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 | 4,052,208,043 | I_kwDOQb6kR87xh8mr | 42,229 | https://github.com/openclaw/openclaw/issues/42229 | https://api.github.com/repos/openclaw/openclaw/issues/42229 | [Bug]: Can't use exec with group:openclaw in gateway with gpt-5.4 | ### Bug type
Regression (worked before, now fails)
### Summary
when i changed in to openai-codex/gpt-5.4 i find that i can't use the exec tools even i used group:openclaw.
### Steps to reproduce
1.set the allow like this:
"tools": {
"profile": "full",
"allow": [
"group:openclaw","exec"
],
"fs... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-10T14:30:17Z | 2026-04-28T04:41:54Z | 2026-04-26T02:50:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | thestrongerthebetter | 166,291,519 | U_kgDOCeloPw | User | false |
openclaw/openclaw | 4,052,229,201 | I_kwDOQb6kR87xiBxR | 42,232 | https://github.com/openclaw/openclaw/issues/42232 | https://api.github.com/repos/openclaw/openclaw/issues/42232 | Feature: Add model parameter to chat.send RPC for per-session model selection | ## Problem
Web UI clients using `chat.send` RPC have no way to set the model for a session programmatically. The `/model` command only works through channel plugin message handlers (Telegram, Discord), not through `chat.send` RPC — the message is passed to the agent as text instead of being intercepted as a command.
... | closed | completed | true | 1 | [] | [] | 2026-03-10T14:33:51Z | 2026-04-28T04:41:54Z | 2026-04-26T02:50:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | coreyjadin-clawd | 257,815,394 | U_kgDOD13zYg | User | false |
openclaw/openclaw | 4,052,232,551 | I_kwDOQb6kR87xiCln | 42,233 | https://github.com/openclaw/openclaw/issues/42233 | https://api.github.com/repos/openclaw/openclaw/issues/42233 | sessions_send frequently times out (status=timeout) and lacks delivery confirmation | @- | closed | completed | true | 3 | [] | [] | 2026-03-10T14:34:24Z | 2026-04-28T04:41:53Z | 2026-04-26T02:50:37Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fonhal | 4,468,037 | MDQ6VXNlcjQ0NjgwMzc= | User | false |
openclaw/openclaw | 4,052,347,980 | I_kwDOQb6kR87xiexM | 42,251 | https://github.com/openclaw/openclaw/issues/42251 | https://api.github.com/repos/openclaw/openclaw/issues/42251 | Session tools: allow tree-scoped access to creator-owned ACP sessions | ## Problem
When `tools.sessions.visibility="tree"`, session tools do not consistently allow the current session to access ACP sessions that it created itself.
This shows up most clearly in:
- `sessions_send(label + agentId)`
- sandboxed flows
- `sessions_list` when the default broad gateway list does not naturally in... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T14:53:41Z | 2026-04-28T04:41:52Z | 2026-04-26T02:50:41Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Wangnov | 48,670,012 | MDQ6VXNlcjQ4NjcwMDEy | User | false |
openclaw/openclaw | 4,329,872,567 | I_kwDOQb6kR88AAAABAhSctw | 71,906 | https://github.com/openclaw/openclaw/issues/71906 | https://api.github.com/repos/openclaw/openclaw/issues/71906 | [Bug] Stale ACPX custom-path override bypasses bundled runtime-deps staging and breaks `openclaw agent` in 2026.4.24 | ## Summary
After updating to `OpenClaw 2026.4.24`, `openclaw agent` failed during CLI bootstrap with:
```text
[plugins] acpx failed to load from /opt/homebrew/lib/node_modules/openclaw/dist/extensions/acpx/index.js: Error: Cannot find module 'acpx/runtime'
Require stack:
- /opt/homebrew/lib/node_modules/openclaw/dist... | closed | completed | true | 1 | [] | [] | 2026-04-26T02:50:42Z | 2026-04-28T04:36:37Z | 2026-04-26T03:54:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | riosbotchen-source | 263,091,329 | U_kgDOD650gQ | User | false |
openclaw/openclaw | 4,052,420,774 | I_kwDOQb6kR87xiwim | 42,266 | https://github.com/openclaw/openclaw/issues/42266 | https://api.github.com/repos/openclaw/openclaw/issues/42266 | [Bug]: iMessage inbound reply/quote context not parsed - missing reply_to fields | ### Bug Report
**Surface:** iMessage (macOS)
**Version:** 2026.3.8
### Description
When a user sends a reply/quoted message via iMessage, OpenClaw receives the message but cannot access the quoted message content. The `inbound_meta` context only contains basic fields (`message_id`, `sender_id`, `timestamp`, `was_men... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T15:05:29Z | 2026-04-28T04:41:51Z | 2026-04-26T02:50:45Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | tianzhanda | 85,002,033 | MDQ6VXNlcjg1MDAyMDMz | User | false |
openclaw/openclaw | 4,052,447,187 | I_kwDOQb6kR87xi2_T | 42,270 | https://github.com/openclaw/openclaw/issues/42270 | https://api.github.com/repos/openclaw/openclaw/issues/42270 | [Bug]: OpenClaw 2026.3.8 regression: empty agent payloads + websocket 1006 closures with local LM Studio backend | ### Bug type
Regression (worked before, now fails)
### Summary
## Summary
After upgrading from OpenClaw 2026.3.7 to 2026.3.8, agent responses intermittently fail.
Observed failure signatures include:
• stopReason=stop with empty payload
• stopReason=length with empty payload
• stopReason=error with "LLM request... | closed | completed | true | 11 | [
"bug",
"regression"
] | [] | 2026-03-10T15:09:41Z | 2026-04-28T04:41:51Z | 2026-04-26T02:50:49Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | SamaritanOC | 262,858,916 | U_kgDOD6ropA | User | false |
openclaw/openclaw | 4,052,651,141 | I_kwDOQb6kR87xjoyF | 42,300 | https://github.com/openclaw/openclaw/issues/42300 | https://api.github.com/repos/openclaw/openclaw/issues/42300 | [Bug]: `openclaw cron runs <jobId>` position arg silently ignored — requires `--id` flag not documented in help | ## Environment
- OpenClaw: v2026.3.8
- OS: macOS 24.6.0 (x64)
- Channel: Feishu
## What happened
When running `openclaw cron runs <jobId>` with the job ID as a positional argument, the command silently falls back to listing all runs (returns empty) instead of filtering by job.
The correct usage requires `--id <jobId... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T15:44:56Z | 2026-04-28T04:41:50Z | 2026-04-26T02:50:54Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Hollychou924 | 128,659,251 | U_kgDOB6svMw | User | false |
openclaw/openclaw | 4,053,016,425 | I_kwDOQb6kR87xlB9p | 42,363 | https://github.com/openclaw/openclaw/issues/42363 | https://api.github.com/repos/openclaw/openclaw/issues/42363 | TypeError: value.includes is not a function in containsEnvVarReference | # Bug Report: TypeError in containsEnvVarReference
## Summary
`openclaw doctor` crashes with `TypeError: value.includes is not a function` when processing config files containing numeric or boolean values.
## Environment
- **OpenClaw Version:** 2026.3.8 (from config `lastTouchedVersion`)
- **Node Version:** 25.6.1
... | closed | completed | true | 1 | [] | [] | 2026-03-10T16:55:25Z | 2026-04-28T04:41:49Z | 2026-04-26T02:50:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | TrevorHinesley | 1,079,481 | MDQ6VXNlcjEwNzk0ODE= | User | false |
openclaw/openclaw | 4,053,084,259 | I_kwDOQb6kR87xlShj | 42,376 | https://github.com/openclaw/openclaw/issues/42376 | https://api.github.com/repos/openclaw/openclaw/issues/42376 | [Bug] QQchat Plugin - Session Aborted Issue with Multi-Instance NapCat Setup | ## Summary
When using the QQchat plugin with multiple NapCat instances (multi-instance mode), incoming messages are received but outgoing replies frequently fail with "Request was aborted" error. Even in single-instance mode, the issue persists intermittently.
## Environment
- **OpenClaw Version**: Latest (via npm)
- ... | closed | completed | true | 1 | [] | [] | 2026-03-10T17:07:51Z | 2026-04-28T04:41:48Z | 2026-04-26T02:51:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | zhqfdn | 25,156,863 | MDQ6VXNlcjI1MTU2ODYz | User | false |
openclaw/openclaw | 4,053,194,696 | I_kwDOQb6kR87xltfI | 42,401 | https://github.com/openclaw/openclaw/issues/42401 | https://api.github.com/repos/openclaw/openclaw/issues/42401 | Feature: expose provider rate-limit headroom in models status / probe | ## Summary
When running `openclaw models status --probe`, the gateway already makes live API calls to each configured provider. The provider responses include rate-limit headers (`x-ratelimit-remaining-requests`, `x-ratelimit-remaining-tokens`, `x-ratelimit-reset-requests`, etc.) but these are currently consumed inter... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T17:29:08Z | 2026-04-28T04:41:47Z | 2026-04-26T02:51:06Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nikolaincode | 32,498,635 | MDQ6VXNlcjMyNDk4NjM1 | User | false |
openclaw/openclaw | 4,053,287,975 | I_kwDOQb6kR87xmEQn | 42,410 | https://github.com/openclaw/openclaw/issues/42410 | https://api.github.com/repos/openclaw/openclaw/issues/42410 | [Feature]: Feishu: multi-bot @mention matching fails due to app-scoped open_id — need fallback identity resolution | ### Summary
When multiple Feishu bot accounts share a group chat, @mention matching fails because open_id is app-scoped, and the only workaround (requireMention: false) causes all bots to respond to every message.
### Problem to solve
Feishu open_id is app-scoped: the same bot/user has a different open_id under each... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-10T17:46:46Z | 2026-04-28T04:41:46Z | 2026-04-26T02:51:10Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | cfu0547-Caroline | 259,740,529 | U_kgDOD3tTcQ | User | false |
openclaw/openclaw | 4,053,302,276 | I_kwDOQb6kR87xmHwE | 42,412 | https://github.com/openclaw/openclaw/issues/42412 | https://api.github.com/repos/openclaw/openclaw/issues/42412 | Feature Request: Skill Usage Statistics | ## Description
It would be great to have a way to track which skills are being used and how frequently. This would help users identify and remove unused skills.
## Use Case
When users install many skills, they often don't know which ones they actually use. Having usage statistics would help with:
- Identifying frequ... | closed | not_planned | true | 1 | [] | [] | 2026-03-10T17:49:41Z | 2026-04-28T04:41:46Z | 2026-04-26T02:51:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | shaozhengkun123 | 224,505,443 | U_kgDODWGuYw | User | false |
openclaw/openclaw | 4,053,788,893 | I_kwDOQb6kR87xn-jd | 42,469 | https://github.com/openclaw/openclaw/issues/42469 | https://api.github.com/repos/openclaw/openclaw/issues/42469 | Raw tool payloads leak into chat; fallback appears unreliable after provider errors | ### Bug type
Regression (worked before, now fails)
### Summary
In the Control UI, raw internal tool payloads sometimes leak directly into user-visible chat, and provider failures such as Codex server_error do not reliably recover via fallback.
### Steps to reproduce
Open OpenClaw Control UI on Windows with OpenCla... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-10T19:31:01Z | 2026-04-28T04:41:45Z | 2026-04-26T02:51:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rbrtwtrs | 26,334,448 | MDQ6VXNlcjI2MzM0NDQ4 | User | false |
openclaw/openclaw | 4,054,373,266 | I_kwDOQb6kR87xqNOS | 42,514 | https://github.com/openclaw/openclaw/issues/42514 | https://api.github.com/repos/openclaw/openclaw/issues/42514 | [Bug]: AnyClaw dashboard loads on :19001 but disconnects from gateway (1006), forced ws://localhost:18789 after update | ### Bug type
Regression (worked before, now fails)
### Summary
Title:
AnyClaw dashboard loads on :19001 but disconnects from gateway (1006), forced ws://localhost:18789 after update
Body:
```md
## Summary
After updating AnyClaw/OpenClaw on March 10, 2026, the browser dashboard loads, but it cannot connect to the g... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-10T21:41:44Z | 2026-04-28T04:41:44Z | 2026-04-26T02:51:22Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mankeluvsit | 272,113 | MDQ6VXNlcjI3MjExMw== | User | false |
openclaw/openclaw | 4,054,977,063 | I_kwDOQb6kR87xsgon | 42,600 | https://github.com/openclaw/openclaw/issues/42600 | https://api.github.com/repos/openclaw/openclaw/issues/42600 | Telegram topic reply continuity breaks across session reset: missing reply-parent hydration / no .reset transcript fallback | ## Summary
Telegram forum-topic replies can lose immediate conversational continuity when the active topic session has been reset/rotated, even when the user is replying directly to a recent assistant message.
This is **not** a normal model-context exhaustion case. The active session can be fresh (`compactionCount: 0`... | closed | completed | true | 1 | [] | [] | 2026-03-11T00:34:41Z | 2026-04-28T04:41:43Z | 2026-04-26T02:51:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 0xRapi | 258,445,411 | U_kgDOD2eQYw | User | false |
openclaw/openclaw | 4,076,830,177 | I_kwDOQb6kR87y_33h | 46,602 | https://github.com/openclaw/openclaw/issues/46602 | https://api.github.com/repos/openclaw/openclaw/issues/46602 | Feature: Show TTS provider/backend info in /status command | ### Summary
Add TTS backend visibility to the /status slash command, including model name, provider, and whether a custom baseUrl is in use.
### Problem to solve
When using a custom `baseUrl` under `messages.tts.openai` (e.g. a local NeuTTS server at `http://127.0.0.1:18801` instead of OpenAI), the `/status` command... | closed | completed | true | 2 | [
"enhancement"
] | [] | 2026-03-14T21:53:07Z | 2026-04-28T04:41:42Z | 2026-04-26T02:53:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | leekuangtao | 261,539,576 | U_kgDOD5bG-A | User | false |
openclaw/openclaw | 4,017,673,841 | I_kwDOQb6kR87veNZx | 33,270 | https://github.com/openclaw/openclaw/issues/33270 | https://api.github.com/repos/openclaw/openclaw/issues/33270 | [Feature]: Discord self-only profile updates + runtime health signaling via auto presence | ### Summary
Add a Discord-only `self-profile` action (strict self scope) and an optional `autoPresence` monitor that uses bot status/activity as a runtime health/quota signal.
### Problem to solve
OpenClaw currently lacks a safe, first-class way to update Discord bot self profile fields (nickname/avatar) from tool a... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-03T16:46:06Z | 2026-04-28T04:41:41Z | 2026-04-26T02:53:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | StdCarrot | 1,968,929 | MDQ6VXNlcjE5Njg5Mjk= | User | false |
openclaw/openclaw | 4,055,122,124 | I_kwDOQb6kR87xtEDM | 42,630 | https://github.com/openclaw/openclaw/issues/42630 | https://api.github.com/repos/openclaw/openclaw/issues/42630 | Talk Mode: Support on-device TTS (iOS AVSpeechSynthesizer) as alternative to ElevenLabs | ## Feature Request
**Current behavior:** Talk Mode only supports ElevenLabs for text-to-speech synthesis, requiring audio to be streamed from the cloud on every response.
**Proposed behavior:** Add support for on-device TTS as a Talk Mode provider option, starting with iOS `AVSpeechSynthesizer`. This would allow the ... | closed | completed | true | 1 | [] | [] | 2026-03-11T01:29:40Z | 2026-04-28T04:41:40Z | 2026-04-26T02:53:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | charlesnjohn-bot | 258,833,599 | U_kgDOD218vw | User | false |
openclaw/openclaw | 4,055,149,630 | I_kwDOQb6kR87xtKw- | 42,639 | https://github.com/openclaw/openclaw/issues/42639 | https://api.github.com/repos/openclaw/openclaw/issues/42639 | [Feature]: Add Perplexity Agent API as a built-in provider (openai-responses) | ### Summary
Add the [Perplexity Agent API](https://docs.perplexity.ai/docs/agent-api/quickstart) as a built-in `openai-responses` provider with support for its `preset` parameter, which bundles a model, tools, system prompt, and multi-step config into a single request field that can't be sent through manual provider c... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-11T01:39:33Z | 2026-04-28T04:41:39Z | 2026-04-26T02:53:32Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kesku | 62,210,496 | MDQ6VXNlcjYyMjEwNDk2 | User | false |
openclaw/openclaw | 4,055,174,077 | I_kwDOQb6kR87xtQu9 | 42,649 | https://github.com/openclaw/openclaw/issues/42649 | https://api.github.com/repos/openclaw/openclaw/issues/42649 | Memory MVP: implement hybrid retrieval scoring with explainable matches | ## Summary
Build the retrieval layer for v1 memory recall.
## Goal
Retrieve the right memory conservatively using hybrid ranking.
## Ranking inputs
- semantic similarity
- keyword/entity match
- recency
- confidence/freshness
## Requirements
- retrieval explanations for why each record matched
- recency is a ranking... | closed | completed | true | 1 | [] | [] | 2026-03-11T01:48:05Z | 2026-04-28T04:41:39Z | 2026-04-26T02:53:37Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | vinced45 | 101,286 | MDQ6VXNlcjEwMTI4Ng== | User | false |
openclaw/openclaw | 4,055,305,173 | I_kwDOQb6kR87xtwvV | 42,686 | https://github.com/openclaw/openclaw/issues/42686 | https://api.github.com/repos/openclaw/openclaw/issues/42686 | [Feature]: Feature Request: Per-agent lane isolation (agents.defaults.lane or per-agent lane config) | ### Summary
## Problem
When running 20+ agents on a single gateway, all agents share a single `main` lane
with a global `maxConcurrent` cap. If one agent runs a long-running task (e.g. an
ACP coding session), it occupies a slot in the shared lane, degrading response times
for all other agents.
## Desired Behavior
S... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-11T02:27:30Z | 2026-04-28T04:41:38Z | 2026-04-26T02:53:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | lijihua2017 | 33,613,942 | MDQ6VXNlcjMzNjEzOTQy | User | false |
openclaw/openclaw | 4,055,504,443 | I_kwDOQb6kR87xuhY7 | 42,751 | https://github.com/openclaw/openclaw/issues/42751 | https://api.github.com/repos/openclaw/openclaw/issues/42751 | WebChat UI: Messages not refreshing in real-time, requires manual refresh | ## Description
The WebChat UI (Control Dashboard at http://127.0.0.1:18789/) is not refreshing messages in real-time. Users need to manually click the refresh button to see new messages.
## Expected Behavior
Messages should appear in real-time as they arrive via WebSocket, with smooth continuous growth of message bu... | closed | completed | true | 2 | [] | [] | 2026-03-11T03:34:02Z | 2026-04-28T04:41:37Z | 2026-04-26T02:53:46Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mouseroser | 8,221,932 | MDQ6VXNlcjgyMjE5MzI= | User | false |
openclaw/openclaw | 4,055,511,301 | I_kwDOQb6kR87xujEF | 42,752 | https://github.com/openclaw/openclaw/issues/42752 | https://api.github.com/repos/openclaw/openclaw/issues/42752 | Slack: userTokenReadOnly: false has no effect on read operations — reads always prefer userToken | ## Description
Setting `userTokenReadOnly: false` in the Slack channel config does not cause read operations to use the bot token. Reads always prefer the user token regardless of this setting, causing `missing_scope` errors when the user token lacks `channels:history` / `groups:history`.
## Root Cause
In `plugin-sd... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T03:36:08Z | 2026-04-28T04:41:36Z | 2026-04-26T02:53:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sinzin91 | 4,664,043 | MDQ6VXNlcjQ2NjQwNDM= | User | false |
openclaw/openclaw | 4,055,570,916 | I_kwDOQb6kR87xuxnk | 42,767 | https://github.com/openclaw/openclaw/issues/42767 | https://api.github.com/repos/openclaw/openclaw/issues/42767 | [Bug]: Running long-running tasks will continuously display "running" even if the model status shows "ready". | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When I use a tool to test the target for a slightly longer period, say over 20 minutes, the background process of the tool stops, but OpenClaw TUI still shows "running." The development tool also exhibits this behavior: no background proces... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T03:55:12Z | 2026-04-28T04:41:36Z | 2026-04-26T02:53:55Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Archerkong | 40,537,619 | MDQ6VXNlcjQwNTM3NjE5 | User | false |
openclaw/openclaw | 4,055,728,653 | I_kwDOQb6kR87xvYIN | 42,792 | https://github.com/openclaw/openclaw/issues/42792 | https://api.github.com/repos/openclaw/openclaw/issues/42792 | node run: add --header flag for Cloudflare Zero Trust service tokens | ## Problem
When a gateway is behind Cloudflare Zero Trust (or similar reverse proxy with auth), `openclaw node run` cannot connect because the WebSocket handshake gets a 302 redirect to the auth page. There's no way to pass custom HTTP headers on the WebSocket connection.
## Current workaround
SSH tunnels to bypass ... | closed | not_planned | true | 4 | [] | [] | 2026-03-11T04:35:25Z | 2026-04-28T04:41:35Z | 2026-04-26T02:54:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kevin-freiburger | 130,078,314 | U_kgDOB8DWag | User | false |
openclaw/openclaw | 4,057,007,387 | I_kwDOQb6kR87x0QUb | 43,032 | https://github.com/openclaw/openclaw/issues/43032 | https://api.github.com/repos/openclaw/openclaw/issues/43032 | Docs: Memory system deep dive guide | ## Gap
`docs/concepts/memory.md` covers basics (~600 words) but lacks depth on:
- Memory best practices (what to store, storage strategies)
- `memory_search` query optimization
- Memory lifecycle management (cleanup, archival, migration)
- Relationship between daily logs, MEMORY.md, and session context
## Proposed: `... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T09:28:39Z | 2026-04-28T04:41:34Z | 2026-04-26T02:54:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | chengjialu8888 | 261,118,124 | U_kgDOD5BYrA | User | false |
openclaw/openclaw | 4,057,038,742 | I_kwDOQb6kR87x0X-W | 43,042 | https://github.com/openclaw/openclaw/issues/43042 | https://api.github.com/repos/openclaw/openclaw/issues/43042 | [Bug]: unable to load apply_patch | ### Bug type
Regression (worked before, now fails)
### Summary
Openclaw unable to load and use apply_patch
### Steps to reproduce
openclaw.json
```json
"tools": {
"exec": {
"ask": "off",
"applyPatch": {
"enabled": true,
"allowModels": [
"kimi-k2.5"
]
}
... | closed | not_planned | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-11T09:34:23Z | 2026-04-28T04:41:32Z | 2026-04-26T02:54:14Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Subway2023 | 147,013,944 | U_kgDOCMNBOA | User | false |
openclaw/openclaw | 4,057,379,377 | I_kwDOQb6kR87x1rIx | 43,099 | https://github.com/openclaw/openclaw/issues/43099 | https://api.github.com/repos/openclaw/openclaw/issues/43099 | [Feature]: Feature request: add a first-class directory listing tool for the workspace | ### Summary
OpenClaw currently provides file tools like read, write, and edit, but there does not seem to be a first-class tool
for listing directory contents.
This creates an awkward gap in filesystem navigation:
- agents can read/write/edit files
- but cannot cleanly enumerate files/folders in a directory with... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-11T10:37:02Z | 2026-04-28T04:41:32Z | 2026-04-26T02:54:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | LaskerSaw | 130,273,415 | U_kgDOB8PQhw | User | false |
openclaw/openclaw | 4,057,461,702 | I_kwDOQb6kR87x1_PG | 43,113 | https://github.com/openclaw/openclaw/issues/43113 | https://api.github.com/repos/openclaw/openclaw/issues/43113 | [Bug]: Sub-agent dies with 'Network connection lost' during large tool call write — all work lost | ### Bug type
Network/streaming bug affecting sub-agents
### Summary
Sub-agents that accumulate large context (600K-800K+ tokens) during research-heavy tasks consistently die with "Network connection lost" when attempting to write large file content via the `write` tool. All work is lost — no partial results persisted.... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T10:52:17Z | 2026-04-28T04:41:31Z | 2026-04-26T02:54:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | a1j9o94 | 4,657,703 | MDQ6VXNlcjQ2NTc3MDM= | User | false |
openclaw/openclaw | 4,057,739,438 | I_kwDOQb6kR87x3DCu | 43,149 | https://github.com/openclaw/openclaw/issues/43149 | https://api.github.com/repos/openclaw/openclaw/issues/43149 | web_fetch: add defaultHeaders config option | **Feature request:** Add a `defaultHeaders` option to `tools.web.fetch` config.
Cloudflare Markdown for Agents (https://blog.cloudflare.com/markdown-for-agents/) serves native Markdown (80% fewer tokens) to clients sending `Accept: text/markdown`. `web_fetch` cannot send this header automatically.
Proposed config:
``... | closed | completed | true | 1 | [] | [] | 2026-03-11T11:45:28Z | 2026-04-28T04:41:30Z | 2026-04-26T02:54:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | k1dw41 | 266,730,321 | U_kgDOD-X7UQ | User | false |
openclaw/openclaw | 4,058,023,731 | I_kwDOQb6kR87x4Icz | 43,180 | https://github.com/openclaw/openclaw/issues/43180 | https://api.github.com/repos/openclaw/openclaw/issues/43180 | [Bug]: No hooks found. Windows下pnpm安装找不到hooks | ### Bug type
Regression (worked before, now fails)
### Summary
首先
pnpm add -g openclaw@latest
然后
PS C:\Users\Administrator> openclaw hooks list
🦞 OpenClaw 2026.3.8 (3caab92) — Built by lobsters, for humans. Don't question the hierarchy.
20:27:28 [plugins] qqbot: loaded without install/load-path provenance; treat... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-03-11T12:44:37Z | 2026-04-28T04:41:29Z | 2026-04-26T02:54:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 7568168 | 10,483,599 | MDQ6VXNlcjEwNDgzNTk5 | User | false |
openclaw/openclaw | 4,058,208,286 | I_kwDOQb6kR87x41ge | 43,201 | https://github.com/openclaw/openclaw/issues/43201 | https://api.github.com/repos/openclaw/openclaw/issues/43201 | Feature: one-shot thinking level via /think <level> <message> | ## Summary
Currently, `/think <level>` changes the session's default thinking level persistently. I'd like a **one-shot** mode: `/think <level> <message>` applies the specified thinking level to that single message only, without modifying the session default.
This is useful when your session default is `medium` (for ... | closed | not_planned | true | 2 | [] | [] | 2026-03-11T13:21:44Z | 2026-04-28T04:41:29Z | 2026-04-26T02:54:36Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Astro-Han | 255,364,436 | U_kgDODziNVA | User | false |
openclaw/openclaw | 4,058,390,633 | I_kwDOQb6kR87x5iBp | 43,237 | https://github.com/openclaw/openclaw/issues/43237 | https://api.github.com/repos/openclaw/openclaw/issues/43237 | Web UI pasted images are silently dropped instead of being processed by Media Understanding or text-only models | ### Description
When a user pastes an image into the Web UI chat input, the frontend successfully converts it to a Base64 data URL and sends it in the `attachments` array of the `chat.send` RPC. However, the image is silently dropped by the Gateway/backend pipeline, and neither the primary text-only model nor the confi... | closed | completed | true | 1 | [] | [] | 2026-03-11T13:55:51Z | 2026-04-28T04:41:28Z | 2026-04-26T02:54:41Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | DavidLyn | 13,762,942 | MDQ6VXNlcjEzNzYyOTQy | User | false |
openclaw/openclaw | 4,058,452,969 | I_kwDOQb6kR87x5xPp | 43,245 | https://github.com/openclaw/openclaw/issues/43245 | https://api.github.com/repos/openclaw/openclaw/issues/43245 | message tool: bound sessions allow same-provider cross-context sends by default | ## Summary
When the `message` tool is bound to an active chat session, OpenClaw still allows an explicit target on the **same provider** to override the bound conversation by default.
In practice, that means a Slack session bound to `#channel-a` can send to `#channel-b` if the model picks the wrong explicit `target`,... | closed | not_planned | true | 2 | [] | [] | 2026-03-11T14:07:12Z | 2026-04-28T04:41:27Z | 2026-04-26T02:54:46Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jdrhyne | 7,828,464 | MDQ6VXNlcjc4Mjg0NjQ= | User | false |
openclaw/openclaw | 4,059,083,583 | I_kwDOQb6kR87x8LM_ | 43,308 | https://github.com/openclaw/openclaw/issues/43308 | https://api.github.com/repos/openclaw/openclaw/issues/43308 | [Feature]: Add observability and tracing integration hooks (Langfuse or similar) | ### Summary
Add first-class observability/tracing hooks to OpenClaw so users can integrate with Langfuse or similar platforms to inspect agent runs, model calls, tool execution, latency, failures, and token/cost usage.
### Problem to solve
OpenClaw is powerful for multi-step agent workflows, but it is currently h... | closed | completed | true | 3 | [
"enhancement"
] | [] | 2026-03-11T15:55:48Z | 2026-04-28T04:41:26Z | 2026-04-26T02:54:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | panxiongfei | 7,157,839 | MDQ6VXNlcjcxNTc4Mzk= | User | false |
openclaw/openclaw | 4,059,104,122 | I_kwDOQb6kR87x8QN6 | 43,313 | https://github.com/openclaw/openclaw/issues/43313 | https://api.github.com/repos/openclaw/openclaw/issues/43313 | fix(media): classify audio-only MP4/M4A containers as audio, not video | ## Problem
Audio-only MP4 and M4A files are silently skipped by the transcription pipeline because the `file-type` library cannot distinguish an audio-only MPEG-4 container from a video one — both return `video/mp4`. This causes `kindFromMime()` to return `"video"`, and `selectAttachments({ capability: "audio" })` ski... | closed | completed | true | 2 | [] | [] | 2026-03-11T15:59:38Z | 2026-04-28T04:41:25Z | 2026-04-26T02:56:17Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mantisai-bot | 259,690,819 | U_kgDOD3qRQw | User | false |
openclaw/openclaw | 4,059,681,027 | I_kwDOQb6kR87x-dED | 43,387 | https://github.com/openclaw/openclaw/issues/43387 | https://api.github.com/repos/openclaw/openclaw/issues/43387 | [Bug]: Cybercentry Access to ClawHub | ### Bug type
Regression (worked before, now fails)
### Summary
I was previously able to log in to ClawHub using GitHub. I can no longer log in or search for the nine skills I uploaded. Please could you support me with this? Many thanks.
### Steps to reproduce
1. Click on Sign with GitHub.
2. Auth ClawHub.
3. Clic... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T17:43:21Z | 2026-04-28T04:41:25Z | 2026-04-26T02:56:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Cybercentry | 194,429,610 | U_kgDOC5bCqg | User | false |
openclaw/openclaw | 4,060,558,540 | I_kwDOQb6kR87yBzTM | 43,457 | https://github.com/openclaw/openclaw/issues/43457 | https://api.github.com/repos/openclaw/openclaw/issues/43457 | [Bug]: Browser downloads don't work for JavaScript-initiated blob downloads | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When using the openclaw browser profile, file downloads initiated by JavaScript (blob URLs, window.open(), dynamic anchor clicks) don't save to disk. The download either opens in the browser, gets blocked, or silently fails.
### Steps to r... | closed | completed | true | 5 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T20:41:19Z | 2026-04-28T04:41:24Z | 2026-04-26T02:56:24Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | clkr | 365,996 | MDQ6VXNlcjM2NTk5Ng== | User | false |
openclaw/openclaw | 4,060,878,996 | I_kwDOQb6kR87yDBiU | 43,492 | https://github.com/openclaw/openclaw/issues/43492 | https://api.github.com/repos/openclaw/openclaw/issues/43492 | Cross-provider fallback retry: prompt-detected images bypass resolveRetryImages guard | ## Summary
`detectAndLoadPromptImages` in `src/agents/pi-embedded-runner/run/attempt.ts` scans prompt text for image file paths (e.g., `./image.png`, `file://...`, `[Image: source:...]`) and loads them from disk independently of the `images` parameter passed through `agentCommand`.
When #43485 strips explicitly-suppl... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T21:49:58Z | 2026-04-28T04:41:23Z | 2026-04-26T02:56:28Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mitchmcalister | 209,334 | MDQ6VXNlcjIwOTMzNA== | User | false |
openclaw/openclaw | 4,060,915,668 | I_kwDOQb6kR87yDKfU | 43,494 | https://github.com/openclaw/openclaw/issues/43494 | https://api.github.com/repos/openclaw/openclaw/issues/43494 | [Feature] Add reasoningDefault config option and Slack workaround | ## Problem
1. **Slack intercepts / commands at UI level**: When users type / in Slack, Slack's own command palette appears before the message is sent, blocking OpenClaw directives.
2. **No reasoningDefault config**: There's thinkingDefault in config but no equivalent for reasoning visibility.
3. **Context bloat**: W... | closed | not_planned | true | 2 | [] | [] | 2026-03-11T21:57:19Z | 2026-04-28T04:41:22Z | 2026-04-26T02:56:32Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Wingnuut | 195,316,867 | U_kgDOC6RMgw | User | false |
openclaw/openclaw | 4,060,969,482 | I_kwDOQb6kR87yDXoK | 43,496 | https://github.com/openclaw/openclaw/issues/43496 | https://api.github.com/repos/openclaw/openclaw/issues/43496 | Feature: Add steer/action support for ACP runtime sessions (sessions_spawn runtime=acp) | ## Problem
ACP runtime sessions (created via `sessions_spawn` with `runtime: "acp"`) do not support steering or action commands like subagent sessions do.
The `subagents` tool only works with `runtime: "subagent"` sessions, not ACP harness sessions. Attempting to steer an ACP session returns:
```
error: "Unknown suba... | closed | completed | true | 1 | [] | [] | 2026-03-11T22:08:39Z | 2026-04-28T04:41:21Z | 2026-04-26T02:56:36Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kai-the-man[bot] | 259,478,402 | BOT_kgDOD3dTgg | Bot | false |
openclaw/openclaw | 4,061,224,314 | I_kwDOQb6kR87yEV16 | 43,518 | https://github.com/openclaw/openclaw/issues/43518 | https://api.github.com/repos/openclaw/openclaw/issues/43518 | [Bug] sessions_spawn task parameter does not deliver to ACP runtime sessions - sessions spawn with empty history | ## Description
The `task` parameter in `sessions_spawn` with `runtime: "acp"` does not reliably deliver the task to the spawned ACP session. Sessions are created successfully but arrive with empty message history, causing the agent to have no knowledge of its assigned task.
## Reproduction Steps
1. Spawn an ACP sess... | closed | not_planned | true | 4 | [] | [] | 2026-03-11T23:24:30Z | 2026-04-28T04:41:21Z | 2026-04-26T02:56:40Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kai-the-man[bot] | 259,478,402 | BOT_kgDOD3dTgg | Bot | false |
openclaw/openclaw | 4,061,272,258 | I_kwDOQb6kR87yEhjC | 43,522 | https://github.com/openclaw/openclaw/issues/43522 | https://api.github.com/repos/openclaw/openclaw/issues/43522 | WhatsApp: expose onWhatsApp() number check + message delivery ACK | ## Feature Request
### Problem
When sending WhatsApp messages to Brazilian numbers, there's a known issue with the **"nono dígito"** (9th digit): some older accounts were registered without it. Sending to the wrong format delivers without error but the message never arrives.
Currently there's no way to:
1. **Validat... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T23:43:16Z | 2026-04-28T04:41:20Z | 2026-04-26T02:56:44Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Donna-Nummus | 267,162,625 | U_kgDOD-yUAQ | User | false |
openclaw/openclaw | 4,061,285,957 | I_kwDOQb6kR87yEk5F | 43,524 | https://github.com/openclaw/openclaw/issues/43524 | https://api.github.com/repos/openclaw/openclaw/issues/43524 | Daily reset doesn't trigger session-memory hook flush | ## Summary
Daily/scheduled session reset (when `evaluateSessionFreshness()` detects a stale session) does not trigger the `session-memory` hook to save conversation context to `~/.openclaw/workspace/memory/`. Only manual `/new` or `/reset` commands trigger the hook.
## Root Cause
In `src/auto-reply/reply/session.ts`, ... | closed | not_planned | true | 4 | [] | [] | 2026-03-11T23:48:41Z | 2026-04-28T04:41:19Z | 2026-04-26T02:56:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | big91987 | 37,739,196 | MDQ6VXNlcjM3NzM5MTk2 | User | false |
openclaw/openclaw | 4,061,305,828 | I_kwDOQb6kR87yEpvk | 43,528 | https://github.com/openclaw/openclaw/issues/43528 | https://api.github.com/repos/openclaw/openclaw/issues/43528 | New session startup duplicates workspace context via Project Context injection + Session Startup reads | Hi, I'm seeing duplicated workspace bootstrap content in new sessions created via `/new` or `/reset`.
## What I observed
In a fresh main session, some workspace files appear to be included twice in the effective context:
- once via default Project Context / workspace bootstrap injection
- once again because the start... | closed | completed | true | 4 | [] | [] | 2026-03-11T23:55:58Z | 2026-04-28T04:41:18Z | 2026-04-26T02:56:52Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | anyran | 11,988,991 | MDQ6VXNlcjExOTg4OTkx | User | false |
openclaw/openclaw | 4,061,507,056 | I_kwDOQb6kR87yFa3w | 43,558 | https://github.com/openclaw/openclaw/issues/43558 | https://api.github.com/repos/openclaw/openclaw/issues/43558 | Maintainer request: please delete old issue #43554 and keep #43557 | Hi — could a maintainer please permanently delete issue #43554?
Reason:
- it briefly contained incorrect information before being redacted
- it has already been replaced by the sanitized issue #43557, which should remain open
Requested action:
- delete #43554
- keep #43557
Thanks.
| closed | not_planned | true | 1 | [] | [] | 2026-03-12T01:11:50Z | 2026-04-28T04:41:18Z | 2026-04-26T02:56:56Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | federicocorradi | 258,326,063 | U_kgDOD2W-Lw | User | false |
openclaw/openclaw | 4,061,596,706 | I_kwDOQb6kR87yFwwi | 43,579 | https://github.com/openclaw/openclaw/issues/43579 | https://api.github.com/repos/openclaw/openclaw/issues/43579 | feat(tts): add Deepgram Aura as first-class TTS provider | ## Summary
Add Deepgram Aura TTS as a first-class provider alongside OpenAI, ElevenLabs, and Edge TTS. Deepgram offers competitive pricing and low-latency speech synthesis via their Aura-2 voice models.
## Motivation
- Deepgram is already used for STT (`runtime.stt.transcribeAudioFile`), so users may already have AP... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T01:50:49Z | 2026-04-28T04:41:17Z | 2026-04-26T02:57:00Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | jbold | 374,123 | MDQ6VXNlcjM3NDEyMw== | User | false |
openclaw/openclaw | 4,061,648,579 | I_kwDOQb6kR87yF9bD | 43,587 | https://github.com/openclaw/openclaw/issues/43587 | https://api.github.com/repos/openclaw/openclaw/issues/43587 | [Feature]: Slack allowBots: "mentions" mode (parity with Discord) | ## Feature Description
Add `"mentions"` as a valid value for Slack's `allowBots` config, matching Discord's existing implementation.
### Problem
When multiple agents (each with their own Slack bot) share a channel, there is no way to allow bot-to-bot communication while preventing infinite reply loops. The current `... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T02:12:24Z | 2026-04-28T04:41:16Z | 2026-04-26T02:57:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | raw34 | 1,996,768 | MDQ6VXNlcjE5OTY3Njg= | User | false |
openclaw/openclaw | 4,061,951,265 | I_kwDOQb6kR87yHHUh | 43,654 | https://github.com/openclaw/openclaw/issues/43654 | https://api.github.com/repos/openclaw/openclaw/issues/43654 | Browser tool consistently times out - unable to start browser sessions | ## Bug Description
The browser tool consistently times out when attempting to start browser sessions, making it completely unusable.
## Environment
- **OpenClaw**: 2026.3.8 (3caab92)
- **OS**: Ubuntu 24.04.4 LTS (ARM64)
- **Node**: v22.22.0
- **Platform**: AWS EC2 (Graviton 3)
## Reproduction Steps
1. Call `browser(... | closed | completed | true | 2 | [] | [] | 2026-03-12T03:57:38Z | 2026-04-28T04:41:15Z | 2026-04-26T02:57:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kobewang1234 | 20,372,811 | MDQ6VXNlcjIwMzcyODEx | User | false |
openclaw/openclaw | 4,062,203,698 | I_kwDOQb6kR87yIE8y | 43,696 | https://github.com/openclaw/openclaw/issues/43696 | https://api.github.com/repos/openclaw/openclaw/issues/43696 | Feature request: feishu_calendar tool — create events, invite attendees, book meeting rooms | ## Summary
Add a `feishu_calendar` tool to the Feishu extension, enabling the agent to create calendar events, invite attendees, and book meeting rooms via natural language.
## Use case
Users in Feishu group chats want to say things like:
> 邀请群里的所有人，预定一个2点的会议，标题：test
Currently the bot cannot do this because there ... | closed | not_planned | true | 2 | [] | [] | 2026-03-12T05:11:20Z | 2026-04-28T04:41:15Z | 2026-04-26T02:57:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | PenguinMiaou | 209,870,260 | U_kgDODIJdtA | User | false |
openclaw/openclaw | 4,062,377,773 | I_kwDOQb6kR87yIvct | 43,729 | https://github.com/openclaw/openclaw/issues/43729 | https://api.github.com/repos/openclaw/openclaw/issues/43729 | [Feature Request] feishu_bitable: native attachment upload and record deletion support | ### Problem
Currently, the `feishu_bitable` toolset supports basic CRUD operations but lacks:
1. **Attachment Uploads**: `feishu_bitable_update_record` requires a `file_token` for attachment fields, but there's no tool to upload a file to Bitable to get this token (the existing `feishu_doc.upload_file` only works for d... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T05:55:23Z | 2026-04-28T04:41:14Z | 2026-04-26T02:57:15Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | fanfanfancheung | 128,835,875 | U_kgDOB63hIw | User | false |
openclaw/openclaw | 4,062,430,389 | I_kwDOQb6kR87yI8S1 | 43,738 | https://github.com/openclaw/openclaw/issues/43738 | https://api.github.com/repos/openclaw/openclaw/issues/43738 | [Feature]: Adaptive session reset mode — daily reset gated by idle time (AND logic) | ### Summary
A new `mode: "adaptive"` session reset that requires BOTH `atHour` to have passed AND the session to have been idle for `idleMinutes` before resetting. Currently when both are configured, whichever fires first wins (OR logic).
### Problem to solve
Current reset modes are either time-based (`daily` — hard... | closed | not_planned | true | 2 | [] | [] | 2026-03-12T06:07:28Z | 2026-04-28T04:41:13Z | 2026-04-26T02:57:19Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mantisai-bot | 259,690,819 | U_kgDOD3qRQw | User | false |
openclaw/openclaw | 4,062,526,844 | I_kwDOQb6kR87yJT18 | 43,758 | https://github.com/openclaw/openclaw/issues/43758 | https://api.github.com/repos/openclaw/openclaw/issues/43758 | [Feature]: Optimize the console log: log the information when OpenClaw sends a message | ### Summary
Adding console logs when messages are sent.
### Problem to solve
Make the console more easy to see the running status.
### Proposed solution
Now there is log information when receiving messages, while there is no log when sending ones.
### Alternatives considered
_No response_
### Impact
Low (enhan... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-12T06:33:05Z | 2026-04-28T04:41:12Z | 2026-04-26T02:57:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | dingxl18gz | 158,808,045 | U_kgDOCXc37Q | User | false |
openclaw/openclaw | 4,062,661,330 | I_kwDOQb6kR87yJ0rS | 43,778 | https://github.com/openclaw/openclaw/issues/43778 | https://api.github.com/repos/openclaw/openclaw/issues/43778 | feat(tui): improve UX when gateway is disconnected or restarting | ## Problem
When the gateway restarts (e.g. after `openclaw gateway restart`), the TUI shows `not connected to gateway — message not sent` with no guidance on what to do next. Users are left wondering whether to wait, retry, or restart.
## Proposed Improvements
### 1. In-TUI reconnect hint
When the TUI detects a disc... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T07:06:49Z | 2026-04-28T04:41:11Z | 2026-04-26T02:57:27Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | happydog-intj | 212,984 | MDQ6VXNlcjIxMjk4NA== | User | false |
openclaw/openclaw | 4,063,608,489 | I_kwDOQb6kR87yNb6p | 43,932 | https://github.com/openclaw/openclaw/issues/43932 | https://api.github.com/repos/openclaw/openclaw/issues/43932 | Rate limit errors surfaced to chat should include model/provider/task context | ## Problem
When an LLM API provider returns a 429 rate limit error, the gateway surfaces this to the chat channel as:
> ⚠️ API rate limit reached. Please try again later.
This message contains zero actionable context. The user has no way to know:
- **Which model/provider** hit the rate limit (e.g. `google/gemini-2.... | closed | not_planned | true | 2 | [] | [] | 2026-03-12T10:17:39Z | 2026-04-28T04:41:11Z | 2026-04-26T02:57:31Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | AlphaComposite | 25,360,554 | MDQ6VXNlcjI1MzYwNTU0 | User | false |
openclaw/openclaw | 4,329,882,277 | I_kwDOQb6kR88AAAABAhTCpQ | 71,910 | https://github.com/openclaw/openclaw/issues/71910 | https://api.github.com/repos/openclaw/openclaw/issues/71910 | [Bug]: WebChat reasoning/thinking content leaks as duplicate gateway-injected transcript message (v2026.4.24 regression) | # Summary
When `reasoningLevel: "on"` and `thinking=high`, the webchat displays reasoning/thinking content **twice**:
1. Once as the intended thinking blocks on the assistant message (correct)
2. Once again as a separate gateway-injected transcript message with `provider: "openclaw"`, `model: "gateway-injected"`, forma... | closed | completed | true | 1 | [] | [] | 2026-04-26T02:57:42Z | 2026-04-29T04:37:24Z | 2026-04-26T10:42:38Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | devnyros1 | 207,947,494 | U_kgDODGUG5g | User | false |
openclaw/openclaw | 4,329,882,810 | I_kwDOQb6kR88AAAABAhTEug | 71,911 | https://github.com/openclaw/openclaw/issues/71911 | https://api.github.com/repos/openclaw/openclaw/issues/71911 | Discord DM messages delivered twice: real-time + queued (follow-up to #71306) | ## Problem
Follow-up to #71306 (closed by Codex review). The duplicate delivery bug persists in **v2026.4.24** and is **specific to Discord DMs** — guild channel messages are unaffected.
### Observed behavior
Every Discord DM message is delivered to the agent twice:
1. Once as a real-time turn (session idle → `run-n... | closed | completed | true | 4 | [] | [] | 2026-04-26T02:58:05Z | 2026-04-29T04:37:24Z | 2026-04-26T10:42:42Z | CONTRIBUTOR | null | 20260429T060025Z | 2026-04-29T06:00:25Z | kagura-agent | 268,167,063 | U_kgDOD_vnlw | User | false |
openclaw/openclaw | 4,329,883,097 | I_kwDOQb6kR88AAAABAhTF2Q | 71,912 | https://github.com/openclaw/openclaw/issues/71912 | https://api.github.com/repos/openclaw/openclaw/issues/71912 | [Bug]: delivery-mirror sends 673 messages in infinite loop during long-running task | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
执行长时间运行的 Docker 任务时，delivery-mirror 机制进入无限循环，6 分钟内发送了 673 条重复消息（"抱歉让你等了！让我检查下当前进度："），每 400-500ms 一条，直到 Gateway 被 SIGTERM 终止。
### Steps to reproduce
1、在 QQBot 会话中执行长时间任务（如 sleep 15 && docker ps）
2、delivery-mir... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-26T02:58:18Z | 2026-04-29T04:37:23Z | 2026-04-26T10:42:46Z | NONE | null | 20260429T060025Z | 2026-04-29T06:00:25Z | wangyc6003 | 96,985,387 | U_kgDOBcfhKw | User | false |
openclaw/openclaw | 4,329,760,105 | I_kwDOQb6kR88AAAABAhLlaQ | 71,877 | https://github.com/openclaw/openclaw/issues/71877 | https://api.github.com/repos/openclaw/openclaw/issues/71877 | Remote macOS skill eligibility ignores system.which bins object response, so apple-notes stays missing memo | ### Summary
On a Linux/container Gateway with a connected macOS node, remote macOS skill eligibility still reports `apple-notes` as missing `memo` even though the Mac node successfully exposes `memo` via `system.which`.
The issue appears to be that the remote bin probe parser only accepts `bins` as an array or `stdou... | closed | completed | true | 1 | [] | [] | 2026-04-26T01:39:27Z | 2026-04-28T04:41:08Z | 2026-04-26T02:58:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | miguelarios | 24,581,058 | MDQ6VXNlcjI0NTgxMDU4 | User | false |
openclaw/openclaw | 4,063,699,951 | I_kwDOQb6kR87yNyPv | 43,946 | https://github.com/openclaw/openclaw/issues/43946 | https://api.github.com/repos/openclaw/openclaw/issues/43946 | Configurable LLM request timeout per provider/model (Ollama cold-start causes silent fallback) | ## Summary
When using local Ollama models, the first request after model load triggers a cold-start that takes **~13-46 seconds** (depending on model size). The default LLM request timeout in OpenClaw appears too short for this scenario, causing a **timeout-based fallback** (status 408) to the next model in the fallba... | closed | completed | true | 3 | [] | [] | 2026-03-12T10:32:16Z | 2026-04-28T04:41:07Z | 2026-04-26T02:59:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Meli73 | 17,142,494 | MDQ6VXNlcjE3MTQyNDk0 | User | false |
openclaw/openclaw | 4,063,820,417 | I_kwDOQb6kR87yOPqB | 43,964 | https://github.com/openclaw/openclaw/issues/43964 | https://api.github.com/repos/openclaw/openclaw/issues/43964 | Bug: Feishu group chat voice message fails with error 234001 | ## Bug: Feishu group chat voice message fails with error 234001
### Description
When using TTS to send voice messages to Feishu group chats, the API returns error code 234001 (Invalid request param). However, sending voice messages to private chats (DM) works correctly.
### Steps to Reproduce
1. Configure OpenClaw wi... | closed | completed | true | 4 | [] | [] | 2026-03-12T10:56:48Z | 2026-04-28T04:41:06Z | 2026-04-26T02:59:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | yanhui84 | 15,568,725 | MDQ6VXNlcjE1NTY4NzI1 | User | false |
openclaw/openclaw | 4,063,892,607 | I_kwDOQb6kR87yOhR_ | 43,972 | https://github.com/openclaw/openclaw/issues/43972 | https://api.github.com/repos/openclaw/openclaw/issues/43972 | Increase maxPingPongTurns cap for agent-to-agent conversations | ## Context
We're running a multi-agent setup where agents need to have real conversations with each other — not just fire-and-forget notifications, but back-and-forth discussion (e.g. a health agent and a planning agent coordinating, or agents discussing topics during reflection time).
## Current Behavior
`session.a... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T11:10:46Z | 2026-04-28T04:41:05Z | 2026-04-26T02:59:43Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | bendavieshe3 | 654,464 | MDQ6VXNlcjY1NDQ2NA== | User | false |
openclaw/openclaw | 4,063,997,501 | I_kwDOQb6kR87yO649 | 43,990 | https://github.com/openclaw/openclaw/issues/43990 | https://api.github.com/repos/openclaw/openclaw/issues/43990 | [Bug]: Rocky Linux cannot install OpenClaw. | ### Bug type
Regression (worked before, now fails)
### Summary
The installation process has stalled.
```
$ npm install -g openclaw@latest
npm warn deprecated node-domexception@1.0.0: Use your platform's native DOMException instead
⠹
```
Not only did it report an error, but it also occupied all of the machine's CPU ... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-12T11:31:06Z | 2026-04-28T04:41:04Z | 2026-04-26T02:59:48Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | pptfz | 44,946,546 | MDQ6VXNlcjQ0OTQ2NTQ2 | User | false |
openclaw/openclaw | 4,064,023,614 | I_kwDOQb6kR87yPBQ- | 43,991 | https://github.com/openclaw/openclaw/issues/43991 | https://api.github.com/repos/openclaw/openclaw/issues/43991 | [Feature]: Support HarmonyOS Node App for OpenClaw,Please help to review | ### Summary
OpenClaw currently lacks a dedicated client for HarmonyOS. While the Android app exists, HarmonyOS users need a solution that leverages the HarmonyOS Node.js runtime/ArkTS stack to ensure seamless gateway authentication and native-like performance on the NEXT platform.
### Problem to solve
I have develop... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-12T11:35:55Z | 2026-04-28T04:41:04Z | 2026-04-26T02:59:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | seasonmac | 3,897,889 | MDQ6VXNlcjM4OTc4ODk= | User | false |
openclaw/openclaw | 4,064,495,685 | I_kwDOQb6kR87yQ0hF | 44,052 | https://github.com/openclaw/openclaw/issues/44052 | https://api.github.com/repos/openclaw/openclaw/issues/44052 | [Bug]: isolated+agentturn cron jobs hang indefinitely on linux arm64 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
**OpenClaw** version : 2026.3.8
os: linux nanopi-r6s arm64
nodejs: v24.14.0
Issue Summary
Symptom: Cron tasks using the isolated + agentTurn mode are marked as running but never actually execute.
Comparison: * The main + systemEvent mode ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T12:59:36Z | 2026-04-28T04:41:03Z | 2026-04-26T02:59:58Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | WYnneaoapc | 17,985,104 | MDQ6VXNlcjE3OTg1MTA0 | User | false |
openclaw/openclaw | 4,064,644,308 | I_kwDOQb6kR87yRYzU | 44,070 | https://github.com/openclaw/openclaw/issues/44070 | https://api.github.com/repos/openclaw/openclaw/issues/44070 | [Bug]: openclaw gateway install failure | ### Bug type
Regression (worked before, now fails)
### Summary
Gateway install failed: Error: systemctl daemon-reload failed: Reload daemon failed: Transport endpoint is not connected
### Steps to reproduce
execute curl -fsSL https://openclaw.ai/install.sh | bash
### Expected behavior
install gateway
### Actual... | closed | not_planned | true | 6 | [
"bug",
"regression"
] | [] | 2026-03-12T13:24:15Z | 2026-04-28T04:41:02Z | 2026-04-26T03:00:03Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | FangJYun | 25,579,255 | MDQ6VXNlcjI1NTc5MjU1 | User | false |
openclaw/openclaw | 4,064,823,827 | I_kwDOQb6kR87ySEoT | 44,084 | https://github.com/openclaw/openclaw/issues/44084 | https://api.github.com/repos/openclaw/openclaw/issues/44084 | cron list crashes with TypeError: value.slice is not a function when payload.model is an object | ## Description
`openclaw cron list` throws `TypeError: value.slice is not a function` when any cron job has `payload.model` stored as an object `{primary, fallbacks}` instead of a plain string.
## Steps to reproduce
1. Create a cron job with a model fallback config, e.g.:
```json
"model": {
"primary": "op... | closed | completed | true | 5 | [] | [] | 2026-03-12T13:51:41Z | 2026-04-28T04:41:01Z | 2026-04-26T03:00:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | mtsogias | 244,073,807 | U_kgDODoxFTw | User | false |
openclaw/openclaw | 4,065,002,636 | I_kwDOQb6kR87ySwSM | 44,110 | https://github.com/openclaw/openclaw/issues/44110 | https://api.github.com/repos/openclaw/openclaw/issues/44110 | [Bug]: tool_choice: "auto" sent even when tools.allow: [] is configured | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When an agent is configured with tools.allow: [] (empty allowlist), OpenClaw still includes tool_choice: "auto" in the API request to the model provider. This causes failures on providers that don't support tool_choice unless their vLLM dep... | closed | completed | true | 4 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T14:17:32Z | 2026-04-28T04:41:00Z | 2026-04-26T03:00:13Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | braydonbarrettfitness-stack | 267,675,327 | U_kgDOD_Rmvw | User | false |
openclaw/openclaw | 4,065,099,943 | I_kwDOQb6kR87yTICn | 44,127 | https://github.com/openclaw/openclaw/issues/44127 | https://api.github.com/repos/openclaw/openclaw/issues/44127 | ACP liveness probe delivered as undifferentiated user prompt — agent executes it as a command | ## Summary
OpenClaw's liveness probe for ACP/ACPX binding sessions is delivered via `session/prompt` with no system metadata to distinguish it from a real user message. As a result, the agent receives the probe as an ordinary instruction and executes it literally.
## Steps to reproduce
1. Configure a persistent ACP ... | closed | completed | true | 1 | [] | [] | 2026-03-12T14:31:51Z | 2026-04-28T04:40:59Z | 2026-04-26T03:00:18Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | aaamosh | 253,971,641 | U_kgDODyNMuQ | User | false |
openclaw/openclaw | 4,065,312,295 | I_kwDOQb6kR87yT74n | 44,149 | https://github.com/openclaw/openclaw/issues/44149 | https://api.github.com/repos/openclaw/openclaw/issues/44149 | [Feature Request] 飞书插件增加「下载消息附件」工具 | ### Summary
[Feature Request] 飞书插件增加「下载消息附件」工具
需求描述：
背景： 飞书聊天中经常需要机器人接收并处理群内发送的文件（如合同、文档等），但目前飞书插件只暴露了云文档/云空间/知识库等工具，没有提供从聊天消息中下载附件的能力。
期望： 新增一个工具（如 feishu_download_resource），允许机器人根据 message_id 和 file_key 下载群消息中的文件附件，并保存到本地或上传到云空间。
技术参考： 插件内部已有实现 downloadMessageResourceFeishu()（见 extensions/feishu/src/media.ts），只需... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-12T15:04:55Z | 2026-04-28T04:40:59Z | 2026-04-26T03:00:23Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Jackrmle | 135,839,681 | U_kgDOCBi_wQ | User | false |
openclaw/openclaw | 4,329,886,020 | I_kwDOQb6kR88AAAABAhTRRA | 71,913 | https://github.com/openclaw/openclaw/issues/71913 | https://api.github.com/repos/openclaw/openclaw/issues/71913 | [Bug]: [2026.4.24 regression] Gateway crashes in a restart loop on EC2 — unhandled CIAO ANNOUNCEMENT CANCELLED from bonjour plugin | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading to 2026.4.24, the gateway enters an infinite crash-restart loop on EC2. The gateway starts successfully, but the bundled bonjour plugin attempts mDNS multicast announcement, which EC2 does not support. The an... | closed | completed | true | 1 | [
"bug",
"regression"
] | [
"vincentkoc"
] | 2026-04-26T03:00:26Z | 2026-04-28T04:39:44Z | 2026-04-26T03:15:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ericzh-ops | 274,634,528 | U_kgDOEF6XIA | User | false |
openclaw/openclaw | 4,065,670,250 | I_kwDOQb6kR87yVTRq | 44,197 | https://github.com/openclaw/openclaw/issues/44197 | https://api.github.com/repos/openclaw/openclaw/issues/44197 | [Bug]: 飞书插件无法接收消息 - 长连接模式不工作 | ### Bug type
Regression (worked before, now fails)
### Summary
飞书插件配置完成后，机器人可以发送消息但无法接收消息。用户通过飞书发送消息给机器人，OpenClaw 没有收到，也没有创建对应的会话。
### Steps to reproduce
1. 配置飞书应用并发布
2. 添加事件订阅（im.message.receive_v1 等）
3. 使用 natapp 配置内网穿透
4. 用户在飞书中给机器人发送消息
5. 检查 OpenClaw sessions_list，发现没有飞书会话
### Expected behavior
用户在飞书中给机器人发送消... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-12T15:59:18Z | 2026-04-28T04:40:58Z | 2026-04-26T03:00:28Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Shannnnne | 266,429,455 | U_kgDOD-FkDw | User | false |
openclaw/openclaw | 4,065,684,543 | I_kwDOQb6kR87yVWw_ | 44,203 | https://github.com/openclaw/openclaw/issues/44203 | https://api.github.com/repos/openclaw/openclaw/issues/44203 | [Bug]: Edit tool fails everytime because of "Missing required parameters" | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Every call to edit tool fails with the error message "[tools] edit failed: Missing required parameters: oldText (oldText or old_string), newText (newText or new_string). Supply correct parameters before retrying.", forcing the model to use ... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T16:01:38Z | 2026-04-28T04:40:57Z | 2026-04-26T03:00:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | hexoctal-sur | 13,842,432 | MDQ6VXNlcjEzODQyNDMy | User | false |
openclaw/openclaw | 4,066,117,223 | I_kwDOQb6kR87yXAZn | 44,261 | https://github.com/openclaw/openclaw/issues/44261 | https://api.github.com/repos/openclaw/openclaw/issues/44261 | [Bug]: Control UI (webchat) messages not mirrored to external channels (Discord, Telegram, etc.) | ## Bug Description
When sending messages from Control UI (webchat), agent replies are no longer being mirrored to external channels (Discord, Telegram, etc.). This is a regression from v3.2.
**Previously (v3.2):**
- Replies from Control UI would be duplicated to Discord
- Allowed conversation continuity on mobile dev... | closed | not_planned | true | 8 | [] | [
"vincentkoc"
] | 2026-03-12T17:19:36Z | 2026-04-28T04:40:56Z | 2026-04-26T03:00:38Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kikkishook | 258,604,130 | U_kgDOD2n8Yg | User | false |
openclaw/openclaw | 4,066,360,721 | I_kwDOQb6kR87yX72R | 44,298 | https://github.com/openclaw/openclaw/issues/44298 | https://api.github.com/repos/openclaw/openclaw/issues/44298 | Audit docs credential examples for secret-scanner-safe placeholders | ## Problem
Credential examples in docs can accidentally trip secret scanners when placeholders are too close to real secret formats. A recent example was a literal `BEGIN PRIVATE KEY` placeholder in docs that triggered the `detect-private-key` pre-commit hook.
## Why It Matters
- Docs-only changes can fail CI for no... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T18:07:00Z | 2026-04-28T04:40:56Z | 2026-04-26T03:00:42Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | aniruddhaadak80 | 127,435,065 | U_kgDOB5iBOQ | User | false |
openclaw/openclaw | 4,066,888,455 | I_kwDOQb6kR87yZ8sH | 44,352 | https://github.com/openclaw/openclaw/issues/44352 | https://api.github.com/repos/openclaw/openclaw/issues/44352 | [Bug]: Memory search shows "no such module: fts5" on Debian systems | **Describe the bug**
OpenClaw memory system reports:
> fts unavailable: no such module: fts5
This happens on Debian-based systems (tested on current Umbrel/Debian 13).
**Diagnostics performed**
- `openclaw memory status --deep` consistently shows FTS: unavailable
- System sqlite3 has ENABLE_FTS5 compiled in
- sqlite... | closed | not_planned | true | 5 | [] | [] | 2026-03-12T19:52:04Z | 2026-04-28T04:40:54Z | 2026-04-26T03:00:53Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | bedair81 | 238,657,937 | U_kgDODjmhkQ | User | false |
openclaw/openclaw | 4,066,986,647 | I_kwDOQb6kR87yaUqX | 44,361 | https://github.com/openclaw/openclaw/issues/44361 | https://api.github.com/repos/openclaw/openclaw/issues/44361 | fix(backup): archive integrity cross-check fails on Windows (tar entry path mismatch) | ## Summary
The post-tar archive integrity cross-check in `backupCreateCommand` fails on Windows when exclude patterns are active (two-step archive path). The check compares `buildBackupArchivePath()` output (forward-slash POSIX paths) against `listArchiveEntries()` output, but on Windows the paths don't match — causin... | closed | not_planned | true | 4 | [] | [] | 2026-03-12T20:13:02Z | 2026-04-28T04:40:53Z | 2026-04-26T03:00:57Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | brettdavies | 26,392,883 | MDQ6VXNlcjI2MzkyODgz | User | false |
openclaw/openclaw | 4,067,034,134 | I_kwDOQb6kR87yagQW | 44,365 | https://github.com/openclaw/openclaw/issues/44365 | https://api.github.com/repos/openclaw/openclaw/issues/44365 | BUG: install.sh prefix override causes silent update failures with sudo npm i -g | ### Bug type
Regression (worked before, now fails)
### Summary
install.sh silently sets prefix=~/.npm-global in ~/.npmrc on Linux when the system npm prefix isn't user-writable. Subsequent sudo npm i -g openclaw@latest installs to the system prefix (/usr/lib/node_modules/) instead of ~/.npm-global, because sudo igno... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-12T20:22:05Z | 2026-04-28T04:40:52Z | 2026-04-26T03:01:02Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kAIborg24 | 261,855,322 | U_kgDOD5uYWg | User | false |
openclaw/openclaw | 4,067,205,736 | I_kwDOQb6kR87ybKJo | 44,381 | https://github.com/openclaw/openclaw/issues/44381 | https://api.github.com/repos/openclaw/openclaw/issues/44381 | feat: retry same model on transient errors before falling to next candidate | ## Problem
When a model returns a transient error (HTTP 408 timeout, 429 rate limit), the fallback system immediately moves to the next candidate (usually a different provider). This means a single transient blip on Anthropic causes the entire session to fall over to OpenAI, even though retrying the same model would l... | closed | not_planned | true | 2 | [] | [] | 2026-03-12T20:54:42Z | 2026-04-28T04:40:52Z | 2026-04-26T03:01:07Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nicolasgrasset | 347,560 | MDQ6VXNlcjM0NzU2MA== | User | false |
openclaw/openclaw | 4,067,475,795 | I_kwDOQb6kR87ycMFT | 44,400 | https://github.com/openclaw/openclaw/issues/44400 | https://api.github.com/repos/openclaw/openclaw/issues/44400 | Feature request: channels.discord.connectStagger config option | ## Problem
When the gateway starts with multiple Discord bot accounts (e.g., 9 agents), all bots connect simultaneously via `Promise.all()`. This overwhelms Discord's API with concurrent connections, triggering 503 'upstream overflow' errors. Failed bots then enter a 5-second auto-restart loop, creating bursts of retr... | closed | completed | true | 1 | [] | [] | 2026-03-12T21:50:52Z | 2026-04-28T04:40:49Z | 2026-04-26T03:02:51Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | ghimbib | 71,830,060 | MDQ6VXNlcjcxODMwMDYw | User | false |
openclaw/openclaw | 4,067,564,556 | I_kwDOQb6kR87ychwM | 44,406 | https://github.com/openclaw/openclaw/issues/44406 | https://api.github.com/repos/openclaw/openclaw/issues/44406 | iMessage relay integration — solving the macOS-only bottleneck for OpenClaw agents | ### The problem
iMessage in OpenClaw has always required a Mac physically running Messages.app — either via the legacy `imsg` CLI bridge (now deprecated, see #35568) or BlueBubbles (still macOS-dependent). This creates a hard constraint for anyone running OpenClaw on Linux, a VPS, or a containerized setup: iMessage is... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T22:10:02Z | 2026-04-28T04:40:48Z | 2026-04-26T03:02:55Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | strang-em-bot | 267,720,410 | U_kgDOD_UW2g | User | false |
openclaw/openclaw | 4,067,889,287 | I_kwDOQb6kR87ydxCH | 44,447 | https://github.com/openclaw/openclaw/issues/44447 | https://api.github.com/repos/openclaw/openclaw/issues/44447 | [Bug] v2026.3.11: Discord thread context re-injected every turn (context pollution + cache bust) | ## Bug Summary
In OpenClaw v2026.3.11, Discord threads re-inject volatile metadata on every conversation turn, causing context contamination, unnecessary token increase, and prompt cache loss.
## Observed Symptoms
- In Discord threads, the following are re-injected every turn: `message_id`, `reply_to_id`, `timestamp... | closed | not_planned | true | 4 | [] | [] | 2026-03-12T23:40:33Z | 2026-04-28T04:40:48Z | 2026-04-26T03:02:59Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,067,894,290 | I_kwDOQb6kR87ydyQS | 44,449 | https://github.com/openclaw/openclaw/issues/44449 | https://api.github.com/repos/openclaw/openclaw/issues/44449 | [Bug] v2026.3.11: Discord thread metadata re-injected every turn (context pollution) | ## Bug Summary
En OpenClaw v2026.3.11, los threads de Discord reinyectan metadata volátil en cada turno de conversación, causando contaminación de contexto, aumento innecesario de tokens y pérdida de cacheo de prompts.
## Síntomas Observados
- En threads Discord se reinyectan por cada turno: `message_id`, `reply_to_... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T23:42:24Z | 2026-04-28T04:40:47Z | 2026-04-26T03:03:04Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,067,894,324 | I_kwDOQb6kR87ydyQ0 | 44,450 | https://github.com/openclaw/openclaw/issues/44450 | https://api.github.com/repos/openclaw/openclaw/issues/44450 | [Bug] v2026.3.11: CLI agent --channel discord --to routes to agent:*:main instead of live session | ## Bug Summary
En OpenClaw v2026.3.11, el CLI `openclaw agent --channel discord --to ...` resuelve incorrectamente la sesión `agent:*:main` en lugar de la sesión viva asociada al canal/thread Discord especificado.
## Síntomas Observados
- `openclaw agent --channel discord --to ...` parece entregar pero resuelve sesi... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T23:42:25Z | 2026-04-28T04:40:46Z | 2026-04-26T03:03:08Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,067,894,357 | I_kwDOQb6kR87ydyRV | 44,451 | https://github.com/openclaw/openclaw/issues/44451 | https://api.github.com/repos/openclaw/openclaw/issues/44451 | [Bug] v2026.3.11: Session resolver in compact lacks delivery-context lookup | ## Bug Summary
En OpenClaw v2026.3.11, el session resolver dentro del comando `agent` en `dist/compact-1mmJ_KWL.js` no realiza lookup por `delivery context`, causando que las sesiones se resuelvan incorrectamente a `agent:*:main` en lugar de la sesión viva del canal objetivo.
## Síntomas Observados
- El comando `age... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T23:42:26Z | 2026-04-28T04:40:45Z | 2026-04-26T03:03:12Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,067,909,438 | I_kwDOQb6kR87yd18- | 44,456 | https://github.com/openclaw/openclaw/issues/44456 | https://api.github.com/repos/openclaw/openclaw/issues/44456 | [Bug]: gog skill fails with "No session found: gog" even though gog works via CLI | ### Bug type
Regression (worked before, now fails)
### Summary
OpenClaw 2026.3.11 on Ubuntu VPS.
gog CLI works and Google OAuth is successful.
openclaw skills check shows gog as ready.
But when running in the browser:
/skill gog gmail labels list
the UI returns:
No session found: gog
Gateway logs show:
sessio... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-12T23:48:01Z | 2026-04-28T04:40:45Z | 2026-04-26T03:03:16Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | brgamble2008-spec | 267,790,882 | U_kgDOD_YqIg | User | false |
openclaw/openclaw | 4,067,981,861 | I_kwDOQb6kR87yeHol | 44,468 | https://github.com/openclaw/openclaw/issues/44468 | https://api.github.com/repos/openclaw/openclaw/issues/44468 | Allow contextWindow/maxTokens overrides in agents.defaults.models | ## Problem
When users want to override `contextWindow` for a model (e.g. setting `openai-codex/gpt-5.4` to 1M tokens), the only path is:
1. Add a full provider entry in `models.providers.openai-codex`
2. Include `baseUrl`, `api`, `name` (all required by schema validation) even though the built-in provider already def... | closed | not_planned | true | 3 | [] | [] | 2026-03-13T00:11:02Z | 2026-04-28T04:40:44Z | 2026-04-26T03:03:21Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kiki830621 | 17,489,494 | MDQ6VXNlcjE3NDg5NDk0 | User | false |
openclaw/openclaw | 4,068,005,300 | I_kwDOQb6kR87yeNW0 | 44,472 | https://github.com/openclaw/openclaw/issues/44472 | https://api.github.com/repos/openclaw/openclaw/issues/44472 | [Patch] Runtime bundle fixes for premature success bug (compact, pi-embedded, reply) | ## Summary
This issue documents specific runtime patches applied to address the "premature success" bug in subagent workflows (see related issue).
## Patched Files
The following files in `dist/` directory were modified:
### 1. compact-1mmJ_KWL.js
**Change:** Modified internal completion event wording
- **Before:** ... | closed | not_planned | true | 1 | [] | [] | 2026-03-13T00:18:08Z | 2026-04-28T04:40:43Z | 2026-04-26T03:03:25Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,068,022,524 | I_kwDOQb6kR87yeRj8 | 44,478 | https://github.com/openclaw/openclaw/issues/44478 | https://api.github.com/repos/openclaw/openclaw/issues/44478 | Telegram and TUI stay in separate sessions; outbound cross-surface mirroring does not occur | # Bug: Telegram and TUI stay in separate sessions; outbound cross-surface mirroring does not occur
## Version
- OpenClaw `2026.3.11 (29dc654)`
## Environment
- Linux `6.17.0-14-generic`
- Gateway local and reachable
- Telegram channel enabled and healthy
- `session.dmScope = "main"`
## Summary
Messages and replies w... | closed | completed | true | 2 | [] | [] | 2026-03-13T00:23:59Z | 2026-04-28T04:40:42Z | 2026-04-26T03:03:29Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | rockyiacool432-prog | 265,914,013 | U_kgDOD9mGnQ | User | false |
openclaw/openclaw | 4,068,085,936 | I_kwDOQb6kR87yehCw | 44,488 | https://github.com/openclaw/openclaw/issues/44488 | https://api.github.com/repos/openclaw/openclaw/issues/44488 | [Feature]: Feature Request: 优化心跳会话的显示标签 | ### Summary
**问题现状：**
- 执行 `/new` 后，控制面板聊天窗口右上角显示 "Main Session"
- 当系统发送心跳轮询（heartbeat poll）时，标签变为 "heartbeat"
- 这让用户困惑，误以为是会话状态变化
**期望行为：**
- 心跳轮询不应改变会话标签的显示
- 标签应保持显示 "Main Session" 或用户可识别的会话名称
- 心跳是后台系统行为，对用户应透明
**建议方案（任选其一）：**
1. **心跳会话继承主会话标签** — heartbeat 请求使用与主会话相同的标签
2. **新增配置项** — 如 `ui.showHeartbeatLabel: ... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-13T00:45:36Z | 2026-04-28T04:40:41Z | 2026-04-26T03:03:33Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | kentfanye | 62,164,032 | MDQ6VXNlcjYyMTY0MDMy | User | false |
openclaw/openclaw | 4,068,212,218 | I_kwDOQb6kR87ye_36 | 44,513 | https://github.com/openclaw/openclaw/issues/44513 | https://api.github.com/repos/openclaw/openclaw/issues/44513 | Feature: global default for reasoning visibility (reasoningLevel) | ## Problem
When a model is reasoning-capable (e.g. has `reasoning: true` in the catalog or provider metadata), the default `reasoningLevel` resolves to `"on"` via `resolveReasoningDefault()`. This causes every reasoning-capable model (Anthropic Sonnet/Opus, local models with `reasoning: true`) to send thinking/reasoni... | closed | not_planned | true | 2 | [] | [] | 2026-03-13T01:26:08Z | 2026-04-28T04:40:41Z | 2026-04-26T03:03:37Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | paulomcg | 6,182,687 | MDQ6VXNlcjYxODI2ODc= | User | false |
openclaw/openclaw | 4,068,519,775 | I_kwDOQb6kR87ygK9f | 44,580 | https://github.com/openclaw/openclaw/issues/44580 | https://api.github.com/repos/openclaw/openclaw/issues/44580 | Bug: preserve fallback thinking signatures at replay boundary | ## Summary
Session-history replay can mishandle provider-native thinking signatures during model fallback.
## Expected behavior
- Anthropic signed thinking is preserved when replay remains compatible, including Gemini fallback
- incompatible signatures are stripped only when replay target is `openai-completions`
- w... | closed | not_planned | true | 1 | [] | [] | 2026-03-13T03:07:21Z | 2026-04-28T04:40:39Z | 2026-04-26T03:03:46Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | nashborges | 9,395,917 | MDQ6VXNlcjkzOTU5MTc= | User | false |
openclaw/openclaw | 4,068,529,477 | I_kwDOQb6kR87ygNVF | 44,587 | https://github.com/openclaw/openclaw/issues/44587 | https://api.github.com/repos/openclaw/openclaw/issues/44587 | [Bug]: Premature success in subagent flows - wording and verification fix | ## Problem
Internal event wording in OpenClaw runtime contributed to a "premature success / missing final verification" pattern where agents treated "subagent completed successfully" as equivalent to "external objective completed".
## Symptoms
- Agents responding "done/ready/complete" without verifying if external r... | closed | not_planned | true | 2 | [] | [] | 2026-03-13T03:10:14Z | 2026-04-28T04:40:38Z | 2026-04-26T03:03:50Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,068,632,568 | I_kwDOQb6kR87ygmf4 | 44,616 | https://github.com/openclaw/openclaw/issues/44616 | https://api.github.com/repos/openclaw/openclaw/issues/44616 | Bug: Anthropic OAuth refresh intermittently fails across chats/topics | # Bug Report: Anthropic OAuth refresh fails intermittently across Telegram topics
- Date (UTC): 2026-03-13 03:40:11
- OpenClaw version: OpenClaw 2026.3.11 (29dc654)
- Default model: openai-codex/gpt-5.3-codex
## Symptom
Agent fails before reply in some chats/topics with:
`OAuth token refresh failed for anthropic: Fai... | closed | not_planned | true | 10 | [] | [] | 2026-03-13T03:41:19Z | 2026-04-28T04:40:38Z | 2026-04-26T03:03:54Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | IIIyban | 137,302,231 | U_kgDOCC8Q1w | User | false |
openclaw/openclaw | 4,068,636,231 | I_kwDOQb6kR87ygnZH | 44,618 | https://github.com/openclaw/openclaw/issues/44618 | https://api.github.com/repos/openclaw/openclaw/issues/44618 | [Bug]: inline /model <provider/model> <message> is stripped but not applied as a one-shot model override | ## Summary
`/model <provider/model> <message>` inside a normal chat message is currently stripped from the prompt, but the selected model is **not** applied to that message.
This shows up most obviously on chat surfaces like Telegram and Discourse where users reasonably expect:
- `/model openai/gpt-5.2 hello` to ans... | closed | not_planned | true | 2 | [] | [] | 2026-03-13T03:42:38Z | 2026-04-28T04:40:37Z | 2026-04-26T03:03:58Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | constansino | 65,108,260 | MDQ6VXNlcjY1MTA4MjYw | User | false |
openclaw/openclaw | 4,068,890,399 | I_kwDOQb6kR87yhlcf | 44,672 | https://github.com/openclaw/openclaw/issues/44672 | https://api.github.com/repos/openclaw/openclaw/issues/44672 | [Bug]: macOS app can stay stuck on generic 'pairing required' after node->operator upgrade approval | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When the macOS app is already paired as a node and then requests operator access for the local control/chat UI, the app only shows a generic "pairing required" state, does not make it clear that this is a `node -> operator` role upgrade for... | closed | completed | true | 2 | [] | [] | 2026-03-13T04:59:27Z | 2026-04-28T04:40:36Z | 2026-04-26T03:04:03Z | CONTRIBUTOR | null | 20260428T060020Z | 2026-04-28T06:00:20Z | constansino | 65,108,260 | MDQ6VXNlcjY1MTA4MjYw | User | false |
openclaw/openclaw | 4,068,937,596 | I_kwDOQb6kR87yhw98 | 44,677 | https://github.com/openclaw/openclaw/issues/44677 | https://api.github.com/repos/openclaw/openclaw/issues/44677 | [Bug]: Feishu plugin returns undefined tenant_access_token | ### Bug type
Regression (worked before, now fails)
### Summary
Feishu plugin tools (feishu_app_scopes, feishu_doc, etc.) return error: Cannot destructure property 'tenant_access_token' of '(intermediate value)' as it is undefined. This happens even though the Feishu app_id and app_secret are valid and the token API ... | closed | not_planned | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-13T05:11:18Z | 2026-04-28T04:40:35Z | 2026-04-26T03:04:07Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | sajdijjsid | 150,767,141 | U_kgDOCPyGJQ | User | false |
openclaw/openclaw | 4,068,962,873 | I_kwDOQb6kR87yh3I5 | 44,679 | https://github.com/openclaw/openclaw/issues/44679 | https://api.github.com/repos/openclaw/openclaw/issues/44679 | Thinking block auto-recovery PR #22270 was never merged (circular closure) | ## Summary
PR #22270 / Issue #22233 contain a working fix for auto-recovery when Anthropic rejects requests due to modified `thinking`/`redacted_thinking` blocks. However, both were closed in a circular loop — neither was actually merged.
## Impact
This is the single most common unrecoverable session failure for Ant... | closed | completed | true | 1 | [] | [] | 2026-03-13T05:18:03Z | 2026-04-28T04:40:35Z | 2026-04-26T03:04:11Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | alexph-dev | 129,837,470 | U_kgDOB70png | User | false |
openclaw/openclaw | 4,134,555,912 | I_kwDOQb6kR872cFEI | 54,429 | https://github.com/openclaw/openclaw/issues/54429 | https://api.github.com/repos/openclaw/openclaw/issues/54429 | Gateway Service Installation Failure: Missing systemd Service File | # Gateway Service Installation Failure: Missing systemd Service File
## Issue Type
- [x] Bug Report
- [ ] Feature Request
- [ ] Documentation Issue
## Description
The OpenClaw Gateway service fails to install with a systemd error indicating the service unit file does not exist. This affects users attempting to run t... | closed | not_planned | true | 1 | [
"clownfish"
] | [] | 2026-03-25T11:08:36Z | 2026-04-28T04:32:20Z | 2026-04-26T03:04:37Z | NONE | null | 20260428T060020Z | 2026-04-28T06:00:20Z | 43622283 | 42,956,557 | MDQ6VXNlcjQyOTU2NTU3 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.