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 | 4,150,317,136 | I_kwDOQb6kR873YNBQ | 55,640 | https://github.com/openclaw/openclaw/issues/55640 | https://api.github.com/repos/openclaw/openclaw/issues/55640 | Bug: messaging tool suppress mechanism skips when tool fails then retries in same turn, causing duplicate delivery | ## Summary
When a `message` tool call **fails** (e.g. `Network request for 'sendMessage' failed`) and is then **retried** in the same turn, both the retry tool call AND the text reply are delivered to the user. This produces duplicate messages.
## Root cause (traced in source)
In `shouldSuppressMessagingToolReplies`... | open | null | false | 1 | [] | [] | 2026-03-27T07:45:01Z | 2026-05-04T22:00:35Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | yyxiaohui | 235,090,347 | U_kgDODgMxqw | User | false |
openclaw/openclaw | 4,150,711,948 | I_kwDOQb6kR873ZtaM | 55,678 | https://github.com/openclaw/openclaw/issues/55678 | https://api.github.com/repos/openclaw/openclaw/issues/55678 | [Feature]: UI 用量显示为 0，实际在消耗 token | ### Summary
UI 用量显示为 0，实际在消耗 token
### Problem to solve
项目 说明
问题 UI 用量显示为 0，实际在消耗 token
原因 usageProviders 列表没有 dashscope
### Proposed solution
建议方案 添加 DashScope 用量 API 支持
优先级 中 - 高
### Alternatives considered
_No response_
### Impact
影响 中国/亚太用户无法追踪成本
优先级 中 - 高
### Evidence/examples
_No response_
### ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-27T08:47:28Z | 2026-05-04T22:00:43Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | tangyusheng | 133,026,816 | U_kgDOB-3UAA | User | false |
openclaw/openclaw | 4,150,866,101 | I_kwDOQb6kR873aTC1 | 55,691 | https://github.com/openclaw/openclaw/issues/55691 | https://api.github.com/repos/openclaw/openclaw/issues/55691 | Feature Request: Support message recall/edit events to cancel running tasks | ## Description
Currently, when a user recalls (deletes) or edits a message on a messaging platform (Feishu, Telegram, Discord, WhatsApp, etc.), the agent continues processing the original message. This wastes compute/tokens, blocks the session from accepting new instructions, and breaks the natural conversational flow... | open | null | false | 1 | [] | [] | 2026-03-27T09:11:26Z | 2026-05-04T22:00:45Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | icytail-a11y | 251,263,718 | U_kgDODvn65g | User | false |
openclaw/openclaw | 4,151,236,526 | I_kwDOQb6kR873bteu | 55,718 | https://github.com/openclaw/openclaw/issues/55718 | https://api.github.com/repos/openclaw/openclaw/issues/55718 | [Feature]: [Feature Request] Support custom display labels for channel accounts in control-ui session list | ### Summary
Add a `displayLabel` configuration field for channel accounts to customize how sessions appear in the control-ui session selector.
### Problem to solve
Currently, the control-ui session selector displays session names derived from session keys (e.g., `feishu:g-oc_xxx`, `telegram:direct:xxx`) or `origin.l... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-27T10:07:30Z | 2026-05-04T22:00:47Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | aoisw | 225,109,537 | U_kgDODWrmIQ | User | false |
openclaw/openclaw | 4,151,534,448 | I_kwDOQb6kR873c2Nw | 55,743 | https://github.com/openclaw/openclaw/issues/55743 | https://api.github.com/repos/openclaw/openclaw/issues/55743 | Add an "Approve and Run" action for assistant-suggested terminal commands in the Control UI | ## Summary
When the assistant suggests shell commands in chat, the current workflow often requires the user to manually copy/paste those commands into Terminal. This creates unnecessary friction, especially during troubleshooting, setup, and iterative admin work.
A better UX would let the Control UI show suggested co... | open | null | false | 1 | [] | [] | 2026-03-27T10:54:01Z | 2026-05-04T22:00:53Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | TEEllis22 | 3,057,055 | MDQ6VXNlcjMwNTcwNTU= | User | false |
openclaw/openclaw | 4,162,159,112 | I_kwDOQb6kR874FYII | 56,698 | https://github.com/openclaw/openclaw/issues/56698 | https://api.github.com/repos/openclaw/openclaw/issues/56698 | Bug: Hook bootstrap virtual files accumulate on every restart (no dedup) | ## Description
Virtual files injected by hooks via `event.context.bootstrapFiles.push()` accumulate on every gateway restart/session reload. There is **no dedup mechanism** at either the Hook layer or the framework layer.
## Steps to Reproduce
1. Install the `self-improving-agent` skill (which registers a `self-impr... | closed | completed | true | 1 | [] | [] | 2026-03-29T01:33:09Z | 2026-05-01T04:43:16Z | 2026-04-28T14:42:26Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Rickvai | 270,566,030 | U_kgDOECCCjg | User | false |
openclaw/openclaw | 4,152,108,612 | I_kwDOQb6kR873fCZE | 55,771 | https://github.com/openclaw/openclaw/issues/55771 | https://api.github.com/repos/openclaw/openclaw/issues/55771 | [Feature]: add block/blockReason to PluginHookBeforeModelResolveResult | ### Summary
## Problem
`before_tool_call` supports `{ block: true, blockReason: "..." }` to prevent tool execution, but `before_model_resolve` has no equivalent mechanism. Plugins that need to prevent a model call (e.g., budget enforcement when budget is exhausted) can only throw an exception, which is caught by the ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-27T12:15:17Z | 2026-05-03T12:30:38Z | null | NONE | null | 20260503T150031Z | 2026-05-03T15:00:31Z | amavashev | 6,786,184 | MDQ6VXNlcjY3ODYxODQ= | User | false |
openclaw/openclaw | 4,164,875,026 | I_kwDOQb6kR874PvMS | 57,139 | https://github.com/openclaw/openclaw/issues/57139 | https://api.github.com/repos/openclaw/openclaw/issues/57139 | sessions_history should support reading archived/reset sessions | ## Problem
When a session is reset via `/new` or daily auto-reset, the old session transcript is archived as a `.jsonl.reset.*` file on disk. However, the `sessions_history` tool only returns history for the **current active session**. There is no agent-facing tool to read archived session transcripts.
This creates t... | closed | not_planned | true | 3 | [] | [] | 2026-03-29T16:00:37Z | 2026-05-01T04:43:15Z | 2026-04-28T14:42:36Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | dwbutler | 746,336 | MDQ6VXNlcjc0NjMzNg== | User | false |
openclaw/openclaw | 4,152,452,363 | I_kwDOQb6kR873gWUL | 55,792 | https://github.com/openclaw/openclaw/issues/55792 | https://api.github.com/repos/openclaw/openclaw/issues/55792 | [Feature]: Catch up on missed inbound messages after gateway restart | ## Problem
When the gateway restarts (config-triggered SIGUSR1, crash, manual restart), inbound messages sent during the downtime window are silently dropped. The Discord/Telegram/etc. provider reconnects and starts processing new events, but never checks the channel history for messages it missed while offline.
**Im... | open | null | false | 1 | [] | [] | 2026-03-27T13:01:36Z | 2026-05-04T22:01:05Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | Kaspre | 36,520,309 | MDQ6VXNlcjM2NTIwMzA5 | User | false |
openclaw/openclaw | 4,152,650,110 | I_kwDOQb6kR873hGl- | 55,806 | https://github.com/openclaw/openclaw/issues/55806 | https://api.github.com/repos/openclaw/openclaw/issues/55806 | Tool call observability: input sanitization, error categorization, and context writeback | ## Problem
Tool call logging captures some information but lacks key observability features needed for production auditing:
1. **Input parameters are not sanitized** - full arguments are logged, potentially exposing sensitive data
2. **Errors are not categorized** - only raw error messages, no structured error types
... | open | null | false | 1 | [] | [] | 2026-03-27T13:27:19Z | 2026-05-04T22:01:06Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | laterequation | 138,646,637 | U_kgDOCEOUbQ | User | false |
openclaw/openclaw | 4,153,311,994 | I_kwDOQb6kR873joL6 | 55,840 | https://github.com/openclaw/openclaw/issues/55840 | https://api.github.com/repos/openclaw/openclaw/issues/55840 | [Feature]: Bring back Chrome Extension Relay support (removed in 2026.3.22) | ### Summary
## Summary
The Chrome Extension-based browser relay was removed in version 2026.3.22 in favor of Playwright-based browser automation. This issue requests the restoration of the Extension Relay feature.
## Why This Matters
### Features Lost
- **Integrated Extension Relay** on port 18792 (no separate proc... | open | null | false | 3 | [
"enhancement"
] | [] | 2026-03-27T14:39:21Z | 2026-05-04T22:01:14Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | evernation | 161,215,019 | U_kgDOCZvyKw | User | false |
openclaw/openclaw | 4,153,456,240 | I_kwDOQb6kR873kLZw | 55,850 | https://github.com/openclaw/openclaw/issues/55850 | https://api.github.com/repos/openclaw/openclaw/issues/55850 | [Bug]: Control UI content clips under the collapsed sidebar in Safari on macOS | ## Bug type
Behavior bug (incorrect output/state without crash)
## Summary
In Safari on macOS, the OpenClaw Control UI can render main content underneath the collapsed left sidebar rail, clipping the left edge of multiple views.
## Steps to reproduce
1. Open the OpenClaw Control UI in Safari at `http://127.0.0.1:1878... | open | null | false | 2 | [] | [] | 2026-03-27T14:57:23Z | 2026-05-04T22:01:15Z | null | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | betamod | 8,882,726 | MDQ6VXNlcjg4ODI3MjY= | User | false |
openclaw/openclaw | 4,153,630,188 | I_kwDOQb6kR873k13s | 55,858 | https://github.com/openclaw/openclaw/issues/55858 | https://api.github.com/repos/openclaw/openclaw/issues/55858 | feat: per-channel debounce/cap/drop queue options (mode already supported) | ## Feature Request
Different chat interfaces have fundamentally different input patterns, making a single global queue config a poor fit for all channels simultaneously.
## Problem
Currently queue settings (mode, debounce, cap, drop) are global. This means you're always picking the least-bad tradeoff across all chan... | open | null | false | 2 | [] | [] | 2026-03-27T15:19:14Z | 2026-05-03T07:44:27Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | cwil2072 | 14,353,178 | MDQ6VXNlcjE0MzUzMTc4 | User | false |
openclaw/openclaw | 4,153,885,130 | I_kwDOQb6kR873l0HK | 55,880 | https://github.com/openclaw/openclaw/issues/55880 | https://api.github.com/repos/openclaw/openclaw/issues/55880 | Scheduled auto-upgrade cron can fail without visible notification, and gateway recovery does not restore macOS LaunchAgent gateway | ### Summary
Scheduled auto-upgrade cron can fail without a reliable visible notification, and post-upgrade gateway recovery does not restore a macOS LaunchAgent gateway.
### Environment
- OpenClaw version: `2026.3.24 (cff6dc9)`
- OS: macOS
- Gateway managed as LaunchAgent
- Gateway port: `18789`
- Cron job name: `he... | open | null | false | 3 | [] | [] | 2026-03-27T15:53:33Z | 2026-05-03T08:38:51Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | samrogers-com | 29,523,222 | MDQ6VXNlcjI5NTIzMjIy | User | false |
openclaw/openclaw | 4,153,890,235 | I_kwDOQb6kR873l1W7 | 55,882 | https://github.com/openclaw/openclaw/issues/55882 | https://api.github.com/repos/openclaw/openclaw/issues/55882 | [Bug]: Agent can drop promised outputs after task switching (need durable pending-deliverables queue; no gateway restart needed) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When the agent is mid-work on a user request and its attention is switched to a higher-priority task (e.g. handling a high-frequency hook such as `autonomy-guard` confirmations, config changes, gateway restarts... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T15:54:22Z | 2026-05-02T17:54:50Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | yww325 | 7,898,118 | MDQ6VXNlcjc4OTgxMTg= | User | false |
openclaw/openclaw | 4,153,954,078 | I_kwDOQb6kR873mE8e | 55,888 | https://github.com/openclaw/openclaw/issues/55888 | https://api.github.com/repos/openclaw/openclaw/issues/55888 | [Feature]: 🚀 [Performance Insight] Unlocking 26.7k Context Window on M4 Pro: Fixing the 8k Compaction Lag (64GB RAM Only) | ### Summary
Optimization report for OpenClaw 2026.3.24 (cff6dc9) on Mac Mini M4 Pro (64GB RAM). Identified a hidden ~4k token system buffer in the memoryFlush logic that causes unnecessary background compaction (20-30s latency) at only 8k tokens. Successfully increased usable context to 26.7k tokens with <1,000ms late... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-27T16:02:48Z | 2026-05-03T07:44:29Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | Vic-HA | 182,946,027 | U_kgDOCueI6w | User | false |
openclaw/openclaw | 4,154,338,477 | I_kwDOQb6kR873niyt | 55,915 | https://github.com/openclaw/openclaw/issues/55915 | https://api.github.com/repos/openclaw/openclaw/issues/55915 | [Feature]: Cloud Relay — openclaw login + heartbeat + relay for remote mobile access | ## Summary
Add cloud relay support so Gateways can register with a user's cloud account and accept relayed mobile connections from anywhere — without Tailscale or same-network requirements. Three components: `openclaw login` (OAuth), background heartbeat, and WebSocket relay acceptance.
## Problem to solve
Currently... | open | null | false | 1 | [] | [] | 2026-03-27T16:45:27Z | 2026-05-02T17:30:30Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | ovai-felix | 185,858,100 | U_kgDOCxP4NA | User | false |
openclaw/openclaw | 4,154,358,419 | I_kwDOQb6kR873nnqT | 55,917 | https://github.com/openclaw/openclaw/issues/55917 | https://api.github.com/repos/openclaw/openclaw/issues/55917 | [Bug]: Telegram documents sometimes arrive only as <media:document> instead of a real attachment | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
I’m seeing intermittent Telegram inbound attachment failures in OpenClaw.
What happens
• A document is sent to the bot in Telegram.
• The message reaches the session, but instead of a usable attachment/path, ... | open | null | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-27T16:47:58Z | 2026-05-03T07:04:06Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | IgorRuzhnikov | 66,778,078 | MDQ6VXNlcjY2Nzc4MDc4 | User | false |
openclaw/openclaw | 4,154,627,531 | I_kwDOQb6kR873opXL | 55,935 | https://github.com/openclaw/openclaw/issues/55935 | https://api.github.com/repos/openclaw/openclaw/issues/55935 | Heartbeat: Use ISO 8601 for Current time and separate trigger timestamp | ## Problem
The heartbeat prompt includes a `Current time` line that uses a locale-dependent format:
> Current time: Friday, March 27th, 2026 — 04:00 (Europe/Oslo) / 2026-03-27 03:00 UTC
This has two issues:
1. **Inconsistent formatting** — Does not follow ISO 8601, the international standard for date/time exchange.... | open | null | false | 1 | [] | [] | 2026-03-27T17:25:33Z | 2026-05-02T17:54:51Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | TommoT2 | 106,430,507 | U_kgDOBlgAKw | User | false |
openclaw/openclaw | 4,154,946,745 | I_kwDOQb6kR873p3S5 | 55,966 | https://github.com/openclaw/openclaw/issues/55966 | https://api.github.com/repos/openclaw/openclaw/issues/55966 | Telegram inbound media ENOSPC leaves zero-byte voice file and later crashes transcription with EOF | ## Summary
When the gateway hits `ENOSPC` while downloading inbound Telegram voice media, the media fetch path fails, but a zero-byte media file can still be left behind in the inbound media store. The local transcription command then crashes with a raw EOF/decode error when it is later invoked on that file.
## Eviden... | closed | completed | true | 2 | [] | [] | 2026-03-27T18:16:37Z | 2026-05-04T04:34:22Z | 2026-05-02T02:58:36Z | CONTRIBUTOR | null | 20260504T060031Z | 2026-05-04T06:00:31Z | OpenCodeEngineer | 261,470,075 | U_kgDOD5W3ew | User | false |
openclaw/openclaw | 4,155,882,315 | I_kwDOQb6kR873tbtL | 56,034 | https://github.com/openclaw/openclaw/issues/56034 | https://api.github.com/repos/openclaw/openclaw/issues/56034 | Session maintenance prunes active/running sessions | ## Bug
Session maintenance in `enforce` mode can prune sessions that are still actively running. It appears the pruning logic only considers age and entry count, but does not check whether a session is currently in-flight (status `running`).
## Impact
In setups with high subagent throughput (e.g. a cron job spawning... | open | null | false | 1 | [] | [] | 2026-03-27T20:57:22Z | 2026-05-03T08:24:27Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | 4gent04 | 266,393,869 | U_kgDOD-DZDQ | User | false |
openclaw/openclaw | 4,156,407,004 | I_kwDOQb6kR873vbzc | 56,072 | https://github.com/openclaw/openclaw/issues/56072 | https://api.github.com/repos/openclaw/openclaw/issues/56072 | Daily session reset silently discards context without memory flush or compaction | ## Problem
The daily session reset (default: 4 AM local time) archives the session transcript without triggering any memory preservation mechanism. Context from the previous session is silently lost unless the agent happened to write it to memory files in real-time.
This is especially impactful for users on large con... | open | null | false | 2 | [] | [] | 2026-03-27T23:13:19Z | 2026-05-03T06:07:09Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | jlian | 2,320,572 | MDQ6VXNlcjIzMjA1NzI= | User | false |
openclaw/openclaw | 4,156,523,005 | I_kwDOQb6kR873v4H9 | 56,078 | https://github.com/openclaw/openclaw/issues/56078 | https://api.github.com/repos/openclaw/openclaw/issues/56078 | [Bug]: Cron Job Output Not Saved in OpenClaw 2026.3.24 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
Cron Job runs in OpenClaw 2026.3.24 do not save output files because no delivery channel is detected, even though the job executes successfully.
### Steps to reproduce
1.Use OpenClaw 2026.3
2.Go to CONTROL → Cron Jobs
3.Cr... | open | null | false | 2 | [
"bug",
"regression"
] | [] | 2026-03-27T23:47:36Z | 2026-05-02T17:30:32Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | nobbywfc | 3,085,680 | MDQ6VXNlcjMwODU2ODA= | User | false |
openclaw/openclaw | 4,156,730,114 | I_kwDOQb6kR873wqsC | 56,096 | https://github.com/openclaw/openclaw/issues/56096 | https://api.github.com/repos/openclaw/openclaw/issues/56096 | Bug: Telegram sendChatAction infinite retry loop with no backoff | ## Severity: High
Makes the bot completely unresponsive until manual process kill.
## Steps to reproduce
1. Send a message while the configured model returns 529 (overloaded)
2. Reply fails, failover also fails
3. `sendChatAction` ("typing...") enters infinite retry loop every ~3 seconds
4. Loop persists across SIGUSR... | open | null | false | 4 | [] | [] | 2026-03-28T00:49:52Z | 2026-05-03T10:18:46Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | cal-rufus | 270,013,050 | U_kgDOEBgSeg | User | false |
openclaw/openclaw | 4,156,771,006 | I_kwDOQb6kR873w0q- | 56,102 | https://github.com/openclaw/openclaw/issues/56102 | https://api.github.com/repos/openclaw/openclaw/issues/56102 | [Bug]: ACP server rejects MCP protocolVersion: 2025-11-25 from VS Code 1.113 / Cursor | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
OpenClaw version 2026.3.24 does not support new Visual Studio Code and Cursor
Visual studio code
2026-03-27 21:35:24.125 [info] Stopping server OpenClaw (Clau)
2026-03-27 21:35:24.155 [info] Starting server OpenClaw (Clau)
2... | open | null | false | 1 | [
"bug"
] | [] | 2026-03-28T01:02:27Z | 2026-05-03T09:12:23Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | moliveto | 1,890,103 | MDQ6VXNlcjE4OTAxMDM= | User | false |
openclaw/openclaw | 4,156,827,983 | I_kwDOQb6kR873xClP | 56,106 | https://github.com/openclaw/openclaw/issues/56106 | https://api.github.com/repos/openclaw/openclaw/issues/56106 | Transcript JSONL encoding corrupted on Windows (GBK/UTF-8 mix) | # Bug Report: Transcript JSONL encoding corrupted on Windows
## Version
OpenClaw 2026.3.23 (ccfeecb)
## Environment
- OS: Windows_NT 10.0.26200
- Node.js: v24.14.0
- Channel: QQBot
## Problem
Chinese characters in chat transcripts (`.jsonl` files under `agents/main/sessions/`) are stored corrupted. Characters appear... | open | null | false | 1 | [] | [] | 2026-03-28T01:18:45Z | 2026-05-03T07:04:08Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | lujun2508 | 267,008,723 | U_kgDOD-o60w | User | false |
openclaw/openclaw | 4,156,899,735 | I_kwDOQb6kR873xUGX | 56,110 | https://github.com/openclaw/openclaw/issues/56110 | https://api.github.com/repos/openclaw/openclaw/issues/56110 | Feature Request: Add STATE.md to auto-loaded workspace bootstrap files | ## Problem
After context compaction, agents lose their working state — what they were doing, what's in progress, what decisions were just made. The current recovery requires reading multiple files manually, which is error-prone and often skipped.
Anthropic's own research on long-running agents ([Effective Harnesses f... | open | null | false | 2 | [] | [] | 2026-03-28T01:45:33Z | 2026-05-03T08:38:52Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | horakhty05 | 227,078,737 | U_kgDODYjyUQ | User | false |
openclaw/openclaw | 4,156,976,417 | I_kwDOQb6kR873xm0h | 56,118 | https://github.com/openclaw/openclaw/issues/56118 | https://api.github.com/repos/openclaw/openclaw/issues/56118 | Node browser proxy: Chrome MCP existing-session (user profile) fails with Connection closed — consent dialog never appears | ## Summary
When using the node browser proxy to access a Chrome MCP existing-session (`user` profile) on a remote node, the connection fails with `McpError: MCP error -32000: Connection closed`. Chrome never shows the consent dialog that would authorize the connection.
The same `user` profile works perfectly when acc... | open | null | false | 1 | [] | [] | 2026-03-28T02:13:42Z | 2026-05-02T16:01:19Z | null | NONE | null | 20260502T180029Z | 2026-05-02T18:00:29Z | mgrandau | 7,772,830 | MDQ6VXNlcjc3NzI4MzA= | User | false |
openclaw/openclaw | 4,157,084,622 | I_kwDOQb6kR873yBPO | 56,131 | https://github.com/openclaw/openclaw/issues/56131 | https://api.github.com/repos/openclaw/openclaw/issues/56131 | Bug: session reset archives are not discoverable via sessions_history, breaking context recovery after reset | ## Summary
When an active session is reset/rotated, the prior transcript may still exist on disk as a `*.jsonl.reset.*` archive, but the session tools (`sessions_history`, `sessions_list`) do not surface that prior context chain. If later system exec-completion messages arrive in a fresh session and the user says "con... | closed | completed | true | 1 | [] | [] | 2026-03-28T02:54:19Z | 2026-05-05T04:25:35Z | 2026-05-03T03:21:08Z | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | octyai | 261,415,581 | U_kgDOD5TinQ | User | false |
openclaw/openclaw | 4,343,405,464 | I_kwDOQb6kR88AAAABAuMbmA | 73,577 | https://github.com/openclaw/openclaw/issues/73577 | https://api.github.com/repos/openclaw/openclaw/issues/73577 | Config validator silently reverts providers.<id> writes with no operator-visible reason (2026.4.25) | ## Summary
Adding a `providers.<id>` block (with `apiKey` + `baseUrl`) to `~/.openclaw/openclaw.json` causes the gateway to revert to last-known-good on the next reload, with **no operator-visible reason**. The rejected config is preserved on disk (good), but no log line, doctor check, or CLI output reveals which fiel... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T13:11:10Z | 2026-05-03T04:34:57Z | 2026-04-30T21:31:01Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | devicemanager | 1,844,069 | MDQ6VXNlcjE4NDQwNjk= | User | false |
openclaw/openclaw | 4,343,419,061 | I_kwDOQb6kR88AAAABAuNQtQ | 73,578 | https://github.com/openclaw/openclaw/issues/73578 | https://api.github.com/repos/openclaw/openclaw/issues/73578 | session status `Context:` line conflates cumulative session tokens with per-call context window | ## Bug
The status card line `📚 Context: X/Y (Z%)` displays a cumulative session-token counter against the per-call context window size, which produces nonsensical "utilization" percentages on long-running sessions.
## Reproduction
In a multi-day persistent session (Telegram-direct, agent=main, model=`claude-cli/cla... | closed | completed | true | 2 | [] | [] | 2026-04-28T13:13:11Z | 2026-05-01T04:40:24Z | 2026-04-29T04:08:16Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | DashLabsDev | 263,981,165 | U_kgDOD7wIbQ | User | false |
openclaw/openclaw | 4,343,463,353 | I_kwDOQb6kR88AAAABAuP9uQ | 73,581 | https://github.com/openclaw/openclaw/issues/73581 | https://api.github.com/repos/openclaw/openclaw/issues/73581 | [Bug]: Agent processing lane can stall for minutes without timeout recovery, plus memory-core dreaming cron race condition on Gateway restart | ## Environment
- **OpenClaw Version**: v2026.4.26 (be8c246)
- **Node**: v22.22.2
- **OS**: Linux 6.17.0-22-generic (Ubuntu/x64)
- **Channel**: Feishu (WebSocket mode)
- **Model**: bailian/qwen3.6-plus
- **Config**: systemd user service with `Restart=always`
---
## Issue 1: Agent Processing Lane Stalls Without Timeou... | closed | not_planned | true | 7 | [] | [] | 2026-04-28T13:20:18Z | 2026-05-03T04:34:44Z | 2026-04-30T22:52:21Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | WS-Q0758 | 274,219,387 | U_kgDOEFhBew | User | false |
openclaw/openclaw | 4,344,027,749 | I_kwDOQb6kR88AAAABAuyaZQ | 73,607 | https://github.com/openclaw/openclaw/issues/73607 | https://api.github.com/repos/openclaw/openclaw/issues/73607 | agents.defaults.params and agents.defaults.models[].params stripped from openclaw.json by gateway runtime (v2026.4.26) | ### Bug type
Configuration not honored / runtime drops user-provided config field
### Beta release blocker
No
### Summary
`agents.defaults.params` and `agents.defaults.models["<provider>/<model>"].params` are documented overrides (`docs/concepts/model-providers.md:88`, `docs/concepts/model-providers.md:90`) — acce... | closed | completed | true | 3 | [] | [] | 2026-04-28T14:41:14Z | 2026-05-02T04:28:20Z | 2026-04-29T12:02:46Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | quangtran88 | 107,824,159 | U_kgDOBm1EHw | User | false |
openclaw/openclaw | 4,343,876,260 | I_kwDOQb6kR88AAAABAupKpA | 73,601 | https://github.com/openclaw/openclaw/issues/73601 | https://api.github.com/repos/openclaw/openclaw/issues/73601 | [Bug]: Google Live browser Talk stuck at "Connecting Talk…" and failed sessions wedge embedded runner | **Severity:** **High** — failed Talk attempts wedge the embedded runner so subsequent text chats fail to dispatch until a full gateway restart
**Affects:** Anyone configuring `talk.provider: "google"` on 2026.4.26 Control UI Talk
**OpenClaw version:** 2026.4.26 (`be8c246`)
**Source file (per #73460):** `ui/src/ui/chat/... | closed | completed | true | 2 | [] | [] | 2026-04-28T14:19:43Z | 2026-05-03T04:37:43Z | 2026-04-30T11:13:05Z | CONTRIBUTOR | null | 20260503T060030Z | 2026-05-03T06:00:30Z | Spolen23 | 215,900,770 | U_kgDODN5iYg | User | false |
openclaw/openclaw | 4,344,064,139 | I_kwDOQb6kR88AAAABAu0oiw | 73,609 | https://github.com/openclaw/openclaw/issues/73609 | https://api.github.com/repos/openclaw/openclaw/issues/73609 | ACPX stale child session can respawn and inject old task output into Telegram thread | ### Summary
After fixing the local ACPX `npx` adapter problem, a Claude ACP child session continued to inject stale background-task output into the requester Telegram thread. Killing the ACP wrapper process was not sufficient: the gateway respawned the wrapper because the persisted ACPX session state remained open.
T... | closed | completed | true | 3 | [] | [] | 2026-04-28T14:46:49Z | 2026-05-03T04:37:29Z | 2026-04-30T13:50:01Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | joerod26 | 277,153,564 | U_kgDOEIUHHA | User | false |
openclaw/openclaw | 4,344,068,055 | I_kwDOQb6kR88AAAABAu031w | 73,610 | https://github.com/openclaw/openclaw/issues/73610 | https://api.github.com/repos/openclaw/openclaw/issues/73610 | Telegram polling provider startup loops when deleteWebhook cleanup fails | ## Summary
On Windows with OpenClaw `2026.4.26 (be8c246)`, enabling the Telegram channel can make the gateway intermittently unresponsive during startup because the Telegram polling runtime treats `deleteWebhook` cleanup failure as a setup retry loop.
This happens even when no Telegram webhook URL is configured and t... | closed | completed | true | 2 | [] | [] | 2026-04-28T14:47:26Z | 2026-05-02T04:28:21Z | 2026-04-29T11:43:18Z | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | kubu0014 | 102,464,866 | U_kgDOBht9Yg | User | false |
openclaw/openclaw | 4,343,599,652 | I_kwDOQb6kR88AAAABAuYSJA | 73,587 | https://github.com/openclaw/openclaw/issues/73587 | https://api.github.com/repos/openclaw/openclaw/issues/73587 | Gateway crash loop in 2026.4.26: `logging.file` paths starting with `~` are no longer expanded | ## Summary
After updating from `2026.4.24` to `2026.4.26`, the gateway enters a tight crash-loop on startup if `openclaw.json` has `logging.file` set to a path beginning with `~`. Under launchd this produces a respawn-loop every ~1.85s that takes the entire stack down (gateway, channels, sidecars, crons all unreachabl... | closed | completed | false | 3 | [] | [] | 2026-04-28T13:39:49Z | 2026-05-04T01:50:03Z | 2026-05-04T01:50:03Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | gabrielexito-stack | 263,964,070 | U_kgDOD7vFpg | User | false |
openclaw/openclaw | 4,344,092,630 | I_kwDOQb6kR88AAAABAu2X1g | 73,612 | https://github.com/openclaw/openclaw/issues/73612 | https://api.github.com/repos/openclaw/openclaw/issues/73612 | [Bug]: Error: Unknown memory embedding provider: voyage | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
after upgrading to v2026.4.26
~$ openclaw memory status
🦞 OpenClaw 2026.4.26 (be8c246) — I've seen your commit messages. We'll work on that together.
[openclaw] Failed to start CLI: Error: Unknown memory embedding provid... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-04-28T14:51:27Z | 2026-05-01T04:43:14Z | 2026-04-28T14:55:59Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Arslan-Mehmood1 | 51,626,734 | MDQ6VXNlcjUxNjI2NzM0 | User | false |
openclaw/openclaw | 4,343,780,208 | I_kwDOQb6kR88AAAABAujTcA | 73,595 | https://github.com/openclaw/openclaw/issues/73595 | https://api.github.com/repos/openclaw/openclaw/issues/73595 | Runtime injection tags (`<system-reminder>`, `<previous_response>`) leak verbatim to delivery channels when model is in degraded state | ## Summary
When an agent session is in a degraded state (after a model timeout, broken parentId chain, or stuck-in-processing recovery), the model's reply may include the runtime injection tags it received in its prompt — `<system-reminder>...</system-reminder>`, `<previous_response>...</previous_response>` — verbatim... | closed | completed | true | 3 | [] | [] | 2026-04-28T14:05:31Z | 2026-05-01T04:41:34Z | 2026-04-28T21:41:05Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | gabrielexito-stack | 263,964,070 | U_kgDOD7vFpg | User | false |
openclaw/openclaw | 4,344,110,620 | I_kwDOQb6kR88AAAABAu3eHA | 73,615 | https://github.com/openclaw/openclaw/issues/73615 | https://api.github.com/repos/openclaw/openclaw/issues/73615 | [Bug]: echoTranscript regression in v2026.4.25+ — transcript echo no longer sent (Telegram) | ## Summary
`tools.media.audio.echoTranscript` stopped working after upgrading from v2026.4.21 to v2026.4.26. The transcript echo message (`📝 "{transcript}"`) is no longer sent to Telegram before agent processing, despite being correctly configured and working on v2026.4.21.
This is a recurring regression — the same ... | closed | duplicate | true | 3 | [
"duplicate",
"close:duplicate",
"dedupe:child"
] | [] | 2026-04-28T14:54:21Z | 2026-05-03T04:37:07Z | 2026-04-30T14:14:56Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | M-Lietz | 222,629,290 | U_kgDODUUNqg | User | false |
openclaw/openclaw | 4,343,892,445 | I_kwDOQb6kR88AAAABAuqJ3Q | 73,602 | https://github.com/openclaw/openclaw/issues/73602 | https://api.github.com/repos/openclaw/openclaw/issues/73602 | [Bug]: WhatsApp flaps and Telegram polling stalls on WSL2 in 2026.4.26 | ### Bug type
Channel reliability / WSL2 regression
### Summary
On WSL2 with OpenClaw 2026.4.26, both configured messaging channels are unreliable:
- WhatsApp is reported as linked/connected/healthy, but repeatedly disconnects, fails WebSocket handshakes, hits DNS `EAI_AGAIN`, and sometimes receives WhatsApp-side `4... | open | null | false | 4 | [] | [] | 2026-04-28T14:22:00Z | 2026-05-03T10:19:44Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | Vksh07 | 74,851,514 | MDQ6VXNlcjc0ODUxNTE0 | User | false |
openclaw/openclaw | 4,344,146,532 | I_kwDOQb6kR88AAAABAu5qZA | 73,617 | https://github.com/openclaw/openclaw/issues/73617 | https://api.github.com/repos/openclaw/openclaw/issues/73617 | [Bug]: Browser runtime ignores root browser.executablePath on 2026.4.26 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
OpenClaw’s browser status/start path is not honoring the configured browser.executablePath even though /usr/bin/google-chrome exists and CDP is reachable on port 18800.
### Steps to reproduce
```bash
openclaw config set br... | closed | completed | true | 4 | [
"bug",
"regression"
] | [] | 2026-04-28T14:59:53Z | 2026-05-01T04:39:48Z | 2026-04-28T15:05:33Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | civiltox | 97,610,456 | U_kgDOBdFq2A | User | false |
openclaw/openclaw | 4,162,197,784 | I_kwDOQb6kR874FhkY | 56,708 | https://github.com/openclaw/openclaw/issues/56708 | https://api.github.com/repos/openclaw/openclaw/issues/56708 | Feature: Allow reaction events to optionally trigger agent turns | ## Summary
Emoji reactions on Slack (and likely other channels) are received by the gateway and queued as system events, but they only surface to the agent on the **next turn** (triggered by a new message or heartbeat). They don't trigger their own agent turn.
## Use Case
When a user reacts to an agent's message wit... | closed | not_planned | true | 1 | [] | [] | 2026-03-29T01:55:30Z | 2026-05-01T04:43:13Z | 2026-04-28T15:10:13Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | potterdigital | 197,414,865 | U_kgDOC8RP0Q | User | false |
openclaw/openclaw | 4,165,099,474 | I_kwDOQb6kR874Ql_S | 57,174 | https://github.com/openclaw/openclaw/issues/57174 | https://api.github.com/repos/openclaw/openclaw/issues/57174 | Feature: Per-agent compaction configuration (thresholds, mode, reserveTokens) | ## Summary
Currently `agents.defaults.compaction` applies **uniformly** to all agents. In multi-agent deployments, agents have fundamentally different context-retention needs. An orchestrator agent needs long memory; a specialist agent should compact aggressively after task completion.
## The Problem
We run 17 agent... | closed | not_planned | true | 4 | [] | [] | 2026-03-29T16:55:02Z | 2026-05-01T04:43:12Z | 2026-04-28T15:10:28Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | 1052326311 | 65,798,732 | MDQ6VXNlcjY1Nzk4NzMy | User | false |
openclaw/openclaw | 4,165,747,329 | I_kwDOQb6kR874TEKB | 57,288 | https://github.com/openclaw/openclaw/issues/57288 | https://api.github.com/repos/openclaw/openclaw/issues/57288 | WhatsApp reactions are not received as inbound events | ## Summary
WhatsApp message reactions (e.g. 👍, ❤️) are not delivered to the agent as inbound events. Only text messages, media, and replies generate inbound events.
## Steps to Reproduce
1. Configure WhatsApp channel with `actions.reactions: true` in `openclaw.json`
2. Send a message from the agent to a group chat
... | closed | not_planned | true | 1 | [] | [] | 2026-03-29T20:39:08Z | 2026-05-01T04:43:11Z | 2026-04-28T15:10:49Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | mgonto | 723,723 | MDQ6VXNlcjcyMzcyMw== | User | false |
openclaw/openclaw | 4,166,919,301 | I_kwDOQb6kR874XiSF | 57,448 | https://github.com/openclaw/openclaw/issues/57448 | https://api.github.com/repos/openclaw/openclaw/issues/57448 | [Feature]: DM conversation handoff for ACP agent sessions | ### Summary
Allow `--bind here` to work in DM conversations so a spawned ACP agent can take over the chat and revert when done.
### Problem to solve
ACP child agents spawned from a Telegram DM cannot interact with the user directly. The parent agent must relay every message back and forth, doubling latency, doubling... | closed | completed | true | 1 | [] | [] | 2026-03-30T03:36:46Z | 2026-05-01T04:43:10Z | 2026-04-28T15:10:59Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | alexanderkreidich | 126,781,073 | U_kgDOB46GkQ | User | false |
openclaw/openclaw | 4,167,534,835 | I_kwDOQb6kR874Z4jz | 57,497 | https://github.com/openclaw/openclaw/issues/57497 | https://api.github.com/repos/openclaw/openclaw/issues/57497 | Feature Request: Support PostgreSQL for Session Storage | ### Summary
核心内容：
OpenClaw 的会话数据目前存在 JSON 文件里，建议支持 PostgreSQL 存储 session 数据，便于统一管理和查询。
# OpenClaw Feature Request 反馈模板
---
## Feature Request: Support PostgreSQL for Session Storage
### Issue Type
Feature Request
### Description
Currently OpenClaw stores session data (conversations, messages) in JSON files:
`~/.... | closed | not_planned | true | 2 | [
"enhancement"
] | [] | 2026-03-30T05:30:05Z | 2026-05-01T04:43:10Z | 2026-04-28T15:11:10Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | kktttl | 266,642,194 | U_kgDOD-SjEg | User | false |
openclaw/openclaw | 4,174,282,917 | I_kwDOQb6kR874zoCl | 57,952 | https://github.com/openclaw/openclaw/issues/57952 | https://api.github.com/repos/openclaw/openclaw/issues/57952 | feat: configurable CommandLane.Nested concurrency — unblock parallel hook/cron/sessions_send execution | ## Summary
The `CommandLane.Nested` lane is hardcoded to `maxConcurrent: 1` with no configuration surface. All webhook hook sessions (`/hooks/agent`), cron isolated agent runs, and `sessions_send` inter-agent calls route through this lane, meaning **only one can execute at a time across the entire gateway**.
This cre... | closed | completed | true | 1 | [] | [] | 2026-03-30T21:48:48Z | 2026-05-01T04:43:09Z | 2026-04-28T15:12:02Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | warren-t-c[bot] | 266,866,348 | BOT_kgDOD-gOrA | Bot | false |
openclaw/openclaw | 4,177,896,611 | I_kwDOQb6kR875BaSj | 58,298 | https://github.com/openclaw/openclaw/issues/58298 | https://api.github.com/repos/openclaw/openclaw/issues/58298 | Feature request: message tool support for Feishu interactive cards | ## Summary
The `message` tool currently only supports plain text and media sends for Feishu. There is no way for agents to send interactive cards with buttons, which means the built-in `ocf1` card action callback routing cannot be utilized by agents.
## Background
OpenClaw already has a complete interactive card sys... | closed | not_planned | true | 1 | [] | [] | 2026-03-31T10:17:55Z | 2026-05-01T04:43:07Z | 2026-04-28T15:12:28Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | SwordImmortal | 31,211,064 | MDQ6VXNlcjMxMjExMDY0 | User | false |
openclaw/openclaw | 4,183,982,628 | I_kwDOQb6kR875YoIk | 58,741 | https://github.com/openclaw/openclaw/issues/58741 | https://api.github.com/repos/openclaw/openclaw/issues/58741 | [Feature]: Feishu: Support session idleTimeout / sessionMaxAge configuration | ### Summary
Feishu channel lacks session idleTimeout — sessions accumulate forever in sessions.json
### Problem to solve
When using Feishu as a channel, OpenClaw sessions never expire. Over time this causes:
1. Session accumulation — sessions.json grows unbounded (70+ entries on a 3-agent single-user setup)
2. Memor... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-04-01T04:29:41Z | 2026-05-01T04:43:06Z | 2026-04-28T15:12:47Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | danielcrazy0517 | 174,254,650 | U_kgDOCmLqOg | User | false |
openclaw/openclaw | 4,187,350,792 | I_kwDOQb6kR875lecI | 59,070 | https://github.com/openclaw/openclaw/issues/59070 | https://api.github.com/repos/openclaw/openclaw/issues/59070 | [Bug]: GatewayRequestError: invalid config: agents: Unrecognized key: "defaultId" + permanent "unsaved config changes" when changing model on new agent | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When creating a new agent and attempting to change its Primary model from "Inherit default" to Grok (xai/grok-4) in the Agents UI, the save fails with GatewayRequestError: invalid config: agents: Unrecognized k... | closed | not_planned | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-04-01T14:25:32Z | 2026-05-01T04:39:55Z | 2026-04-28T15:13:41Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | wilkieeee | 272,845,606 | U_kgDOEENLJg | User | false |
openclaw/openclaw | 4,166,701,399 | I_kwDOQb6kR874WtFX | 57,427 | https://github.com/openclaw/openclaw/issues/57427 | https://api.github.com/repos/openclaw/openclaw/issues/57427 | Env vars from .env not passed to sandbox exec calls — generic provider placeholder not resolved | ## Description
Environment variables defined in `~/.openclaw/.env` are not available inside sandbox `exec` calls. When a generic OpenShell provider is configured (e.g., `ncbi` with credential key `NCBI_API_KEY`), the env var inside the sandbox resolves to the OpenShell placeholder `openshell:resolve:env:NCBI_API_KEY` ... | open | null | false | 2 | [] | [] | 2026-03-30T02:53:47Z | 2026-05-02T05:56:01Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | teosborne | 4,041,297 | MDQ6VXNlcjQwNDEyOTc= | User | false |
openclaw/openclaw | 4,166,847,420 | I_kwDOQb6kR874XQu8 | 57,442 | https://github.com/openclaw/openclaw/issues/57442 | https://api.github.com/repos/openclaw/openclaw/issues/57442 | [Feature Request] Add elapsed/response time back to session_status display | ## Feature Request
### Problem
After upgrading to 2026.3.28, the `session_status` output no longer shows elapsed/response time. Previous versions displayed this information in the status card.
### Expected Behavior
The status card (shown via `session_status` tool or `/status` command) should include elapsed/response ... | open | null | false | 1 | [] | [] | 2026-03-30T03:22:47Z | 2026-05-02T05:56:02Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | lyhs-hf | 259,303,213 | U_kgDOD3SnLQ | User | false |
openclaw/openclaw | 4,166,864,073 | I_kwDOQb6kR874XUzJ | 57,443 | https://github.com/openclaw/openclaw/issues/57443 | https://api.github.com/repos/openclaw/openclaw/issues/57443 | [Bug]: Tool JSON Schema patternProperties causes 400 errors on BytePlus Ark (doubao) — schema cleaning should be universal, not provider-specific | ## Summary
`normalizeToolParameters()` only strips `patternProperties` (and other JSON Schema-only keywords) for Google/Gemini providers via `cleanSchemaForGemini()`. BytePlus Ark (doubao-seed, doubao-pro, etc.) also rejects `patternProperties` with HTTP 400, but is not covered by the cleaning logic.
This is the same... | open | null | false | 1 | [] | [] | 2026-03-30T03:25:55Z | 2026-05-02T05:37:07Z | null | CONTRIBUTOR | null | 20260502T060142Z | 2026-05-02T06:01:42Z | wujiaming88 | 7,887,545 | MDQ6VXNlcjc4ODc1NDU= | User | false |
openclaw/openclaw | 4,166,912,762 | I_kwDOQb6kR874Xgr6 | 57,447 | https://github.com/openclaw/openclaw/issues/57447 | https://api.github.com/repos/openclaw/openclaw/issues/57447 | sessions_send blocked by visibility guard even when a2a policy allows cross-agent messaging | ## Problem
When `tools.sessions.visibility` is set to `tree` (the default) and `tools.agentToAgent.enabled=true` with appropriate `allow` patterns, `sessions_send` is still blocked for cross-agent messaging. The visibility guard in `createSessionVisibilityGuard` rejects all cross-agent access when `visibility !== "all... | open | null | false | 4 | [] | [] | 2026-03-30T03:35:27Z | 2026-05-02T05:56:04Z | null | CONTRIBUTOR | null | 20260502T060142Z | 2026-05-02T06:01:42Z | zijiess | 126,048,304 | U_kgDOB4NYMA | User | false |
openclaw/openclaw | 4,167,068,569 | I_kwDOQb6kR874YGuZ | 57,461 | https://github.com/openclaw/openclaw/issues/57461 | https://api.github.com/repos/openclaw/openclaw/issues/57461 | MCP tools return data in structuredContent only — Cursor and Claude.ai only read content | ## Problem
The MCP channel tools (`conversations_list`, `messages_read`, `conversation_get`, `attachments_fetch`, `events_poll`, etc.) return their actual data payloads in `structuredContent`, with only a summary count in `content`:
```json
{
"content": [{"type": "text", "text": "conversations: 1"}],
"structuredC... | closed | completed | false | 4 | [] | [] | 2026-03-30T04:03:28Z | 2026-05-04T06:42:30Z | 2026-05-04T06:42:30Z | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | chrysb | 120,388 | MDQ6VXNlcjEyMDM4OA== | User | false |
openclaw/openclaw | 4,190,804,191 | I_kwDOQb6kR875ypjf | 59,319 | https://github.com/openclaw/openclaw/issues/59319 | https://api.github.com/repos/openclaw/openclaw/issues/59319 | [Bug]: [2026.4.1] boot-md injects DAILY.md multiple times per message | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
After upgrading to OpenClaw 2026.4.1 (da64a97), the boot-md plugin injects the same workspace file (e.g. memory/YYYY-MM-DD.md mapped as DAILY.md) 3–5× per message instead of once. Each injection appears as a se... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-02T01:15:04Z | 2026-05-01T04:43:05Z | 2026-04-28T15:16:05Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | swaminadhankj | 206,019,326 | U_kgDODEea_g | User | false |
openclaw/openclaw | 4,168,221,053 | I_kwDOQb6kR874cgF9 | 57,557 | https://github.com/openclaw/openclaw/issues/57557 | https://api.github.com/repos/openclaw/openclaw/issues/57557 | subagents/runs.json read on every RPC call causes O(n) gateway latency | ## Problem
The gateway reads `/root/.openclaw/subagents/runs.json` synchronously (`fs.readFileSync` + `JSON.parse`) via `getSubagentRunsSnapshotForRead()` on **every incoming RPC call**. As completed subagent runs accumulate (`cleanup: 'keep'` is the default), the file grows unbounded. At 68 runs / 433KB, this caused ... | closed | completed | true | 1 | [] | [] | 2026-03-30T07:24:46Z | 2026-05-03T04:37:45Z | 2026-04-30T11:12:20Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | potterdigital | 197,414,865 | U_kgDOC8RP0Q | User | false |
openclaw/openclaw | 4,168,311,224 | I_kwDOQb6kR874c2G4 | 57,562 | https://github.com/openclaw/openclaw/issues/57562 | https://api.github.com/repos/openclaw/openclaw/issues/57562 | [Bug]: bug-report-openclaw-upgrade-config-migration.pdf | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
bug-report-openclaw-upgrade-config-migration: error: config validation failed: channels.feishu.accounts.main: invalid config: must not have additional properties
### Steps to reproduce
Updated and Never Start... | closed | not_planned | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-30T07:38:05Z | 2026-05-04T04:34:31Z | 2026-05-02T02:22:55Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | geogleo | 60,126,952 | MDQ6VXNlcjYwMTI2OTUy | User | false |
openclaw/openclaw | 4,168,380,840 | I_kwDOQb6kR874dHGo | 57,567 | https://github.com/openclaw/openclaw/issues/57567 | https://api.github.com/repos/openclaw/openclaw/issues/57567 | Bug: Configuration Migration Failure During Upgrade (v3.24 → v3.28) | ## Summary
After upgrading from v3.24 to v3.28, the OpenClaw gateway failed to start due to corrupted configuration file generated by the migration script. The `openclaw doctor --fix` command was also unable to repair the corrupted JSON, requiring manual intervention.
## Environment
- **OpenClaw version**: v2026.3.2... | open | null | false | 4 | [] | [] | 2026-03-30T07:47:52Z | 2026-05-02T02:12:01Z | null | NONE | null | 20260502T030029Z | 2026-05-02T03:00:29Z | geogleo | 60,126,952 | MDQ6VXNlcjYwMTI2OTUy | User | false |
openclaw/openclaw | 4,191,662,645 | I_kwDOQb6kR87517I1 | 59,435 | https://github.com/openclaw/openclaw/issues/59435 | https://api.github.com/repos/openclaw/openclaw/issues/59435 | [Bug] openclaw update 命令会错误覆盖 ai.openclaw.gateway.plist 的配置路径 | ## Bug 描述
**报告者**: Claude Code（AI 助手，在多代理配置排错时发现）
**版本**: OpenClaw 2026.4.1
在运行 `openclaw update` 更新多实例 OpenClaw Gateway 时，`ai.openclaw.gateway.plist` 会被错误地重新生成，导致主 Gateway 实例配置混乱。
## 问题背景
系统部署了三个独立的 OpenClaw Gateway 实例：
- Friday: `~/.openclaw/openclaw.json`，端口 18789，Label `ai.openclaw.gateway`
- Andy: `~/.opencla... | closed | completed | true | 1 | [] | [] | 2026-04-02T05:03:19Z | 2026-05-01T04:43:04Z | 2026-04-28T15:16:23Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | JAANG131 | 203,095,831 | U_kgDODBr_Fw | User | false |
openclaw/openclaw | 4,168,414,131 | I_kwDOQb6kR874dPOz | 57,570 | https://github.com/openclaw/openclaw/issues/57570 | https://api.github.com/repos/openclaw/openclaw/issues/57570 | [Bug]: openclaw config no response | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
The openclaw config command is stuck and not responding
> node scripts/run-node.mjs config
...
no reponse ...
### Steps to reproduce
v2026.3.30
openclaw config
### Expected behavior
config successfully
### Actual beh... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-30T07:53:08Z | 2026-05-03T04:37:44Z | 2026-04-30T11:12:27Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | XCZAMICD1215 | 15,889,158 | MDQ6VXNlcjE1ODg5MTU4 | User | false |
openclaw/openclaw | 4,193,079,198 | I_kwDOQb6kR8757U-e | 59,571 | https://github.com/openclaw/openclaw/issues/59571 | https://api.github.com/repos/openclaw/openclaw/issues/59571 | [Bug]: feishu_doc 插件加载正常，但工具未暴露给 agent 调用 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
飞书文档权限已开通、插件已加载、channels.feishu.tools.doc 已配置为 true，但 agent 无法调用 feishu_doc 工具。尝试通过飞书聊天发送 wiki 链接，agent 无法识别并读取文档内容。
环境：
OpenClaw 版本：2026.4.1（npm 安装）
操作系统：macOS
飞书版本：Feishu ([feishu.cn](http://feishu.cn/))
Node... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-04-02T09:05:30Z | 2026-05-01T04:43:04Z | 2026-04-28T15:16:29Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | fengzoujun-coder | 273,055,264 | U_kgDOEEZ-IA | User | false |
openclaw/openclaw | 4,168,710,325 | I_kwDOQb6kR874eXi1 | 57,588 | https://github.com/openclaw/openclaw/issues/57588 | https://api.github.com/repos/openclaw/openclaw/issues/57588 | [Feature]: Security: fetch-auth bearer token fallback attaches tokens to arbitrary URLs without fail-closed defaults | ### Summary
**Summary**
The bearer auth fallback helper in src/plugin-sdk/fetch-auth.ts attaches tokens to URLs returning 401/403 without fail-closed defaults. The requireHttps and shouldAttachAuth mitigations are opt-in, meaning callers that don't explicitly configure restrictions will send tokens to any URL — includ... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-30T08:32:45Z | 2026-05-02T04:19:09Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | anshumanbh | 9,063,340 | MDQ6VXNlcjkwNjMzNDA= | User | false |
openclaw/openclaw | 4,168,810,105 | I_kwDOQb6kR874ev55 | 57,592 | https://github.com/openclaw/openclaw/issues/57592 | https://api.github.com/repos/openclaw/openclaw/issues/57592 | TUI: Support inline image display in iTerm2 / Kitty / Sixel terminals | ## Feature Request
### Summary
Add inline image rendering support to `openclaw tui` for terminals that support image protocols (iTerm2, Kitty, Sixel, etc.).
### Motivation
When the assistant generates or references images (AI renderings, avatars, charts, screenshots), the TUI currently shows a placeholder like `[imag... | open | null | false | 1 | [] | [] | 2026-03-30T08:46:23Z | 2026-05-02T03:22:42Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | xunuoi | 1,924,536 | MDQ6VXNlcjE5MjQ1MzY= | User | false |
openclaw/openclaw | 4,169,049,881 | I_kwDOQb6kR874fqcZ | 57,607 | https://github.com/openclaw/openclaw/issues/57607 | https://api.github.com/repos/openclaw/openclaw/issues/57607 | [Bug]: groupPolicy allowlist silently drops pending history for non-allowlisted group senders | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When `channels.mattermost.groupPolicy` is set to `"allowlist"`, messages from senders not in `groupAllowFrom` are dropped at the access-check stage **before** `recordPendingHistoryEntryIfEnabled()` is called. T... | open | null | false | 1 | [] | [] | 2026-03-30T09:19:18Z | 2026-05-02T03:02:43Z | null | CONTRIBUTOR | null | 20260502T060142Z | 2026-05-02T06:01:42Z | JasonWang1124 | 56,307,673 | MDQ6VXNlcjU2MzA3Njcz | User | false |
openclaw/openclaw | 4,169,440,684 | I_kwDOQb6kR874hJ2s | 57,638 | https://github.com/openclaw/openclaw/issues/57638 | https://api.github.com/repos/openclaw/openclaw/issues/57638 | feat: cron.defaults for model, delivery, and contextTokens | ## Feature Request: Cron Default Model & Delivery Settings
### Problem
When creating cron jobs with `payload.kind: "agentTurn"`, there's no way to set fleet-wide defaults for `model` or `delivery`. If a cron author forgets to specify `model`, the job inherits the agent's primary model (often an expensive tier like So... | open | null | false | 1 | [] | [] | 2026-03-30T10:12:45Z | 2026-05-02T04:19:10Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | Kyzcreig | 9,063,726 | MDQ6VXNlcjkwNjM3MjY= | User | false |
openclaw/openclaw | 4,169,448,984 | I_kwDOQb6kR874hL4Y | 57,639 | https://github.com/openclaw/openclaw/issues/57639 | https://api.github.com/repos/openclaw/openclaw/issues/57639 | Feature request: maxPerSession limit for subagents | ## Problem
Currently `subagents.maxConcurrent` limits how many subagents can run **simultaneously**, but there's no cap on the **total number of subagents spawned over the lifetime of a session**.
In practice, a Discord channel with `requireMention: false` and an active coding agent spawned 100+ subagents over time (... | open | null | false | 1 | [] | [] | 2026-03-30T10:13:50Z | 2026-05-02T04:19:11Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | fffchaobu | 173,046,355 | U_kgDOClB6Uw | User | false |
openclaw/openclaw | 4,169,655,350 | I_kwDOQb6kR874h-Q2 | 57,658 | https://github.com/openclaw/openclaw/issues/57658 | https://api.github.com/repos/openclaw/openclaw/issues/57658 | [Feature]: Support for custom roles with configurable scopes | ### Summary
Currently OpenClaw only supports system-defined roles (operator, admin, etc.) with
hardcoded scopes. There is no way to create custom roles with user-defined scope
combinations.
### Problem to solve
I need to create a "read-only" role (or custom named role like "operatorRW") that
all... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-03-30T10:46:08Z | 2026-05-02T04:19:12Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | foilfoil | 272,185,269 | U_kgDOEDk3tQ | User | false |
openclaw/openclaw | 4,170,548,796 | I_kwDOQb6kR874lYY8 | 57,700 | https://github.com/openclaw/openclaw/issues/57700 | https://api.github.com/repos/openclaw/openclaw/issues/57700 | Hook delivery error system events leak into agent sessions, causing dedup noise in channels | ## Problem
When a hook session completes and its delivery encounters an error (or produces a delivery artifact), the gateway injects a `Hook Hook (error): <content>` system event into the target agent session. The agent then processes this system event as if it were a real message.
Despite extensive AGENTS.md instruc... | open | null | false | 1 | [] | [] | 2026-03-30T12:59:03Z | 2026-05-02T04:19:14Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | weretere | 38,572,760 | MDQ6VXNlcjM4NTcyNzYw | User | false |
openclaw/openclaw | 4,170,625,141 | I_kwDOQb6kR874lrB1 | 57,705 | https://github.com/openclaw/openclaw/issues/57705 | https://api.github.com/repos/openclaw/openclaw/issues/57705 | feat(sessions): expose sub-agent concurrency budget in session_status or new sessions_capacity tool | ## Problem
When running multiple concurrent cron heartbeats (e.g., adversarial-swarm spawning 7-9 sub-agents + hypothesis-discovery spawning 3 sub-agents), agents have no way to know:
1. How many sub-agent slots are currently available
2. Whether a `sessions_spawn` call will succeed before making it
3. The current co... | open | null | false | 1 | [] | [] | 2026-03-30T13:10:09Z | 2026-05-02T05:28:06Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | gasvn | 20,515,144 | MDQ6VXNlcjIwNTE1MTQ0 | User | false |
openclaw/openclaw | 4,170,709,236 | I_kwDOQb6kR874l_j0 | 57,708 | https://github.com/openclaw/openclaw/issues/57708 | https://api.github.com/repos/openclaw/openclaw/issues/57708 | RFC: Markdown → rich_text block auto-conversion for Slack messages | ## Context
Following up on #12602 and the Block Kit plumbing work by @vincentkoc (passthrough in #44592, interactive replies in #57473, table blocks in #57591).
The infrastructure for sending Block Kit is solid. The next layer is **auto-conversion** — making agent responses look better in Slack without requiring agen... | open | null | false | 1 | [] | [] | 2026-03-30T13:21:37Z | 2026-05-02T02:12:03Z | null | CONTRIBUTOR | null | 20260502T030029Z | 2026-05-02T03:00:29Z | hsiaoa | 70,124,331 | MDQ6VXNlcjcwMTI0MzMx | User | false |
openclaw/openclaw | 4,171,300,470 | I_kwDOQb6kR874oP52 | 57,745 | https://github.com/openclaw/openclaw/issues/57745 | https://api.github.com/repos/openclaw/openclaw/issues/57745 | Skill spec: conditional activation based on tool/toolset availability | ## Feature Request
### Summary
Add support for conditional skill activation in the AgentSkills spec — skills that show or hide themselves based on which tools are available in the current session.
### Motivation
hermes-agent ([NousResearch/hermes-agent](https://github.com/NousResearch/hermes-agent)) implements this c... | open | null | false | 1 | [] | [] | 2026-03-30T14:36:29Z | 2026-05-02T05:28:07Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | Ryno2390 | 16,257,434 | MDQ6VXNlcjE2MjU3NDM0 | User | false |
openclaw/openclaw | 4,171,361,427 | I_kwDOQb6kR874oeyT | 57,752 | https://github.com/openclaw/openclaw/issues/57752 | https://api.github.com/repos/openclaw/openclaw/issues/57752 | [Bug]: 新版本使用kimi2.5模型的时候，一直报错，不能使用 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
你好，
我在使用 OpenClaw v2026.3.28 版本时，遇到了一个稳定复现的 Bug：
1. 问题表现：
- 选择任意 Kimi/Moonshot 系列模型（kimi/kimi-2.5、moonshot/kimi-k2.5、kimi-code 等）
- 发送消息后，界面弹出红色错误：
`事件顺序意外，先收到 message_start，后才收到 "message_stop"`
- 消息实际已成... | open | null | false | 5 | [
"bug",
"regression"
] | [] | 2026-03-30T14:44:08Z | 2026-05-02T05:28:08Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | chenhuant | 261,808,867 | U_kgDOD5ri4w | User | false |
openclaw/openclaw | 4,171,403,988 | I_kwDOQb6kR874opLU | 57,754 | https://github.com/openclaw/openclaw/issues/57754 | https://api.github.com/repos/openclaw/openclaw/issues/57754 | Android: Add message copy and text selection to chat screen | ### Summary
Add the ability to copy or selectively copy any chat message on the Android app.
### Problem
Currently, only code blocks support text selection and copy on the Android chat screen. When using the app on a phone, it is very common to want to copy a message (or part of it) to use elsewhere — paste into ano... | open | null | false | 1 | [] | [] | 2026-03-30T14:49:03Z | 2026-05-02T02:12:05Z | null | NONE | null | 20260502T030029Z | 2026-05-02T03:00:29Z | bdhwan | 5,001,286 | MDQ6VXNlcjUwMDEyODY= | User | false |
openclaw/openclaw | 4,171,550,149 | I_kwDOQb6kR874pM3F | 57,775 | https://github.com/openclaw/openclaw/issues/57775 | https://api.github.com/repos/openclaw/openclaw/issues/57775 | Windows headless node host supports exec approvals via CLI, but nodes describe / Control UI do not advertise system.execApprovals.get/set | ## Environment
- OpenClaw version: `2026.3.28`
- Gateway host: macOS
- Remote node host: Windows (`win32`)
- Node type: headless node host started via `openclaw node run`
## Expected behavior
A Windows headless node host should advertise exec approvals commands, consistent with the docs and implementation:
- `syste... | open | null | false | 1 | [] | [] | 2026-03-30T15:06:55Z | 2026-05-02T03:02:48Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | RTKOP | 126,468,604 | U_kgDOB4nB_A | User | false |
openclaw/openclaw | 4,171,747,691 | I_kwDOQb6kR874p9Fr | 57,790 | https://github.com/openclaw/openclaw/issues/57790 | https://api.github.com/repos/openclaw/openclaw/issues/57790 | Gateway shutdown/restart does not fire `session_end` for active sessions | ## Summary
When the OpenClaw gateway process is terminated (SIGTERM/SIGINT/kill) or restarted, `session_end` hooks are never fired for currently active sessions. This leaves sessions in an incomplete state in any downstream system that relies on session lifecycle hooks (e.g., claude-mem plugin).
## Root Cause
`sessi... | open | null | false | 1 | [] | [] | 2026-03-30T15:32:32Z | 2026-05-02T03:02:50Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | GigiTiti-Kai | 146,460,333 | U_kgDOCLrOrQ | User | false |
openclaw/openclaw | 4,172,354,208 | I_kwDOQb6kR874sRKg | 57,846 | https://github.com/openclaw/openclaw/issues/57846 | https://api.github.com/repos/openclaw/openclaw/issues/57846 | Mattermost: separate requireMention for channel vs thread messages | ## Feature Request
### Problem
Currently `requireMention` is a single boolean per channel/group. When set to `true`, the agent requires @mention in **both** top-level channel messages and thread replies.
For Mattermost with Collapsed Reply Threads (CRT), this is problematic:
- In **channels**: @mention makes sense —... | open | null | false | 1 | [] | [] | 2026-03-30T16:50:34Z | 2026-05-02T03:02:52Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | alvin223dev | 260,335,034 | U_kgDOD4Rlug | User | false |
openclaw/openclaw | 4,172,703,823 | I_kwDOQb6kR874tmhP | 57,858 | https://github.com/openclaw/openclaw/issues/57858 | https://api.github.com/repos/openclaw/openclaw/issues/57858 | coding-agent skill: Claude Code dispatch fails silently due to CLAUDECODE nesting guard | ## Problem
When OpenClaw spawns Claude Code as a subprocess via the coding-agent skill, the child process inherits `CLAUDECODE=1` from the parent environment. Claude Code sees this and silently exits, thinking it's a nested session. Result: 0% success rate on background agent dispatch.
## Evidence
On 2026-03-30, all... | open | null | false | 1 | [] | [] | 2026-03-30T17:32:58Z | 2026-05-02T05:37:10Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | UBIpromoter | 46,543,177 | MDQ6VXNlcjQ2NTQzMTc3 | User | false |
openclaw/openclaw | 4,173,926,594 | I_kwDOQb6kR874yRDC | 57,933 | https://github.com/openclaw/openclaw/issues/57933 | https://api.github.com/repos/openclaw/openclaw/issues/57933 | Feature: Per-Session Health Polling & Live Status Indicators | ## Problem
When a session is processing a request (especially waiting for a slow model or a long tool call), the Control UI shows no indication of whether the session is alive, thinking, stalled, or crashed. Users experience "dead air" — minutes of silence with no feedback — which is indistinguishable from a crash or ... | open | null | false | 1 | [] | [] | 2026-03-30T20:37:38Z | 2026-05-02T03:02:53Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | apoapostolov | 4,083,812 | MDQ6VXNlcjQwODM4MTI= | User | false |
openclaw/openclaw | 4,174,456,199 | I_kwDOQb6kR8740SWH | 57,967 | https://github.com/openclaw/openclaw/issues/57967 | https://api.github.com/repos/openclaw/openclaw/issues/57967 | macOS: gateway LaunchAgent plist is written with 0644 instead of 0600 | ## Summary
On macOS, the gateway LaunchAgent plist appears to be written with mode `0644` instead of `0600`.
In our setup this showed up as a recurring permission hardening issue for:
`~/Library/LaunchAgents/ai.openclaw.gateway.plist`
## Expected
The LaunchAgent plist should be owner-readable/writable only:
- exp... | closed | completed | false | 2 | [] | [] | 2026-03-30T22:29:34Z | 2026-05-03T23:12:57Z | 2026-05-03T23:12:57Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | l0hde | 128,327,689 | U_kgDOB6YgCQ | User | false |
openclaw/openclaw | 4,174,547,471 | I_kwDOQb6kR8740ooP | 57,975 | https://github.com/openclaw/openclaw/issues/57975 | https://api.github.com/repos/openclaw/openclaw/issues/57975 | [Bug]: The search for the AI is not working, it does not search | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
The search for the AI isn't working, it doesn't search, I used tavily, it couldn't search, I asked, it couldn't search
### Steps to reproduce
i dont know
### Expected behavior
.
### Actual behavior
.
### OpenClaw versi... | open | null | false | 1 | [
"bug",
"bug:crash"
] | [] | 2026-03-30T22:54:33Z | 2026-05-02T03:02:55Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | pcqvix1 | 231,783,992 | U_kgDODdC-OA | User | false |
openclaw/openclaw | 4,174,678,472 | I_kwDOQb6kR8741InI | 57,992 | https://github.com/openclaw/openclaw/issues/57992 | https://api.github.com/repos/openclaw/openclaw/issues/57992 | [Bug]: Sandbox container creation silently fails - no error logged, agent zombies | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When spawning a sandboxed subagent (`sessions_spawn` with a named agent that inherits `agents.defaults.sandbox.mode: "non-main"`), the spawn is accepted and a workspace directory is created, but the Docker container creation step silently f... | open | null | false | 1 | [] | [] | 2026-03-30T23:34:28Z | 2026-05-02T04:19:22Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | marcbal77 | 8,789,169 | MDQ6VXNlcjg3ODkxNjk= | User | false |
openclaw/openclaw | 4,174,717,974 | I_kwDOQb6kR8741SQW | 57,996 | https://github.com/openclaw/openclaw/issues/57996 | https://api.github.com/repos/openclaw/openclaw/issues/57996 | QMD per-agent SQLite caches cause extreme disk I/O on multi-agent deployments | **Environment**
- macOS 26.3, MacBook Air (M4), 460GB SSD
- OpenClaw latest stable
- 19 persistent agents running concurrently
**Problem**
Each agent gets its own QMD SQLite cache at `~/.openclaw/agents/<name>/qmd/xdg-cache/qmd/index.sqlite`. With 19 agents, the combined cache totals 2.5GB on disk (main agent alone i... | open | null | false | 2 | [] | [
"vincentkoc"
] | 2026-03-30T23:47:21Z | 2026-05-02T05:28:10Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | Orionation | 260,579,663 | U_kgDOD4ghTw | User | false |
openclaw/openclaw | 4,174,961,637 | I_kwDOQb6kR8742Nvl | 58,026 | https://github.com/openclaw/openclaw/issues/58026 | https://api.github.com/repos/openclaw/openclaw/issues/58026 | normalizeExtraMemoryPaths does not expand tilde (~) in extraPaths | ## Bug
`normalizeExtraMemoryPaths()` in the memory search module does not expand `~/` prefixes in `agents.defaults.memorySearch.extraPaths`. The function uses `path.isAbsolute()` which returns `false` for `~/` paths, causing the path to be resolved relative to each agent's workspace directory.
## Expected Behavior
`... | open | null | false | 1 | [] | [] | 2026-03-31T01:02:07Z | 2026-05-02T05:28:11Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | stadman | 10,135,297 | MDQ6VXNlcjEwMTM1Mjk3 | User | false |
openclaw/openclaw | 4,174,964,794 | I_kwDOQb6kR8742Og6 | 58,028 | https://github.com/openclaw/openclaw/issues/58028 | https://api.github.com/repos/openclaw/openclaw/issues/58028 | Feature Request: Add /resume command to restore previous conversation context | ## Summary
Add a `/resume` slash command that allows users to restore a previous conversation session, similar to the `/resume` command in Claude Code.
## Problem
When a user runs `/clear` to reset the conversation context (e.g., because the context window got too long or the AI started getting confused), there is n... | open | null | false | 1 | [] | [] | 2026-03-31T01:03:09Z | 2026-05-02T05:37:12Z | null | NONE | null | 20260502T060142Z | 2026-05-02T06:01:42Z | TonyHe-Lab | 251,595,513 | U_kgDODv8K-Q | User | false |
openclaw/openclaw | 4,344,231,702 | I_kwDOQb6kR88AAAABAu-3Fg | 73,625 | https://github.com/openclaw/openclaw/issues/73625 | https://api.github.com/repos/openclaw/openclaw/issues/73625 | [Bug]: iMessage attachments not coming through - permissions boundary for multi-agent setup | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
I am experiencing imessage attachments are no longer coming through once i setup a multi-agent deployment.
The image tool's allowed directory list is built per agent, using each agent's workspace directory as... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-28T15:13:25Z | 2026-05-01T04:43:03Z | 2026-04-28T15:18:57Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | aflo101 | 32,745,330 | MDQ6VXNlcjMyNzQ1MzMw | User | false |
openclaw/openclaw | 4,344,282,714 | I_kwDOQb6kR88AAAABAvB-Wg | 73,631 | https://github.com/openclaw/openclaw/issues/73631 | https://api.github.com/repos/openclaw/openclaw/issues/73631 | [Bug]: WebSocket handshake-timeout on reconnect causes Control UI to stay disconnected for minutes (nginx reverse proxy) | ## Summary
After a browser refresh (or any WebSocket reconnect), the Control UI stays disconnected for **several minutes** while the client retries. Each WebSocket handshake attempt fails with `handshake-timeout` (12–63 seconds per attempt) before eventually succeeding.
## Environment
- OpenClaw version: 2026.4.26
-... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T15:21:09Z | 2026-05-01T04:43:02Z | 2026-04-28T15:25:32Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | jkobject | 17,239,533 | MDQ6VXNlcjE3MjM5NTMz | User | false |
openclaw/openclaw | 4,336,972,016 | I_kwDOQb6kR88AAAABAoDw8A | 72,885 | https://github.com/openclaw/openclaw/issues/72885 | https://api.github.com/repos/openclaw/openclaw/issues/72885 | Bug: memory_search mislabels session transcript hits as corpus=memory | ## Summary
When `memory_search` returns session-transcript hits from the built-in memory backend, the tool response labels those results with `corpus: "memory"` even when the underlying hit has `source: "sessions"`.
This makes session-backed hits indistinguishable from durable memory-file hits for downstream consumer... | open | null | false | 2 | [] | [] | 2026-04-27T15:40:17Z | 2026-05-03T09:13:13Z | null | NONE | null | 20260503T120038Z | 2026-05-03T12:00:38Z | funmerlin | 258,679,497 | U_kgDOD2siyQ | User | false |
openclaw/openclaw | 4,344,322,942 | I_kwDOQb6kR88AAAABAvEbfg | 73,634 | https://github.com/openclaw/openclaw/issues/73634 | https://api.github.com/repos/openclaw/openclaw/issues/73634 | WebChat: Restore browser-based voice input (MediaRecorder / getUserMedia) | ## Type
Feature Request
## Summary
The WebChat UI previously supported browser-based voice input using the MediaRecorder / getUserMedia APIs. This functionality appears to have been removed and is no longer available in the WebChat interface.
## Current Behavior
- The WebChat UI (accessed via browser) has **no microp... | closed | not_planned | true | 1 | [] | [] | 2026-04-28T15:27:48Z | 2026-05-01T04:43:01Z | 2026-04-28T15:39:13Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | TRI-TF | 168,457,053 | U_kgDOCgpzXQ | User | false |
openclaw/openclaw | 4,344,336,435 | I_kwDOQb6kR88AAAABAvFQMw | 73,636 | https://github.com/openclaw/openclaw/issues/73636 | https://api.github.com/repos/openclaw/openclaw/issues/73636 | [Bug]: Remote CLI TUI stuck in device-pairing-required loop after device wipe | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Remote CLI `openclaw tui` (against a tailnet-fronted gateway) is locked into a `device pairing required` loop after a gateway-side device wipe: every connect opens a brand-new pending pair request that expires sil... | closed | completed | true | 1 | [] | [] | 2026-04-28T15:29:58Z | 2026-05-01T04:43:01Z | 2026-04-28T15:43:30Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | chepurko | 206,400 | MDQ6VXNlcjIwNjQwMA== | User | false |
openclaw/openclaw | 4,229,493,313 | I_kwDOQb6kR878GPJB | 63,550 | https://github.com/openclaw/openclaw/issues/63550 | https://api.github.com/repos/openclaw/openclaw/issues/63550 | Multi-User/Partition Support for Memory Search | ### Summary
Add support for multi-user/partition isolation in the memory-lancedb plugin, allowing each user partition to have isolated memory storage and search results.
### Problem to solve
The `memory-lancedb` plugin uses a single shared LanceDB table for all users. When OpenClaw is configured with user partitions... | open | null | false | 2 | [
"enhancement"
] | [] | 2026-04-09T04:49:36Z | 2026-05-02T06:09:48Z | null | NONE | null | 20260502T090030Z | 2026-05-02T09:00:30Z | XingchenYu | 12,799,238 | MDQ6VXNlcjEyNzk5MjM4 | User | false |
openclaw/openclaw | 4,344,459,845 | I_kwDOQb6kR88AAAABAvMyRQ | 73,653 | https://github.com/openclaw/openclaw/issues/73653 | https://api.github.com/repos/openclaw/openclaw/issues/73653 | Gateway AgentParamsSchema rejects adapter-injected metadata fields (paperclip case) | ## Summary
`AgentParamsSchema` (in `dist/protocol-*.js`, e.g. `protocol-Hjar_s3V.js` in 2026.4.26) is closed with `additionalProperties: false` and only allows the 35 keys listed below. Adapters that legitimately want to attach context metadata to a wake — Paperclip's `@paperclipai/adapter-openclaw-gateway` is the cas... | closed | not_planned | true | 2 | [] | [] | 2026-04-28T15:48:16Z | 2026-05-03T04:36:16Z | 2026-04-30T15:31:42Z | NONE | null | 20260503T060030Z | 2026-05-03T06:00:30Z | bryangauvin | 36,140,319 | MDQ6VXNlcjM2MTQwMzE5 | User | false |
openclaw/openclaw | 4,085,348,156 | I_kwDOQb6kR87zgXc8 | 48,558 | https://github.com/openclaw/openclaw/issues/48558 | https://api.github.com/repos/openclaw/openclaw/issues/48558 | Feature Request: Native support for Anthropic Memory Tool (memory_20250818) | ## Summary
OpenClaw should natively support Anthropic's Memory Tool (`memory_20250818`) as a first-class tool type for agents. This would give Claude agents self-directed persistent memory across conversations — Claude decides what to remember and retrieves it on demand.
## What is the Memory Tool?
Anthropic's [Memo... | closed | not_planned | true | 3 | [] | [] | 2026-03-16T23:35:16Z | 2026-05-01T04:43:00Z | 2026-04-28T15:57:49Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | Orionation | 260,579,663 | U_kgDOD4ghTw | User | false |
openclaw/openclaw | 4,184,959,666 | I_kwDOQb6kR875cWqy | 58,842 | https://github.com/openclaw/openclaw/issues/58842 | https://api.github.com/repos/openclaw/openclaw/issues/58842 | [Bug]: model hanging | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
I have an old computer with access to 11.9 gb ram and I am trying to got run openclaw locally, the model I’m running is llama3.2:1b. When I run the model separately it runs fine at a pretty good speed but when I try to run it t... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-04-01T07:58:34Z | 2026-05-01T04:42:59Z | 2026-04-28T15:57:56Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | TwoShot2 | 89,351,851 | MDQ6VXNlcjg5MzUxODUx | User | false |
openclaw/openclaw | 4,195,409,873 | I_kwDOQb6kR876EN_R | 59,767 | https://github.com/openclaw/openclaw/issues/59767 | https://api.github.com/repos/openclaw/openclaw/issues/59767 | [Bug]: Voice Wake trigger word matching fails on Mac mini (Mac16,10) macOS 26.2 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
# Voice Wake Bug Report - Mac mini (Mac16,10) macOS 26.2
## 问题描述
Voice Wake 功能在 Mac mini 上无法正常工作：麦克风和语音识别均正常，但触发词(wake word)始终无法匹配成功。
## 环境信息
- **设备**: Mac mini (Mac16,10)
- **macOS**: 26.2 (Build 25C56)
- **OpenClaw**: 20... | closed | not_planned | true | 2 | [
"bug",
"regression"
] | [] | 2026-04-02T15:36:10Z | 2026-05-01T04:42:58Z | 2026-04-28T15:58:14Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | quanmiao99 | 133,095,482 | U_kgDOB-7gOg | User | false |
openclaw/openclaw | 4,196,961,278 | I_kwDOQb6kR876KIv- | 59,909 | https://github.com/openclaw/openclaw/issues/59909 | https://api.github.com/repos/openclaw/openclaw/issues/59909 | Add TUI footer display configuration options | ## Problem
The TUI footer currently always shows the full session name, which can be long (e.g., `agent:main:tui-e868c846-aedf-4d02-a1e4-ec73044d0141`). There's no way to hide or customize which elements are displayed.
According to docs: "The current agent + session are always visible in the footer."
## Proposed Sol... | closed | not_planned | true | 1 | [] | [] | 2026-04-02T21:00:29Z | 2026-05-01T04:42:57Z | 2026-04-28T15:58:32Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | xinye1017 | 113,286,396 | U_kgDOBsCc_A | User | false |
openclaw/openclaw | 4,344,380,019 | I_kwDOQb6kR88AAAABAvH6cw | 73,645 | https://github.com/openclaw/openclaw/issues/73645 | https://api.github.com/repos/openclaw/openclaw/issues/73645 | [Bug]: sidecars.channels stalls 145–625s on startup due to redundant loadOpenClawPlugins call from shouldSuppressBuiltInModel | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
On every gateway startup, the sidecars.channels phase stalls for 145–625 seconds before the gateway becomes ready.
### Steps to reproduce
1. Start the gateway with OPENCLAW_GATEWAY_STARTUP_TRACE=1.
2. Observe the sidecar... | closed | completed | true | 2 | [
"bug",
"bug:crash"
] | [] | 2026-04-28T15:36:36Z | 2026-05-01T04:42:57Z | 2026-04-28T15:41:44Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | 0822Jett | 280,067,816 | U_kgDOELF-6A | User | false |
openclaw/openclaw | 4,344,516,005 | I_kwDOQb6kR88AAAABAvQNpQ | 73,656 | https://github.com/openclaw/openclaw/issues/73656 | https://api.github.com/repos/openclaw/openclaw/issues/73656 | Bug: subagent initial visible message hides delegated task | ## Summary
Subagent sessions currently begin with a visible user message that does not include the delegated task:
```text
[Subagent Context] You are running as a subagent (depth 1/1). Results auto-announce to your requester; do not busy-poll for status.
Begin. Your assigned task is in the system prompt under Your R... | open | null | false | 1 | [] | [] | 2026-04-28T15:56:55Z | 2026-05-03T08:25:54Z | null | NONE | null | 20260503T090142Z | 2026-05-03T09:01:42Z | ramitrkar-hash | 280,062,852 | U_kgDOELFrhA | User | false |
openclaw/openclaw | 4,344,524,821 | I_kwDOQb6kR88AAAABAvQwFQ | 73,657 | https://github.com/openclaw/openclaw/issues/73657 | https://api.github.com/repos/openclaw/openclaw/issues/73657 | [Bug]: cli-runner reports bare alias as agentMeta.model; persisted to sessions.json; primary candidate not alias-resolved → cascading 404 + auth profile cooldown | ## Summary
For agents using a CLI backend (e.g. `claude-cli/sonnet`), every successful turn poisons the persisted session entry with a bare model alias (`'sonnet'`, `'haiku'`, `'opus'`). On the next request the bare alias is sent literally to api.anthropic.com → HTTP 404 `model_not_found`. A single 404 puts the auth p... | closed | completed | true | 2 | [] | [] | 2026-04-28T15:58:28Z | 2026-05-01T04:42:07Z | 2026-04-28T19:42:34Z | NONE | null | 20260501T113057Z | 2026-05-01T11:30:57Z | hashslingers | 129,774,036 | U_kgDOB7wx1A | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.