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 | 4,017,688,433 | I_kwDOQb6kR87veQ9x | 33,278 | https://github.com/openclaw/openclaw/issues/33278 | https://api.github.com/repos/openclaw/openclaw/issues/33278 | /status should show configured fallback models, not only active fallback | ## Current behavior
`/status` only shows the fallback model when it is *actively* being used (i.e. when the primary model is rate-limited and the system has fallen back):
```
🧠 Model: anthropic/claude-opus-4-6 · 🔑 token (anthropic:default)
↪️ Fallback: openai-codex/gpt-5.3-codex (selected model unavailable)
```
Wh... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T16:49:04Z | 2026-04-19T04:24:02Z | 2026-04-16T04:26:40Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | braminatorh | 263,757,029 | U_kgDOD7ic5Q | User | false |
openclaw/openclaw | 4,017,684,181 | I_kwDOQb6kR87veP7V | 33,274 | https://github.com/openclaw/openclaw/issues/33274 | https://api.github.com/repos/openclaw/openclaw/issues/33274 | Feature: Built-in Dead Man Switch for change control operations | ## Summary
OpenClaw has no native safety mechanism for agents making configuration
changes. When an agent modifies `openclaw.json` and something goes wrong,
there is no automatic recovery — the gateway goes down and stays down until
a human intervenes. This issue proposes a built-in Dead Man Switch (DMS)
to solve t... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T16:48:13Z | 2026-04-19T04:24:04Z | 2026-04-16T04:26:38Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | RichCorbett | 259,269,566 | U_kgDOD3Qjvg | User | false |
openclaw/openclaw | 4,017,679,639 | I_kwDOQb6kR87veO0X | 33,272 | https://github.com/openclaw/openclaw/issues/33272 | https://api.github.com/repos/openclaw/openclaw/issues/33272 | Custom provider reasoning: true + thinkingDefault sends reasoning_effort to unsupported endpoints | ## Bug Description
When a custom provider (e.g., `custom-ollama-com` using `api: "openai-completions"`) has models configured with `"reasoning": true` and the global `thinkingDefault` is set to anything other than `"off"`, OpenClaw injects `reasoning_effort` into the chat completions request body. This breaks tool cal... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-03T16:47:15Z | 2026-04-25T04:19:09Z | 2026-04-22T04:30:23Z | NONE | null | 20260425T060021Z | 2026-04-25T06:00:21Z | lorenzo-atelier | 263,712,792 | U_kgDOD7fwGA | User | false |
openclaw/openclaw | 4,017,676,976 | I_kwDOQb6kR87veOKw | 33,271 | https://github.com/openclaw/openclaw/issues/33271 | https://api.github.com/repos/openclaw/openclaw/issues/33271 | POST /hooks/wake silently drops requests in 2026.3.2 (returns 200 but never wakes session) | ## Description
`POST /hooks/wake` accepts webhook requests and returns `200 OK` with `{"ok":true,"mode":"now"}`, but the session is never actually woken. No system event is enqueued, no heartbeat is triggered, and no log entry is produced for the wake attempt.
Mapped hooks via `POST /hooks/<name>` with `"action": "wa... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T16:46:42Z | 2026-04-19T04:24:05Z | 2026-04-16T04:26:34Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | bmilne1981 | 138,170,065 | U_kgDOCDxO0Q | User | false |
openclaw/openclaw | 4,017,664,070 | I_kwDOQb6kR87veLBG | 33,268 | https://github.com/openclaw/openclaw/issues/33268 | https://api.github.com/repos/openclaw/openclaw/issues/33268 | CLI `openclaw approvals allowlist add --agent "*"` writes to literal wildcard key instead of expanding to concrete agents | **Environment:**
- OpenClaw: v2026.3.2 (85377a2)
- macOS: 15.7.3 (arm64)
**Issue:**
CLI command with `--agent "*"` writes allowlist entries to a literal `agents["*"]` key that runtime ignores, instead of expanding to concrete agent ids (main, minimax, etc.).
**Reproduction:**
```bash
openclaw approvals allowlist add ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T16:43:53Z | 2026-04-19T04:24:06Z | 2026-04-16T04:26:33Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | aboveUAV | 261,767,811 | U_kgDOD5pCgw | User | false |
openclaw/openclaw | 4,017,641,790 | I_kwDOQb6kR87veFk- | 33,263 | https://github.com/openclaw/openclaw/issues/33263 | https://api.github.com/repos/openclaw/openclaw/issues/33263 | [Bug]: Webchat duplicates assistant replies via delivery-mirror transcript entry | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
In Control UI Webchat, a single assistant turn is often rendered/stored as two assistant messages (normal model reply + openclaw/delivery-mirror).
### Steps to reproduce
1. Run OpenClaw 2026.3.2 on macOS.
2. Onboard/openai-codex OAuth and... | closed | not_planned | true | 5 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-03T16:38:52Z | 2026-04-20T05:33:02Z | 2026-04-16T04:26:31Z | NONE | null | 20260420T060038Z | 2026-04-20T06:00:38Z | ryou0903 | 131,286,605 | U_kgDOB9NGTQ | User | false |
openclaw/openclaw | 4,017,607,828 | I_kwDOQb6kR87vd9SU | 33,261 | https://github.com/openclaw/openclaw/issues/33261 | https://api.github.com/repos/openclaw/openclaw/issues/33261 | [Bug]: /skills page shows only main agent workspace skills in multi-agent setup (misleading global scope) | ## Summary
In a multi-agent setup, the Control UI **Skills** page (`/skills`) appears to be global, but the `openclaw-workspace` group effectively points to the **main agent's workspace skills** only.
<img width="1788" height="1023" alt="Image" src="https://github.com/user-attachments/assets/25f789cf-e921-432e-a440-f6... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T16:31:39Z | 2026-04-19T04:24:08Z | 2026-04-16T04:26:29Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | Undertone0809 | 72,488,598 | MDQ6VXNlcjcyNDg4NTk4 | User | false |
openclaw/openclaw | 4,017,597,109 | I_kwDOQb6kR87vd6q1 | 33,259 | https://github.com/openclaw/openclaw/issues/33259 | https://api.github.com/repos/openclaw/openclaw/issues/33259 | [Bug]:Control UI shows stale version after upgrade | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
After upgrading OpenClaw from 2026.2.21-2 to 2026.3.2, the version label
in the Control UI (top right) still shows 2026.2.21-2 with a green badge.
The gateway correctly reports 2026.3.2 via CLI (openclaw --version) and
the installed pac... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-03T16:29:19Z | 2026-04-19T04:24:08Z | 2026-04-16T04:26:27Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | pmoc2001 | 752,957 | MDQ6VXNlcjc1Mjk1Nw== | User | false |
openclaw/openclaw | 3,910,911,360 | I_kwDOQb6kR87pG8WA | 11,352 | https://github.com/openclaw/openclaw/issues/11352 | https://api.github.com/repos/openclaw/openclaw/issues/11352 | Auth profile cooldown backoff too aggressive for transient rate limits | ## Problem
When a single transient HTTP 429 rate limit error occurs (e.g., per-minute burst limit), the auth profile cooldown system enters an exponential backoff spiral that can lock out the provider for 20+ minutes, even when the actual API quota is barely used (e.g., 10% of Max plan).
## What happens
1. Profile A... | closed | not_planned | true | 10 | [
"bug",
"stale"
] | [] | 2026-02-07T18:32:12Z | 2026-04-16T04:26:26Z | 2026-04-13T04:26:18Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | sazora | 54,324,712 | MDQ6VXNlcjU0MzI0NzEy | User | false |
openclaw/openclaw | 4,017,583,693 | I_kwDOQb6kR87vd3ZN | 33,256 | https://github.com/openclaw/openclaw/issues/33256 | https://api.github.com/repos/openclaw/openclaw/issues/33256 | [Bug]: googlechat broken in 2026.3.3 | ### Bug type
Regression (worked before, now fails)
### Summary
googlechat was working, updated from 2026.2.26 to 2026.3.3, no longer working.
Running `openclaw doctor` shows:
`Google Chat: failed (unknown) - Cannot convert undefined or null to object`
the webhook returns 401 to the caller now.
My googlechat confi... | closed | not_planned | true | 8 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-03T16:26:36Z | 2026-04-19T04:24:09Z | 2026-04-16T04:26:25Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | donbowman | 5,131,923 | MDQ6VXNlcjUxMzE5MjM= | User | false |
openclaw/openclaw | 3,965,452,821 | I_kwDOQb6kR87sXAIV | 21,329 | https://github.com/openclaw/openclaw/issues/21329 | https://api.github.com/repos/openclaw/openclaw/issues/21329 | Slack streaming truncation: messages split/truncated despite streaming: false | ## Description
Slack messages are truncated mid-stream, then the remainder posts as a separate message. The first message shows "(edited)" suffix. This occurs even with `channels.slack.streaming: false` and `blockStreamingDefault: "on"` in config.
## Environment
- **OpenClaw version:** 2026.2.19-2
- **Channel:** Sla... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-02-19T22:27:48Z | 2026-04-16T04:26:25Z | 2026-04-13T04:26:37Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | chachi-max | 258,696,601 | U_kgDOD2tlmQ | User | false |
openclaw/openclaw | 3,963,935,107 | I_kwDOQb6kR87sRNmD | 21,084 | https://github.com/openclaw/openclaw/issues/21084 | https://api.github.com/repos/openclaw/openclaw/issues/21084 | Session JSONL files not persisted on crash — data loss risk | ## Feature Request / Bug
When the gateway process crashes unexpectedly, active session JSONL files may not be persisted to disk. This results in complete loss of the conversation log for any in-flight sessions.
## Context
On Feb 19, 2026, our gateway crashed due to an unhandled promise rejection (#21082). The sessio... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-02-19T16:45:13Z | 2026-04-16T04:26:25Z | 2026-04-13T04:26:34Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | mafresh-max | 258,871,817 | U_kgDOD24SCQ | User | false |
openclaw/openclaw | 3,969,347,734 | I_kwDOQb6kR87sl3CW | 22,003 | https://github.com/openclaw/openclaw/issues/22003 | https://api.github.com/repos/openclaw/openclaw/issues/22003 | Session context bleeding between channels causes 401 auth errors | ## Bug Description
When using both webchat and Telegram simultaneously, session context bleeds between channels causing authentication failures and tool execution errors.
## Error Messages
-
## Environment
- OpenClaw version: 2026.2.15
- Channels enabled: webchat, telegram, discord, slack
- Primary agent: main (Floy... | closed | not_planned | true | 8 | [
"stale"
] | [] | 2026-02-20T16:27:49Z | 2026-04-16T04:26:24Z | 2026-04-13T04:26:41Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | shawnhyd3-crypto | 236,213,060 | U_kgDODhRTRA | User | false |
openclaw/openclaw | 3,981,900,014 | I_kwDOQb6kR87tVvju | 25,157 | https://github.com/openclaw/openclaw/issues/25157 | https://api.github.com/repos/openclaw/openclaw/issues/25157 | feat: make embedding batch timeout configurable for remote/local providers (Ollama) | ## Problem
When using a local embedding provider served via Ollama with `provider: "openai"` and `remote.baseUrl` pointing to an Ollama instance, the embedding batch timeout is hardcoded to **2 minutes** (`EMBEDDING_BATCH_TIMEOUT_REMOTE_MS = 2 * 60000`).
This is too short for larger local models (e.g. `qwen3-embeddin... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-02-24T06:52:36Z | 2026-04-16T04:26:23Z | 2026-04-13T04:26:52Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ItamarCoh3n | 26,421,901 | MDQ6VXNlcjI2NDIxOTAx | User | false |
openclaw/openclaw | 4,017,105,416 | I_kwDOQb6kR87vcCoI | 33,120 | https://github.com/openclaw/openclaw/issues/33120 | https://api.github.com/repos/openclaw/openclaw/issues/33120 | Dashboard: Agent model selector shows "Inherit default" when model uses object format | ## Description
When an agent's model config uses the object format with primary + fallbacks:
```json
{
"primary": "anthropic/claude-opus-4-6",
"fallbacks": ["openai-codex/gpt-5.3-codex", "anthropic/claude-sonnet-4-6"]
}
```
The **Model Selection** dropdown on the Agents page shows `Inherit default (openai-codex/... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T14:49:57Z | 2026-04-19T04:24:10Z | 2026-04-16T04:26:22Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | cristian-martin | 39,029,499 | MDQ6VXNlcjM5MDI5NDk5 | User | false |
openclaw/openclaw | 3,991,888,453 | I_kwDOQb6kR87t72JF | 26,877 | https://github.com/openclaw/openclaw/issues/26877 | https://api.github.com/repos/openclaw/openclaw/issues/26877 | BOOTSTRAP.md [MISSING] confuses bound agents into thinking they are new | ## Problem
When a bound agent (e.g. `fleet-update`) starts a session, OpenClaw injects workspace context files including `BOOTSTRAP.md`. If the file doesn't exist, it renders as:
```
[MISSING] Expected at: /home/claudedev/clawd/agents/fleet-update/BOOTSTRAP.md
```
The LLM interprets `[MISSING]` as "something expecte... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-02-25T21:20:09Z | 2026-04-16T04:26:22Z | 2026-04-13T04:27:00Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | vovapetry | 144,446,973 | U_kgDOCJwV_Q | User | false |
openclaw/openclaw | 3,990,140,311 | I_kwDOQb6kR87t1LWX | 26,624 | https://github.com/openclaw/openclaw/issues/26624 | https://api.github.com/repos/openclaw/openclaw/issues/26624 | False positive warning: 'Config token differs from service token' when tokens match | ## Bug Description
`openclaw gateway restart` displays a warning about token mismatch even when the config and service file tokens are identical:
```
⚠️ Config token differs from service token. The daemon will use the old token after restart.
Run `openclaw gateway install --force` to sync the token.
```
## Reproduct... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-02-25T15:29:04Z | 2026-04-16T04:26:22Z | 2026-04-13T04:26:58Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jbrianfrancis-ir | 127,874,124 | U_kgDOB580TA | User | false |
openclaw/openclaw | 3,992,037,661 | I_kwDOQb6kR87t8akd | 26,895 | https://github.com/openclaw/openclaw/issues/26895 | https://api.github.com/repos/openclaw/openclaw/issues/26895 | [Bug] Legacy telegramslash session key is re-created after status on OpenClaw 2026.2.24 post-migration | ### Summary
After successful migration, Telegram slash commands re-create legacy session key format (telegramslash214522323) instead of canonical format (agentmaintelegramslash214522323). Running 'openclaw doctor --fix' canonicalizes the key, but it reappears after the next slash command interaction.
### Steps to rep... | closed | not_planned | true | 8 | [
"bug",
"stale"
] | [] | 2026-02-25T21:55:38Z | 2026-04-16T04:26:21Z | 2026-04-13T04:27:02Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | fireniko | 211,585,889 | U_kgDODJyLYQ | User | false |
openclaw/openclaw | 4,017,099,262 | I_kwDOQb6kR87vcBH- | 33,119 | https://github.com/openclaw/openclaw/issues/33119 | https://api.github.com/repos/openclaw/openclaw/issues/33119 | [Bug]: Control UI (webchat) cannot display image attachments sent by AI agent | ### Bug type
Regression (worked before, now fails)
### Summary
When the AI agent sends image attachments via the read tool (reading local image files), the images are not displayed in the Control UI (webchat) interface. The tool executes successfully and returns "Read image file", but the image does not appear in th... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-03T14:48:41Z | 2026-04-19T04:24:11Z | 2026-04-16T04:26:20Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | duyiliu | 4,182,886 | MDQ6VXNlcjQxODI4ODY= | User | false |
openclaw/openclaw | 4,016,979,892 | I_kwDOQb6kR87vbj-0 | 33,102 | https://github.com/openclaw/openclaw/issues/33102 | https://api.github.com/repos/openclaw/openclaw/issues/33102 | TUI: Add config support for --deliver flag default | ## Description
The `openclaw tui` command has a `--deliver` flag that defaults to `false`. This means assistant replies are stored but not printed to stdout by default, which is confusing for users.
## Expected behavior
Users should be able to set a default value for `--deliver` in the OpenClaw config (e.g., `tui.de... | open | null | false | 5 | [] | [] | 2026-03-03T14:26:51Z | 2026-05-05T03:11:03Z | null | NONE | null | 20260505T060034Z | 2026-05-05T06:00:34Z | samoclah | 261,493,458 | U_kgDOD5YS0g | User | false |
openclaw/openclaw | 4,016,915,657 | I_kwDOQb6kR87vbUTJ | 33,095 | https://github.com/openclaw/openclaw/issues/33095 | https://api.github.com/repos/openclaw/openclaw/issues/33095 | Feature: Agent interrupt and wall broadcast | ## Summary
There is currently no way for an orchestrator to interrupt a running
sub-agent mid-task, nor to broadcast a message to all active agents
simultaneously via the bulletin wall.
## Use Case
In a live orchestration scenario:
- A P0 incident fires while several sub-agents are mid-task
- The orchestrator needs ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T14:15:00Z | 2026-04-19T04:24:11Z | 2026-04-16T04:26:18Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | RichCorbett | 259,269,566 | U_kgDOD3Qjvg | User | false |
openclaw/openclaw | 3,994,482,647 | I_kwDOQb6kR87uFvfX | 27,398 | https://github.com/openclaw/openclaw/issues/27398 | https://api.github.com/repos/openclaw/openclaw/issues/27398 | [Feature Request] Improve Tailnet detection to support Headscale/custom IP ranges | ### Problem
Currently, `bind: "tailnet"` fails to recognize the Tailscale interface when using **Headscale** or customized IP prefixes (e.g., `192.168.100.0/24`). It appears to rely strictly on the official `100.64.0.0/10` range.
### Context
- **OS:** macOS (using `utun` interfaces)
- **Network:** Headscale with custo... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-02-26T09:52:25Z | 2026-04-16T04:26:18Z | 2026-04-13T04:27:08Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | JasprW | 5,129,532 | MDQ6VXNlcjUxMjk1MzI= | User | false |
openclaw/openclaw | 3,993,867,966 | I_kwDOQb6kR87uDZa- | 27,314 | https://github.com/openclaw/openclaw/issues/27314 | https://api.github.com/repos/openclaw/openclaw/issues/27314 | [Bug]: openclaw configure to add Telegram channel destroys workspace, sessions, memory, etc. | ### Summary
### Description
Running `openclaw configure` to add a Telegram bot channel resulted in complete, unrecoverable data loss. The workspace was re-seeded with fresh bootstrap templates, all session history was deleted, memory databases were wiped, and API auth credentials were removed.
There was **no warning... | closed | not_planned | true | 8 | [
"bug",
"stale"
] | [] | 2026-02-26T07:46:49Z | 2026-04-16T04:26:18Z | 2026-04-13T04:27:06Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | warmonger0 | 11,595,517 | MDQ6VXNlcjExNTk1NTE3 | User | false |
openclaw/openclaw | 3,997,757,447 | I_kwDOQb6kR87uSPAH | 27,919 | https://github.com/openclaw/openclaw/issues/27919 | https://api.github.com/repos/openclaw/openclaw/issues/27919 | openclaw configure/doctor silently overwrites workspace files (SOUL.md, AGENTS.md etc.) without backup | ## Bug
### Summary
`openclaw configure` and `openclaw doctor` silently overwrite all workspace `.md` files (SOUL.md, AGENTS.md, IDENTITY.md, USER.md, TOOLS.md, etc.) with default templates — **without warning, confirmation, or backup**.
This caused complete loss of 3 weeks of customized agent configuration and worksp... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-02-26T21:07:14Z | 2026-04-16T04:26:17Z | 2026-04-13T04:27:14Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ukr-coder | 246,043,930 | U_kgDODqpVGg | User | false |
openclaw/openclaw | 4,016,915,213 | I_kwDOQb6kR87vbUMN | 33,094 | https://github.com/openclaw/openclaw/issues/33094 | https://api.github.com/repos/openclaw/openclaw/issues/33094 | Feature: Sub-agent QoS / priority control | ## Summary
When spawning sub-agents, there is currently no way to signal priority or
resource weight. All sub-agents are treated equally regardless of the
criticality of the task they're running.
## Use Case
In a multi-agent orchestration setup (e.g. an Orchestrator spawning
several sub-agents in parallel), some ta... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T14:14:54Z | 2026-04-19T04:24:12Z | 2026-04-16T04:26:16Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | RichCorbett | 259,269,566 | U_kgDOD3Qjvg | User | false |
openclaw/openclaw | 4,004,400,246 | I_kwDOQb6kR87urkx2 | 29,678 | https://github.com/openclaw/openclaw/issues/29678 | https://api.github.com/repos/openclaw/openclaw/issues/29678 | [WhatsApp] Reasoning/thinking blocks not visible despite verbose=full and reasoning=on | ### Description
With both `reasoning: on` and `verbose: full` enabled via slash commands, the agent's thinking/reasoning blocks are not transmitted to WhatsApp. The runtime reports `Reasoning: on (hidden unless on/stream)` — the "hidden" behavior appears to override the verbose=full setting.
### Steps to reproduce
1... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-02-28T10:26:21Z | 2026-04-16T04:26:16Z | 2026-04-13T04:27:47Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | NicodemPL | 45,172,249 | MDQ6VXNlcjQ1MTcyMjQ5 | User | false |
openclaw/openclaw | 4,004,397,993 | I_kwDOQb6kR87urkOp | 29,677 | https://github.com/openclaw/openclaw/issues/29677 | https://api.github.com/repos/openclaw/openclaw/issues/29677 | [WhatsApp] messageId not exposed to agent — cannot update reactions dynamically (progress tracking) | ### Description
The `message action=react` tool requires a `messageId` parameter, but the inbound context metadata (`openclaw.inbound_meta.v1`) does not include the WhatsApp message ID of the triggering message. This makes it impossible for agents to react to inbound messages with emoji.
### Error
Calling `message a... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-02-28T10:25:21Z | 2026-04-16T04:26:16Z | 2026-04-13T04:27:46Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | NicodemPL | 45,172,249 | MDQ6VXNlcjQ1MTcyMjQ5 | User | false |
openclaw/openclaw | 4,005,756,268 | I_kwDOQb6kR87uwv1s | 30,152 | https://github.com/openclaw/openclaw/issues/30152 | https://api.github.com/repos/openclaw/openclaw/issues/30152 | Model allowlist silently drops models not in catalog, even when provider is implicitly configured | ## Bug
When `agents.defaults.models` contains models from providers configured via auth profiles (implicit discovery) rather than explicit `models.providers` entries, `buildAllowedModelSet` silently drops those models from the allowed set. This causes "model not allowed" errors for models the user explicitly added to ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T00:09:51Z | 2026-04-16T04:26:15Z | 2026-04-13T04:28:01Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | derrickburns | 583,930 | MDQ6VXNlcjU4MzkzMA== | User | false |
openclaw/openclaw | 4,016,907,915 | I_kwDOQb6kR87vbSaL | 33,093 | https://github.com/openclaw/openclaw/issues/33093 | https://api.github.com/repos/openclaw/openclaw/issues/33093 | [Bug] Browser Relay: 'tab not found' after navigation even though /json/list shows target | # Upstream issue draft — Browser Relay `tab not found` after navigation (extension relay)
## Summary
When using the Chrome extension Browser Relay (profile `chrome`, driver=extension, loopback relay ports 18791/18792), `tabs()` (and direct `/json/list`) continues to list attached targets with valid `targetId`s, but `s... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T14:13:38Z | 2026-04-19T04:24:13Z | 2026-04-16T04:26:14Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | john-openclaw-bot | 264,207,373 | U_kgDOD798DQ | User | false |
openclaw/openclaw | 4,016,903,739 | I_kwDOQb6kR87vbRY7 | 33,092 | https://github.com/openclaw/openclaw/issues/33092 | https://api.github.com/repos/openclaw/openclaw/issues/33092 | Cron: recover missed jobs after gateway restart | ## Problem
When the gateway restarts (e.g. `openclaw gateway restart`, config change, update, crash), any cron job whose scheduled time falls during the brief downtime window is **silently skipped**. The `nextRunAtMs` advances to the next occurrence with no indication the job was missed.
This means a 2-second restart... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T14:12:51Z | 2026-04-19T04:24:14Z | 2026-04-16T04:26:12Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | ryanlowdermilk | 342,360 | MDQ6VXNlcjM0MjM2MA== | User | false |
openclaw/openclaw | 4,007,180,219 | I_kwDOQb6kR87u2Le7 | 30,667 | https://github.com/openclaw/openclaw/issues/30667 | https://api.github.com/repos/openclaw/openclaw/issues/30667 | [Bug]: Agent models.json cache not synced when openclaw.json provider baseUrl changes | ## Summary
When a user updates `models.providers.*.baseUrl` in `~/.openclaw/openclaw.json`, the cached model configuration at `~/.openclaw/agents/<id>/agent/models.json` is **not updated**. The gateway continues to use the stale `baseUrl` from the cached file, causing persistent `Connection error.` with no actionable ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T13:21:01Z | 2026-04-16T04:26:12Z | 2026-04-13T04:28:15Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ksp-apoc | 256,353,461 | U_kgDOD0ektQ | User | false |
openclaw/openclaw | 4,016,874,043 | I_kwDOQb6kR87vbKI7 | 33,091 | https://github.com/openclaw/openclaw/issues/33091 | https://api.github.com/repos/openclaw/openclaw/issues/33091 | Gemini Flash reasoning content delivered as bare text to iMessage channel | ## Summary
When using `google/gemini-3-flash-preview` with `thinkingDefault: "off"`, the model embeds chain-of-thought reasoning as regular assistant text content. This text is delivered verbatim to messaging channels (iMessage) via `message()` tool calls.
## Observed behavior
Flash includes reasoning in two forms:
... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-03T14:06:54Z | 2026-04-19T04:24:14Z | 2026-04-16T04:26:11Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | Starefossen | 968,267 | MDQ6VXNlcjk2ODI2Nw== | User | false |
openclaw/openclaw | 4,007,192,875 | I_kwDOQb6kR87u2Okr | 30,672 | https://github.com/openclaw/openclaw/issues/30672 | https://api.github.com/repos/openclaw/openclaw/issues/30672 | Bedrock provider fails when api.anthropic.com is unreachable (geo-restriction) | ## Summary
When using AWS Bedrock as the model provider, OpenClaw's embedded agent still makes a direct HTTP request to `https://api.anthropic.com/api/oauth/usage`. If this endpoint is unreachable (e.g. due to Anthropic's geographic restrictions), the entire agent run fails — even though the Bedrock model inference it... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T13:25:12Z | 2026-04-16T04:26:11Z | 2026-04-13T04:28:18Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | outstandingcandy | 5,536,666 | MDQ6VXNlcjU1MzY2NjY= | User | false |
openclaw/openclaw | 4,007,183,513 | I_kwDOQb6kR87u2MSZ | 30,669 | https://github.com/openclaw/openclaw/issues/30669 | https://api.github.com/repos/openclaw/openclaw/issues/30669 | Edit tool does not expand ~ (tilde) in file paths | ## Problem
The `Edit` tool (used for precise file edits) does not expand `~` in `file_path`/`path` parameters. This causes `File not found` errors when targeting files like `~/.codex/config.toml`.
Other tools handle this fine:
- `exec` — shell expands `~` naturally
- `Read` — resolves `~` to `$HOME`
The Edit tool by... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-01T13:22:20Z | 2026-04-16T04:26:11Z | 2026-04-13T04:28:17Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Chrisr6records | 256,422,608 | U_kgDOD0iy0A | User | false |
openclaw/openclaw | 4,007,206,236 | I_kwDOQb6kR87u2R1c | 30,675 | https://github.com/openclaw/openclaw/issues/30675 | https://api.github.com/repos/openclaw/openclaw/issues/30675 | web_search tool returns missing_gemini_api_key for concurrent subagent calls | ## Description
When a subagent (Gemini Flash model) fires multiple parallel `web_search` tool calls in a single turn, the first call succeeds but subsequent concurrent calls return:
```json
{"error": "missing_gemini_api_key", "message": "web_search (gemini) needs an API key..."}
```
The API key is correctly configur... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T13:30:14Z | 2026-04-16T04:26:10Z | 2026-04-13T04:28:20Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | CountofGlamorgan | 59,939,121 | MDQ6VXNlcjU5OTM5MTIx | User | false |
openclaw/openclaw | 4,016,863,257 | I_kwDOQb6kR87vbHgZ | 33,089 | https://github.com/openclaw/openclaw/issues/33089 | https://api.github.com/repos/openclaw/openclaw/issues/33089 | [Bug]: HELP_MY_CLAW_NO_SOUL ！！每次醒来文件读取不完整，希望可以修改每次/new后的系统发出的第一句话 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
每次/new /reset之后，都会发【A new session was started via /new or /reset. Execute your Session Startup sequence now - read the required files before responding to the user. Then greet the user in your configured persona, if one is provided. Be your... | closed | not_planned | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-03T14:04:53Z | 2026-04-19T04:24:15Z | 2026-04-16T04:26:09Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | foreverxdord | 44,799,061 | MDQ6VXNlcjQ0Nzk5MDYx | User | false |
openclaw/openclaw | 4,007,233,084 | I_kwDOQb6kR87u2YY8 | 30,685 | https://github.com/openclaw/openclaw/issues/30685 | https://api.github.com/repos/openclaw/openclaw/issues/30685 | Tool surface mismatch across runtime lanes (session_status available in main, missing in subagent lane) | ## Summary
Tool availability is inconsistent between runtime lanes. In our environment, `session_status` is available in main agent sessions but not available in a local-model subagent lane, causing deterministic tool-call tests to fail even when model behavior is otherwise correct.
## Environment
- OpenClaw: 2026.2.2... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-01T13:42:58Z | 2026-04-16T04:26:09Z | 2026-04-13T04:28:22Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,007,241,557 | I_kwDOQb6kR87u2adV | 30,689 | https://github.com/openclaw/openclaw/issues/30689 | https://api.github.com/repos/openclaw/openclaw/issues/30689 | Gateway dashboard shows old version after update | ## Description
After updating OpenClaw from 2026.2.25 to 2026.2.26, the Control UI dashboard still shows the old version number.
## Steps to Reproduce
1. Updated via `openclaw update`, `npm install -g openclaw@latest`, full reinstall, and reboot
2. CLI reports correct version: `openclaw --version` → `2026.2.26`
3. `... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-01T13:46:51Z | 2026-04-16T04:26:08Z | 2026-04-13T04:28:24Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jarvis-drakon | 257,379,654 | U_kgDOD1dNRg | User | false |
openclaw/openclaw | 4,007,323,417 | I_kwDOQb6kR87u2ucZ | 30,711 | https://github.com/openclaw/openclaw/issues/30711 | https://api.github.com/repos/openclaw/openclaw/issues/30711 | [Bug]: Sandbox exec: workdir path not mapped + stdout lost on background transition | ## Summary
Sandbox exec has two reliability issues that cause bun (and likely other runtimes) to produce no captured output:
1. **`workdir` parameter doesn't map host↔container paths** — passing `workdir: "/workspace"` fails because the Gateway validates the path on the host side, where `/workspace` doesn't exist.
2.... | closed | not_planned | true | 3 | [
"bug",
"stale"
] | [] | 2026-03-01T14:20:24Z | 2026-04-16T04:26:07Z | 2026-04-13T04:28:27Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | aaajiao | 49,885 | MDQ6VXNlcjQ5ODg1 | User | false |
openclaw/openclaw | 4,007,298,866 | I_kwDOQb6kR87u2ocy | 30,704 | https://github.com/openclaw/openclaw/issues/30704 | https://api.github.com/repos/openclaw/openclaw/issues/30704 | Bug: Internal tool-call trace leaked into user-visible chat message (Olvid group) | ### Summary
In Olvid group chat, internal tool-call trace text was accidentally sent as a user-visible message.
Example leaked fragment:
```
NO_REPLY +#+#+#+#+#+assistant to=functions.olvid_list_groups recipient_name... json {"olvidChannelAccountId":""}
```
### Expected
Only user-facing natural language should be del... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T14:11:33Z | 2026-04-16T04:26:07Z | 2026-04-13T04:28:25Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | siludose | 21,173,093 | MDQ6VXNlcjIxMTczMDkz | User | false |
openclaw/openclaw | 4,016,838,674 | I_kwDOQb6kR87vbBgS | 33,086 | https://github.com/openclaw/openclaw/issues/33086 | https://api.github.com/repos/openclaw/openclaw/issues/33086 | SSRF protection blocks Discord CDN when behind Clash Verge TUN (fake-ip) | ## Problem
When running OpenClaw behind **Clash Verge** with TUN mode enabled (fake-ip strategy), `cdn.discordapp.com` resolves to `198.18.x.x` (Clash's fake-ip range). OpenClaw's SSRF protection treats this as a private/internal IP and blocks the fetch.
This means **all Discord image and file attachments are blocked... | closed | not_planned | true | 11 | [
"stale"
] | [] | 2026-03-03T14:00:22Z | 2026-04-19T04:24:16Z | 2026-04-16T04:26:06Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | anxingjian | 14,119,645 | MDQ6VXNlcjE0MTE5NjQ1 | User | false |
openclaw/openclaw | 4,007,335,044 | I_kwDOQb6kR87u2xSE | 30,713 | https://github.com/openclaw/openclaw/issues/30713 | https://api.github.com/repos/openclaw/openclaw/issues/30713 | Bug: memory index command stuck in infinite loop | ## Environment
- OpenClaw Version: 2026.2.26
- macOS: Darwin 22.6.0
- Node: v24.13.1
## Description
The `openclaw memory index` command gets stuck in an infinite loop and never completes. Process runs but shows 0% CPU usage.
## Steps to Reproduce
1. Run `openclaw memory index --verbose --force`
2. Observe it never co... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T14:24:46Z | 2026-04-16T04:26:06Z | 2026-04-13T04:28:29Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Meisenburg | 75,205,246 | MDQ6VXNlcjc1MjA1MjQ2 | User | false |
openclaw/openclaw | 4,007,346,171 | I_kwDOQb6kR87u2z_7 | 30,714 | https://github.com/openclaw/openclaw/issues/30714 | https://api.github.com/repos/openclaw/openclaw/issues/30714 | [Feature Request] 跨渠道消息不同步 - 网页版与飞书聊天记录不互通 | ## 问题描述
用户在网页版 (webchat) 与 OpenClaw 对话后，在飞书中无法查看聊天记录，反之亦然。
## 期望行为
不同渠道的消息应该能够跨渠道同步，或者至少在统一界面查看历史记录。
## 环境信息
- OpenClaw 版本: 2026.2.25
- 渠道: webchat, feishu
- 模型: minimax/MiniMax-M2.5
## 补充说明
这是功能性需求，而非 bug。当前实现中，不同渠道的消息是分开存储的，导致用户需要在不同平台切换查看历史记录，体验不连贯。 | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T14:28:21Z | 2026-04-16T04:26:05Z | 2026-04-13T04:28:31Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ultrazhl98 | 72,202,522 | MDQ6VXNlcjcyMjAyNTIy | User | false |
openclaw/openclaw | 4,016,831,877 | I_kwDOQb6kR87va_2F | 33,085 | https://github.com/openclaw/openclaw/issues/33085 | https://api.github.com/repos/openclaw/openclaw/issues/33085 | BlueBubbles webhook: 'unable to parse message payload' on reactions/tapbacks | ## Description
After upgrading to OpenClaw 2026.3.2, the BlueBubbles webhook handler rejects certain incoming message types with:
```
[bluebubbles] webhook rejected: unable to parse message payload
```
This happens consistently for:
- **Reactions/tapbacks** (Liked, Emphasized, Laughed at, etc.)
- **Image attachments... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T13:59:02Z | 2026-04-19T04:24:16Z | 2026-04-16T04:26:04Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | Orionation | 260,579,663 | U_kgDOD4ghTw | User | false |
openclaw/openclaw | 4,007,471,376 | I_kwDOQb6kR87u3SkQ | 30,741 | https://github.com/openclaw/openclaw/issues/30741 | https://api.github.com/repos/openclaw/openclaw/issues/30741 | Discord: attachments silently fail to attach / no response triggered | ## Bug Report
### Summary
When sending messages with attachments in Discord, some attachments silently fail — the file does not get processed and no agent response is triggered. No error is shown to the user and no indication the attachment was dropped.
### Behavior
- User sends a Discord message with a file attachme... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-01T15:15:31Z | 2026-04-16T04:26:04Z | 2026-04-13T04:28:32Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | MiscMich | 76,935,191 | MDQ6VXNlcjc2OTM1MTkx | User | false |
openclaw/openclaw | 4,016,820,649 | I_kwDOQb6kR87va9Gp | 33,083 | https://github.com/openclaw/openclaw/issues/33083 | https://api.github.com/repos/openclaw/openclaw/issues/33083 | Bug: exec event timestamps appear incorrect (off by ~11.5 hours) | ## Bug Report
**Observed behavior:**
When a session starts, the exec failure log shows a timestamp that is approximately 11.5 hours behind the actual current time.
**Example:**
- Exec failure log timestamp: `2026-03-02 18:34:16 PST`
- Actual time when session started: `2026-03-03 05:53 PST`
- Discrepancy: ~11 hours 1... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-03T13:56:53Z | 2026-04-19T04:24:17Z | 2026-04-16T04:26:03Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | zhendalf | 5,859,177 | MDQ6VXNlcjU4NTkxNzc= | User | false |
openclaw/openclaw | 4,007,553,882 | I_kwDOQb6kR87u3mta | 30,764 | https://github.com/openclaw/openclaw/issues/30764 | https://api.github.com/repos/openclaw/openclaw/issues/30764 | Feature: User-configurable session lifecycle hooks (onBeforeReset, onSessionStart) |
## Summary
Add user-configurable script-level hooks for session lifecycle events, allowing deterministic file operations at session boundaries without relying on LLM compliance.
## Related Issues
- #17707 (RFC: Agent Lifecycle Hooks — plugin API level, currently stale)
- #27351 (workspace file dereference + session... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T15:52:37Z | 2026-04-16T04:26:03Z | 2026-04-13T04:28:36Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | cwmine | 4,034,274 | MDQ6VXNlcjQwMzQyNzQ= | User | false |
openclaw/openclaw | 4,007,525,065 | I_kwDOQb6kR87u3frJ | 30,760 | https://github.com/openclaw/openclaw/issues/30760 | https://api.github.com/repos/openclaw/openclaw/issues/30760 | [Feature]: modelByChannel should inherit parent channel ID for Discord threads | ## Problem
When a Discord channel is configured in `channels.modelByChannel`, the override only applies to messages sent directly in that channel. Discord threads spawn as separate sessions with their own IDs, and `modelByChannel` performs an exact key match — so the parent channel override never applies to thread ses... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T15:40:41Z | 2026-04-16T04:26:03Z | 2026-04-13T04:28:34Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ddonathan | 1,904,043 | MDQ6VXNlcjE5MDQwNDM= | User | false |
openclaw/openclaw | 4,007,562,298 | I_kwDOQb6kR87u3ow6 | 30,768 | https://github.com/openclaw/openclaw/issues/30768 | https://api.github.com/repos/openclaw/openclaw/issues/30768 | [Bug]: 企业微信重复消息 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
我在企业微信会话中，openclaw 有时候会多次给我发一样的消息，[是否跟飞书的重复信息问题是一样的？](https://github.com/openclaw/openclaw/issues/10557)
<img width="3347" height="3530" alt="Image" src="https://github.com/user-attachments/assets/514af732-db72-4895-97f5-1609a2abf151" />
... | closed | not_planned | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-01T15:56:21Z | 2026-04-16T04:26:02Z | 2026-04-13T04:28:38Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | qq31311137 | 38,181,726 | MDQ6VXNlcjM4MTgxNzI2 | User | false |
openclaw/openclaw | 4,016,820,336 | I_kwDOQb6kR87va9Bw | 33,082 | https://github.com/openclaw/openclaw/issues/33082 | https://api.github.com/repos/openclaw/openclaw/issues/33082 | Cron delivery drops TTS tool media (voice notes never reach Telegram) | ## Summary
When a cron job's agent calls the `tts` tool, the generated audio file is lost in the cron delivery pipeline. The text briefing arrives in Telegram, but the voice note does not. The same TTS flow works correctly for non-cron (inbound message) runs.
## Steps to reproduce
1. Create a cron job with `--announ... | closed | completed | true | 6 | [
"stale"
] | [] | 2026-03-03T13:56:49Z | 2026-04-30T04:45:13Z | 2026-04-25T06:49:23Z | NONE | null | 20260430T060018Z | 2026-04-30T06:00:18Z | klokie | 168,309 | MDQ6VXNlcjE2ODMwOQ== | User | false |
openclaw/openclaw | 4,007,583,401 | I_kwDOQb6kR87u3t6p | 30,771 | https://github.com/openclaw/openclaw/issues/30771 | https://api.github.com/repos/openclaw/openclaw/issues/30771 | openclaw-mem0: Support agent-scoped userId for multi-agent setups | ## Problem
In multi-agent setups, the openclaw-mem0 plugin uses a single global `userId` from config for both auto-recall and auto-capture. This means all agents share the same memory pool — Smith (NEIC support bot) sees infrastructure conversations from Parker, Jackie gets Smith's customer interactions, etc.
## Curr... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T16:04:16Z | 2026-04-16T04:26:01Z | 2026-04-13T04:28:40Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jpjanze | 45,051,594 | MDQ6VXNlcjQ1MDUxNTk0 | User | false |
openclaw/openclaw | 4,016,779,463 | I_kwDOQb6kR87vazDH | 33,068 | https://github.com/openclaw/openclaw/issues/33068 | https://api.github.com/repos/openclaw/openclaw/issues/33068 | feat: content script fallback channel for browser relay extension | ## Summary
The browser relay Chrome extension is currently a pure CDP proxy — all automation intelligence lives server-side in Playwright. This works for most sites but fails on:
- **Government sites** with framesets and strict CSP that interfere with CDP
- **React/Vue apps** with controlled inputs that ignore \`DOM.... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T13:48:42Z | 2026-04-19T04:24:18Z | 2026-04-16T04:26:00Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | saturn-bugatt | 250,167,540 | U_kgDODulA9A | User | false |
openclaw/openclaw | 4,007,634,520 | I_kwDOQb6kR87u36ZY | 30,783 | https://github.com/openclaw/openclaw/issues/30783 | https://api.github.com/repos/openclaw/openclaw/issues/30783 | DingTalk渠道消息延迟-工具调用阻塞消息队列 | 问题描述：
使用 DingTalk 渠道时，消息响应延迟严重。经常需要发送 2-3 条消息才有回复。
日志分析：
工具调用（exec/curl）耗时 10-20 秒，阻塞了消息队列。
环境：
- OpenClaw 版本：2026.2.26
- 渠道：DingTalk
- 模型：dashscope-coding/qwen3.5-plus
日志示例：
13:05:13 - 收到消息
13:05:31 - 工具 exec 开始（18 秒延迟）
13:06:26 - 回复发送
总计：73 秒
期望行为：
- 用户消息应优先于工具响应
- 工具调用应异步，不阻塞消息队列
临时解决方案：
- 工具调用设置超时（5-10 秒）
- 长任... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T16:22:23Z | 2026-04-16T04:26:00Z | 2026-04-13T04:28:43Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | wjfmy200-web | 261,499,455 | U_kgDOD5YqPw | User | false |
openclaw/openclaw | 4,007,596,440 | I_kwDOQb6kR87u3xGY | 30,773 | https://github.com/openclaw/openclaw/issues/30773 | https://api.github.com/repos/openclaw/openclaw/issues/30773 | [Bug]: Write tool reports error in isolated cron agentTurn jobs despite file being successfully created | ### Summary
In isolated `agentTurn` cron jobs, the `write` tool consistently reports a failure (shown as `lastError` and `lastDeliveryError` in cron job state), but the file is **actually created successfully** on disk. This causes `consecutiveErrors` to increment on every run even though the job completed its work co... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-01T16:08:46Z | 2026-04-16T04:26:00Z | 2026-04-13T04:28:41Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | gchahrouri | 37,080,676 | MDQ6VXNlcjM3MDgwNjc2 | User | false |
openclaw/openclaw | 4,007,639,012 | I_kwDOQb6kR87u37fk | 30,785 | https://github.com/openclaw/openclaw/issues/30785 | https://api.github.com/repos/openclaw/openclaw/issues/30785 | Feature request: `maxToolCallsBeforeReply` config option for main session | ## Problem
In long interactive sessions, the agent can go on extended tool-call streaks (50+ consecutive calls) without sending any user-visible message.
The user is locked out: each message they send to get the agent's attention just triggers another restart of the same runaway loop.
This is especially frustrating ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T16:23:57Z | 2026-04-16T04:25:59Z | 2026-04-13T04:28:45Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | henritheowl | 258,452,454 | U_kgDOD2er5g | User | false |
openclaw/openclaw | 4,016,778,597 | I_kwDOQb6kR87vay1l | 33,067 | https://github.com/openclaw/openclaw/issues/33067 | https://api.github.com/repos/openclaw/openclaw/issues/33067 | bug: downloads fail path validation when explicit output path is outside DEFAULT_DOWNLOAD_DIR | ## Summary
`saveDownloadPayload` in `pw-tools-core.downloads.ts` uses `DEFAULT_DOWNLOAD_DIR` as the root directory for `writeViaSiblingTempPath`, but user-specified explicit download paths are not constrained to that directory. This causes:
1. **Path validation failures** when the explicit output path is outside `DEF... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T13:48:32Z | 2026-04-19T04:24:18Z | 2026-04-16T04:25:58Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | saturn-bugatt | 250,167,540 | U_kgDODulA9A | User | false |
openclaw/openclaw | 4,007,639,789 | I_kwDOQb6kR87u37rt | 30,787 | https://github.com/openclaw/openclaw/issues/30787 | https://api.github.com/repos/openclaw/openclaw/issues/30787 | UNSCHEDULED_REMINDER_NOTE appended to agent text instead of delivered as system annotation | ## Problem
When an agent mentions a reminder or future action (e.g. 'I'll ping you when X is done') without scheduling a cron job, OpenClaw appends a hardcoded note to the end of the agent's reply:
> Note: I did not schedule a reminder in this turn, so this will not trigger automatically.
This note is appended as pl... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T16:24:14Z | 2026-04-16T04:25:58Z | 2026-04-13T04:28:47Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ThalynLabs | 38,703,365 | MDQ6VXNlcjM4NzAzMzY1 | User | false |
openclaw/openclaw | 4,007,648,401 | I_kwDOQb6kR87u39yR | 30,797 | https://github.com/openclaw/openclaw/issues/30797 | https://api.github.com/repos/openclaw/openclaw/issues/30797 | [Bug]: Message instability after extended sessions - memory accumulation | ### Summary
After working for extended periods (30+ minutes of continuous edits), the system becomes unstable:
- Messages duplicate (text and audio sent twice)
- Response lag increases significantly
- System becomes sluggish and aggressive (reboots itself)
- Requires gateway restart to recover
- Last configuration is ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T16:27:34Z | 2026-04-16T04:25:57Z | 2026-04-13T04:28:50Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | PritonEdu | 260,367,740 | U_kgDOD4TlfA | User | false |
openclaw/openclaw | 4,016,732,066 | I_kwDOQb6kR87vanei | 33,059 | https://github.com/openclaw/openclaw/issues/33059 | https://api.github.com/repos/openclaw/openclaw/issues/33059 | [Bug] slug-generator missing Google API key - auth not inherited from main agent | **Description:** The internal `slug-generator` (session:temp:slug-generator) fails with:
```
FailoverError: No API key found for provider "google". Auth store: /home/moltbot/.openclaw/agents/main/agent/auth-profiles.json (agentDir: /home/moltbot/.openclaw/agents/main/agent). Configure auth for this agent (openclaw agen... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T13:39:49Z | 2026-04-19T04:24:19Z | 2026-04-16T04:25:56Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | kai-the-man[bot] | 259,478,402 | BOT_kgDOD3dTgg | Bot | false |
openclaw/openclaw | 4,007,668,659 | I_kwDOQb6kR87u4Cuz | 30,807 | https://github.com/openclaw/openclaw/issues/30807 | https://api.github.com/repos/openclaw/openclaw/issues/30807 | ACP runtime enabled/loaded but sessions_spawn(acp) can still fail as backend unavailable | ## Summary
ACP runtime can appear enabled/loaded (`acpx` plugin loaded) but `sessions_spawn` with `runtime:"acp"` still fails with runtime-backend errors (e.g., "ACP runtime backend is not configured" / spawn failures).
## Environment
- OpenClaw: 2026.2.26
- Host: macOS 26.3 arm64
- Gateway local loopback
- `acpx` plu... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T16:35:36Z | 2026-04-16T04:25:56Z | 2026-04-13T04:28:53Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,007,663,847 | I_kwDOQb6kR87u4Bjn | 30,804 | https://github.com/openclaw/openclaw/issues/30804 | https://api.github.com/repos/openclaw/openclaw/issues/30804 | [Bug] 钉钉AI卡片模式未发送最终响应(isFinalize=true) | ## 问题描述
在钉钉AI卡片模式下，回复只显示思考过程，没有完成卡片最终响应。
## 复现步骤
1. 通过钉钉发送消息给机器人
2. 机器人执行工具（如查询日历、打开浏览器等）
3. 工具结果流式更新到卡片成功
4. 但最终响应（isFinalize=true）没有发送到卡片
## 预期行为
根据钉钉官方文档，AI卡片模式应在最后一次流更新中使用 isFinalize=true 标记完成卡片。
## 实际行为
日志显示：
- ✅ 创建卡片成功
- ✅ 工具结果流式更新到卡片成功
- ❌ 没有发送最终响应（isFinalize=true）
## 日志片段
```
2026-03-01T16:32:16.187Z [... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T16:33:57Z | 2026-04-16T04:25:56Z | 2026-04-13T04:28:52Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | aidescend | 138,131,715 | U_kgDOCDu5Aw | User | false |
openclaw/openclaw | 4,007,669,075 | I_kwDOQb6kR87u4C1T | 30,808 | https://github.com/openclaw/openclaw/issues/30808 | https://api.github.com/repos/openclaw/openclaw/issues/30808 | Feature: runtime-lane tool capability manifest (main vs subagent vs ACP) | ## Feature request
Add a first-class tool capability manifest by runtime lane (main, subagent, ACP).
## Problem
Tool availability differs by lane, but there is no single authoritative capability view. This causes benchmark confusion and false model blame when a tool is missing in one lane.
## Proposal
Add one or more... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T16:35:46Z | 2026-04-16T04:25:55Z | 2026-04-13T04:28:55Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,016,717,208 | I_kwDOQb6kR87vaj2Y | 33,057 | https://github.com/openclaw/openclaw/issues/33057 | https://api.github.com/repos/openclaw/openclaw/issues/33057 | Bug: Heartbeat 觸發頻率異常 - 應為 30 分鐘卻在數秒內連續觸發 | ## 問題描述
Heartbeat 設定為每 30 分鐘觸發一次，但實際上在短時間內連續觸發多次。
## 重現步驟
1. 設定 `heartbeat.every: 30m`
2. 開始新 session
3. 觀察 heartbeat 觸發頻率
## 實際行為
在約 1.5 小時內收到 **30 次** heartbeat，而非預期的 3 次。
Transcript 時間戳顯示：
```
12:08:05 - heartbeat
12:08:24 - heartbeat (19秒後)
12:08:34 - heartbeat (10秒後)
12:08:46 - heartbeat (12秒後)
12:09:38 - he... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T13:36:45Z | 2026-04-19T04:24:20Z | 2026-04-16T04:25:54Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | JardiniTsai | 257,678,972 | U_kgDOD1vefA | User | false |
openclaw/openclaw | 4,007,670,483 | I_kwDOQb6kR87u4DLT | 30,810 | https://github.com/openclaw/openclaw/issues/30810 | https://api.github.com/repos/openclaw/openclaw/issues/30810 | Feature: browser.snapshot graceful fallback to screenshot+metadata | ## Feature request
Add graceful fallback when `browser.snapshot` fails (timeout/CDP issues): return screenshot + minimal extracted metadata instead of hard failure.
## Problem
In real use, `browser.open` and `browser.screenshot` can succeed while `browser.snapshot` intermittently fails. This causes brittle automation ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T16:36:16Z | 2026-04-16T04:25:54Z | 2026-04-13T04:28:58Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,007,670,024 | I_kwDOQb6kR87u4DEI | 30,809 | https://github.com/openclaw/openclaw/issues/30809 | https://api.github.com/repos/openclaw/openclaw/issues/30809 | Feature: doctor lane-parity check for core tool availability | ## Feature request
Add a doctor/self-test mode that verifies runtime-lane parity for core tools.
## Problem
Main lane and subagent/ACP lanes can drift in effective capabilities. This is hard to detect early and causes late failures.
## Proposal
New command:
- `openclaw doctor --lane-parity`
Behavior:
1. Run a small ... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T16:36:05Z | 2026-04-16T04:25:54Z | 2026-04-13T04:28:57Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,016,688,116 | I_kwDOQb6kR87vacv0 | 33,053 | https://github.com/openclaw/openclaw/issues/33053 | https://api.github.com/repos/openclaw/openclaw/issues/33053 | QQ Bot - cron announce only receives system messages, cannot receive normal messages | ### Bug type
Regression (worked before, now fails)
### Summary
QQ Bot can only receive system messages, but cannot receive normal messages when cron tasks run
### Steps to reproduce
1. Configure cron task with isolated session + --announce
2. Run the cron task manually
3. Check QQ for received messages
### Expect... | closed | not_planned | true | 2 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-03T13:30:49Z | 2026-04-19T04:24:20Z | 2026-04-16T04:25:53Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | JasonKong1990 | 18,439,086 | MDQ6VXNlcjE4NDM5MDg2 | User | false |
openclaw/openclaw | 4,007,671,016 | I_kwDOQb6kR87u4DTo | 30,811 | https://github.com/openclaw/openclaw/issues/30811 | https://api.github.com/repos/openclaw/openclaw/issues/30811 | Feature: candidate model probation mode with automatic fallback | ## Feature request
Add a built-in candidate-model probation mode for safe local-model promotion.
## Problem
Users need to evaluate new local models in production-like conditions without risking global degradation. Current setup is manual and error-prone.
## Proposal
Introduce a model probation policy mode, e.g.:
- ro... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T16:36:30Z | 2026-04-16T04:25:53Z | 2026-04-13T04:29:00Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,007,690,057 | I_kwDOQb6kR87u4H9J | 30,816 | https://github.com/openclaw/openclaw/issues/30816 | https://api.github.com/repos/openclaw/openclaw/issues/30816 | [Bug]: Discord Timeout and duplicate delivery | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Delivery of cron jobs into Discord cause the same briefings to be sent more than once. I added a message key and also advanced options for the cron configuration but still occurs.
### Steps to reproduce
1. Configure a cron to delivery... | closed | not_planned | true | 3 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-01T16:44:09Z | 2026-04-16T04:25:52Z | 2026-04-13T04:29:02Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | waynetaylor | 2,912,112 | MDQ6VXNlcjI5MTIxMTI= | User | false |
openclaw/openclaw | 4,016,670,454 | I_kwDOQb6kR87vaYb2 | 33,049 | https://github.com/openclaw/openclaw/issues/33049 | https://api.github.com/repos/openclaw/openclaw/issues/33049 | ACP session deliveryContext missing 'to' field — announce not routing back to requester channel (Telegram forum/topic) | ## Summary
When spawning an ACP session via `sessions_spawn` with `runtime="acp"` from a Telegram forum topic session, the ACP session's `deliveryContext` only contains `{ "channel": "telegram" }` — the `to` field (e.g. `telegram:-1003827078672`) is never inherited from the requester session. As a result, task complet... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-03T13:26:52Z | 2026-04-19T04:24:21Z | 2026-04-16T04:25:51Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | ClawKPA | 263,955,455 | U_kgDOD7uj_w | User | false |
openclaw/openclaw | 4,007,770,444 | I_kwDOQb6kR87u4blM | 30,833 | https://github.com/openclaw/openclaw/issues/30833 | https://api.github.com/repos/openclaw/openclaw/issues/30833 | Plugin CLI commands not registered when invoked via node dist/index.js | ## Bug
Plugins that register CLI commands via `api.registerCli()` fail with `error: unknown command '<name>'` when the CLI is invoked through `node dist/index.js` (the library entry point).
### Root cause
`src/index.ts` has an `if (isMainModule(...))` block that calls `buildProgram()` + `program.parseAsync()` direct... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-01T17:18:19Z | 2026-04-16T04:25:51Z | 2026-04-13T04:29:06Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | ashbrener | 116,340 | MDQ6VXNlcjExNjM0MA== | User | false |
openclaw/openclaw | 4,007,765,392 | I_kwDOQb6kR87u4aWQ | 30,832 | https://github.com/openclaw/openclaw/issues/30832 | https://api.github.com/repos/openclaw/openclaw/issues/30832 | Loop continuation can stall until webchat session is foregrounded/opened | ## Summary
Long-running/looped work appears to stall until the webchat thread is foregrounded (opened), then pending continuation executes immediately.
## Observed behavior
From user timeline/screenshot:
- A continuation-worthy event arrives around **11:36**.
- Next loop action/response does **not** execute until user... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-03-01T17:16:20Z | 2026-04-16T04:25:51Z | 2026-04-13T04:29:04Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Cuttingwater | 226,908,709 | U_kgDODYZaJQ | User | false |
openclaw/openclaw | 4,007,803,784 | I_kwDOQb6kR87u4juI | 30,844 | https://github.com/openclaw/openclaw/issues/30844 | https://api.github.com/repos/openclaw/openclaw/issues/30844 | [Bug]: gpt-5.3-codex is hard-routed to openai-codex OAuth provider in 2026.2.26, blocking API-key usage on openai provider | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`openai/gpt-5.3-codex` is unconditionally normalized to `openai-codex/gpt-5.3-codex` at runtime, forcing OAuth-path behavior and failing preflight with `Failed to extract accountId from token`, even when valid `OPENAI_API_KEY` is configured... | closed | not_planned | true | 4 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-01T17:33:02Z | 2026-04-16T04:25:50Z | 2026-04-13T04:29:08Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | EmpireCreator | 12,735,079 | MDQ6VXNlcjEyNzM1MDc5 | User | false |
openclaw/openclaw | 4,016,668,091 | I_kwDOQb6kR87vaX27 | 33,048 | https://github.com/openclaw/openclaw/issues/33048 | https://api.github.com/repos/openclaw/openclaw/issues/33048 | mem0 sqlite bindings still broken in 2026.3.2 (regression from 2026.2.26) | # Bug Report: mem0/SQLite bindings still broken in 2026.3.2
## Bug Type
- [x] Regression (worked in previous version)
- [ ] Crash
- [ ] Behavior
## Summary
After updating to **OpenClaw 2026.3.2**, the mem0-based memory tools (`memory_search`, `memory_list`, `memory_get`, etc.) are still failing with a native sqlite3... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T13:26:19Z | 2026-04-19T04:24:22Z | 2026-04-16T04:25:49Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | Valkster70 | 129,894,487 | U_kgDOB74IVw | User | false |
openclaw/openclaw | 4,007,847,660 | I_kwDOQb6kR87u4ubs | 30,852 | https://github.com/openclaw/openclaw/issues/30852 | https://api.github.com/repos/openclaw/openclaw/issues/30852 | [RFC] Native Tezit Protocol Integration — Encrypted File Sharing for Agents | ## Summary
This RFC proposes adding native Tezit Protocol support to OpenClaw, giving every agent encrypted file sharing and knowledge bundling out of the box — no external services, no plugin installation, no configuration. Agents would be able to create, encrypt, share, search, and delete knowledge bundles ("tezits"... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-01T17:51:22Z | 2026-04-16T04:25:49Z | 2026-04-13T04:29:09Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | vibewrk | 175,494,844 | U_kgDOCnXWvA | User | false |
openclaw/openclaw | 4,008,131,020 | I_kwDOQb6kR87u5znM | 30,934 | https://github.com/openclaw/openclaw/issues/30934 | https://api.github.com/repos/openclaw/openclaw/issues/30934 | Version label can be pinned by stale OPENCLAW_BUNDLED_VERSION env override | # BUG-058 Draft: Version label can be pinned by stale OPENCLAW_BUNDLED_VERSION
## Summary
OpenClaw can report a stale version label when `OPENCLAW_BUNDLED_VERSION` is present in environment, even when module/package/build metadata reflects a newer runtime. In practice this causes confusing post-update state where git/... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-01T19:50:44Z | 2026-04-16T04:25:48Z | 2026-04-13T04:29:11Z | CONTRIBUTOR | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Jackten | 2,895,479 | MDQ6VXNlcjI4OTU0Nzk= | User | false |
openclaw/openclaw | 4,016,628,726 | I_kwDOQb6kR87vaOP2 | 33,043 | https://github.com/openclaw/openclaw/issues/33043 | https://api.github.com/repos/openclaw/openclaw/issues/33043 | [Bug]: 1. [Bug] Feishu TTS audio truncated when reply from Feishu to Feishu | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
1. ## Bug Description
2.
3. When using the Feishu channel, TTS audio is truncated when:
4. - User sends message via Feishu
5. - AI replies with TTS audio to Feishu
6. - Only a few words are played (e.g., 2-3 words)
7.
8. However, when u... | closed | not_planned | true | 2 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-03T13:17:39Z | 2026-04-19T04:24:23Z | 2026-04-16T04:25:47Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | cattear1120 | 244,715,720 | U_kgDODpYQyA | User | false |
openclaw/openclaw | 4,010,719,260 | I_kwDOQb6kR87vDrgc | 31,623 | https://github.com/openclaw/openclaw/issues/31623 | https://api.github.com/repos/openclaw/openclaw/issues/31623 | Add kickstart cron docs + explicit startup hint | **Title:** Docs/CLI should clarify launchctl kickstart semantics and limitations
**Labels:** documentation, enhancement, macos
---
## Description
The current documentation and CLI behavior around `launchctl kickstart -k` for gateway restarts could be clearer about when it works vs. when `bootout + bootstrap` is req... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-02T11:21:16Z | 2026-04-16T04:25:47Z | 2026-04-13T04:29:19Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | aboveUAV | 261,767,811 | U_kgDOD5pCgw | User | false |
openclaw/openclaw | 4,016,622,473 | I_kwDOQb6kR87vaMuJ | 33,042 | https://github.com/openclaw/openclaw/issues/33042 | https://api.github.com/repos/openclaw/openclaw/issues/33042 | [Slack]: Drop unnecessary per-message metadata blocks in Slack threads | Every Slack thread message injects the same metadata blocks into context:
```
Conversation info (untrusted metadata):
{
"message_id": "1772543223.976569",
"reply_to_id": "1772490038.636569",
"sender_id": "U07.....L15",
"conversation_label": "#all-slack",
"sender": "Ahmed Mansour",
"timestamp": "Tue 2026-03... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T13:16:18Z | 2026-04-19T04:24:24Z | 2026-04-16T04:25:45Z | CONTRIBUTOR | null | 20260419T052035Z | 2026-04-19T05:20:35Z | sourman | 3,827,766 | MDQ6VXNlcjM4Mjc3NjY= | User | false |
openclaw/openclaw | 4,010,720,136 | I_kwDOQb6kR87vDruI | 31,624 | https://github.com/openclaw/openclaw/issues/31624 | https://api.github.com/repos/openclaw/openclaw/issues/31624 | Add dry-run mode | **Title:** openclaw update preflight improvements (disk space check, stall detection, fallback path)
**Labels:** enhancement, cli, ux
---
## Description
`openclaw update` could benefit from preflight checks and better handling of common failure scenarios, particularly:
1. Disk space validation before starting
2. St... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-02T11:21:25Z | 2026-04-16T04:25:45Z | 2026-04-13T04:29:21Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | aboveUAV | 261,767,811 | U_kgDOD5pCgw | User | false |
openclaw/openclaw | 4,016,606,985 | I_kwDOQb6kR87vaI8J | 33,039 | https://github.com/openclaw/openclaw/issues/33039 | https://api.github.com/repos/openclaw/openclaw/issues/33039 | Slack button interactions don't trigger session/heartbeat | ## Problem
When a user clicks a Slack button (with `openclaw:` prefix in `action_id`), the interaction is captured and logged:
```
[slack] interaction action=openclaw:pdd_action_TEST8 type=button user=U04N0A45W6S channel=C0AFP9EPSKC
```
The event is enqueued via `enqueueSystemEvent(formatSlackInteractionSystemEvent(... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-03T13:12:39Z | 2026-04-19T04:24:25Z | 2026-04-16T04:25:43Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | visshehi | 162,566,191 | U_kgDOCbCQLw | User | false |
openclaw/openclaw | 4,010,721,393 | I_kwDOQb6kR87vDsBx | 31,625 | https://github.com/openclaw/openclaw/issues/31625 | https://api.github.com/repos/openclaw/openclaw/issues/31625 | Session stuck in permanent 400 error loop after truncated tool_use — no auto-recovery | ## Bug Description
When an assistant response is terminated mid-generation during a tool_use block (stopReason: terminated), the transcript repair mechanism inserts a synthetic tool_result — but the corresponding tool_use was never fully saved. This leaves the session transcript in a permanently broken state.
Every s... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-02T11:21:43Z | 2026-04-16T04:25:44Z | 2026-04-13T04:29:23Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | TheDuke-Code | 205,517,862 | U_kgDODD_0Jg | User | false |
openclaw/openclaw | 4,010,802,261 | I_kwDOQb6kR87vD_xV | 31,634 | https://github.com/openclaw/openclaw/issues/31634 | https://api.github.com/repos/openclaw/openclaw/issues/31634 | [Feature]: Telegram — media-aware debounce for caption messages | ### Problem
When a user sends a media attachment with a separate text comment in Telegram DM, it arrives as two messages:
1. The media (no text)
2. The text comment
OpenClaw processes them sequentially, triggering two agent turns. The first turn sees only the attachment with no context and wastes tokens responding "I... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T11:38:39Z | 2026-04-16T04:25:43Z | 2026-04-13T04:29:26Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | Mquarmoc | 179,647,526 | U_kgDOCrU0Jg | User | false |
openclaw/openclaw | 4,010,792,884 | I_kwDOQb6kR87vD9e0 | 31,633 | https://github.com/openclaw/openclaw/issues/31633 | https://api.github.com/repos/openclaw/openclaw/issues/31633 | Discord channel handler uses default media roots instead of agent-scoped roots | ## Bug Description
When an agent sends a message with a local image attachment via Discord, the media path validation uses `getDefaultMediaLocalRoots()` instead of `getAgentScopedMediaLocalRoots()`. This causes `LocalMediaAccessError` for images stored in agent workspaces (`workspace-*` directories), even though the a... | closed | not_planned | true | 3 | [
"stale"
] | [] | 2026-03-02T11:36:11Z | 2026-04-16T04:25:43Z | 2026-04-13T04:29:24Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | jorsby | 264,672,960 | U_kgDOD8aWwA | User | false |
openclaw/openclaw | 4,016,580,173 | I_kwDOQb6kR87vaCZN | 33,025 | https://github.com/openclaw/openclaw/issues/33025 | https://api.github.com/repos/openclaw/openclaw/issues/33025 | Telegram DM streaming (sendMessageDraft / editMessageText preview) never triggers | ## Bug: Telegram DM streaming never triggers
### Environment
- **OpenClaw:** 2026.3.2
- **Node:** v24.12.0
- **Platform:** Ubuntu 24.04 (arm64)
- **Telegram plugin:** enabled, grammY (bundled)
- **Channel:** Telegram Bot API, private DM chat
### Config (relevant)
```json5
channels: {
telegram: {
streaming: "par... | closed | not_planned | true | 6 | [
"stale"
] | [] | 2026-03-03T13:06:23Z | 2026-04-19T04:24:25Z | 2026-04-16T04:25:42Z | CONTRIBUTOR | null | 20260419T052035Z | 2026-04-19T05:20:35Z | xdanger | 7,087 | MDQ6VXNlcjcwODc= | User | false |
openclaw/openclaw | 4,010,898,044 | I_kwDOQb6kR87vEXJ8 | 31,647 | https://github.com/openclaw/openclaw/issues/31647 | https://api.github.com/repos/openclaw/openclaw/issues/31647 | Feishu: message tool rejects send with 'Use target instead of to/channelId' | @/Users/tszhang/.openclaw-ops/workspace/tmp/openclaw-feishu-message-tool-bug.md | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T12:00:38Z | 2026-04-16T04:25:42Z | 2026-04-13T04:29:28Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | iron6909 | 137,274,451 | U_kgDOCC6kUw | User | false |
openclaw/openclaw | 4,010,914,065 | I_kwDOQb6kR87vEbER | 31,652 | https://github.com/openclaw/openclaw/issues/31652 | https://api.github.com/repos/openclaw/openclaw/issues/31652 | [Feature]: Feishu/Lark: Support im.message.reaction.created_v1 event for incoming reaction notifications | ## Summary
OpenClaw's Feishu extension does not handle the `im.message.reaction.created_v1` event, so the agent cannot receive notifications when users react to messages with emojis.
## Problem
When a user adds a reaction (emoji) to a bot message in Feishu:
1. Feishu sends the `im.message.reaction.created_v1` event ... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T12:03:55Z | 2026-04-16T04:25:41Z | 2026-04-13T04:29:29Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | liutiexiu | 6,936,635 | MDQ6VXNlcjY5MzY2MzU= | User | false |
openclaw/openclaw | 4,016,570,426 | I_kwDOQb6kR87vaAA6 | 33,021 | https://github.com/openclaw/openclaw/issues/33021 | https://api.github.com/repos/openclaw/openclaw/issues/33021 | Telegram 群组消息重复接收问题 (2026.3.2) | # OpenClaw 重复消息问题诊断报告
## 版本信息
- OpenClaw 版本: 2026.3.2 (最新)
- 系统: macOS Darwin 25.3.0 (arm64)
- Node: v23.11.0
## 问题描述
在 Telegram 群组中，当使用 Bot 发送消息时，同一消息被重复接收多次。问题与已关闭的 Issue #10377 相似，但当前版本仍有问题。
## 复现场景
1. Telegram 群组配置了 OpenClaw Bot
2. 发送消息后，收到重复的消息
3. 可能与消息确认机制有关
## 诊断信息 (openclaw doctor --deep)
### Gateway 状态
- ... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T13:04:10Z | 2026-04-19T04:24:26Z | 2026-04-16T04:25:40Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | survivor998 | 226,792,843 | U_kgDODYSViw | User | false |
openclaw/openclaw | 4,010,925,887 | I_kwDOQb6kR87vEd8_ | 31,655 | https://github.com/openclaw/openclaw/issues/31655 | https://api.github.com/repos/openclaw/openclaw/issues/31655 | Dependabot: vulnerable transitive dependencies (tar, fast-xml-parser) | ## Summary
Several npm dependencies have known vulnerabilities that show up in Dependabot alerts for projects using OpenClaw as a peerDependency.
## Vulnerabilities
### HIGH (4 CVEs) — `tar`
- **GHSA-5955-9wpr-37jh**: Arbitrary File Read/Write via Hardlink Target Escape
- **GHSA-r628-mhmh-qjhw**: Arbitrary File Crea... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T12:06:20Z | 2026-04-16T04:25:40Z | 2026-04-13T04:29:33Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | floriansiebert | 22,081,639 | MDQ6VXNlcjIyMDgxNjM5 | User | false |
openclaw/openclaw | 4,010,957,812 | I_kwDOQb6kR87vElv0 | 31,663 | https://github.com/openclaw/openclaw/issues/31663 | https://api.github.com/repos/openclaw/openclaw/issues/31663 | bug: Chrome extension relay — tabOperationLocks not cleared on relay disconnect (zombie lock) | ## Summary
`tabOperationLocks` is never cleared in `onRelayClosed()` in `assets/chrome-extension/background.js`. If a tab attach/detach operation is in progress when the relay WebSocket disconnects, the lock remains set permanently.
## Behaviour
After relay reconnect, clicking the badge to re-attach a tab does nothi... | closed | not_planned | true | 8 | [
"stale"
] | [] | 2026-03-02T12:13:41Z | 2026-04-16T04:25:39Z | 2026-04-13T04:29:37Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | yodason | 61,774,516 | MDQ6VXNlcjYxNzc0NTE2 | User | false |
openclaw/openclaw | 4,010,948,932 | I_kwDOQb6kR87vEjlE | 31,661 | https://github.com/openclaw/openclaw/issues/31661 | https://api.github.com/repos/openclaw/openclaw/issues/31661 | feat: Auto-label group/channel sessions for dashboard readability | ## Problem
When viewing sessions in the OpenClaw dashboard, group/channel sessions (Discord, Slack, etc.) show an empty `(optional)` label field. The `displayName` does contain the channel name (e.g. `discord:1472892609285328988#config`), but it appears in small text below — not in the prominent label column.
This ma... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T12:11:23Z | 2026-04-16T04:25:39Z | 2026-04-13T04:29:35Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | proto-genesys-x | 260,201,181 | U_kgDOD4Ja3Q | User | false |
openclaw/openclaw | 4,010,958,855 | I_kwDOQb6kR87vEmAH | 31,664 | https://github.com/openclaw/openclaw/issues/31664 | https://api.github.com/repos/openclaw/openclaw/issues/31664 | bug: agent runner stalls silently on rate limit / model switch mid-stream (attempt.ts race condition) | ## Summary
When an API rate limit or model fallback (e.g. Claude → Gemini) occurs mid-stream, the agent state machine in `pi-embedded-runner/run/attempt.ts` can stall permanently with no error surfaced to the user. The session goes silent and never resumes, even after the rate limit clears.
## Behaviour
- Agent is m... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T12:13:57Z | 2026-04-16T04:25:38Z | 2026-04-13T04:29:39Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | yodason | 61,774,516 | MDQ6VXNlcjYxNzc0NTE2 | User | false |
openclaw/openclaw | 4,016,023,827 | I_kwDOQb6kR87vX6kT | 32,921 | https://github.com/openclaw/openclaw/issues/32921 | https://api.github.com/repos/openclaw/openclaw/issues/32921 | Feature Request: Built-in Backup and Restore System | ## Summary
Add a native backup/restore system to protect user configurations, memories, skills, and other critical data with one command, reducing the risk of data loss and simplifying disaster recovery.
## Problem
**Current situation:** Users must implement their own backup solutions:
- Write custom shell scripts
-... | closed | not_planned | true | 5 | [
"stale"
] | [] | 2026-03-03T10:57:10Z | 2026-04-19T04:24:27Z | 2026-04-16T04:25:37Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | SonicBotMan | 258,392,144 | U_kgDOD2bAUA | User | false |
openclaw/openclaw | 4,010,989,456 | I_kwDOQb6kR87vEteQ | 31,671 | https://github.com/openclaw/openclaw/issues/31671 | https://api.github.com/repos/openclaw/openclaw/issues/31671 | sessions_send corrupts target session channel from 'telegram' to 'webchat' | ## Bug Description
`sessions_send` from one agent to another changes the target session's `channel` field in `sessions.json` from `"telegram"` to `"webchat"`. This breaks Telegram message routing — the target agent stops receiving Telegram messages from the user.
## Steps to Reproduce
1. Agent A (main) has a running... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T12:21:46Z | 2026-04-16T04:25:37Z | 2026-04-13T04:29:41Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | RicAgent | 257,661,236 | U_kgDOD1uZNA | User | false |
openclaw/openclaw | 4,011,112,441 | I_kwDOQb6kR87vFLf5 | 31,689 | https://github.com/openclaw/openclaw/issues/31689 | https://api.github.com/repos/openclaw/openclaw/issues/31689 | [Bug]: vllm + qwen3-coder-next 无法使用,报connection error | ### Bug type
Regression (worked before, now fails)
### Summary
我用vllm本地部署了 qwen3-coder-next ,并且用/usr/bin/curl http://127.0.0.1:8000/v1/chat/completions -H "Content-Type: application/json" -d '{
"model": "qwen3-coder",
"messages": [
{"role": "system", "content": "你是一个有用的助手。"},
{"role": "user",... | closed | not_planned | true | 7 | [
"bug",
"stale",
"regression"
] | [] | 2026-03-02T12:51:51Z | 2026-04-16T04:25:36Z | 2026-04-13T04:29:48Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | abxis | 57,876,454 | MDQ6VXNlcjU3ODc2NDU0 | User | false |
openclaw/openclaw | 4,011,013,413 | I_kwDOQb6kR87vEzUl | 31,678 | https://github.com/openclaw/openclaw/issues/31678 | https://api.github.com/repos/openclaw/openclaw/issues/31678 | Feature Request: Add compaction.keepRecent config to preserve recent messages | # Feature Request: Add compaction.keepRecent config to preserve recent messages
## Problem
When context compaction is triggered, important recent conversation context may be lost, causing the agent to "forget" what was just discussed.
## Proposed Solution
Add a `compaction.keepRecent` configuration option t... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-02T12:27:51Z | 2026-04-16T04:25:36Z | 2026-04-13T04:29:43Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | RAFOLIE | 129,680,436 | U_kgDOB7rENA | User | false |
openclaw/openclaw | 4,015,985,971 | I_kwDOQb6kR87vXxUz | 32,917 | https://github.com/openclaw/openclaw/issues/32917 | https://api.github.com/repos/openclaw/openclaw/issues/32917 | Feature Request: Built-in Watchdog for Gateway Auto-Recovery | ## Summary
Add an optional built-in watchdog mechanism to automatically detect and recover from Gateway crashes, reducing manual intervention and improving system reliability.
## Problem
**Current situation:** Users must implement their own monitoring scripts to detect and restart crashed Gateway instances. This req... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T10:48:58Z | 2026-04-19T04:24:29Z | 2026-04-16T04:25:35Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | SonicBotMan | 258,392,144 | U_kgDOD2bAUA | User | false |
openclaw/openclaw | 4,011,166,813 | I_kwDOQb6kR87vFYxd | 31,694 | https://github.com/openclaw/openclaw/issues/31694 | https://api.github.com/repos/openclaw/openclaw/issues/31694 | Feature Request: Skill Orders - Making Agents as Powerful as Command Line | # Feature Request: Skill Orders - Making Agents as Powerful as Command Line
## Problem Background
Current issues with OpenClaw agent:
- Execute one action at a time, lacking continuous execution capability
- Requires users to repeatedly say "continue" to complete tasks
- Cannot autonomously think, decompose tasks, an... | closed | not_planned | true | 4 | [
"stale"
] | [] | 2026-03-02T13:03:02Z | 2026-04-16T04:25:35Z | 2026-04-13T04:29:50Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | RAFOLIE | 129,680,436 | U_kgDOB7rENA | User | false |
openclaw/openclaw | 4,015,964,619 | I_kwDOQb6kR87vXsHL | 32,908 | https://github.com/openclaw/openclaw/issues/32908 | https://api.github.com/repos/openclaw/openclaw/issues/32908 | Feature Request: Smart Configuration Merge Protection | ## Summary
Prevent accidental loss of sensitive configuration fields (tokens, credentials) when updating configuration by implementing smart merge instead of overwrite.
## Problem
**Real-world incident:** When updating a remote instance's configuration, a simplified config snippet was applied that **overwrote the en... | closed | not_planned | true | 2 | [
"stale"
] | [] | 2026-03-03T10:43:36Z | 2026-04-19T04:24:30Z | 2026-04-16T04:25:34Z | NONE | null | 20260419T052035Z | 2026-04-19T05:20:35Z | SonicBotMan | 258,392,144 | U_kgDOD2bAUA | User | false |
openclaw/openclaw | 4,011,208,016 | I_kwDOQb6kR87vFi1Q | 31,704 | https://github.com/openclaw/openclaw/issues/31704 | https://api.github.com/repos/openclaw/openclaw/issues/31704 | [Bug]: Bot misunderstoods who is who in Whatsapp | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Disclaimer: This is a problem that always ocurred, independent of the version. I think it is because OpenClaw was designed to talk only with his human, and not with many ppl as i use. BUT, most of time it works well talking with strangers, ... | closed | not_planned | true | 4 | [
"bug",
"stale",
"bug:behavior"
] | [] | 2026-03-02T13:12:13Z | 2026-04-16T04:25:34Z | 2026-04-13T04:29:52Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | octaivermatt | 262,254,098 | U_kgDOD6GuEg | User | false |
openclaw/openclaw | 4,011,259,484 | I_kwDOQb6kR87vFvZc | 31,710 | https://github.com/openclaw/openclaw/issues/31710 | https://api.github.com/repos/openclaw/openclaw/issues/31710 | Discord provider repeatedly restarted by health-monitor (reason: stuck) causing duplicate messages | ## Bug Report
**Version:** 2026.3.1
## Symptoms
- Discord provider (all agents) repeatedly restarted every ~5 minutes by health-monitor with `reason: stuck`
- Each restart creates a new event listener without cleaning up the previous one
- Result: incoming messages appear duplicated in both sub-agents' view (same mes... | closed | not_planned | true | 7 | [
"stale"
] | [] | 2026-03-02T13:23:35Z | 2026-04-16T04:25:33Z | 2026-04-13T04:29:54Z | NONE | null | 20260417T163222Z | 2026-04-17T16:32:22Z | agent-eli | 263,353,023 | U_kgDOD7Jyvw | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.