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 | 4,058,192,166 | I_kwDOQb6kR87x4xkm | 43,196 | https://github.com/openclaw/openclaw/issues/43196 | https://api.github.com/repos/openclaw/openclaw/issues/43196 | [Bug]: Feishu old-message replay/late-processing after gateway restart causes stale replies | ## Summary
After gateway restart/recovery, Feishu DM may replay or late-deliver an **old inbound message**, causing the agent to respond to stale content (user sees "why did you only reply now?").
## Problem
In Feishu direct chat, a previously handled message can be processed again after restart/reconnect path (or a... | closed | completed | true | 1 | [] | [] | 2026-03-11T13:18:44Z | 2026-04-27T04:51:18Z | 2026-04-24T22:18:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | DragonYang-code | 56,200,279 | MDQ6VXNlcjU2MjAwMjc5 | User | false |
openclaw/openclaw | 4,058,198,937 | I_kwDOQb6kR87x4zOZ | 43,197 | https://github.com/openclaw/openclaw/issues/43197 | https://api.github.com/repos/openclaw/openclaw/issues/43197 | Pilot Protocol channel plugin | ## Proposal: Pilot Protocol Channel Plugin
### What is Pilot Protocol?
Pilot Protocol is **not a messaging app**. It is a L3/L4 network stack for autonomous agents — a virtual overlay network layered on top of IP/TCP/UDP that gives agents first-class network citizenship: 48-bit addresses, 16-bit ports, reliable strea... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T13:19:58Z | 2026-04-27T04:51:18Z | 2026-04-24T22:18:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | TeoSlayer | 45,018,530 | MDQ6VXNlcjQ1MDE4NTMw | User | false |
openclaw/openclaw | 4,058,206,296 | I_kwDOQb6kR87x41BY | 43,200 | https://github.com/openclaw/openclaw/issues/43200 | https://api.github.com/repos/openclaw/openclaw/issues/43200 | Bug / Feature Request: message tool cross-session messages not recorded in target session history (per-channel-peer) | ## Problem
When using `session.dmScope: "per-channel-peer"`, each user gets an isolated session. When the agent uses the `message` tool from User A's session to send a message to User B:
1. ✅ The message is successfully delivered to User B's chat (e.g., Telegram DM)
2. ✅ User B's reply correctly enters User B's own s... | closed | completed | true | 2 | [] | [] | 2026-03-11T13:21:21Z | 2026-04-27T04:51:17Z | 2026-04-24T22:18:39Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Senorsen | 4,176,091 | MDQ6VXNlcjQxNzYwOTE= | User | false |
openclaw/openclaw | 4,058,365,158 | I_kwDOQb6kR87x5bzm | 43,233 | https://github.com/openclaw/openclaw/issues/43233 | https://api.github.com/repos/openclaw/openclaw/issues/43233 | [Bug]: [telegram] Polling stall detected (no getUpdates for 107.67s); forcing restart. | ### Bug type
Regression (worked before, now fails)
### Summary
After openclaw update to 2026.3.8 `3caab92` telegram channel stopped working
the crash happens immediately after sending a direct message to bot.
### Steps to reproduce
1. Configure telegram channel
2. [optional] openclaw daemon restart
3. Send messag... | closed | completed | true | 11 | [
"bug",
"regression"
] | [] | 2026-03-11T13:50:58Z | 2026-04-27T04:51:16Z | 2026-04-24T22:18:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lrepno | 5,663,956 | MDQ6VXNlcjU2NjM5NTY= | User | false |
openclaw/openclaw | 4,058,472,600 | I_kwDOQb6kR87x52CY | 43,247 | https://github.com/openclaw/openclaw/issues/43247 | https://api.github.com/repos/openclaw/openclaw/issues/43247 | [Bug]: opencode-go provider missing from PROVIDER_ENV_API_KEY_CANDIDATES | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The opencode-go provider exists in the pi-ai catalog with models like kimi-k2.5, glm-5, and minimax-m2.5, but OpenClaw's resolveEnvApiKey function cannot resolve the API key for this provider because it's missing from the PROVIDER_ENV_API_K... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T14:10:47Z | 2026-04-27T04:51:15Z | 2026-04-24T22:18:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | K-NRS | 20,605,111 | MDQ6VXNlcjIwNjA1MTEx | User | false |
openclaw/openclaw | 4,058,573,950 | I_kwDOQb6kR87x6Ox- | 43,255 | https://github.com/openclaw/openclaw/issues/43255 | https://api.github.com/repos/openclaw/openclaw/issues/43255 | cron run --force enqueues but never executes (no runs entry, no delivery) | # Bug report: `cron run --force` enqueues but never executes
## Summary
Manual cron trigger appears to be accepted (`ok: true`, `enqueued: true`), but run execution never starts:
- no entries in `cron runs` for the job,
- no message delivery,
- job state does not show completion/error for the forced run.
## Environme... | closed | completed | true | 1 | [] | [] | 2026-03-11T14:27:44Z | 2026-04-27T04:51:15Z | 2026-04-24T22:18:46Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tomeks691 | 85,071,303 | MDQ6VXNlcjg1MDcxMzAz | User | false |
openclaw/openclaw | 4,058,669,638 | I_kwDOQb6kR87x6mJG | 43,263 | https://github.com/openclaw/openclaw/issues/43263 | https://api.github.com/repos/openclaw/openclaw/issues/43263 | Tool calls should validate required parameters before execution | ## Problem
OpenClaw tools do not validate required parameters before execution, leading to silent failures that are only discovered later in logs.
### Examples
**write tool missing path parameter:**
**edit tool with non-matching text:**
## Impact
- Important data updates are lost silently
- Failures are only vi... | closed | completed | true | 1 | [] | [] | 2026-03-11T14:44:16Z | 2026-04-27T04:51:14Z | 2026-04-24T22:18:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | gumi-ink | 267,238,280 | U_kgDOD-27iA | User | false |
openclaw/openclaw | 4,058,697,311 | I_kwDOQb6kR87x6s5f | 43,267 | https://github.com/openclaw/openclaw/issues/43267 | https://api.github.com/repos/openclaw/openclaw/issues/43267 | Bug: memory-openviking requests abort when OpenViking process is alive but HTTP endpoints hang | ## Summary
When using `memory-openviking` in local mode, `memory_store` / `auto-capture` can fail with:
- `AbortError: This operation was aborted`
- or `extract returned 0 memories`
In our case, the OpenViking process is still alive and TCP port `1933` is connectable, but HTTP endpoints such as `/health` and `/api/v... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T14:49:13Z | 2026-04-27T04:51:13Z | 2026-04-24T22:18:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | JerryLeibniz | 58,539,255 | MDQ6VXNlcjU4NTM5MjU1 | User | false |
openclaw/openclaw | 4,058,785,212 | I_kwDOQb6kR87x7CW8 | 43,278 | https://github.com/openclaw/openclaw/issues/43278 | https://api.github.com/repos/openclaw/openclaw/issues/43278 | [Bug]: 执行任务卡死，不再反应 | ### Bug type
Regression (worked before, now fails)
### Summary
通过telegram发出指令。再执行一段时间后，就卡死了。不再接收指令，也不再返回任何信息。直到重启gateway。
### Steps to reproduce
1. 通过telegram发出一个指令
2. openclaw执行一段时间，可能遇到问题，或者其他问题，如大模型中断（但我有多个模型备份）。openclaw停止执行，不再返回任何信息，也不再接受指令。
### Expected behavior
持续输出，并返回最终结果。并等待接受新的指令。
### Actual behavior... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T15:03:39Z | 2026-04-27T04:51:12Z | 2026-04-24T22:18:57Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tianlichunhong | 13,270,039 | MDQ6VXNlcjEzMjcwMDM5 | User | false |
openclaw/openclaw | 4,058,850,432 | I_kwDOQb6kR87x7SSA | 43,282 | https://github.com/openclaw/openclaw/issues/43282 | https://api.github.com/repos/openclaw/openclaw/issues/43282 | [Bug]: System event injection into sessions with thinking enabled corrupts thinking blocks, causing LLM rejection | ## Description
When a cron job with `sessionTarget: "main"` injects a system event into a session where the model has thinking enabled (`thinking: medium` or higher), the injection process modifies or fails to preserve `thinking`/`redacted_thinking` content blocks in the conversation history. Anthropic's API then reje... | closed | completed | true | 2 | [] | [] | 2026-03-11T15:14:21Z | 2026-04-27T04:51:11Z | 2026-04-24T22:18:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ThalynLabs | 38,703,365 | MDQ6VXNlcjM4NzAzMzY1 | User | false |
openclaw/openclaw | 4,058,855,772 | I_kwDOQb6kR87x7Tlc | 43,283 | https://github.com/openclaw/openclaw/issues/43283 | https://api.github.com/repos/openclaw/openclaw/issues/43283 | [Bug]: Tools unavailable despite correct config | ### Bug type
Regression (worked before, now fails)
### Summary
## Description
All OpenClaw Agent Tools (`exec`, `read`, `write`, `subagents`, etc.) return "Tool not found", even though the configuration is correctly set to `"profile": "full"`. This causes complete system blockage.
## Environment
- **OpenClaw Versio... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T15:15:21Z | 2026-04-27T04:51:10Z | 2026-04-24T22:19:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | RICK1328 | 168,750,068 | U_kgDOCg7r9A | User | false |
openclaw/openclaw | 4,058,857,673 | I_kwDOQb6kR87x7UDJ | 43,284 | https://github.com/openclaw/openclaw/issues/43284 | https://api.github.com/repos/openclaw/openclaw/issues/43284 | [Bug]: Chat tool card sidebar does not show full command for long exec commands | ### Bug type
Regression (worked before, now fails)
### Summary
Long exec commands are truncated in the chat tool card sidebar, making it difficult to see the full command that was executed.
### Steps to reproduce
Configure an agent with a long exec command (e.g., a curl command with multiple flags)
Execute the com... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T15:15:41Z | 2026-04-27T04:51:09Z | 2026-04-24T22:19:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | soumojit-D48 | 207,209,292 | U_kgDODFnDTA | User | false |
openclaw/openclaw | 4,058,873,964 | I_kwDOQb6kR87x7YBs | 43,287 | https://github.com/openclaw/openclaw/issues/43287 | https://api.github.com/repos/openclaw/openclaw/issues/43287 | Browser tool via node host crBrowser tool via node host bypasses gateway's relay-aware HTTP API, fails with Target.attachToBrowserTarget on extension relay | ### Bug type
Regression (worked before, now fails)
### Summary
When the agent browser tool routes a request via target="node" to a node host on the same machine as the Chrome extension relay, it fails with Target.attachToBrowserTarget: Not allowed. The gateway's own browser control HTTP API (port 18791) handles the ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T15:18:49Z | 2026-04-27T04:51:09Z | 2026-04-24T22:19:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | naomihoogeweij-cpu | 263,912,999 | U_kgDOD7r-Jw | User | false |
openclaw/openclaw | 4,058,899,321 | I_kwDOQb6kR87x7eN5 | 43,289 | https://github.com/openclaw/openclaw/issues/43289 | https://api.github.com/repos/openclaw/openclaw/issues/43289 | [feishu-openclaw-plugin] dmPolicy: "pairing" 模式下配对状态未正确持久化/读取 | ## 问题描述
当设置 `dmPolicy: "pairing"` 时，用户完成配对后，配对状态没有被正确持久化或读取，导致每次发送消息都会重新触发配对流程，不停发送配对码。
## 环境信息
- OpenClaw 版本: `0.1.8-fix.3`
- 飞书插件版本: `@larksuiteoapi/feishu-openclaw-plugin@2026.3.8`
- 配置: `dmPolicy: "pairing"`
## 复现步骤
1. 配置 `channels.feishu.accounts.default.dmPolicy: "pairing"`
2. 用户首次与机器人对话，输入配对码完成配对
3. 用户再次发送消... | closed | completed | true | 1 | [] | [] | 2026-03-11T15:23:18Z | 2026-04-27T04:51:08Z | 2026-04-24T22:19:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | LINK-HOPE | 32,232,613 | MDQ6VXNlcjMyMjMyNjEz | User | false |
openclaw/openclaw | 4,058,899,951 | I_kwDOQb6kR87x7eXv | 43,290 | https://github.com/openclaw/openclaw/issues/43290 | https://api.github.com/repos/openclaw/openclaw/issues/43290 | Browser tool should support multi-session reuse of a single CDP instance (avoid SingletonLock conflicts) | ### Summary
Allow multiple OpenClaw sessions to reuse a single existing CDP browser instance instead of trying to launch separate Chromium processes against the same profile.
### Problem to solve
In multi-session use (DMs, channels, threads, subagents), different sessions may invoke the browser tool concurrently. To... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-11T15:23:24Z | 2026-04-27T04:51:07Z | 2026-04-24T22:19:13Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | capjason | 5,365,501 | MDQ6VXNlcjUzNjU1MDE= | User | false |
openclaw/openclaw | 4,059,003,623 | I_kwDOQb6kR87x73rn | 43,301 | https://github.com/openclaw/openclaw/issues/43301 | https://api.github.com/repos/openclaw/openclaw/issues/43301 | [Feature] Add ClawHub skill search in Control UI Skills panel | ### Summary
Add a search feature to the Control UI Skills panel to browse and search skills from ClawHub
### Problem to solve
Currently, the Control UI Skills panel only shows installed skills and allows basic enable/disable/install operations. Users cannot search for new skills from ClawHub directly in the UI - the... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-11T15:41:16Z | 2026-04-27T04:51:06Z | 2026-04-24T22:19:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | songsoso | 11,565,193 | MDQ6VXNlcjExNTY1MTkz | User | false |
openclaw/openclaw | 4,059,007,123 | I_kwDOQb6kR87x74iT | 43,302 | https://github.com/openclaw/openclaw/issues/43302 | https://api.github.com/repos/openclaw/openclaw/issues/43302 | Feature Request: Interactive interrupt and correction during task execution | ## 🦞 Feature Request: Interactive Interrupt & Correction
### Problem Statement
Currently, when OpenClaw is running a task, there is no way to interrupt and correct it mid-execution if:
- The assistant is going in the wrong direction
- There is a misunderstanding of the task requirements
- The user notices a better a... | closed | completed | true | 1 | [] | [] | 2026-03-11T15:41:51Z | 2026-04-27T04:51:06Z | 2026-04-24T22:19:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lileilei-camera | 7,010,843 | MDQ6VXNlcjcwMTA4NDM= | User | false |
openclaw/openclaw | 4,059,140,220 | I_kwDOQb6kR87x8ZB8 | 43,318 | https://github.com/openclaw/openclaw/issues/43318 | https://api.github.com/repos/openclaw/openclaw/issues/43318 | [Bug]: sessions_send hardcodes INTERNAL_MESSAGE_CHANNEL, causing reply routing to flip from external channel to webchat | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
`sessions_send` injects messages into the target session with `channel: INTERNAL_MESSAGE_CHANNEL` ("webchat") hardcoded. For main sessions that normally route replies to an external channel (e.g. Discord), this causes the session's `lastCha... | closed | completed | true | 5 | [] | [] | 2026-03-11T16:06:09Z | 2026-04-27T04:51:04Z | 2026-04-24T22:19:22Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rbutera | 6,047,293 | MDQ6VXNlcjYwNDcyOTM= | User | false |
openclaw/openclaw | 4,059,222,864 | I_kwDOQb6kR87x8tNQ | 43,329 | https://github.com/openclaw/openclaw/issues/43329 | https://api.github.com/repos/openclaw/openclaw/issues/43329 | [Bug]: Twilio outbound conversation calls drop immediately because initial message is spoken before media stream is active | ## Summary
On OpenClaw `2026.3.8`, Twilio outbound voice calls in `conversation` mode connect and then drop almost immediately after answer. Local investigation suggests the initial outbound message is being spoken too early, before the Twilio media stream is active, which triggers a TwiML `<Say>` fallback that interf... | closed | completed | true | 1 | [] | [] | 2026-03-11T16:20:58Z | 2026-04-27T04:51:03Z | 2026-04-24T22:19:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | owaiskl | 111,141,936 | U_kgDOBp_kMA | User | false |
openclaw/openclaw | 4,059,373,561 | I_kwDOQb6kR87x9R_5 | 43,354 | https://github.com/openclaw/openclaw/issues/43354 | https://api.github.com/repos/openclaw/openclaw/issues/43354 | Health monitor: stale-socket restarts entire gateway, causing cron job failures and excessive downtime | ## Problem
The channel health monitor triggers full gateway restarts on `stale-socket` detection, causing cascading failures:
- **58 gateway restarts in a single day** on a production instance
- Cron jobs killed mid-execution or missed entirely (gateway down at trigger time)
- All channels restart even when only one ... | closed | completed | true | 1 | [] | [] | 2026-03-11T16:46:22Z | 2026-04-27T04:51:03Z | 2026-04-24T22:19:27Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | microblue | 1,555,783 | MDQ6VXNlcjE1NTU3ODM= | User | false |
openclaw/openclaw | 4,059,415,962 | I_kwDOQb6kR87x9cWa | 43,361 | https://github.com/openclaw/openclaw/issues/43361 | https://api.github.com/repos/openclaw/openclaw/issues/43361 | Sub-agent completion events leak to Feishu group main thread instead of staying in original thread | ## Description
When a sub-agent (e.g. Codex spawned via `sessions_spawn`) completes a task that was initiated from within a **Feishu group thread**, the completion callback message is posted to the **group main timeline** instead of the **original thread** where the task was started.
## Steps to Reproduce
1. In a Fe... | closed | completed | true | 1 | [] | [] | 2026-03-11T16:53:17Z | 2026-04-27T04:51:02Z | 2026-04-24T22:19:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mayorx | 27,191,939 | MDQ6VXNlcjI3MTkxOTM5 | User | false |
openclaw/openclaw | 4,059,509,291 | I_kwDOQb6kR87x9zIr | 43,368 | https://github.com/openclaw/openclaw/issues/43368 | https://api.github.com/repos/openclaw/openclaw/issues/43368 | [Bug]: iClawd iOS 2.0.1 ElevenLabs TTS fails with 'Cannot read property Base64 of undefined' | ## Summary
iClawd iOS app (v2.0.1, TestFlight) fails to use ElevenLabs TTS and falls back to system voice with error:
```
ElevenLabs TTS Error
Falling back to system voice.
Error: Cannot read property 'Base64' of undefined
```
## Environment
- **iClawd version:** 2.0.1 (TestFlight)
- **iPhone:** iOS 18.7
- **Gatewa... | closed | completed | true | 3 | [] | [] | 2026-03-11T17:10:20Z | 2026-04-27T04:51:01Z | 2026-04-24T22:19:32Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | charlenemercury | 133,576,629 | U_kgDOB_Y3tQ | User | false |
openclaw/openclaw | 4,059,542,085 | I_kwDOQb6kR87x97JF | 43,372 | https://github.com/openclaw/openclaw/issues/43372 | https://api.github.com/repos/openclaw/openclaw/issues/43372 | Embedded agent error: 'No tool call found for function call output with call_id' under tool-heavy / retry conditions | ## Summary
OpenClaw intermittently throws embedded-agent/tool lifecycle errors like:
- `400 No tool call found for function call output with call_id ...`
In practice, this shows up during long, tool-heavy sessions and causes noisy user-visible error spam, even when underlying work is still progressing.
## Why this l... | closed | completed | true | 1 | [] | [] | 2026-03-11T17:17:10Z | 2026-04-27T04:51:00Z | 2026-04-24T22:19:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | joesinvestments | 257,641,472 | U_kgDOD1tMAA | User | false |
openclaw/openclaw | 4,059,542,569 | I_kwDOQb6kR87x97Qp | 43,373 | https://github.com/openclaw/openclaw/issues/43373 | https://api.github.com/repos/openclaw/openclaw/issues/43373 | [Bug]: feishu-openclaw-plugin probe.js has no caching — burns ~43k API calls/month per account on health checks | ## Summary
The `feishu-openclaw-plugin` extension's `probe.js` calls `GET /open-apis/bot/v3/info` on **every gateway health check** with no caching. With the gateway checking every minute, this burns ~43,200 API calls/month **per Feishu account** — easily exceeding the free tier quota of 10,000/month.
This is a separ... | closed | completed | true | 1 | [] | [] | 2026-03-11T17:17:16Z | 2026-04-27T04:51:00Z | 2026-04-24T22:19:37Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Hollychou924 | 128,659,251 | U_kgDOB6svMw | User | false |
openclaw/openclaw | 4,059,544,724 | I_kwDOQb6kR87x97yU | 43,376 | https://github.com/openclaw/openclaw/issues/43376 | https://api.github.com/repos/openclaw/openclaw/issues/43376 | feat(agents): dynamic store override based on provider metadata | ## Problem
PR #41870 hardcodes Azure OpenAI as the only provider that doesn't support the \`store\` parameter:
```javascript
const store = provider === 'azure' ? false : (params.store ?? true);
```
This approach has several issues:
1. **Maintenance burden** — Every new provider that doesn't support `store` requires... | closed | completed | true | 1 | [] | [] | 2026-03-11T17:17:43Z | 2026-04-27T04:50:59Z | 2026-04-24T22:19:40Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | JonathanJing | 17,068,507 | MDQ6VXNlcjE3MDY4NTA3 | User | false |
openclaw/openclaw | 4,059,790,252 | I_kwDOQb6kR87x-3us | 43,395 | https://github.com/openclaw/openclaw/issues/43395 | https://api.github.com/repos/openclaw/openclaw/issues/43395 | Workspace data not migrated when upgrading to per-agent workspace paths (workspace → workspace-{agentId}) | ## Summary
Upgrading from v2026.2.26 to v2026.3.8 changed the workspace directory convention from `~/.openclaw/workspace` to `~/.openclaw/workspace-main` (and presumably `workspace-{agentId}` for other agents). However, existing files in the old `workspace/` directory are not migrated to the new path, resulting in a c... | closed | not_planned | true | 2 | [] | [] | 2026-03-11T18:05:18Z | 2026-04-27T04:50:58Z | 2026-04-24T22:19:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | littlemac51 | 257,225,191 | U_kgDOD1Tx5w | User | false |
openclaw/openclaw | 4,059,809,393 | I_kwDOQb6kR87x-8Zx | 43,398 | https://github.com/openclaw/openclaw/issues/43398 | https://api.github.com/repos/openclaw/openclaw/issues/43398 | [Feature]: Parallel Web Search as Built-in web_search and web_fetch | ### Summary
Current native web search options are limited. I propose adding [Parallel](https://parallela.i) as a native search and fetch tool.
### Problem to solve
Agents need token-efficient search, and they need to fetch complex web sites. Parallel provides token-efficient search responses and handles javascript h... | closed | not_planned | true | 1 | [
"enhancement"
] | [] | 2026-03-11T18:09:05Z | 2026-04-27T04:50:57Z | 2026-04-24T22:19:45Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | NormallyGaussian | 3,804,139 | MDQ6VXNlcjM4MDQxMzk= | User | false |
openclaw/openclaw | 4,059,812,905 | I_kwDOQb6kR87x-9Qp | 43,400 | https://github.com/openclaw/openclaw/issues/43400 | https://api.github.com/repos/openclaw/openclaw/issues/43400 | Fix: Fallback mechanism never triggers due to per-model timeout equaling global run timeout | ## Bug Description
In the current implementation of OpenClaw, the model fallback mechanism fails to trigger when an LLM provider hangs. The agent instead waits for 60 seconds and then completely aborts the run with an `embedded run timeout` error.
## Root Cause
In `dist/compact-*.js`, there is a global run timeout (`e... | closed | not_planned | true | 1 | [] | [] | 2026-03-11T18:09:46Z | 2026-04-27T04:50:57Z | 2026-04-24T22:19:48Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | conanwhf | 14,151,735 | MDQ6VXNlcjE0MTUxNzM1 | User | false |
openclaw/openclaw | 4,059,840,184 | I_kwDOQb6kR87x_D64 | 43,402 | https://github.com/openclaw/openclaw/issues/43402 | https://api.github.com/repos/openclaw/openclaw/issues/43402 | Feature Request: Pass TELEGRAM_MESSAGE_THREAD_ID to Skills | ## Description
Currently, when a Skill is invoked in a Telegram Topic (Thread), OpenClaw passes `TELEGRAM_CHAT_ID` to the Skill environment, but it does NOT pass `TELEGRAM_MESSAGE_THREAD_ID`.
This causes Skills that need to send messages back to the correct Topic to fail, as they have no way to know which Topic the m... | closed | completed | true | 1 | [] | [] | 2026-03-11T18:15:32Z | 2026-04-27T04:50:56Z | 2026-04-24T22:19:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | soulcolor23 | 97,148,327 | U_kgDOBcpdpw | User | false |
openclaw/openclaw | 4,059,938,169 | I_kwDOQb6kR87x_b15 | 43,408 | https://github.com/openclaw/openclaw/issues/43408 | https://api.github.com/repos/openclaw/openclaw/issues/43408 | [Bug]: case-insensitive host-mounted workspace duplicates MEMORY.md/memory.md in bootstrap and memory indexing | ## Summary
On a macOS host with a **host-mounted workspace inside the OpenClaw container**, `MEMORY.md` and `memory.md` can resolve to the **same underlying file** while still being collected as two logical paths.
This causes OpenClaw to inject the same long-term memory content **twice** into Project Context (bootstr... | closed | completed | true | 3 | [] | [] | 2026-03-11T18:36:14Z | 2026-04-27T04:50:55Z | 2026-04-24T22:19:53Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhanglg12 | 165,030,701 | U_kgDOCdYrLQ | User | false |
openclaw/openclaw | 4,060,131,283 | I_kwDOQb6kR87yAK_T | 43,418 | https://github.com/openclaw/openclaw/issues/43418 | https://api.github.com/repos/openclaw/openclaw/issues/43418 | Feature Request: Add `before_dispatch` plugin hook for universal pre-LLM message interception | ## Summary
Plugin authors currently lack a reliable way to **block or intercept inbound messages before LLM processing** across all channels. This is critical for security plugins (e.g., identity/authentication gates) that need to prevent unauthenticated users from consuming LLM tokens.
## Problem
The existing plugi... | closed | completed | true | 1 | [] | [] | 2026-03-11T19:12:46Z | 2026-04-27T04:50:55Z | 2026-04-24T22:19:56Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | loveyana | 87,516,818 | MDQ6VXNlcjg3NTE2ODE4 | User | false |
openclaw/openclaw | 4,060,317,012 | I_kwDOQb6kR87yA4VU | 43,435 | https://github.com/openclaw/openclaw/issues/43435 | https://api.github.com/repos/openclaw/openclaw/issues/43435 | Mattermost: media attachments in intermediate multi-turn turns are silently dropped | ## Summary
In multi-turn responses with edit-in-place streaming active, the skip logic in `deliver()` (introduced in #43041) returns early before media handling. If an intermediate turn payload included `mediaUrls`, those attachments would be dropped.
## Context
`onPartialReply` only carries text, so streaming turns... | closed | completed | true | 1 | [] | [] | 2026-03-11T19:50:20Z | 2026-04-27T04:50:54Z | 2026-04-24T22:19:58Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | teconomix | 6,959,299 | MDQ6VXNlcjY5NTkyOTk= | User | false |
openclaw/openclaw | 4,060,345,767 | I_kwDOQb6kR87yA_Wn | 43,437 | https://github.com/openclaw/openclaw/issues/43437 | https://api.github.com/repos/openclaw/openclaw/issues/43437 | Telegram DM chat path can break while slash commands and outbound send still work | ## Summary
Telegram DMs can stop behaving like normal chat while the bot still appears partially healthy:
- ordinary inbound Telegram messages do not get normal conversational replies
- slash commands still work
- outbound sends still work
- sending `HEARTBEAT` gets consumed/disappears
We saw this on **Fred** today, a... | closed | completed | true | 1 | [] | [] | 2026-03-11T19:56:11Z | 2026-04-27T04:50:53Z | 2026-04-24T22:20:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SunDevilATB | 170,658,350 | U_kgDOCiwKLg | User | false |
openclaw/openclaw | 4,060,418,446 | I_kwDOQb6kR87yBRGO | 43,442 | https://github.com/openclaw/openclaw/issues/43442 | https://api.github.com/repos/openclaw/openclaw/issues/43442 | Auto-load recognized bootstrap files when they exist in workspace | ## The Problem
OpenClaw recognizes several workspace files as "bootstrap files": SOUL.md, IDENTITY.md, USER.md, HEARTBEAT.md, BOOTSTRAP.md, and MEMORY.md. However, only AGENTS.md and TOOLS.md auto-load into agent sessions. All other files require the user to manually enable the `bootstrap-extra-files` hook in `opencla... | closed | completed | true | 1 | [] | [] | 2026-03-11T20:11:15Z | 2026-04-27T04:50:52Z | 2026-04-24T22:20:03Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | felixlatipac | 265,658,163 | U_kgDOD9WfMw | User | false |
openclaw/openclaw | 4,060,441,946 | I_kwDOQb6kR87yBW1a | 43,444 | https://github.com/openclaw/openclaw/issues/43444 | https://api.github.com/repos/openclaw/openclaw/issues/43444 | [Feature]: Configurable list of custom context files for auto-reading (bypass VALID_BOOTSTRAP_NAMES) | ### Summary
Allow users to specify arbitrary files (e.g., `PROJECT.md`, `CODING_STANDARDS.md`) in the agent configuration to be automatically read and injected into the agent's system prompt as bootstrap context.
### Problem to solve
Currently, the list of valid context (bootstrap) files in OpenClaw is strictly hard... | closed | not_planned | true | 3 | [
"enhancement"
] | [] | 2026-03-11T20:15:53Z | 2026-04-27T04:50:51Z | 2026-04-24T22:20:06Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Ar3ss12 | 150,044,748 | U_kgDOCPGATA | User | false |
openclaw/openclaw | 4,060,527,560 | I_kwDOQb6kR87yBrvI | 43,452 | https://github.com/openclaw/openclaw/issues/43452 | https://api.github.com/repos/openclaw/openclaw/issues/43452 | [Bug]: Cron manual run is enqueued but never appears in cron runs history (main/systemEvent job) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Issue: openclaw cron run <jobId> returns enqueued: true with a new runId, but openclaw cron runs --id <jobId> does not show a new finished entry (only old runs).
Affected job: 948faf38-7992-4172-aa63-0df4c923b120 (lifeos-whatsapp-daily-summ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-11T20:34:36Z | 2026-04-27T04:50:51Z | 2026-04-24T22:20:08Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | spaceriderhub | 267,450,825 | U_kgDOD_D5yQ | User | false |
openclaw/openclaw | 4,060,612,583 | I_kwDOQb6kR87yCAfn | 43,464 | https://github.com/openclaw/openclaw/issues/43464 | https://api.github.com/repos/openclaw/openclaw/issues/43464 | fix(ssrf): Telegram media download IPv4 fallback regression from 45b74fb56c | ## Bug Description
Telegram media downloads (PDFs, images, voice, etc.) fail with `TypeError: fetch failed → AggregateError [ETIMEDOUT]` in dual-stack environments where IPv6 is enabled but has no internet route (common in Docker/container deployments).
## Root Cause
Commit `45b74fb56c` ("move network fallback to re... | closed | completed | true | 2 | [] | [] | 2026-03-11T20:53:28Z | 2026-04-27T04:50:50Z | 2026-04-24T22:20:11Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mekenthompson | 8,923,913 | MDQ6VXNlcjg5MjM5MTM= | User | false |
openclaw/openclaw | 4,060,684,608 | I_kwDOQb6kR87yCSFA | 43,471 | https://github.com/openclaw/openclaw/issues/43471 | https://api.github.com/repos/openclaw/openclaw/issues/43471 | Cron scheduler doesn't wake for one-shot 'at' jobs | ## Bug Description
The cron scheduler's `nextWakeAtMs` is not updated when one-shot `at` jobs are created. This causes one-shot jobs to be skipped entirely until the next recurring cron job fires.
## Steps to Reproduce
1. Have a recurring cron job scheduled for a future time (e.g., 3:00 PM)
2. Create a one-shot `at`... | closed | completed | true | 1 | [] | [] | 2026-03-11T21:09:24Z | 2026-04-27T04:50:49Z | 2026-04-24T22:20:14Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | guglielmofonda | 54,238,344 | MDQ6VXNlcjU0MjM4MzQ0 | User | false |
openclaw/openclaw | 4,060,713,480 | I_kwDOQb6kR87yCZII | 43,479 | https://github.com/openclaw/openclaw/issues/43479 | https://api.github.com/repos/openclaw/openclaw/issues/43479 | [Bug]: Telegram long-polling stops receiving messages after connection drop, no auto-reconnect | ### Bug type
Regression (worked before, now fails)
### Summary
Version: 2026.3.8
OS: Windows 11 x64
Mode: Local gateway, polling (not webhook)
Issue: After a network hiccup or Telegram API timeout, the bot stops receiving messages entirely. The gateway process stays alive and webchat continues to work, so the issue ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-11T21:16:14Z | 2026-04-27T04:50:48Z | 2026-04-24T22:20:17Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kaokula | 128,977,779 | U_kgDOB7ALcw | User | false |
openclaw/openclaw | 4,060,725,979 | I_kwDOQb6kR87yCcLb | 43,482 | https://github.com/openclaw/openclaw/issues/43482 | https://api.github.com/repos/openclaw/openclaw/issues/43482 | Fallback retry should detect partial tool execution before replaying prompt | ## Problem
When a model fallback retry fires after the previous attempt partially executed (e.g. completed tool calls before timing out or hitting a rate limit), the retry replays the full original prompt to a new model. The new model has no knowledge of the prior attempt's side effects and may repeat non-idempotent a... | closed | completed | true | 1 | [] | [] | 2026-03-11T21:19:06Z | 2026-04-27T04:50:48Z | 2026-04-24T22:20:20Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mitchmcalister | 209,334 | MDQ6VXNlcjIwOTMzNA== | User | false |
openclaw/openclaw | 4,060,729,962 | I_kwDOQb6kR87yCdJq | 43,483 | https://github.com/openclaw/openclaw/issues/43483 | https://api.github.com/repos/openclaw/openclaw/issues/43483 | TTS tool hangs indefinitely in isolated cron sessions (never returns, always times out) | ## Bug report
**Summary:** The `tts` tool never returns when called from an isolated cron session (`sessionTarget: "isolated"`). The job always times out at the configured timeout limit (tested with 120s, 300s). The TTS call appears to hang silently — no error is thrown, it just never resolves.
## Steps to reproduce
... | closed | completed | true | 6 | [] | [] | 2026-03-11T21:20:01Z | 2026-04-27T04:50:47Z | 2026-04-24T22:20:22Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sisutuulenisa | 261,256,991 | U_kgDOD5J3Hw | User | false |
openclaw/openclaw | 4,060,862,941 | I_kwDOQb6kR87yC9nd | 43,490 | https://github.com/openclaw/openclaw/issues/43490 | https://api.github.com/repos/openclaw/openclaw/issues/43490 | MiniMax token usage always returns zeros in 2026.3.8 | ## Summary
After upgrading from 2026.3.2 to 2026.3.8, MiniMax-M2.5 `lastCallUsage` always returns zeros for all token fields. This was working correctly in 2026.3.2.
## Reproduction
```bash
openclaw agent --agent main --message "ping" --json
```
`result.meta.agentMeta.lastCallUsage`:
```json
{
"input": 0,
"outp... | closed | completed | true | 9 | [] | [] | 2026-03-11T21:46:43Z | 2026-04-27T04:50:46Z | 2026-04-24T22:20:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | GiPPe | 1,165,233 | MDQ6VXNlcjExNjUyMzM= | User | false |
openclaw/openclaw | 4,061,131,867 | I_kwDOQb6kR87yD_Rb | 43,508 | https://github.com/openclaw/openclaw/issues/43508 | https://api.github.com/repos/openclaw/openclaw/issues/43508 | Discord button interactions not routing back to agent | ## Problem
Discord button interactions are not being routed back to the agent as inbound messages, despite documentation stating they should be.
## Steps to reproduce
1. Send a message with interactive components (buttons) using the message tool with `components` payload
2. Button appears correctly in Discord
3. User ... | closed | completed | true | 1 | [] | [] | 2026-03-11T22:52:59Z | 2026-04-27T04:50:45Z | 2026-04-24T22:20:27Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cbcampos | 63,550,718 | MDQ6VXNlcjYzNTUwNzE4 | User | false |
openclaw/openclaw | 4,061,267,332 | I_kwDOQb6kR87yEgWE | 43,521 | https://github.com/openclaw/openclaw/issues/43521 | https://api.github.com/repos/openclaw/openclaw/issues/43521 | ACP runtime ignores model parameter on sessions_spawn | ## Bug Description
The `model` parameter on `sessions_spawn(runtime="acp")` is ignored. Sessions always use the agent's default model instead of the specified override.
## Reproduction
```javascript
sessions_spawn({
runtime: "acp",
agentId: "codex",
thread: true,
mode: "run",
model: "openai-codex/gpt-5.2-c... | closed | completed | true | 2 | [] | [] | 2026-03-11T23:41:08Z | 2026-04-27T04:50:45Z | 2026-04-24T22:20:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | kai-the-man[bot] | 259,478,402 | BOT_kgDOD3dTgg | Bot | false |
openclaw/openclaw | 4,061,322,467 | I_kwDOQb6kR87yEtzj | 43,532 | https://github.com/openclaw/openclaw/issues/43532 | https://api.github.com/repos/openclaw/openclaw/issues/43532 | feat: bootstrapFreezeMode for session-stable workspace file injection | ## Problem
Workspace bootstrap files (MEMORY.md, USER.md, SOUL.md, AGENTS.md, TOOLS.md) are re-read and re-injected into the system prompt on every turn. When an agent writes to MEMORY.md mid-session (which is a core part of the memory workflow), the system prompt changes, invalidating the LLM provider's KV prefix cac... | closed | completed | true | 1 | [] | [] | 2026-03-12T00:02:05Z | 2026-04-27T04:50:44Z | 2026-04-24T22:20:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | MrJax77 | 82,862,260 | MDQ6VXNlcjgyODYyMjYw | User | false |
openclaw/openclaw | 4,061,330,303 | I_kwDOQb6kR87yEvt_ | 43,534 | https://github.com/openclaw/openclaw/issues/43534 | https://api.github.com/repos/openclaw/openclaw/issues/43534 | Feature: per-agent allowBundled skill filtering | ## Summary
Currently `skills.allowBundled` is a **global-only** setting — it applies the same whitelist to every agent. In multi-agent setups each sub-agent has a distinct role and only needs a small subset of bundled skills, yet all ready bundled skills are injected into every agent's system prompt.
## Problem
| Se... | closed | completed | true | 2 | [] | [] | 2026-03-12T00:05:08Z | 2026-04-27T04:50:43Z | 2026-04-24T22:20:35Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | youngjin39 | 83,735,994 | MDQ6VXNlcjgzNzM1OTk0 | User | false |
openclaw/openclaw | 4,061,380,986 | I_kwDOQb6kR87yE8F6 | 43,544 | https://github.com/openclaw/openclaw/issues/43544 | https://api.github.com/repos/openclaw/openclaw/issues/43544 | sessions_spawn(runtime="acp") fails: "spawnedBy is only supported for subagent:* sessions" | ## Bug
`sessions_spawn(runtime="acp", agentId="claude", ...)` consistently fails with:
```
{"status":"error","error":"spawnedBy is only supported for subagent:* sessions"}
```
This affects both `thread=true` and non-thread ACP spawns.
## Root Cause
In the session store patch handler (`gateway-cli-*.js`), `spawnedB... | closed | completed | true | 3 | [] | [] | 2026-03-12T00:23:55Z | 2026-04-27T04:50:43Z | 2026-04-24T22:20:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | tpys | 3,313,406 | MDQ6VXNlcjMzMTM0MDY= | User | false |
openclaw/openclaw | 4,061,445,676 | I_kwDOQb6kR87yFL4s | 43,550 | https://github.com/openclaw/openclaw/issues/43550 | https://api.github.com/repos/openclaw/openclaw/issues/43550 | [Bug] Kimi web_search tool uses incorrect tool type format | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Kimi web_search tool fails with 401 error due to incorrect tool type format.
### Steps to reproduce
1. Configure OpenClaw with Kimi API key in openclaw.json
2. Use the web_search tool: web_search(query="latest AI news")
3. Observe error: ... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T00:44:45Z | 2026-04-27T04:50:42Z | 2026-04-24T22:20:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Mike-Feng | 11,884,093 | MDQ6VXNlcjExODg0MDkz | User | false |
openclaw/openclaw | 4,061,447,049 | I_kwDOQb6kR87yFMOJ | 43,551 | https://github.com/openclaw/openclaw/issues/43551 | https://api.github.com/repos/openclaw/openclaw/issues/43551 | [Bug]: openclaw-mem0 plugin crashes gateway — schema rejects all config keys except "enabled" | ### Bug type
Regression (worked before, now fails)
### Summary
plugins.entries.openclaw-mem0 schema validation rejects all keys except enabled — even though the plugin's own ALLOWED_KEYS includes mode, userId, autoRecall, oss etc.
The workaround of using a nested config key also fails validation
Version: OpenClaw 20... | closed | not_planned | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-12T00:45:17Z | 2026-04-27T04:50:41Z | 2026-04-24T22:20:43Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | openoptimus88 | 263,033,022 | U_kgDOD62Qvg | User | false |
openclaw/openclaw | 4,061,517,811 | I_kwDOQb6kR87yFdfz | 43,561 | https://github.com/openclaw/openclaw/issues/43561 | https://api.github.com/repos/openclaw/openclaw/issues/43561 | trusted-proxy mode is broken with nginx — proxy headers are never evaluated, making production multi-node deployments insecure | ## The problem
In `trusted-proxy` mode, the shared-secret fallback fires **before** nginx proxy headers (`X-Forwarded-For`, `X-Real-IP`, etc.) are evaluated. This means any worker node connecting through nginx cannot authenticate — the auth chain rejects it before proxy identity is even checked.
The result: operators... | closed | not_planned | true | 8 | [] | [
"vincentkoc"
] | 2026-03-12T01:16:36Z | 2026-04-27T04:50:40Z | 2026-04-24T22:20:46Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rickstonerz | 123,851,024 | U_kgDOB2HREA | User | false |
openclaw/openclaw | 4,061,571,594 | I_kwDOQb6kR87yFqoK | 43,569 | https://github.com/openclaw/openclaw/issues/43569 | https://api.github.com/repos/openclaw/openclaw/issues/43569 | 409 Conflict loop when polling multiple Telegram bot accounts simultaneously | ## Description
When running multiple Telegram bot accounts under a single OpenClaw gateway instance, frequent `409 Conflict` errors occur on `getUpdates` calls, causing a restart loop.
## Environment
- **OpenClaw version:** 2026.3.8
- **OS:** macOS (Darwin, arm64)
- **Node:** v24.x
- **Telegram accounts:** 4 bot tok... | closed | completed | true | 1 | [] | [] | 2026-03-12T01:39:45Z | 2026-04-27T04:50:39Z | 2026-04-24T22:20:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | jwxxkim | 137,521,442 | U_kgDOCDJpIg | User | false |
openclaw/openclaw | 4,061,628,524 | I_kwDOQb6kR87yF4hs | 43,586 | https://github.com/openclaw/openclaw/issues/43586 | https://api.github.com/repos/openclaw/openclaw/issues/43586 | [Bug]: WebChat上传的图片在聊天记录中消失 | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
WebChat上传的图片在发送后几秒钟内从聊天记录中消失，图片没有保存到media文件夹
### Steps to reproduce
1. 打开WebChat界面
2. 上传一张图片
3. 发送图片
4. 等待几秒钟
5. 图片从聊天记录中消失
### Expected behavior
图片应该持久显示在聊天记录中，刷新页面后仍然可见
### Actual behavior
图片几秒钟后消失，刷新页面后也不回来。上传的blob图片似乎没有保存到media文件夹... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T02:04:30Z | 2026-04-27T04:50:39Z | 2026-04-24T22:20:51Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | brucestarkliu | 249,122,845 | U_kgDODtlQHQ | User | false |
openclaw/openclaw | 4,061,666,004 | I_kwDOQb6kR87yGBrU | 43,590 | https://github.com/openclaw/openclaw/issues/43590 | https://api.github.com/repos/openclaw/openclaw/issues/43590 | [WebChat] Images sent from Control UI are not received by the agent | ### Bug type
Bug
### Summary
When sending images through the WebChat Control UI, the agent does not receive or see the images. The agent's `image` tool cannot find the images, and checking the `~/.openclaw/media/inbound/` directory shows no recent files.
### Steps to reproduce
1. Open WebChat Control UI (http://12... | closed | completed | true | 1 | [
"close:duplicate",
"close:already-fixed",
"dedupe:child"
] | [
"vincentkoc"
] | 2026-03-12T02:18:18Z | 2026-04-27T04:31:45Z | 2026-04-24T22:20:54Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | liuyongbing | 1,701,373 | MDQ6VXNlcjE3MDEzNzM= | User | false |
openclaw/openclaw | 4,061,674,268 | I_kwDOQb6kR87yGDsc | 43,591 | https://github.com/openclaw/openclaw/issues/43591 | https://api.github.com/repos/openclaw/openclaw/issues/43591 | Codex API server_error (500) not recognized by model failover classifier | ## Bug
When the OpenAI Codex API returns a `server_error` (HTTP 500), OpenClaw does **not** trigger model fallback. Instead, the error is surfaced directly to the user, even when a fallback chain is configured.
### Expected behavior
A Codex `server_error` should be classified as a transient/retryable error and trigge... | closed | completed | true | 2 | [] | [] | 2026-03-12T02:21:27Z | 2026-04-27T04:50:38Z | 2026-04-24T22:20:56Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | bradfrosty | 5,558,413 | MDQ6VXNlcjU1NTg0MTM= | User | false |
openclaw/openclaw | 4,061,695,125 | I_kwDOQb6kR87yGIyV | 43,592 | https://github.com/openclaw/openclaw/issues/43592 | https://api.github.com/repos/openclaw/openclaw/issues/43592 | [Feature]: broadcast inbound user messages as WebSocket events | ### Summary
Gateway WebSocket events only broadcast assistant responses — inbound user messages from `chat.send` are not broadcast to other connected clients.
### Problem to solve
When multiple clients connect to the gateway, subscribers cannot see the full conversation flow in real time. If client A subscribes to e... | closed | completed | true | 1 | [
"enhancement"
] | [] | 2026-03-12T02:28:28Z | 2026-04-27T04:50:36Z | 2026-04-24T22:20:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | ColinHuang | 985,443 | MDQ6VXNlcjk4NTQ0Mw== | User | false |
openclaw/openclaw | 4,061,700,027 | I_kwDOQb6kR87yGJ-7 | 43,593 | https://github.com/openclaw/openclaw/issues/43593 | https://api.github.com/repos/openclaw/openclaw/issues/43593 | [Bug]: Menu bar widget reports "Idle/No Activity" when Gateway runs as LaunchAgent | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
The menu bar widget incorrectly displays an "Idle" status due to a log path mismatch when OpenClaw runs as a system service.
### Steps to reproduce
1. Install/Configure: Run OpenClaw as a system service (LaunchAgent).
2. Start: Ensure the... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T02:29:56Z | 2026-04-27T04:50:36Z | 2026-04-24T22:21:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | TARDIS95 | 267,501,670 | U_kgDOD_HAZg | User | false |
openclaw/openclaw | 4,061,721,242 | I_kwDOQb6kR87yGPKa | 43,602 | https://github.com/openclaw/openclaw/issues/43602 | https://api.github.com/repos/openclaw/openclaw/issues/43602 | [Bug]: LaunchAgent silently unloads after macOS sleep/idle — gateway start/restart fail until re-install | ## Description
The gateway LaunchAgent silently unloads after macOS sleep or extended idle. Once unloaded, `openclaw gateway start` and `openclaw gateway restart` fail silently — only `openclaw gateway install` re-registers and restarts the service.
This causes agents to go completely unresponsive until someone manual... | closed | completed | true | 1 | [] | [] | 2026-03-12T02:37:00Z | 2026-04-27T04:50:35Z | 2026-04-24T22:21:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SPFAdvisors | 133,194,757 | U_kgDOB_BkBQ | User | false |
openclaw/openclaw | 4,061,721,961 | I_kwDOQb6kR87yGPVp | 43,603 | https://github.com/openclaw/openclaw/issues/43603 | https://api.github.com/repos/openclaw/openclaw/issues/43603 | [Feature]: Auto-compaction or warning when agent session context exceeds threshold | ## Description
Agent sessions gradually bloat in token usage over extended conversations. When a session reaches 70-80%+ of the context window, the agent becomes sluggish, starts timing out, or freezes entirely. There's no automatic compaction and no warning to the user or operator.
## Environment
- OpenClaw 2026.3.8
... | closed | completed | true | 1 | [] | [] | 2026-03-12T02:37:16Z | 2026-04-27T04:50:34Z | 2026-04-24T22:21:07Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SPFAdvisors | 133,194,757 | U_kgDOB_BkBQ | User | false |
openclaw/openclaw | 4,061,722,670 | I_kwDOQb6kR87yGPgu | 43,604 | https://github.com/openclaw/openclaw/issues/43604 | https://api.github.com/repos/openclaw/openclaw/issues/43604 | [Bug]: Cron jobs report 'ok' status when agent model is unreachable — silent backup failures | ## Description
When a cron job has a hardcoded model override (e.g., `openai-codex/gpt-5.3-codex`) and that model/provider is experiencing an outage, the cron job still reports `lastStatus: ok` even though the agent turn failed to execute the task.
This caused 3 consecutive nights of missed backups (March 8-10) that w... | closed | completed | true | 3 | [] | [] | 2026-03-12T02:37:30Z | 2026-04-27T04:50:34Z | 2026-04-24T22:21:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | SPFAdvisors | 133,194,757 | U_kgDOB_BkBQ | User | false |
openclaw/openclaw | 4,061,758,430 | I_kwDOQb6kR87yGYPe | 43,612 | https://github.com/openclaw/openclaw/issues/43612 | https://api.github.com/repos/openclaw/openclaw/issues/43612 | Feishu: duplicate card delivery for long streaming responses | ## Description
When `streaming: true` and `renderMode: "card"` are both enabled for a Feishu channel, long responses (e.g. multi-table project summaries ~4000+ chars) are delivered **twice**:
1. First as a **single streaming card** (correct, complete content)
2. Then again as **multiple chunked cards** (same content,... | closed | completed | true | 2 | [] | [] | 2026-03-12T02:50:28Z | 2026-04-27T04:50:33Z | 2026-04-24T22:21:12Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lefarcen | 20,859,779 | MDQ6VXNlcjIwODU5Nzc5 | User | false |
openclaw/openclaw | 4,061,783,886 | I_kwDOQb6kR87yGedO | 43,617 | https://github.com/openclaw/openclaw/issues/43617 | https://api.github.com/repos/openclaw/openclaw/issues/43617 | Bug: 消息到达 Gateway 但未持久化到 session 文件 | ### 问题描述
消息到达 Gateway 但未持久化到 session 文件
### 环境
- 操作系统：macOS
- OpenClaw 版本：2026.3.8
- 消息渠道：DingTalk
### 问题详情
1. 用户通过 DingTalk 发送消息
2. Gateway 日志显示消息已到达并成功回复
3. 但 session 文件中不存在该消息
### 复现频率
多次复现 | closed | completed | true | 1 | [] | [] | 2026-03-12T02:59:21Z | 2026-04-27T04:50:32Z | 2026-04-24T22:21:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | InformationX | 23,731,260 | MDQ6VXNlcjIzNzMxMjYw | User | false |
openclaw/openclaw | 4,061,813,270 | I_kwDOQb6kR87yGloW | 43,627 | https://github.com/openclaw/openclaw/issues/43627 | https://api.github.com/repos/openclaw/openclaw/issues/43627 | 技能系统文档需要更新 | The documentation for the skill system needs to be reviewed and updated to reflect the latest changes and best practices. | closed | completed | true | 2 | [] | [] | 2026-03-12T03:09:25Z | 2026-04-27T04:50:31Z | 2026-04-24T22:21:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | pureairtech001-collab | 267,277,913 | U_kgDOD-5WWQ | User | false |
openclaw/openclaw | 4,061,818,538 | I_kwDOQb6kR87yGm6q | 43,628 | https://github.com/openclaw/openclaw/issues/43628 | https://api.github.com/repos/openclaw/openclaw/issues/43628 | [BUG] macOS LaunchAgent + configure wizard creates duplicate gateway process, causing 30+ hour Telegram 409 polling conflict | ### Bug type
Regression (worked before, now fails)
### Summary
Running `openclaw configure` (wizard) while the LaunchAgent is already managing a running gateway process leaves the original process alive. Both instances then attempt to long-poll Telegram simultaneously, producing continuous 409 Conflict errors that s... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-12T03:11:31Z | 2026-04-27T04:50:31Z | 2026-04-24T22:21:21Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | michael-ruffolo | 266,198,036 | U_kgDOD93cFA | User | false |
openclaw/openclaw | 4,061,827,183 | I_kwDOQb6kR87yGpBv | 43,629 | https://github.com/openclaw/openclaw/issues/43629 | https://api.github.com/repos/openclaw/openclaw/issues/43629 | Solution: Proxy configuration for LLM timeout and Telegram in restricted networks (China) | # Solution: Proxy Configuration for LLM Timeout and Telegram in Restricted Networks (China)
## Problem Description
Encountered the following issues when using OpenClaw:
1. **LLM Timeout Error** - `LLM request timed out`
2. **Telegram Connection Failed** - `deleteWebhook failed: Network request failed`
## Environment... | closed | completed | true | 3 | [] | [] | 2026-03-12T03:14:52Z | 2026-04-27T04:50:30Z | 2026-04-24T22:21:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AbnerZhuZhu | 45,137,813 | MDQ6VXNlcjQ1MTM3ODEz | User | false |
openclaw/openclaw | 4,061,833,418 | I_kwDOQb6kR87yGqjK | 43,630 | https://github.com/openclaw/openclaw/issues/43630 | https://api.github.com/repos/openclaw/openclaw/issues/43630 | [Bug]: Assistant fails to output text response after completing file operations (Write/Exec) in Chat UI | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
1. Send a prompt to create a file (e.g., "Create a blank HTML template").<img width="1464" height="542" alt="Image" src="https://github.com/user-attachments/assets/3c017923-3f10-435a-8b5b-201e938cb5af" />
2. Notice that after the tools sho... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T03:17:21Z | 2026-04-27T04:50:29Z | 2026-04-24T22:21:26Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AzraelJi | 40,855,023 | MDQ6VXNlcjQwODU1MDIz | User | false |
openclaw/openclaw | 4,061,864,444 | I_kwDOQb6kR87yGyH8 | 43,633 | https://github.com/openclaw/openclaw/issues/43633 | https://api.github.com/repos/openclaw/openclaw/issues/43633 | [Bug] Feishu bindings with accountId or group peer.kind not working | ### Bug type
Regression (worked before, now fails)
### Summary
Feishu bindings not working - messages routed to default agent instead of configured agentId
### Steps to reproduce
1. Configure multiple Feishu accounts (main, nova, atlas) in openclaw.json
2. Add bindings with peer.kind=group for different group IDs
... | closed | completed | true | 4 | [
"bug"
] | [] | 2026-03-12T03:28:08Z | 2026-04-27T04:50:28Z | 2026-04-24T22:21:29Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | leonzopp | 20,434,972 | MDQ6VXNlcjIwNDM0OTcy | User | false |
openclaw/openclaw | 4,061,932,033 | I_kwDOQb6kR87yHCoB | 43,650 | https://github.com/openclaw/openclaw/issues/43650 | https://api.github.com/repos/openclaw/openclaw/issues/43650 | Bug: hard NO_REPLY behavior produces bad UX in webchat after message-tool delivery | ## Summary
When the assistant sends the real user-facing response via `message(action=send)` to another channel (for example Feishu), the required follow-up `NO_REPLY` can produce bad UX in webchat.
## Environment
- channel: `webchat`
- cross-channel delivery to: `feishu`
- main agent session
## Repro
1. Start in ... | closed | completed | true | 1 | [] | [] | 2026-03-12T03:51:37Z | 2026-04-27T04:50:27Z | 2026-04-24T22:21:31Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | WingsOfPanda | 46,889,518 | MDQ6VXNlcjQ2ODg5NTE4 | User | false |
openclaw/openclaw | 4,061,961,552 | I_kwDOQb6kR87yHJ1Q | 43,657 | https://github.com/openclaw/openclaw/issues/43657 | https://api.github.com/repos/openclaw/openclaw/issues/43657 | [Bug]: openai-completions custom API (school CSU) returns JSON-wrapped response instead of plain text | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When using a custom openai-completions provider, the agent response is returned as a nested JSON-serialized array instead of plain text.
### Steps to reproduce
1. Configure a custom `openai-completions` provider in `~/.openclaw/openclaw.j... | closed | completed | true | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T04:01:03Z | 2026-04-27T04:50:27Z | 2026-04-24T22:21:34Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | BlueBlueKitty | 95,596,887 | U_kgDOBbKxVw | User | false |
openclaw/openclaw | 4,062,043,411 | I_kwDOQb6kR87yHd0T | 43,667 | https://github.com/openclaw/openclaw/issues/43667 | https://api.github.com/repos/openclaw/openclaw/issues/43667 | ACP_SESSION_INIT_FAILED for gemini: sessions ensure/new returns no valid session id | ## Summary
ACP session init fails for `gemini` harness: neither `sessions ensure` nor `sessions new` returns valid session identifiers.
## Environment
- OpenClaw: `2026.3.8 (3caab92)`
- OS: Linux x64
- Channel: Telegram direct chat
- Gemini CLI binary is installed and works locally (`gemini --version` and one-shot pro... | closed | completed | true | 5 | [] | [] | 2026-03-12T04:23:41Z | 2026-04-27T04:50:26Z | 2026-04-24T22:21:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | sd4142395 | 56,948,990 | MDQ6VXNlcjU2OTQ4OTkw | User | false |
openclaw/openclaw | 4,062,170,370 | I_kwDOQb6kR87yH80C | 43,689 | https://github.com/openclaw/openclaw/issues/43689 | https://api.github.com/repos/openclaw/openclaw/issues/43689 | [slack] Unhandled promise rejection after WebSocket 408 on reconnect | **Environment**
- OpenClaw version: 2026.3.8 (3caab92)
- Platform: Linux x64 / Node 22.22.0
- Channel: Slack (Socket Mode)
**Summary**
After a health-monitor-triggered socket restart, the Slack WebSocket connection experienced a pong timeout followed by a HTTP 408 error, which triggered an `Unhandled promise rejectio... | closed | completed | true | 1 | [] | [] | 2026-03-12T05:01:19Z | 2026-04-27T04:50:24Z | 2026-04-24T22:21:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | davetroy | 11,199 | MDQ6VXNlcjExMTk5 | User | false |
openclaw/openclaw | 4,062,173,376 | I_kwDOQb6kR87yH9jA | 43,691 | https://github.com/openclaw/openclaw/issues/43691 | https://api.github.com/repos/openclaw/openclaw/issues/43691 | Anthropic thinking block 'signature' field lost during session persistence — causes API rejection | ## Problem
When using Anthropic models with extended thinking enabled, the `signature` field on `thinking` content blocks is not preserved when messages are serialized to the session JSONL file. When the conversation history is sent back to the Anthropic API in subsequent turns, the API rejects the request:
```
LLM r... | closed | completed | true | 1 | [] | [] | 2026-03-12T05:02:13Z | 2026-04-27T04:50:23Z | 2026-04-24T22:21:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | PenguinMiaou | 209,870,260 | U_kgDODIJdtA | User | false |
openclaw/openclaw | 4,062,186,802 | I_kwDOQb6kR87yIA0y | 43,692 | https://github.com/openclaw/openclaw/issues/43692 | https://api.github.com/repos/openclaw/openclaw/issues/43692 | Proposal: per-model / per-lane behavior sheets | I built a local MVP for **per-model / per-lane behavior sheets** and it seems promising enough to ask whether this should stay workspace-local or become a first-class OpenClaw concept.
## The problem
When model lineups change, behavior drifts in ways that are hard to manage with one giant universal prompt.
The same a... | closed | not_planned | true | 2 | [] | [] | 2026-03-12T05:05:56Z | 2026-04-27T04:50:23Z | 2026-04-24T22:21:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | qpp927 | 79,176,975 | MDQ6VXNlcjc5MTc2OTc1 | User | false |
openclaw/openclaw | 4,062,198,249 | I_kwDOQb6kR87yIDnp | 43,695 | https://github.com/openclaw/openclaw/issues/43695 | https://api.github.com/repos/openclaw/openclaw/issues/43695 | Cron delivery ignores threadId for Telegram forum topics | **Version:** OpenClaw 2026.3.8 (3caab92)
**Bug:** Cron job delivery with `threadId` in the delivery config doesn't respect the thread — messages land in General (topic 1) instead of the specified topic.
**Cron config:**
```json
"delivery": {
"mode": "announce",
"channel": "telegram",
"to": "-1003700845925",
"... | closed | completed | true | 1 | [] | [] | 2026-03-12T05:09:26Z | 2026-04-27T04:50:22Z | 2026-04-24T22:21:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | KeKs0r | 2,970,814 | MDQ6VXNlcjI5NzA4MTQ= | User | false |
openclaw/openclaw | 4,062,244,705 | I_kwDOQb6kR87yIO9h | 43,704 | https://github.com/openclaw/openclaw/issues/43704 | https://api.github.com/repos/openclaw/openclaw/issues/43704 | [Bug]: Feishu streaming card merges unrelated replies when agent produces multiple final messages | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When the agent produces multiple independent final replies (`replies=2`) in a single request, the Feishu streaming card incorrectly merges content from the second reply into the first card, while also sending the second reply as a separate ... | closed | completed | true | 1 | [
"bug"
] | [] | 2026-03-12T05:23:36Z | 2026-04-27T04:50:21Z | 2026-04-24T22:21:52Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cyrasafia | 9,101,789 | MDQ6VXNlcjkxMDE3ODk= | User | false |
openclaw/openclaw | 4,062,254,069 | I_kwDOQb6kR87yIRP1 | 43,706 | https://github.com/openclaw/openclaw/issues/43706 | https://api.github.com/repos/openclaw/openclaw/issues/43706 | [Bug/UX] Control UI loses gateway token after container rebuild | # [Bug/UX] Control UI loses gateway token after container rebuild, causing persistent token mismatch errors
After every container rebuild (upgrade, restart), the Control UI desktop app loses connection and displays: unauthorized: gateway token mismatch
The user must manually re-enter the gateway token each time, even... | closed | completed | true | 1 | [] | [] | 2026-03-12T05:25:53Z | 2026-04-27T04:50:20Z | 2026-04-24T22:21:57Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | YiweiHu3513 | 222,989,777 | U_kgDODUqN0Q | User | false |
openclaw/openclaw | 4,062,375,540 | I_kwDOQb6kR87yIu50 | 43,728 | https://github.com/openclaw/openclaw/issues/43728 | https://api.github.com/repos/openclaw/openclaw/issues/43728 | Heartbeat config written to invalid `gateway.heartbeat` location instead of `agents.defaults.heartbeat` | # GitHub Issue Draft
**Repository:** openclaw/openclaw
**Created:** 2026-03-12
**Status:** Ready to submit
---
## Title
Heartbeat config written to invalid `gateway.heartbeat` location instead of `agents.defaults.heartbeat`
## Labels
bug, config, regression
---
## Summary
When configuring heartbeat settings, O... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T05:54:55Z | 2026-04-27T04:50:20Z | 2026-04-24T22:21:59Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | joemcurry | 1,192,832 | MDQ6VXNlcjExOTI4MzI= | User | false |
openclaw/openclaw | 4,062,447,864 | I_kwDOQb6kR87yJAj4 | 43,741 | https://github.com/openclaw/openclaw/issues/43741 | https://api.github.com/repos/openclaw/openclaw/issues/43741 | Thinking blocks modified during compaction/session replay cause API rejection | ## Summary
When extended thinking is enabled (`thinking=adaptive` or `thinking=on`), Anthropic API calls fail after session compaction because thinking blocks are modified during the compaction/replay process. The API requires thinking blocks to be preserved verbatim.
## Steps to Reproduce
1. Configure an agent with... | closed | completed | true | 1 | [] | [] | 2026-03-12T06:12:02Z | 2026-04-27T04:50:19Z | 2026-04-24T22:22:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | intellectualcreations | 266,452,023 | U_kgDOD-G8Nw | User | false |
openclaw/openclaw | 4,062,474,590 | I_kwDOQb6kR87yJHFe | 43,745 | https://github.com/openclaw/openclaw/issues/43745 | https://api.github.com/repos/openclaw/openclaw/issues/43745 | [Skills] 37 "Skipping skill path that resolves outside its configured root" warnings from built-in extensions | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
```markdown
## 🐛 Bug Description
When running `openclaw doctor` or `openclaw skills list`, I get 37 warnings:
```
[skills] Skipping skill path that resolves outside its configured root.
```
### Steps to reproduce
These warnings come fr... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T06:19:35Z | 2026-04-27T04:50:18Z | 2026-04-24T22:22:04Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | mashuai333 | 24,646,750 | MDQ6VXNlcjI0NjQ2NzUw | User | false |
openclaw/openclaw | 4,062,494,532 | I_kwDOQb6kR87yJL9E | 43,753 | https://github.com/openclaw/openclaw/issues/43753 | https://api.github.com/repos/openclaw/openclaw/issues/43753 | Bug: 2026.3.11 fails to start — @mariozechner/pi-ai missing getOAuthApiKey / getEnvApiKey exports from main index | ## Environment
- OpenClaw: 2026.3.11 (from source, git tag `v2026.3.11`)
- `@mariozechner/pi-ai`: 0.57.1 (latest on npm)
- Node: v22.22.0
- OS: macOS 15 (arm64)
## Error
Immediately on any CLI invocation (`openclaw --version`, `openclaw doctor --fix`, etc.):
```
[openclaw] Failed to start CLI: file:///path/to/openc... | closed | completed | true | 2 | [] | [] | 2026-03-12T06:24:28Z | 2026-04-27T04:50:17Z | 2026-04-24T22:22:07Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | nickjlamb | 129,560,626 | U_kgDOB7jwMg | User | false |
openclaw/openclaw | 4,062,517,591 | I_kwDOQb6kR87yJRlX | 43,755 | https://github.com/openclaw/openclaw/issues/43755 | https://api.github.com/repos/openclaw/openclaw/issues/43755 | [Bug]: Subagent result not delivered to parent session (sessions_spawn with claude) | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When using `sessions_spawn` to call `claude` subagent, the result is captured by gateway but not delivered to parent session, showing `(no output)` instead of actual content.
### Steps to reproduce
1. Use `sessions_spawn(agentId="claude",... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T06:30:33Z | 2026-04-27T04:50:17Z | 2026-04-24T22:22:10Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aoguai | 34,203,474 | MDQ6VXNlcjM0MjAzNDc0 | User | false |
openclaw/openclaw | 4,062,610,597 | I_kwDOQb6kR87yJoSl | 43,768 | https://github.com/openclaw/openclaw/issues/43768 | https://api.github.com/repos/openclaw/openclaw/issues/43768 | [Bug]: sessions_spawn(runtime="subagent") does not honor agents.defaults.subagents.model or explicit model override | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
Native subagents spawned via `sessions_spawn(runtime="subagent")` ignore both `agents.defaults.subagents.model` and explicit `sessions_spawn.model`, and run on the caller/main model (`openai-codex/gpt-5.4`) instead of the requested `openai/... | closed | completed | true | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T06:54:22Z | 2026-04-27T04:50:15Z | 2026-04-24T22:22:15Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | GitsKleeNoob | 256,877,272 | U_kgDOD0-i2A | User | false |
openclaw/openclaw | 4,062,617,216 | I_kwDOQb6kR87yJp6A | 43,771 | https://github.com/openclaw/openclaw/issues/43771 | https://api.github.com/repos/openclaw/openclaw/issues/43771 | [Bug]: Telegram responses are truncated or do not show up | ### Bug type
Regression (worked before, now fails)
### Summary
When i send a message via telegram the response is either truncated or doesnt show up at all. It use to work before. The responses still show up on the web UI.
### Steps to reproduce
I asked the Agent to do a particular task. I thinks happens when it ... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-12T06:56:11Z | 2026-04-27T04:50:14Z | 2026-04-24T22:22:18Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | augmentedstartups | 4,731,260 | MDQ6VXNlcjQ3MzEyNjA= | User | false |
openclaw/openclaw | 4,062,627,343 | I_kwDOQb6kR87yJsYP | 43,772 | https://github.com/openclaw/openclaw/issues/43772 | https://api.github.com/repos/openclaw/openclaw/issues/43772 | /steer command for reliable mid-run subagent steering | ## Problem
Steering active subagents is unreliable:
- `subagents steer` fails with 'unknown target' even when session is active
- `sessions_send` times out frequently
- No user-facing command to steer — requires the agent to mediate
## Proposed Solution
Add a `/steer <message>` slash command that:
1. Finds the active ... | closed | completed | true | 1 | [] | [] | 2026-03-12T06:58:59Z | 2026-04-27T04:50:14Z | 2026-04-24T22:22:20Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | AlphaComposite | 25,360,554 | MDQ6VXNlcjI1MzYwNTU0 | User | false |
openclaw/openclaw | 4,062,633,422 | I_kwDOQb6kR87yJt3O | 43,774 | https://github.com/openclaw/openclaw/issues/43774 | https://api.github.com/repos/openclaw/openclaw/issues/43774 | ClawHub CLI missing license acceptance flag for publish command | ## Description
The ClawHub CLI (v0.7.0) is missing a flag to accept license terms when publishing skills. Users attempting to publish a skill receive an error:
```
acceptLicenseTerms: invalid value
```
This blocks skill publication entirely from the CLI.
## Reproduction Steps
1. Install ClawHub CLI v0.7.0
2. Attem... | closed | not_planned | true | 1 | [] | [] | 2026-03-12T07:00:27Z | 2026-04-27T04:50:13Z | 2026-04-24T22:22:23Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | willificent | 67,238,055 | MDQ6VXNlcjY3MjM4MDU1 | User | false |
openclaw/openclaw | 4,062,668,364 | I_kwDOQb6kR87yJ2ZM | 43,779 | https://github.com/openclaw/openclaw/issues/43779 | https://api.github.com/repos/openclaw/openclaw/issues/43779 | [Bug]: `openclaw message send --channel qqbot --target` fails with `ToolInputError: to required` | ## Bug Description
`openclaw message send --channel qqbot --target <openid>` fails with `ToolInputError: to required`, regardless of the target format used. The `--target` CLI flag is not being mapped to the `to` parameter for the QQ Bot extension channel.
## Environment
- OpenClaw version: 2026.3.8 (3caab92)
- `@op... | closed | completed | true | 4 | [] | [] | 2026-03-12T07:08:18Z | 2026-04-27T04:50:12Z | 2026-04-24T22:22:25Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | TecFancy | 24,548,536 | MDQ6VXNlcjI0NTQ4NTM2 | User | false |
openclaw/openclaw | 4,062,775,894 | I_kwDOQb6kR87yKQpW | 43,802 | https://github.com/openclaw/openclaw/issues/43802 | https://api.github.com/repos/openclaw/openclaw/issues/43802 | [Bug] Subagent completion notification loop causes excessive API calls | ## Bug Description
When using `sessions_spawn` with `delivery.mode: "none"`, the subagent completion notification gets repeatedly sent to the parent agent, causing an infinite loop of API calls.
## Environment
- OpenClaw version: 0.1.8-fix.3
- Model provider: Bailian (Qwen3.5-plus)
- Session target: isolated
## Ste... | closed | completed | true | 1 | [] | [] | 2026-03-12T07:29:35Z | 2026-04-27T04:50:11Z | 2026-04-24T22:22:28Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lkk66 | 45,393,751 | MDQ6VXNlcjQ1MzkzNzUx | User | false |
openclaw/openclaw | 4,062,782,338 | I_kwDOQb6kR87yKSOC | 43,804 | https://github.com/openclaw/openclaw/issues/43804 | https://api.github.com/repos/openclaw/openclaw/issues/43804 | [Bug] doctor --deep says no channel security warnings while security audit --deep reports criticals | ## Summary
`openclaw doctor --deep` reports:
> Security: No channel security warnings detected. Run: openclaw security audit --deep
But immediately running `openclaw security audit --deep` on the same config reports multiple severe findings (including channel exposure criticals).
This is misleading during post-upgra... | closed | completed | true | 1 | [] | [] | 2026-03-12T07:30:43Z | 2026-04-27T04:50:11Z | 2026-04-24T22:22:30Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Mdx2025 | 181,967,107 | U_kgDOCtiZAw | User | false |
openclaw/openclaw | 4,062,792,935 | I_kwDOQb6kR87yKUzn | 43,807 | https://github.com/openclaw/openclaw/issues/43807 | https://api.github.com/repos/openclaw/openclaw/issues/43807 | [Bug] Reasoning mode 'On' sends only thinking content, no final response in Telegram | ## Description
When using `/reasoning on` in Telegram, only the thinking content is displayed/sent, and the final response message is never delivered to the user.
## Environment
- OpenClaw version: 2026.3.8
- Channel: Telegram
- Model: MiniMax-M2.5 (via minimax-portal)
- Reasoning mode: On (via /reasoning command)
... | closed | completed | true | 1 | [] | [] | 2026-03-12T07:32:54Z | 2026-04-27T04:50:10Z | 2026-04-24T22:22:33Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | huhtreebot | 267,431,607 | U_kgDOD_Cutw | User | false |
openclaw/openclaw | 4,062,827,582 | I_kwDOQb6kR87yKdQ- | 43,813 | https://github.com/openclaw/openclaw/issues/43813 | https://api.github.com/repos/openclaw/openclaw/issues/43813 | [Bug]: plugin discovery scans into node_modules/browser_data causing FD exhaustion and spawn EBADF (reproducible on 2026.3.11) | ## Summary
`shouldIgnoreScannedDirectory()` in `src/plugins/discovery.ts` only skips `.bak`/`.backup-`/`.disabled` directories. It does not skip `node_modules`, `.git`, `.venv`, `browser_data`, or any other heavy dependency/build directories.
When a workspace skill or plugin contains `node_modules`, the plugin discov... | closed | completed | true | 2 | [] | [] | 2026-03-12T07:40:46Z | 2026-04-27T04:50:09Z | 2026-04-24T22:22:36Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | zhangyi-3 | 19,740,070 | MDQ6VXNlcjE5NzQwMDcw | User | false |
openclaw/openclaw | 4,062,888,299 | I_kwDOQb6kR87yKsFr | 43,828 | https://github.com/openclaw/openclaw/issues/43828 | https://api.github.com/repos/openclaw/openclaw/issues/43828 | [Bug]: Duplicated number omitted in TUI | ### Bug type
Behavior bug (incorrect output/state without crash)
### Summary
When Openclaw uses bc for calculation, the TUI output will omit duplicate number, but get normal when reopen the TUI
### Steps to reproduce
1. Install given [calculator skill](https://github.com/user-attachments/files/25928307/SKILL.md)
2... | closed | completed | true | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-03-12T07:54:38Z | 2026-04-27T04:50:08Z | 2026-04-24T22:22:38Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Thtgan | 37,025,083 | MDQ6VXNlcjM3MDI1MDgz | User | false |
openclaw/openclaw | 4,062,930,965 | I_kwDOQb6kR87yK2gV | 43,834 | https://github.com/openclaw/openclaw/issues/43834 | https://api.github.com/repos/openclaw/openclaw/issues/43834 | [Feature]: Add compaction.model config to use faster model for summarization | ### Summary
Add `agents.defaults.compaction.model` config option to specify a different (faster) model for compaction summarization.
### Problem to solve
Compaction uses the session's primary model for summarization. When the session uses a slow model like Opus, compaction becomes extremely slow:
- Each chunk summa... | closed | completed | true | 1 | [] | [] | 2026-03-12T08:05:17Z | 2026-04-27T04:50:07Z | 2026-04-24T22:22:41Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | stwith | 35,755,092 | MDQ6VXNlcjM1NzU1MDky | User | false |
openclaw/openclaw | 4,062,960,483 | I_kwDOQb6kR87yK9tj | 43,840 | https://github.com/openclaw/openclaw/issues/43840 | https://api.github.com/repos/openclaw/openclaw/issues/43840 | [Bug]: Filename Becomes URL-Encoded String After Sending Files in Lark/Feishu | ### Bug type
Regression (worked before, now fails)
### Summary
When sending a file (via attachment or upload) in Lark/Feishu, if the original filename contains non-ASCII characters (e.g., Chinese, Japanese, emoji, or special symbols), the file is received with a URL-encoded filename (e.g., %E6%B5%8B%E8%AF%95.txt) in... | closed | completed | true | 1 | [
"bug",
"regression"
] | [] | 2026-03-12T08:12:59Z | 2026-04-27T04:50:07Z | 2026-04-24T22:22:44Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | lijihua2017 | 33,613,942 | MDQ6VXNlcjMzNjEzOTQy | User | false |
openclaw/openclaw | 4,062,978,992 | I_kwDOQb6kR87yLCOw | 43,842 | https://github.com/openclaw/openclaw/issues/43842 | https://api.github.com/repos/openclaw/openclaw/issues/43842 | Browser Relay: tab attachment not persistent across Chrome restarts / sessions | ## Problem
The Browser Relay Chrome extension requires manual re-attachment every time Chrome restarts or the attached tab is closed. This makes it completely unusable for automated workflows (cron jobs, Scout agents, etc.).
## Expected behavior
Browser Relay tab attachment should either:
1. Persist across Chrome re... | closed | completed | true | 1 | [] | [] | 2026-03-12T08:17:04Z | 2026-04-27T04:50:06Z | 2026-04-24T22:22:47Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | Max-Resilient | 64,356,730 | MDQ6VXNlcjY0MzU2NzMw | User | false |
openclaw/openclaw | 4,063,017,750 | I_kwDOQb6kR87yLLsW | 43,850 | https://github.com/openclaw/openclaw/issues/43850 | https://api.github.com/repos/openclaw/openclaw/issues/43850 | [Bug]: Cron jobs with isolated session target timing out | ### Bug type
Regression (worked before, now fails)
### Summary
Cron jobs with sessionTarget=isolated timeout even when task is simple (regression)
### Steps to reproduce
1. Create a cron job with sessionTarget: "isolated" and payload.kind: "agentTurn"
2. Set a simple task (e.g., proactive check that takes ~20s)
3.... | closed | completed | true | 2 | [
"bug",
"regression"
] | [] | 2026-03-12T08:26:10Z | 2026-04-27T04:50:05Z | 2026-04-24T22:22:49Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | superbmw92-dotcom | 265,217,547 | U_kgDOD87mCw | User | false |
openclaw/openclaw | 4,063,021,053 | I_kwDOQb6kR87yLMf9 | 43,851 | https://github.com/openclaw/openclaw/issues/43851 | https://api.github.com/repos/openclaw/openclaw/issues/43851 | feat: Gateway response timeout watchdog for stale sessions | ## Problem
When a session receives a user message but fails to respond (due to hanging, context overflow, or silent failure), there is no mechanism to detect this. The gateway currently uses fire-and-forget delivery — once a message is dispatched to a session, there is no tracking of whether a response was produced.
... | closed | completed | true | 2 | [] | [] | 2026-03-12T08:26:56Z | 2026-04-27T04:37:00Z | 2026-04-24T22:22:52Z | CONTRIBUTOR | null | 20260427T090020Z | 2026-04-27T09:00:20Z | p3nchan | 5,032,148 | MDQ6VXNlcjUwMzIxNDg= | User | false |
openclaw/openclaw | 4,063,043,150 | I_kwDOQb6kR87yLR5O | 43,854 | https://github.com/openclaw/openclaw/issues/43854 | https://api.github.com/repos/openclaw/openclaw/issues/43854 | compaction.model config not passed through safeguard compaction path | ## Bug
`agents.defaults.compaction.model` is set to `anthropic/claude-sonnet-4-6` but compaction always uses the session's primary model instead.
## Root cause
In `reply-DeXK9BLT.js` line 38904, the safeguard compaction path calls:
```js
session.compact(params.customInstructions)
```
The legacy context engine's `c... | closed | completed | true | 2 | [] | [] | 2026-03-12T08:32:27Z | 2026-04-27T04:50:05Z | 2026-04-24T22:22:55Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | allanjeng | 14,968,942 | MDQ6VXNlcjE0OTY4OTQy | User | false |
openclaw/openclaw | 4,063,046,020 | I_kwDOQb6kR87yLSmE | 43,856 | https://github.com/openclaw/openclaw/issues/43856 | https://api.github.com/repos/openclaw/openclaw/issues/43856 | bug(telegram): main bot provider runs as accountId='default' despite token being in named account, requiring group config duplication | ## Summary
When a Telegram bot token is stored in a named account (e.g. `channels.telegram.accounts.rex-system`), the running provider is labeled `[default]` in gateway logs and resolves group/topic policies using `accountId='default'` — not `'rex-system'`. This means any group config stored under the named account is... | closed | completed | true | 1 | [] | [] | 2026-03-12T08:33:05Z | 2026-04-27T04:50:04Z | 2026-04-24T22:22:57Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | rex05ai | 258,758,601 | U_kgDOD2xXyQ | User | false |
openclaw/openclaw | 4,063,046,053 | I_kwDOQb6kR87yLSml | 43,857 | https://github.com/openclaw/openclaw/issues/43857 | https://api.github.com/repos/openclaw/openclaw/issues/43857 | [Bug]:Dashboard Chat page freezes browser — infinite "Loading chat…" on v2026.3.8 and v2026.3.11-beta.1 | ### Bug type
Regression (worked before, now fails)
### Summary
## Description
The Dashboard **Chat page** freezes the browser tab completely after navigating to it. The page shows "Loading chat…" and then becomes unresponsive ("This page is not responding").
The **Overview page works perfectly** — only the Chat p... | closed | completed | true | 3 | [
"bug",
"regression"
] | [] | 2026-03-12T08:33:05Z | 2026-04-27T04:50:03Z | 2026-04-24T22:23:00Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | cha772 | 187,355,305 | U_kgDOCyrQqQ | User | false |
openclaw/openclaw | 4,063,047,778 | I_kwDOQb6kR87yLTBi | 43,858 | https://github.com/openclaw/openclaw/issues/43858 | https://api.github.com/repos/openclaw/openclaw/issues/43858 | Edit tool silently wipes file to 0 bytes while reporting success | ## Bug Description
The `Edit` (file edit / replace exact text) tool silently wipes the target file to 0 bytes while reporting "Successfully replaced text in <file>." The old text is found and matched correctly, but instead of performing an in-place replacement, the entire file content is lost.
## Reproduction
**Envi... | closed | completed | true | 12 | [] | [] | 2026-03-12T08:33:27Z | 2026-04-27T04:50:02Z | 2026-04-24T22:23:02Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | aaajiao | 49,885 | MDQ6VXNlcjQ5ODg1 | User | false |
openclaw/openclaw | 4,063,169,953 | I_kwDOQb6kR87yLw2h | 43,872 | https://github.com/openclaw/openclaw/issues/43872 | https://api.github.com/repos/openclaw/openclaw/issues/43872 | Bug: openclaw browser open can crash OpenClaw.app via browser.proxy / MacNodeBrowserProxy.makeRequest | # Bug: `openclaw browser open` can crash OpenClaw.app via `browser.proxy` / `MacNodeBrowserProxy.makeRequest`
## Summary
On macOS, `openclaw browser open <url>` can crash `OpenClaw.app` instead of returning a normal browser/tool error.
The failure reproduces with both:
- the managed `openclaw` browser profile
- the... | closed | completed | true | 3 | [] | [] | 2026-03-12T08:58:45Z | 2026-04-27T04:50:02Z | 2026-04-24T22:23:05Z | NONE | null | 20260427T090020Z | 2026-04-27T09:00:20Z | caojm93 | 7,261,500 | MDQ6VXNlcjcyNjE1MDA= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.