 | [Bug]: Comprehensive issues: CLI exit code 1, memory leak, crash dumps not uploaded, skill-interceptor auto-recovery | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
Multiple critical issues in QClaw 2026.3.13 on Windows: CLI returns exit code 1 on every command, 5 crash dumps (~33MB each) not uploaded for 2 weeks, memory leak ~1.5GB, and skill-interceptor auto-reenables after every update ... | open | null | false | 2 | [
"bug",
"bug:crash"
] | [] | 2026-04-19T04:09:29Z | 2026-05-03T11:25:27Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | 15285701683 | 235,316,189 | U_kgDODgaj3Q | User | false |
openclaw/openclaw | 3,909,178,301 | I_kwDOQb6kR87pAVO9 | 10,841 | https://github.com/openclaw/openclaw/issues/10841 | https://api.github.com/repos/openclaw/openclaw/issues/10841 | Reminders set for wrong times because agent doesn't know current time | ## Problem
When you ask OpenClaw to "remind me in 30 minutes" or "set an alarm for 3pm", it often gets the time wrong. The agent guesses the current time instead of knowing it precisely.
## Why this happens
The cron scheduler requires exact timestamps (ISO-8601 format), but the agent only sees the timezone in its sy... | closed | not_planned | true | 22 | [
"bug",
"stale"
] | [] | 2026-02-07T02:47:16Z | 2026-04-26T04:29:04Z | 2026-04-24T04:24:56Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | jeanlucthumm | 4,934,853 | MDQ6VXNlcjQ5MzQ4NTM= | User | false |
openclaw/openclaw | 3,941,919,202 | I_kwDOQb6kR87q9Oni | 16,357 | https://github.com/openclaw/openclaw/issues/16357 | https://api.github.com/repos/openclaw/openclaw/issues/16357 | Feature Request: Support prompt_cache_key metadata for provider-side cache routing | ## Problem
When running multiple sessions (e.g., 18+ sessions), identical or similar prompts may be routed to different GPU clusters by OpenRouter, causing cache misses and higher latency/cost.
## Proposed Solution
Expose `prompt_cache_key` (or similar mechanism) in configuration to allow users to:
1. Route similar... | closed | not_planned | true | 13 | [
"stale"
] | [] | 2026-02-14T17:33:18Z | 2026-04-22T04:32:11Z | 2026-04-19T04:22:52Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | lumenclaw-cloud | 260,919,617 | U_kgDOD41RQQ | User | false |
openclaw/openclaw | 3,970,864,356 | I_kwDOQb6kR87srpTk | 22,282 | https://github.com/openclaw/openclaw/issues/22282 | https://api.github.com/repos/openclaw/openclaw/issues/22282 | feat: Proactive model fallback based on rate limit headers (preemptive 429 avoidance) | ## Problem
When a provider hits a rate limit (429), OpenClaw falls back to the next configured model/profile. But by the time you get a 429, it's too late — the user's message fails, and in interactive sessions (Telegram, Discord), the user **cannot send a follow-up message to manually switch models** because the sess... | closed | not_planned | true | 8 | [
"stale"
] | [] | 2026-02-21T00:08:35Z | 2026-04-22T04:32:08Z | 2026-04-19T04:23:01Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | clawrl3000 | 258,632,274 | U_kgDOD2pqUg | User | false |
openclaw/openclaw | 3,998,639,285 | I_kwDOQb6kR87uVmS1 | 28,143 | https://github.com/openclaw/openclaw/issues/28143 | https://api.github.com/repos/openclaw/openclaw/issues/28143 | [Bug] Slow gateway restart/shutdown from stale paired node probe timeouts | ## Description
Gateway restart/shutdown becomes slow and unreliable when a paired node is unresponsive. Remote skill bin probes block the restart path, causing 3-4s delays per probe attempt. Shutdown times out repeatedly.
## Steps to Reproduce
1. Pair a remote node (e.g., GUI app registers as paired device)
2. Make pa... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-02-27T02:03:03Z | 2026-04-22T04:32:06Z | 2026-04-19T04:23:12Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,004,032,403 | I_kwDOQb6kR87uqK-T | 29,506 | https://github.com/openclaw/openclaw/issues/29506 | https://api.github.com/repos/openclaw/openclaw/issues/29506 | Slack: missing_scope error when using channel name (works with channel ID) | ## Description
When using the Slack channel with a channel name (e.g., `all-weis-1m-arr-journey`), OpenClaw returns a `missing_scope` error for both reading and sending messages. However, using the channel ID directly (e.g., `C0AHNJRJSMC`) works perfectly.
## Steps to Reproduce
1. Have a Slack app with the correct s... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-02-28T06:27:39Z | 2026-04-22T04:32:04Z | 2026-04-19T04:23:20Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | remodelai | 134,542,351 | U_kgDOCAT0Dw | User | false |
openclaw/openclaw | 4,004,145,360 | I_kwDOQb6kR87uqmjQ | 29,570 | https://github.com/openclaw/openclaw/issues/29570 | https://api.github.com/repos/openclaw/openclaw/issues/29570 | Feature Request: Heartbeat task batching and HEARTBEAT.md documentation improvements | ## Problem
The current heartbeat system has two pain points:
### 1. No built-in batching mechanism
Users who want to run multiple periodic checks (email, calendar, weather, system health) must either:
- Combine everything into one giant heartbeat prompt (messy, hard to maintain)
- Create multiple cron jobs (increases... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-02-28T07:43:48Z | 2026-04-22T04:32:03Z | 2026-04-19T04:23:22Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | zycaskevin | 223,135,116 | U_kgDODUzFjA | User | false |
openclaw/openclaw | 4,004,153,193 | I_kwDOQb6kR87uqodp | 29,573 | https://github.com/openclaw/openclaw/issues/29573 | https://api.github.com/repos/openclaw/openclaw/issues/29573 | Feature: openclaw-mem0 — allow separate LLM config for fact extraction vs. agent chat | ## Problem
The `openclaw-mem0` plugin uses `oss.llm` for fact extraction during `autoCapture`. In most OSS deployments, this is a small local model (e.g. `llama3.2:3b`) that's well-suited for quick conversational inference but lacks the instruction-following reliability needed for nuanced fact extraction.
The result:... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-02-28T07:49:53Z | 2026-04-22T04:32:02Z | 2026-04-19T04:23:23Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Paul-Kyle | 155,019,651 | U_kgDOCT1pgw | User | false |
openclaw/openclaw | 4,004,533,855 | I_kwDOQb6kR87usFZf | 29,727 | https://github.com/openclaw/openclaw/issues/29727 | https://api.github.com/repos/openclaw/openclaw/issues/29727 | [Bug]: All Sub-Agents Return "allowed: none" Despite Valid Configuration | ### Summary
## Summary
All sub-agents (worker, researcher, media-clipper, esg-analyst, gis-analyzer) fail to spawn with error `forbidden: agentId is not allowed for sessions_spawn (allowed: none)`, despite being properly configured in `openclaw.json`.
## Environment
- **OpenClaw Version:** 2026.2.26
- **OS:** macOS ... | closed | not_planned | true | 6 | [
"bug",
"stale"
] | [] | 2026-02-28T11:56:13Z | 2026-04-22T04:32:00Z | 2026-04-19T04:23:26Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | tamasking | 33,271,990 | MDQ6VXNlcjMzMjcxOTkw | User | false |
openclaw/openclaw | 4,005,300,081 | I_kwDOQb6kR87uvAdx | 30,055 | https://github.com/openclaw/openclaw/issues/30055 | https://api.github.com/repos/openclaw/openclaw/issues/30055 | Feature: Explicit OAuth vs API key selection in model routing + fallback chain | ## Summary
Allow explicit selection of auth method (OAuth vs API key) per model in the primary model config and fallback chain, with visibility into which auth method was used per request.
## Current Behavior
- Provider entries (e.g. `openai`, `anthropic`) silently use whichever auth profile is available
- No way to... | closed | not_planned | true | 8 | [
"stale"
] | [] | 2026-02-28T20:13:59Z | 2026-04-22T04:31:59Z | 2026-04-19T04:23:30Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | adroidian | 7,563,431 | MDQ6VXNlcjc1NjM0MzE= | User | false |
openclaw/openclaw | 4,006,058,630 | I_kwDOQb6kR87ux5qG | 30,243 | https://github.com/openclaw/openclaw/issues/30243 | https://api.github.com/repos/openclaw/openclaw/issues/30243 | Isolated cron sessions have no memory bridge back to originating conversation | ## Problem
When a user sets up a cron job in a conversation (e.g. a Telegram topic), the cron runs in an isolated session. The isolated session produces output and delivers it (e.g. DM to user). But when the user later asks about the output **in the original conversation**, the conversation session has zero context — ... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-01T02:52:57Z | 2026-04-22T04:31:59Z | 2026-04-19T04:23:32Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | 4ier | 5,648,066 | MDQ6VXNlcjU2NDgwNjY= | User | false |
openclaw/openclaw | 4,007,004,121 | I_kwDOQb6kR87u1gfZ | 30,604 | https://github.com/openclaw/openclaw/issues/30604 | https://api.github.com/repos/openclaw/openclaw/issues/30604 | Followup queue delivers same message multiple times when agent is busy | ## Bug
When an inbound message arrives while the agent is busy (long-running tool calls), the followup queue can deliver the same message to the agent multiple times across successive drain cycles.
### Steps to reproduce
1. Agent is busy with a long-running task (e.g., background exec generating files for ~30 minute... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-03-01T11:47:05Z | 2026-04-22T04:29:25Z | 2026-04-19T04:23:38Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | CyglerAI | 257,286,777 | U_kgDOD1XieQ | User | false |
openclaw/openclaw | 4,012,524,968 | I_kwDOQb6kR87vKkWo | 31,993 | https://github.com/openclaw/openclaw/issues/31993 | https://api.github.com/repos/openclaw/openclaw/issues/31993 | feat: Reorder system prompt blocks for better Anthropic prompt cache hit rates | ## Problem
Anthropic's prompt caching is prefix-based — any change in the early system blocks invalidates the entire cache. Currently, OpenClaw places dynamic content (current timestamp, runtime metadata, session state) early in the system prompt, before static blocks like SOUL.md, AGENTS.md, tool definitions, and ski... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-02T17:50:38Z | 2026-04-22T04:31:56Z | 2026-04-19T04:23:49Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | dambrajoe-hue | 264,535,157 | U_kgDOD8R8dQ | User | false |
openclaw/openclaw | 4,017,308,405 | I_kwDOQb6kR87vc0L1 | 33,179 | https://github.com/openclaw/openclaw/issues/33179 | https://api.github.com/repos/openclaw/openclaw/issues/33179 | [Bug]: Control UI reprocesses historical images on session:history (repeated agents/tool-images “Image resized to fit limits” logs) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
# Bug report — `agents/tool-images` resize triggered by `session:history` in Control UI
## Summary
In OpenClaw Control UI (webchat), log lines from `agents/tool-images` (`Image resized to fit limits`) are repeatedly emitted even when the u... | closed | not_planned | true | 6 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-03T15:30:14Z | 2026-04-26T04:29:04Z | 2026-04-24T04:25:31Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | vangelismosca | 204,862,609 | U_kgDODDX0kQ | User | false |
openclaw/openclaw | 4,019,294,004 | I_kwDOQb6kR87vkY80 | 33,642 | https://github.com/openclaw/openclaw/issues/33642 | https://api.github.com/repos/openclaw/openclaw/issues/33642 | Feature request: policy-based proactive compaction (time + activity + threshold triggers) | ## Summary
OpenClaw compaction is currently great for overflow-driven safety, but there is no first-class way to define **proactive compaction policy** (e.g., compact every N hours only if session is active, or compact at lower soft/hard thresholds before overflow pressure).
For long-running production sessions this c... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T23:58:23Z | 2026-04-22T04:31:55Z | 2026-04-19T04:24:04Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,019,318,421 | I_kwDOQb6kR87vke6V | 33,650 | https://github.com/openclaw/openclaw/issues/33650 | https://api.github.com/repos/openclaw/openclaw/issues/33650 | [Bug]: Chrome/Brave extension relay fails to attach: Target.attachToBrowserTarget "Not allowed" (tab list works, control fails) since 2/20 | ### Bug type
Regression (worked before, now fails)
### Summary
The OpenClaw Browser Relay extension can reach the relay and list tabs, but any control action fails with:
browserContext.newCDPSession: Protocol error (Target.attachToBrowserTarget): {"code":-32000,"message":"Not allowed"}
This reproduces on normal ta... | closed | not_planned | true | 3 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T00:08:08Z | 2026-04-22T04:31:54Z | 2026-04-19T04:24:06Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | jacksonh94 | 17,287,428 | MDQ6VXNlcjE3Mjg3NDI4 | User | false |
openclaw/openclaw | 4,019,326,205 | I_kwDOQb6kR87vkgz9 | 33,654 | https://github.com/openclaw/openclaw/issues/33654 | https://api.github.com/repos/openclaw/openclaw/issues/33654 | Feature: Notify on model fallback | ## Problem
When the primary model (e.g., Anthropic Opus) hits a rate limit or temporary outage and OpenClaw silently falls back to a fallback model, the user has no way to know this happened unless they check `/status` after the fact. This makes it difficult to understand unexpected behavior or track usage across prov... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-04T00:11:17Z | 2026-04-26T04:29:03Z | 2026-04-24T04:25:37Z | NONE | null | 20260426T060034Z | 2026-04-26T06:00:34Z | marcxaurelius | 209,353,012 | U_kgDODHp5NA | User | false |
openclaw/openclaw | 4,019,368,073 | I_kwDOQb6kR87vkrCJ | 33,671 | https://github.com/openclaw/openclaw/issues/33671 | https://api.github.com/repos/openclaw/openclaw/issues/33671 | Sending PDF document via Telegram causes Gateway process to hang | ## Description
Sending a PDF file (~16.7 MB) to a Telegram bot agent causes the Gateway process to become unresponsive. After the PDF is sent, the affected agent session stops responding to all subsequent messages (text included). The issue persists until the session is manually reset.
## Environment
- **OS:** macOS... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-04T00:27:52Z | 2026-04-22T04:31:54Z | 2026-04-19T04:24:09Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | kylin19860916 | 229,595,532 | U_kgDODa9ZjA | User | false |
openclaw/openclaw | 4,019,387,581 | I_kwDOQb6kR87vkvy9 | 33,677 | https://github.com/openclaw/openclaw/issues/33677 | https://api.github.com/repos/openclaw/openclaw/issues/33677 | Heartbeat with messageChannel creates unbounded followup run chains | ## Summary
When `heartbeat.target` is set (e.g., `"imessage"`), heartbeat runs execute with `messageChannel=imessage`. After each run completes, OpenClaw creates a followup run because the main session has no response (output was routed to the target channel). Each followup run sees the same minimal context and indepe... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T00:34:37Z | 2026-04-22T04:31:53Z | 2026-04-19T04:24:11Z | CONTRIBUTOR | null | 20260422T060039Z | 2026-04-22T06:00:39Z | jaden-clovervnd | 91,520,439 | U_kgDOBXR9tw | User | false |
openclaw/openclaw | 4,019,397,204 | I_kwDOQb6kR87vkyJU | 33,679 | https://github.com/openclaw/openclaw/issues/33679 | https://api.github.com/repos/openclaw/openclaw/issues/33679 | Feature request: compaction.onFailure policy (halt session when compaction fails) | ## Problem
When compaction times out or fails (safeguard mode chunked summarization), the session continues running with broken/bloated context. The agent keeps accepting messages but every subsequent model call either fails (dangling tool_use blocks, Anthropic validation errors) or runs on degraded context, burning t... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-04T00:38:45Z | 2026-04-22T04:31:52Z | 2026-04-19T04:24:12Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | cygnostik | 5,272,292 | MDQ6VXNlcjUyNzIyOTI= | User | false |
openclaw/openclaw | 4,019,401,858 | I_kwDOQb6kR87vkzSC | 33,682 | https://github.com/openclaw/openclaw/issues/33682 | https://api.github.com/repos/openclaw/openclaw/issues/33682 | message send accepts buffer params but outbound send does not hydrate them into media | ## Summary
`message action=send` exposes `buffer`, `filename`, and `contentType` in its public params, but the outbound send path does not currently hydrate buffered payloads into `media`.
As a result, channels that rely on the normal `media/mediaUrl` send flow do not receive a real attachment input when callers use ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T00:40:48Z | 2026-04-22T04:31:51Z | 2026-04-19T04:24:14Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | raw34 | 1,996,768 | MDQ6VXNlcjE5OTY3Njg= | User | false |
openclaw/openclaw | 4,019,405,137 | I_kwDOQb6kR87vk0FR | 33,685 | https://github.com/openclaw/openclaw/issues/33685 | https://api.github.com/repos/openclaw/openclaw/issues/33685 | Bug Report: setup-podman.sh fails on Fedora Silverblue | ### Bug type
Regression (worked before, now fails)
### Summary
Bug Report: setup-podman.sh fails on Fedora Silverblue due to SELinux, UID mapping, and EnvironmentFile permission conflicts
### Steps to reproduce
Steps to Reproduce
Run setup-podman.sh --quadlet on Fedora Silverblue
Start the service: sudo systemctl... | closed | not_planned | true | 4 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-04T00:42:17Z | 2026-04-22T04:31:51Z | 2026-04-19T04:24:16Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | ronbarbosa | 1,733,836 | MDQ6VXNlcjE3MzM4MzY= | User | false |
openclaw/openclaw | 4,019,483,557 | I_kwDOQb6kR87vlHOl | 33,704 | https://github.com/openclaw/openclaw/issues/33704 | https://api.github.com/repos/openclaw/openclaw/issues/33704 | Plugin loader missing jiti alias for plugin-sdk/keyed-async-queue subpath | ## Bug
The Matrix extension fails to load with:
```
Cannot find module '/path/to/dist/plugin-sdk/index.js/keyed-async-queue'
```
## Root Cause
The jiti plugin loader in `src/plugins/loader.ts` registers aliases for `openclaw/plugin-sdk` and `openclaw/plugin-sdk/account-id`, but **not** for `openclaw/plugin-sdk/keye... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-04T01:09:24Z | 2026-04-22T04:31:48Z | 2026-04-19T04:24:17Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | BestJoester | 89,434,888 | MDQ6VXNlcjg5NDM0ODg4 | User | false |
openclaw/openclaw | 4,019,546,758 | I_kwDOQb6kR87vlWqG | 33,730 | https://github.com/openclaw/openclaw/issues/33730 | https://api.github.com/repos/openclaw/openclaw/issues/33730 | [Bug]: 为啥会这样啊 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
open claw 安装成功后 开始设置的时候出错了
[openclaw] Unhandled promise rejection: TypeError: Invalid URL
### Steps to reproduce
运行这个`iwr -useb https://openclaw.ai/install.ps1 | iex`
Starting setup...
🦞 OpenClaw 2026.3.2 (85377a2) — I'm the assistant ... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-04T01:32:41Z | 2026-04-22T04:31:48Z | 2026-04-19T04:24:19Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | yapeng-wyp | 84,566,810 | MDQ6VXNlcjg0NTY2ODEw | User | false |
openclaw/openclaw | 4,019,554,977 | I_kwDOQb6kR87vlYqh | 33,736 | https://github.com/openclaw/openclaw/issues/33736 | https://api.github.com/repos/openclaw/openclaw/issues/33736 | feat: 支持飞书语音消息发送 (Feishu audio message support) | ## 功能描述
希望在飞书 channel 中支持发送语音消息，而不是作为文件附件。
## 使用场景
- TTS 生成的语音可以直接作为语音气泡发送，用户无需下载即可播放
- 情报播报、语音提醒等场景更加友好
## 技术方案
飞书语音消息需要：
1. 音频格式转换为 OPUS：`ffmpeg -i input.mp3 -acodec libopus -ac 1 -ar 16000 output.opus`
2. 上传文件获取 file_key：`POST /open-apis/im/v1/files` (file_type=opus)
3. 发送消息：`msg_type: audio`, content 包含 file_k... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-04T01:35:40Z | 2026-04-22T04:31:47Z | 2026-04-19T04:24:21Z | NONE | null | 20260422T060039Z | 2026-04-22T06:00:39Z | hadiliu | 1,926,042 | MDQ6VXNlcjE5MjYwNDI= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.