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 | 4,372,452,044 | I_kwDOQb6kR88AAAABBJ5SzA | 76,836 | https://github.com/openclaw/openclaw/issues/76836 | https://api.github.com/repos/openclaw/openclaw/issues/76836 | [Bug]: Assistant reply fragments appear out of order when streamed across multiple tool calls | ## Bug Description
When the assistant produces multiple messages in quick succession (e.g., one per tool call) the resulting Telegram bubbles are sometimes delivered in the wrong order, causing the user to miss the final answer because it appears above the intermediate steps.
## Environment
- **Channel:** Telegram (d... | closed | completed | false | 1 | [
"bug",
"regression"
] | [] | 2026-05-03T17:16:00Z | 2026-05-03T17:22:06Z | 2026-05-03T17:22:07Z | NONE | null | 20260503T180029Z | 2026-05-03T18:00:29Z | atosonik | 266,384,676 | U_kgDOD-C1JA | User | false |
openclaw/openclaw | 4,372,459,670 | I_kwDOQb6kR88AAAABBJ5wlg | 76,838 | https://github.com/openclaw/openclaw/issues/76838 | https://api.github.com/repos/openclaw/openclaw/issues/76838 | [Bug]: Agent ignores MEMORY.md instructions and falls back to default TTS behavior | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
The agent loads MEMORY.md at session startup but does not consistently follow instructions stored there. Specifically, instructions to use local Piper TTS instead of [[tts:...]] commands are ignored after some ... | open | null | false | 3 | [
"bug"
] | [] | 2026-05-03T17:19:29Z | 2026-05-03T21:36:16Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | orianflaust | 59,671,213 | MDQ6VXNlcjU5NjcxMjEz | User | false |
openclaw/openclaw | 4,372,419,336 | I_kwDOQb6kR88AAAABBJ3TCA | 76,829 | https://github.com/openclaw/openclaw/issues/76829 | https://api.github.com/repos/openclaw/openclaw/issues/76829 | [Bug]: Session transcript corruption causes provider cooldown cascade, blocks all requests | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Bug
Corrupted session transcript (blank user messages) causes an unrecoverable error loop that blocks all requests until cooldown expires.
## Logs
[agent/embedded] session file repaired: rewrote 1 assistant message(s)... | open | null | false | 1 | [
"bug"
] | [] | 2026-05-03T17:03:27Z | 2026-05-04T20:34:08Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | mlitanin | 4,940,335 | MDQ6VXNlcjQ5NDAzMzU= | User | false |
openclaw/openclaw | 4,372,172,809 | I_kwDOQb6kR88AAAABBJoQCQ | 76,770 | https://github.com/openclaw/openclaw/issues/76770 | https://api.github.com/repos/openclaw/openclaw/issues/76770 | Feature: expose defer_loading flag for first-party MCP tool schemas | ## Context
OpenClaw exposes ~13 first-party `mcp__openclaw__*` tools (cron, image_generate, memory_get, memory_search, music_generate, session_status, sessions_history, sessions_list, sessions_send, sessions_spawn, sessions_yield, subagents, video_generate, web_fetch, web_search). These appear hard-baked into the runt... | open | null | false | 1 | [] | [] | 2026-05-03T15:20:11Z | 2026-05-03T19:33:39Z | null | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | datus1982 | 278,869,131 | U_kgDOEJ80iw | User | false |
openclaw/openclaw | 4,372,416,092 | I_kwDOQb6kR88AAAABBJ3GXA | 76,828 | https://github.com/openclaw/openclaw/issues/76828 | https://api.github.com/repos/openclaw/openclaw/issues/76828 | [Bug]: deliveryMode="final_only" silently drops long tool-loop replies when no phase=final chunk is emitted | ## Summary
On OpenClaw `2026.5.2 (8b2a6e5)` with `acp.stream.deliveryMode = "final_only"` and a Telegram-backed agent session, when an agent turn includes a long tool loop (8+ tool calls, ~15+ minutes of work) the assistant emits a sequence of legitimate progress-narration TEXT chunks — but every chunk has `textSignat... | closed | completed | false | 2 | [] | [] | 2026-05-03T17:02:31Z | 2026-05-04T10:30:32Z | 2026-05-04T10:30:32Z | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | sene1337 | 261,265,124 | U_kgDOD5KW5A | User | false |
openclaw/openclaw | 4,372,489,476 | I_kwDOQb6kR88AAAABBJ7lBA | 76,848 | https://github.com/openclaw/openclaw/issues/76848 | https://api.github.com/repos/openclaw/openclaw/issues/76848 | WebUI /new dashboard sessions should get friendly labels automatically | # Issue draft: WebUI `/new` dashboard sessions should get friendly labels automatically
## Title
WebUI `/new` creates dashboard-scoped sessions without friendly labels, making `openclaw sessions` hard to read
## Summary
In OpenClaw 2026.5.2, using `/new` in the WebUI / Control UI creates a new dashboard-scoped sessi... | open | null | false | 1 | [] | [] | 2026-05-03T17:33:32Z | 2026-05-04T01:06:50Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | sunnylearnai | 271,660,640 | U_kgDOEDE2YA | User | false |
openclaw/openclaw | 4,372,472,680 | I_kwDOQb6kR88AAAABBJ6jaA | 76,844 | https://github.com/openclaw/openclaw/issues/76844 | https://api.github.com/repos/openclaw/openclaw/issues/76844 | Windows: EPERM fsync error when uploading .docx files via WebChat | ## Describe the bug
Uploading .docx (Word) files via WebChat results in a MediaOffloadError with EPERM: operation not permitted, fsync. The file cannot be saved to disk.
## To Reproduce
1. Upload a .docx file to the OpenClaw WebChat UI
2. The gateway logs show:
`
Failed to save intercepted media to disk: EPERM: ope... | closed | completed | false | 2 | [] | [] | 2026-05-03T17:25:34Z | 2026-05-04T03:26:59Z | 2026-05-04T03:26:59Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | DrJinFX | 128,752,072 | U_kgDOB6yZyA | User | false |
openclaw/openclaw | 4,372,506,898 | I_kwDOQb6kR88AAAABBJ8pEg | 76,850 | https://github.com/openclaw/openclaw/issues/76850 | https://api.github.com/repos/openclaw/openclaw/issues/76850 | Redacted | Opened in the wrong tracker with environment-specific details. Redacted.
| closed | not_planned | false | 1 | [] | [] | 2026-05-03T17:41:16Z | 2026-05-03T17:47:01Z | 2026-05-03T17:47:01Z | NONE | null | 20260503T180029Z | 2026-05-03T18:00:29Z | przadka | 585,411 | MDQ6VXNlcjU4NTQxMQ== | User | false |
openclaw/openclaw | 4,372,461,274 | I_kwDOQb6kR88AAAABBJ522g | 76,841 | https://github.com/openclaw/openclaw/issues/76841 | https://api.github.com/repos/openclaw/openclaw/issues/76841 | [Feature]: Integrate Stigmem for Persistent Multi-Agent/Multi-Org Memory | ### Summary
I am the founder of Stigmem, an open specification and reference implementation for a federated knowledge fabric: a shared, persistent layer where AI agents and humans write typed, traceable facts that travel across tools, platforms, and organizations. Every fact is an immutable record — (entity, relation,... | closed | not_planned | false | 1 | [
"enhancement"
] | [] | 2026-05-03T17:20:11Z | 2026-05-03T17:49:31Z | 2026-05-03T17:49:31Z | NONE | null | 20260503T180029Z | 2026-05-03T18:00:29Z | offbyonce | 26,172,112 | MDQ6VXNlcjI2MTcyMTEy | User | false |
openclaw/openclaw | 4,372,490,147 | I_kwDOQb6kR88AAAABBJ7now | 76,849 | https://github.com/openclaw/openclaw/issues/76849 | https://api.github.com/repos/openclaw/openclaw/issues/76849 | Feature: ship Codex Desktop bootstrap bridge for AGENTS.md and SOUL.md | ## Summary
OpenClaw should ship a built-in bridge for users who want Codex Desktop sessions to inherit their OpenClaw workspace identity and operating context (`AGENTS.md` + `SOUL.md`) without maintaining a separate local plugin by hand.
OpenClaw already treats workspace bootstrap files as first-class agent context, ... | closed | not_planned | false | 1 | [] | [] | 2026-05-03T17:33:49Z | 2026-05-03T17:49:35Z | 2026-05-03T17:49:35Z | CONTRIBUTOR | null | 20260503T180029Z | 2026-05-03T18:00:29Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,372,517,768 | I_kwDOQb6kR88AAAABBJ9TiA | 76,853 | https://github.com/openclaw/openclaw/issues/76853 | https://api.github.com/repos/openclaw/openclaw/issues/76853 | active-memory plugin: Bundled plugin dirName error with QQ session keys containing colons | ## Problem Description
After upgrading to 2026.5.2, active-memory plugin lane task reports error:
```
Error: Bundled plugin dirName must be a single directory: c2c:*****09
```
## Error Analysis
The session key format is `c2c:*****09` (QQ session), which contains a colon. However, the validation logic in `public-sur... | closed | completed | false | 1 | [] | [] | 2026-05-03T17:45:42Z | 2026-05-03T17:55:04Z | 2026-05-03T17:50:36Z | NONE | null | 20260503T180029Z | 2026-05-03T18:00:29Z | ijevin | 2,133,499 | MDQ6VXNlcjIxMzM0OTk= | User | false |
openclaw/openclaw | 4,372,511,945 | I_kwDOQb6kR88AAAABBJ88yQ | 76,852 | https://github.com/openclaw/openclaw/issues/76852 | https://api.github.com/repos/openclaw/openclaw/issues/76852 | [Bug]: 2026.5.2: Periodic Telegram getMe 10s fetch-timeout storm on networks without IPv6 egress (regression vs 2026.4.26) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On hosts where DNS returns AAAA records but the network has no working IPv6 egress, OpenClaw 2026.5.2 emits a continuous storm of [fetch-timeout] log
lines for periodic https://api.telegram.org/bot.../getMe calls every ~6... | open | null | false | 1 | [
"bug",
"regression"
] | [] | 2026-05-03T17:43:10Z | 2026-05-03T22:24:23Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | jplavoiemtl | 36,643,005 | MDQ6VXNlcjM2NjQzMDA1 | User | false |
openclaw/openclaw | 4,372,533,408 | I_kwDOQb6kR88AAAABBJ-QoA | 76,855 | https://github.com/openclaw/openclaw/issues/76855 | https://api.github.com/repos/openclaw/openclaw/issues/76855 | [Bug]: Microsoft TTS on Telegram uses the wrong voice despite correct voice configuration | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Microsoft TTS is configured with voice `de-DE-ConradNeural`. When creating an audiofile directly to Telegram, the set voice is ignored and a female voice is used. This only happens the Telegram TTS flow, while ... | closed | completed | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-05-03T17:52:42Z | 2026-05-03T18:30:09Z | 2026-05-03T18:30:09Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | Jean-Luc-Bot | 266,308,814 | U_kgDOD9-Mzg | User | false |
openclaw/openclaw | 4,372,544,865 | I_kwDOQb6kR88AAAABBJ-9YQ | 76,860 | https://github.com/openclaw/openclaw/issues/76860 | https://api.github.com/repos/openclaw/openclaw/issues/76860 | 2026.5.2 upgrade silently wipes operator-customized gateway.systemd.env contents (EnvironmentFile contents reverted to bundled defaults) | ## Summary
Upgrading from 2026.4.29 to 2026.5.2 (and/or running `openclaw doctor` post-upgrade) silently rewrites `~/.openclaw/gateway.systemd.env` — the file declared as the gateway service's `EnvironmentFile=` in the systemd user unit. Operator-populated secrets are removed; only a small bundled-default subset (in o... | closed | completed | false | 2 | [] | [] | 2026-05-03T17:57:47Z | 2026-05-03T23:35:21Z | 2026-05-03T23:35:21Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | liemnhoang | 58,699,349 | MDQ6VXNlcjU4Njk5MzQ5 | User | false |
openclaw/openclaw | 4,372,541,165 | I_kwDOQb6kR88AAAABBJ-u7Q | 76,857 | https://github.com/openclaw/openclaw/issues/76857 | https://api.github.com/repos/openclaw/openclaw/issues/76857 | [Bug]: 2026.5.2: openai-codex agent runs fail with "LLM request failed: network connection error" on networks without IPv6 egress (regression vs 2026.4.26) | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
On hosts where DNS returns AAAA records but the network has no working IPv6 egress, OpenClaw 2026.5.2 agent runs against the openai-codex provider
repeatedly fail with `LLM request failed: network connection error. rawErro... | closed | completed | false | 2 | [
"bug",
"regression"
] | [] | 2026-05-03T17:56:09Z | 2026-05-04T01:47:34Z | 2026-05-04T01:47:34Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | jplavoiemtl | 36,643,005 | MDQ6VXNlcjM2NjQzMDA1 | User | false |
openclaw/openclaw | 4,372,545,075 | I_kwDOQb6kR88AAAABBJ--Mw | 76,861 | https://github.com/openclaw/openclaw/issues/76861 | https://api.github.com/repos/openclaw/openclaw/issues/76861 | Anthropic reasoning models (claude-opus-4-7, claude-sonnet-4-6) reject temperature parameter; gateway should strip it for reasoning:true models | ## Summary
Anthropic's reasoning models (`claude-opus-4-7`, `claude-sonnet-4-6` — both have `reasoning: true` in the gateway's catalog metadata) deprecated the `temperature` parameter. The gateway forwards `temperature: 0` (or whatever the agent's `params.temperature` is) to ANY model invocation, including these reaso... | open | null | false | 1 | [] | [] | 2026-05-03T17:57:53Z | 2026-05-03T19:33:52Z | null | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | liemnhoang | 58,699,349 | MDQ6VXNlcjU4Njk5MzQ5 | User | false |
openclaw/openclaw | 4,372,549,985 | I_kwDOQb6kR88AAAABBJ_RYQ | 76,865 | https://github.com/openclaw/openclaw/issues/76865 | https://api.github.com/repos/openclaw/openclaw/issues/76865 | dev/source -> stable cutover can leave service wrapper and plugin runtime state inconsistent | ## Summary
During a local dev/source -> stable package cutover to OpenClaw `2026.5.2` on macOS arm64, the gateway became partially inconsistent across service wrapper state, bundled/plugin channel loading, and plugin install metadata. The machine recovered after local manual repair, but the failure mode suggests the u... | open | null | false | 1 | [] | [] | 2026-05-03T17:59:57Z | 2026-05-04T12:27:09Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Secr-et | 150,955,236 | U_kgDOCP9k5A | User | false |
openclaw/openclaw | 4,372,546,488 | I_kwDOQb6kR88AAAABBJ_DuA | 76,862 | https://github.com/openclaw/openclaw/issues/76862 | https://api.github.com/repos/openclaw/openclaw/issues/76862 | [Feature]: Do not close tickets unless the code has been released in a stable version. | ### Summary
I’ve opened several issues, but many were closed with remarks that they’re “in main.” If “main” means the released version, then the issues aren’t fixed. If it means there’s a fix pending release, the tickets should stay open until the fix is actually out, ideally with a follow-up question to confirm the i... | closed | not_planned | false | 2 | [
"enhancement"
] | [] | 2026-05-03T17:58:30Z | 2026-05-03T20:31:27Z | 2026-05-03T20:31:27Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,372,547,536 | I_kwDOQb6kR88AAAABBJ_H0A | 76,863 | https://github.com/openclaw/openclaw/issues/76863 | https://api.github.com/repos/openclaw/openclaw/issues/76863 | [Bug]: 04.29 install freezes at TUI | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
2026.04.29 freezes at the last step of installation, at the launch of TUI.
### Steps to reproduce
Install 2026.04.29. Does not relate to any install options selected. It does this on global node and nvm as well. Does it on v... | closed | completed | false | 2 | [
"bug"
] | [] | 2026-05-03T17:58:52Z | 2026-05-03T20:46:27Z | 2026-05-03T20:46:27Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | vhunyad | 17,292,468 | MDQ6VXNlcjE3MjkyNDY4 | User | false |
openclaw/openclaw | 4,372,570,826 | I_kwDOQb6kR88AAAABBKAiyg | 76,867 | https://github.com/openclaw/openclaw/issues/76867 | https://api.github.com/repos/openclaw/openclaw/issues/76867 | [Bug]: Appearance → Import from tweakcn fails with NetworkError | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Pasting a tweakcn.com theme URL into Appearance → Theme → "Import from tweakcn" immediately fails with `NetworkError when attempting to fetch resource.` and no theme is applied.
### Steps to reproduce
1. Open... | closed | not_planned | false | 1 | [] | [] | 2026-05-03T18:08:52Z | 2026-05-03T18:15:07Z | 2026-05-03T18:15:07Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | jimdawdy-hub | 262,052,366 | U_kgDOD56aDg | User | false |
openclaw/openclaw | 4,372,577,980 | I_kwDOQb6kR88AAAABBKA-vA | 76,868 | https://github.com/openclaw/openclaw/issues/76868 | https://api.github.com/repos/openclaw/openclaw/issues/76868 | [Feature]: enable use of arbitrary extra parameters for local LLMs using OpenAI-compatible API | ### Summary
Remove the limitation of just extra_tokens and temperature to the parameters possible to pass to local LLMs using OpenAI API (custom provider)
### Problem to solve
Many models require use of extra parameters to properly make use of their capabilities, including controlling reasoning and working optimally... | closed | completed | false | 1 | [
"enhancement"
] | [] | 2026-05-03T18:12:10Z | 2026-05-03T18:17:01Z | 2026-05-03T18:17:01Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | syncword | 158,204,276 | U_kgDOCW4BdA | User | false |
openclaw/openclaw | 4,372,587,442 | I_kwDOQb6kR88AAAABBKBjsg | 76,872 | https://github.com/openclaw/openclaw/issues/76872 | https://api.github.com/repos/openclaw/openclaw/issues/76872 | doctor and gateway disagree about plugins.allow when channels.discord is enabled | ## Summary
Doctor reports `plugins.allow: plugin not found: discord (stale config entry ignored; remove it from plugins config)` as a config warning. However, every time the gateway starts, it automatically re-adds `discord` to `plugins.allow` based on the presence of `channels.discord.enabled: true`, making the docto... | closed | completed | false | 2 | [] | [] | 2026-05-03T18:16:32Z | 2026-05-03T21:44:22Z | 2026-05-03T21:44:22Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | groksrc | 9,099,311 | MDQ6VXNlcjkwOTkzMTE= | User | false |
openclaw/openclaw | 4,372,594,339 | I_kwDOQb6kR88AAAABBKB-ow | 76,875 | https://github.com/openclaw/openclaw/issues/76875 | https://api.github.com/repos/openclaw/openclaw/issues/76875 | [Bug]: Agents for some reason now prioritize more on previous messages than current ones | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Upgrading from 4.23 to 5.2, a behavior I found on agents is that the agents somehow focused more on previous messages rather than latest user message.
### Steps to reproduce
1. Start with 4.23 and try to have... | closed | not_planned | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-05-03T18:19:50Z | 2026-05-04T04:30:53Z | 2026-05-04T01:08:56Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | Deskehs | 124,361,319 | U_kgDOB2maZw | User | false |
openclaw/openclaw | 4,372,612,774 | I_kwDOQb6kR88AAAABBKDGpg | 76,877 | https://github.com/openclaw/openclaw/issues/76877 | https://api.github.com/repos/openclaw/openclaw/issues/76877 | [Bug]: 2026.5.2 Agents stop responding mid work, they use tooling and then suddenly nothing.. | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
2026.5.2 Agents stop responding mid-work; they use tooling, and then suddenly nothing until you ask for progress. Due to a lot of bugs, I could not run anything newer than version 2026.04-23 did not have this agent freeze in... | open | null | false | 1 | [
"bug",
"regression"
] | [] | 2026-05-03T18:28:16Z | 2026-05-03T23:49:48Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,372,631,806 | I_kwDOQb6kR88AAAABBKEQ_g | 76,880 | https://github.com/openclaw/openclaw/issues/76880 | https://api.github.com/repos/openclaw/openclaw/issues/76880 | [Bug]: googlechat broken in 2026-05-01 (working in 2026-04-29) Google Chat webhook auth rejected: Failed to retrieve verification certificates: config.headers.has is not a function -- 76742 was not resolved, reopen | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
#76742 was closed as resolved. it does not, i just tested it, the log shows:
2026-05-03T18:33:34.558Z info channels/googlechat {"subsystem":"channels/googlechat"} [default] Google Chat webhook auth rejected: res?.headers.ge... | closed | completed | false | 3 | [
"bug",
"regression"
] | [] | 2026-05-03T18:37:11Z | 2026-05-04T12:16:12Z | 2026-05-04T09:03:21Z | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | donbowman | 5,131,923 | MDQ6VXNlcjUxMzE5MjM= | User | false |
openclaw/openclaw | 4,372,668,475 | I_kwDOQb6kR88AAAABBKGgOw | 76,888 | https://github.com/openclaw/openclaw/issues/76888 | https://api.github.com/repos/openclaw/openclaw/issues/76888 | [Bug]: Queued/orphaned user-message merge can produce stale reply | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
A queued/merged user message run used the latest prompt text, but the model answered an earlier unresolved user turn from the same session, producing a stale visible reply.
### Steps to reproduce
1. In a chan... | open | null | false | 1 | [] | [] | 2026-05-03T18:53:20Z | 2026-05-04T20:34:10Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | hoyanhan | 45,415,044 | MDQ6VXNlcjQ1NDE1MDQ0 | User | false |
openclaw/openclaw | 4,372,671,736 | I_kwDOQb6kR88AAAABBKGs-A | 76,890 | https://github.com/openclaw/openclaw/issues/76890 | https://api.github.com/repos/openclaw/openclaw/issues/76890 | memory-core: synchronous indexing on session start blocks gateway event loop for 30+ seconds | ## Summary
When a new session is created (session key not present in `sessions.json`), some path in `memory-core` plugin and/or `agents.defaults.memorySearch.sync.onSessionStart` causes the gateway Node.js event loop to block synchronously for **30+ seconds**, fully saturating one CPU core. During this window the gate... | open | null | false | 1 | [] | [] | 2026-05-03T18:54:51Z | 2026-05-03T23:43:04Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | huangyingw | 91,757 | MDQ6VXNlcjkxNzU3 | User | false |
openclaw/openclaw | 4,372,671,769 | I_kwDOQb6kR88AAAABBKGtGQ | 76,891 | https://github.com/openclaw/openclaw/issues/76891 | https://api.github.com/repos/openclaw/openclaw/issues/76891 | sessions.list takes 30+ seconds for first request on a brand-new session key (single-entry store) | ## Summary
`sessions.list` RPC takes 28-34 seconds on the gateway when the requested `--session <KEY>` is **not yet present** in `agents/<agent>/sessions/sessions.json`. The store has only ONE existing session entry (`agent:main:main`), so a 33-second response time for listing it is unexpected.
For an *existing* sess... | closed | completed | false | 1 | [] | [] | 2026-05-03T18:54:52Z | 2026-05-03T18:59:59Z | 2026-05-03T18:59:59Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | huangyingw | 91,757 | MDQ6VXNlcjkxNzU3 | User | false |
openclaw/openclaw | 4,372,591,608 | I_kwDOQb6kR88AAAABBKBz-A | 76,874 | https://github.com/openclaw/openclaw/issues/76874 | https://api.github.com/repos/openclaw/openclaw/issues/76874 | WebUI Stop button sometimes remains active after assistant reply has completed | # Issue draft: WebUI Stop button can stay active after assistant reply is already visible
## Title
WebUI Stop button sometimes remains active after the assistant reply has completed
## Summary
In OpenClaw 2026.5.2, the WebUI / Control UI sometimes keeps showing the red **Stop** button even after the assistant reply ... | open | null | false | 2 | [] | [] | 2026-05-03T18:18:30Z | 2026-05-03T21:50:09Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | sunnylearnai | 271,660,640 | U_kgDOEDE2YA | User | false |
openclaw/openclaw | 4,372,682,761 | I_kwDOQb6kR88AAAABBKHYCQ | 76,892 | https://github.com/openclaw/openclaw/issues/76892 | https://api.github.com/repos/openclaw/openclaw/issues/76892 | [Bug]: Control UI Chat History Display Issues | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
1. **Duplicate messages** — during agent streaming output, the UI renders multiple overlapping chat bubbles with truncated/partial text.
2. **History loss on refresh** — after refreshing the page, the entire chat conversatio... | closed | not_planned | false | 3 | [
"bug",
"regression"
] | [] | 2026-05-03T19:00:09Z | 2026-05-03T21:49:44Z | 2026-05-03T21:49:44Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | yusunz | 227,024,514 | U_kgDODYgegg | User | false |
openclaw/openclaw | 4,372,635,190 | I_kwDOQb6kR88AAAABBKEeNg | 76,882 | https://github.com/openclaw/openclaw/issues/76882 | https://api.github.com/repos/openclaw/openclaw/issues/76882 | [Bug]: denyCommands performs exact command-name matching but does not filter command payloads | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
## Description
Running `openclaw security audit` surfaces the warning `gateway.nodes.deny_commands_ineffective`, which notes that `denyCommands` uses exact command-name matching rather than shell-text filterin... | closed | not_planned | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-05-03T18:38:46Z | 2026-05-03T19:12:10Z | 2026-05-03T19:12:10Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | hoo01 | 169,421,229 | U_kgDOChkprQ | User | false |
openclaw/openclaw | 4,372,670,852 | I_kwDOQb6kR88AAAABBKGphA | 76,889 | https://github.com/openclaw/openclaw/issues/76889 | https://api.github.com/repos/openclaw/openclaw/issues/76889 | [Bug]: Discord named-account status/startup fails to resolve token SecretRefs from active runtime snapshot | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
Discord channel startup and channels.status can fail for named Discord accounts when the account token is configured as an environment-backed SecretRef, even when the value exists in .openclaw/.env.
The fail... | closed | completed | false | 3 | [
"bug"
] | [] | 2026-05-03T18:54:27Z | 2026-05-04T21:15:39Z | 2026-05-03T22:56:38Z | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | khaney64 | 13,684,640 | MDQ6VXNlcjEzNjg0NjQw | User | false |
openclaw/openclaw | 4,372,579,193 | I_kwDOQb6kR88AAAABBKBDeQ | 76,869 | https://github.com/openclaw/openclaw/issues/76869 | https://api.github.com/repos/openclaw/openclaw/issues/76869 | CLI backend (claude-cli) does not stream text deltas to Telegram preview — only native API path is wired | ## Summary
When using `cliBackends.claude-cli` (Anthropic Max plan via OAuth) as the agent backend, Telegram channel shows no live text-streaming preview during model response generation. The same agent on the native API path (`anthropic/claude-*-4-6` direct) streams correctly.
## Environment
- OpenClaw: 2026.5.2 (als... | open | null | false | 4 | [] | [] | 2026-05-03T18:12:44Z | 2026-05-04T00:52:32Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | kai-jar | 257,655,971 | U_kgDOD1uEow | User | false |
openclaw/openclaw | 4,372,724,556 | I_kwDOQb6kR88AAAABBKJ7TA | 76,897 | https://github.com/openclaw/openclaw/issues/76897 | https://api.github.com/repos/openclaw/openclaw/issues/76897 | [Bug]: Telegram drops text when assistant reply contains text + MEDIA: (text swallowed, only file delivered) | ### Version
OpenClaw 2026.4.23 (also reproduced on 2026.4.25)
### Affected surface
Telegram direct chat
### Summary
When an assistant reply contains both visible text and a `MEDIA:` attachment in the same text block, Telegram receives only the file attachment. The text portion is silently swallowed — it does not appe... | closed | completed | false | 2 | [] | [] | 2026-05-03T19:18:58Z | 2026-05-03T19:41:18Z | 2026-05-03T19:23:43Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | rollingshmily | 62,282,098 | MDQ6VXNlcjYyMjgyMDk4 | User | false |
openclaw/openclaw | 4,372,766,111 | I_kwDOQb6kR88AAAABBKMdnw | 76,900 | https://github.com/openclaw/openclaw/issues/76900 | https://api.github.com/repos/openclaw/openclaw/issues/76900 | BUG: Discord plugin crashes gateway when channel status/health resolves an unresolved SecretRef token | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
When the Discord plugin is enabled and channels.discord.token is configured as a SecretRef-backed token, Discord account/status resolution can receive the unresolved SecretRef instead of a resolved token value. The Discord toke... | closed | completed | false | 2 | [
"bug",
"bug:crash"
] | [] | 2026-05-03T19:37:24Z | 2026-05-03T22:57:53Z | 2026-05-03T22:56:38Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | kAIborg24 | 261,855,322 | U_kgDOD5uYWg | User | false |
openclaw/openclaw | 4,035,785,974 | I_kwDOQb6kR87wjTT2 | 38,283 | https://github.com/openclaw/openclaw/issues/38283 | https://api.github.com/repos/openclaw/openclaw/issues/38283 | PR Limit Update: Why We Now Cap at 20 Open PRs Per Author | We changed the limit to 20 open PRs per author. If someone goes over 20, the `r: too-many-prs` label gets added and the PR is auto‑closed by our automation. This is a hard limit.
We’ve been hit with waves of spammy PRs: AI slop posted across dozens of PRs, duplicates of the same change, and low‑effort noise that burns... | closed | completed | true | 0 | [] | [] | 2026-03-06T19:24:26Z | 2026-05-03T19:45:09Z | 2026-04-04T20:17:12Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | barnacle-openclaw[bot] | 264,559,031 | BOT_kgDOD8TZtw | Bot | false |
openclaw/openclaw | 4,372,776,485 | I_kwDOQb6kR88AAAABBKNGJQ | 76,905 | https://github.com/openclaw/openclaw/issues/76905 | https://api.github.com/repos/openclaw/openclaw/issues/76905 | [Bug]: Stale Tool-Turn Finalization Delivered After Newer WhatsApp Message | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading OpenClaw from `2026.4.29` to `2026.5.2`, WhatsApp direct-message replies can become stale/out-of-order. A previous tool-backed assistant turn may finish after a newer user message arrives, and that old final ... | open | null | false | 4 | [
"bug"
] | [] | 2026-05-03T19:42:19Z | 2026-05-04T20:34:12Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | sbmilburn | 11,755,869 | MDQ6VXNlcjExNzU1ODY5 | User | false |
openclaw/openclaw | 4,372,775,633 | I_kwDOQb6kR88AAAABBKNC0Q | 76,904 | https://github.com/openclaw/openclaw/issues/76904 | https://api.github.com/repos/openclaw/openclaw/issues/76904 | [Bug]: No failover to configured Microsoft (or other) tts when Minimax tts codeplan usage limit exceeds | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
When minimax code plan tts usage has exceeded it does not failback to failback what is configured.
### Steps to reproduce
Configure minimax tts as primary configure a secondary tts provider, exceed usage limit and it does... | open | null | false | 2 | [
"bug"
] | [] | 2026-05-03T19:41:55Z | 2026-05-03T23:00:47Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | najef1979-code | 233,555,716 | U_kgDODevHBA | User | false |
openclaw/openclaw | 4,372,785,052 | I_kwDOQb6kR88AAAABBKNnnA | 76,907 | https://github.com/openclaw/openclaw/issues/76907 | https://api.github.com/repos/openclaw/openclaw/issues/76907 | [Bug]: npm global upgrade leaves stale Vite chunks, gateway 500s on every HTTP request | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading OpenClaw from 2026.4.23 to 2026.5.2 via `npm install -g`, the gateway starts but every HTTP request returns Internal Server Error due to missing Vite/Rollup chunk files that the new entry points reference.
... | closed | completed | false | 1 | [
"bug",
"regression"
] | [] | 2026-05-03T19:46:05Z | 2026-05-03T19:51:25Z | 2026-05-03T19:51:25Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | EmpireCreator | 12,735,079 | MDQ6VXNlcjEyNzM1MDc5 | User | false |
openclaw/openclaw | 4,372,762,051 | I_kwDOQb6kR88AAAABBKMNww | 76,899 | https://github.com/openclaw/openclaw/issues/76899 | https://api.github.com/repos/openclaw/openclaw/issues/76899 | [Bug] Discord plugin manifest schema rejects supported `directLike` channel config | ## Bug
The Discord runtime/config code supports `channels.discord.channels.<channelId>.directLike`, but the bundled Discord plugin schema can reject that property during gateway startup.
This creates a confusing mismatch:
- `openclaw config validate` passes
- `openclaw config schema` may include `directLike`
- gatew... | closed | completed | false | 1 | [] | [] | 2026-05-03T19:35:54Z | 2026-05-03T19:55:20Z | 2026-05-03T19:54:19Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | Madder | 324,300 | MDQ6VXNlcjMyNDMwMA== | User | false |
openclaw/openclaw | 4,372,798,421 | I_kwDOQb6kR88AAAABBKOb1Q | 76,910 | https://github.com/openclaw/openclaw/issues/76910 | https://api.github.com/repos/openclaw/openclaw/issues/76910 | [memory-lancedb] CLI openclaw ltm hangs - missing apache-arrow in bundled plugin | Bug: CLI openclaw ltm hangs due to bundled LanceDB missing apache-arrow. Gateway memory system works. See: https://github.com/openclaw/openclaw/issues (tracking) | closed | completed | false | 2 | [] | [] | 2026-05-03T19:52:25Z | 2026-05-03T23:30:27Z | 2026-05-03T23:30:29Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | afiqfiles-max | 278,209,291 | U_kgDOEJUjCw | User | false |
openclaw/openclaw | 4,117,639,453 | I_kwDOQb6kR871bjEd | 52,558 | https://github.com/openclaw/openclaw/issues/52558 | https://api.github.com/repos/openclaw/openclaw/issues/52558 | pnpm check fails on main with tsgo errors in irc/mattermost/msteams/nextcloud-talk | ## Summary
`pnpm check` currently fails on `origin/main` because `tsgo` reports TypeScript errors in files outside PR #25260's docs-only diff.
## Reproduction
```bash
pnpm check
```
Observed on `origin/main` at:
- commit: `645c9210b31b8ca7d7b8b56a33912b8174475652`
- subject: `style(format): fix provider test forma... | closed | completed | false | 3 | [
"maintainer"
] | [] | 2026-03-23T01:28:46Z | 2026-05-03T19:58:11Z | 2026-05-03T19:58:11Z | CONTRIBUTOR | null | 20260503T210034Z | 2026-05-03T21:00:34Z | jalehman | 550,978 | MDQ6VXNlcjU1MDk3OA== | User | false |
openclaw/openclaw | 4,372,829,643 | I_kwDOQb6kR88AAAABBKQVyw | 76,915 | https://github.com/openclaw/openclaw/issues/76915 | https://api.github.com/repos/openclaw/openclaw/issues/76915 | [Bug] Assistant final text reply not delivered to messaging channel after message tool dispatch | # [Bug] Assistant final text reply not delivered to messaging channel after message tool dispatch
## Summary
When a turn includes a `message` tool call that successfully delivers content to a messaging channel (e.g. Telegram), any subsequent assistant text reply appears only in the web transcript but is **not deliver... | closed | completed | false | 4 | [] | [] | 2026-05-03T20:06:07Z | 2026-05-04T23:23:23Z | 2026-05-04T01:09:04Z | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | rollingshmily | 62,282,098 | MDQ6VXNlcjYyMjgyMDk4 | User | false |
openclaw/openclaw | 4,372,837,807 | I_kwDOQb6kR88AAAABBKQ1rw | 76,916 | https://github.com/openclaw/openclaw/issues/76916 | https://api.github.com/repos/openclaw/openclaw/issues/76916 | Dreams: add review-queue mode so Deep promotion doesn't autonomously rewrite MEMORY.md | ### Summary
Add a review-queue mode for Dreams Deep-phase promotion so deployments with strong "no autonomous edits to curated memory files" policies can run Dreams meaningfully without Deep autonomously writing to `MEMORY.md`.
### Context
`memory-core`'s Dreaming feature has a great staged design — Light / Deep / R... | open | null | false | 1 | [] | [] | 2026-05-03T20:09:32Z | 2026-05-03T21:50:29Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | dzalexin | 48,893,552 | MDQ6VXNlcjQ4ODkzNTUy | User | false |
openclaw/openclaw | 4,372,663,307 | I_kwDOQb6kR88AAAABBKGMCw | 76,886 | https://github.com/openclaw/openclaw/issues/76886 | https://api.github.com/repos/openclaw/openclaw/issues/76886 | BUG: Codex app-server mirrored final replies can be lost from the active session transcript branch | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When an agent runs through the Codex app-server / CLI-style provider path, final assistant replies are delivered to chat and briefly appended to the session transcript by the Codex transcript mirror. However, t... | closed | completed | false | 2 | [
"bug",
"bug:behavior"
] | [] | 2026-05-03T18:50:58Z | 2026-05-04T22:18:03Z | 2026-05-04T22:18:03Z | NONE | null | 20260505T000034Z | 2026-05-05T00:00:34Z | kAIborg24 | 261,855,322 | U_kgDOD5uYWg | User | false |
openclaw/openclaw | 4,372,883,302 | I_kwDOQb6kR88AAAABBKTnZg | 76,924 | https://github.com/openclaw/openclaw/issues/76924 | https://api.github.com/repos/openclaw/openclaw/issues/76924 | [Bug]: Agent silently freezes for hours when Anthropic API returns "Output blocked by content filtering policy" | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
When the Anthropic API returns an invalid_request_error with message "Output blocked by content filtering policy" on a response containing a tool call, OpenClaw enters a silent retry loop. Neither the agent nor... | closed | completed | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-05-03T20:26:00Z | 2026-05-03T20:30:57Z | 2026-05-03T20:30:57Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | PurpleMarsupial | 254,790,473 | U_kgDODy_LSQ | User | false |
openclaw/openclaw | 4,372,889,018 | I_kwDOQb6kR88AAAABBKT9ug | 76,925 | https://github.com/openclaw/openclaw/issues/76925 | https://api.github.com/repos/openclaw/openclaw/issues/76925 | [Feature]: Refactor xurl/skill-creator/gh-issues for skill-quality framework | ### Summary
Tighten frontmatter descriptions and factor oversized `SKILL.md` bodies in three stock skills (`xurl`, `skill-creator`, `gh-issues`) so they pass a [Tessl-framework](https://tessl.io/blog/common-pitfalls-of-skills-development-and-how-to-fix-them/) skill-quality audit.
### Problem to solve
I run a periodi... | open | null | false | 1 | [] | [] | 2026-05-03T20:28:32Z | 2026-05-04T20:34:14Z | null | NONE | null | 20260504T210031Z | 2026-05-04T21:00:31Z | jaikharbanda | 203,336,087 | U_kgDODB6plw | User | false |
openclaw/openclaw | 4,372,640,352 | I_kwDOQb6kR88AAAABBKEyYA | 76,884 | https://github.com/openclaw/openclaw/issues/76884 | https://api.github.com/repos/openclaw/openclaw/issues/76884 | [Bug]: OpenClaw on native Windows getting notably slower and slower with each new version??? | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrades through v2026.4.24 up to the most recent version, I've had error with openclaw like "...on windows absolute path required..." and significant slowdown in communication and response time with the agent. asked t... | open | null | false | 2 | [
"bug"
] | [] | 2026-05-03T18:41:10Z | 2026-05-04T09:33:11Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | wingraver | 45,340,255 | MDQ6VXNlcjQ1MzQwMjU1 | User | false |
openclaw/openclaw | 4,372,802,488 | I_kwDOQb6kR88AAAABBKOruA | 76,912 | https://github.com/openclaw/openclaw/issues/76912 | https://api.github.com/repos/openclaw/openclaw/issues/76912 | openclaw status crashes on 2026.4.21 due to missing Feishu dependency, while Discord channels repeatedly restart with stale-socket | ## Summary
On macOS, `openclaw@2026.4.21` appears to have two related operator-facing problems:
1. `openclaw status` crashes at startup because the Feishu extension tries to load `@larksuiteoapi/node-sdk`, which is missing from the installed package.
2. Discord channels remain partially functional but repeatedly rest... | closed | completed | false | 2 | [] | [] | 2026-05-03T19:54:16Z | 2026-05-03T20:42:49Z | 2026-05-03T20:42:49Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | papiofficial | 72,460,442 | MDQ6VXNlcjcyNDYwNDQy | User | false |
openclaw/openclaw | 4,372,774,782 | I_kwDOQb6kR88AAAABBKM_fg | 76,903 | https://github.com/openclaw/openclaw/issues/76903 | https://api.github.com/repos/openclaw/openclaw/issues/76903 | [Bug]: Discord package using wrong structure | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
In version 2026.5.2, attempting to install the Discord plugin via `openclaw plugins install discord --force` errors out with compaints of an invalid / unsupported package structure:
```
Extracting /tmp/openclaw-hook-pack-gM... | closed | completed | false | 3 | [
"bug",
"regression"
] | [] | 2026-05-03T19:41:30Z | 2026-05-03T20:43:58Z | 2026-05-03T20:43:58Z | NONE | null | 20260503T210034Z | 2026-05-03T21:00:34Z | jonhartman | 39,955,044 | MDQ6VXNlcjM5OTU1MDQ0 | User | false |
openclaw/openclaw | 4,372,926,733 | I_kwDOQb6kR88AAAABBKWRDQ | 76,931 | https://github.com/openclaw/openclaw/issues/76931 | https://api.github.com/repos/openclaw/openclaw/issues/76931 | sessions.list takes 16-21s on ~160 sessions; row enrichment appears to dominate more than transcript title reads | ## Summary
`sessions.list` is taking ~16–21s on a real multi-agent setup with only ~160 returned sessions, which is slow enough to break the default gateway timeout budget and make the app feel unhealthy.
This does **not** look like pure transport latency. Gateway logs show the backend method itself is taking that lon... | closed | completed | false | 2 | [] | [] | 2026-05-03T20:43:58Z | 2026-05-04T01:53:33Z | 2026-05-04T01:53:33Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | leonlushgit | 256,653,063 | U_kgDOD0w3Bw | User | false |
openclaw/openclaw | 4,372,946,121 | I_kwDOQb6kR88AAAABBKXcyQ | 76,932 | https://github.com/openclaw/openclaw/issues/76932 | https://api.github.com/repos/openclaw/openclaw/issues/76932 | BUG: Tool-result guard counts hidden details and triggers premature overflow-retry compaction | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
In long-running embedded Pi sessions, the tool-result context guard can trigger `overflow-retry` auto-compaction well before the displayed effective context budget is full. The root cause is that the guard esti... | closed | not_planned | false | 1 | [
"bug",
"bug:behavior"
] | [] | 2026-05-03T20:51:25Z | 2026-05-03T21:05:02Z | 2026-05-03T21:05:02Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | kAIborg24 | 261,855,322 | U_kgDOD5uYWg | User | false |
openclaw/openclaw | 4,372,950,711 | I_kwDOQb6kR88AAAABBKXutw | 76,933 | https://github.com/openclaw/openclaw/issues/76933 | https://api.github.com/repos/openclaw/openclaw/issues/76933 | [Discussion]: Memory subsystem stability — proposing a slower beta soak for breaking changes | ### Context
Thanks for OpenClaw, and the maintainer responsiveness — #76778 was fixed in 4 hours after I filed today, which is rare and appreciated.
Sharing some user-side observations on the memory subsystem.
### What I noticed
Six active-memory bugs were filed AND closed on a single day (2026-05-03): #76343, #765... | open | null | false | 1 | [] | [] | 2026-05-03T20:52:58Z | 2026-05-03T22:24:27Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | LaFleurAdvertising | 14,811,991 | MDQ6VXNlcjE0ODExOTkx | User | false |
openclaw/openclaw | 4,372,965,941 | I_kwDOQb6kR88AAAABBKYqNQ | 76,935 | https://github.com/openclaw/openclaw/issues/76935 | https://api.github.com/repos/openclaw/openclaw/issues/76935 | QQ Bot sends verbose/repetitive replies after upgrading to 2026.5.2 | ## Bug Description
After upgrading from OpenClaw 2026.4.29 to 2026.5.2, the QQ Bot channel plugin sends multiple messages per user input and frequently repeats content from previous replies. The same LLM model (deepseek-v4-flash) produced concise, single-message replies on 2026.4.29, so this is not a model behavior ch... | open | null | false | 2 | [] | [] | 2026-05-03T20:59:04Z | 2026-05-03T21:17:36Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | xiongpeng1209 | 33,438,788 | MDQ6VXNlcjMzNDM4Nzg4 | User | false |
openclaw/openclaw | 4,372,981,193 | I_kwDOQb6kR88AAAABBKZlyQ | 76,938 | https://github.com/openclaw/openclaw/issues/76938 | https://api.github.com/repos/openclaw/openclaw/issues/76938 | Context compaction ineffective — agent loses context and loops after truncation | ## Bug Report
### Summary
Context compaction is not working effectively. When session history grows large, OpenClaw truncates messages with a `[truncated in memory]` marker, but the agent loses critical context and starts looping, giving irrelevant answers, or repeating the same response.
### Steps to Reproduce
1. Ha... | open | null | false | 1 | [] | [] | 2026-05-03T21:04:19Z | 2026-05-03T21:08:44Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | qiblatdigital | 192,334,389 | U_kgDOC3bKNQ | User | false |
openclaw/openclaw | 4,373,002,754 | I_kwDOQb6kR88AAAABBKa6Ag | 76,944 | https://github.com/openclaw/openclaw/issues/76944 | https://api.github.com/repos/openclaw/openclaw/issues/76944 | [Bug]: HEARTBEAT_OK/silent sentinel replies are rendered in Control UI webchat | Bug type: Behavior bug (incorrect output/state without crash)
Beta release blocker: No
Summary:
Heartbeat/no-op sentinel replies such as `HEARTBEAT_OK` are displayed in Control UI/webchat instead of being suppressed as silent internal responses.
Steps to reproduce:
1. Use a WebChat direct conversation with an otherwi... | closed | completed | false | 1 | [] | [] | 2026-05-03T21:14:34Z | 2026-05-03T21:20:40Z | 2026-05-03T21:20:40Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | lily-oc | 281,468,826 | U_kgDOEMbfmg | User | false |
openclaw/openclaw | 4,372,985,548 | I_kwDOQb6kR88AAAABBKZ2zA | 76,939 | https://github.com/openclaw/openclaw/issues/76939 | https://api.github.com/repos/openclaw/openclaw/issues/76939 | Plugin recovery during `openclaw update` silently drops plugins (sequential `npm install` prunes siblings) | ## Summary
After running `openclaw update` (2026.4.29 → 2026.5.2), the post-update **plugin recovery** step silently dropped `@openclaw/diagnostics-otel` from `plugins.entries` and left it uninstalled, even though the update reported `Update Result: OK`. The gateway came back up without OTEL log export.
The root caus... | closed | completed | false | 2 | [] | [] | 2026-05-03T21:06:06Z | 2026-05-03T21:21:09Z | 2026-05-03T21:20:59Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | jtgcyber | 129,338,009 | U_kgDOB7WKmQ | User | false |
openclaw/openclaw | 4,373,052,612 | I_kwDOQb6kR88AAAABBKd8xA | 76,953 | https://github.com/openclaw/openclaw/issues/76953 | https://api.github.com/repos/openclaw/openclaw/issues/76953 | Gateway sessions.list ~1s baseline and event_loop_delay warnings on 2026.5.2 | ### Summary
After updating to OpenClaw `2026.5.2`, Gateway shows persistent `sessions.list` latency around 0.9–1.0s with occasional 3–4s spikes, plus repeated `event_loop_delay` liveness warnings. This persists after disabling heartbeats/autonomy loops and reducing stale session count.
This is happening on a resized ... | closed | not_planned | false | 1 | [] | [] | 2026-05-03T21:38:33Z | 2026-05-03T21:39:39Z | 2026-05-03T21:39:39Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | bryan493 | 235,377,663 | U_kgDODgeT_w | User | false |
openclaw/openclaw | 4,373,002,853 | I_kwDOQb6kR88AAAABBKa6ZQ | 76,946 | https://github.com/openclaw/openclaw/issues/76946 | https://api.github.com/repos/openclaw/openclaw/issues/76946 | [Bug]: Bootstrap truncation warning is injected into webchat runtime context | Bug type: Behavior bug (incorrect output/state without crash)
Beta release blocker: No
Summary:
A bootstrap truncation warning is injected into the webchat/runtime context when workspace startup files exceed bootstrap limits, which can expose internal configuration details and require manual rereads to recover importa... | closed | completed | false | 2 | [] | [] | 2026-05-03T21:14:36Z | 2026-05-04T00:34:06Z | 2026-05-04T00:34:06Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | lily-oc | 281,468,826 | U_kgDOEMbfmg | User | false |
openclaw/openclaw | 4,373,051,861 | I_kwDOQb6kR88AAAABBKd51Q | 76,952 | https://github.com/openclaw/openclaw/issues/76952 | https://api.github.com/repos/openclaw/openclaw/issues/76952 | Docs/Feature Request: Explain Realtime Talk voices, voice-agent role, and mobile/phone bridge options | ## Summary
Now that Realtime Talk is working, I love the feature. It feels genuinely useful and low-latency. A few user-facing docs and feature improvements would make it easier to understand and more practical while away from the desktop.
This is related to, but separate from, the OpenAI Platform API credit / Codex ... | open | null | false | 2 | [] | [] | 2026-05-03T21:38:09Z | 2026-05-05T11:17:45Z | null | NONE | null | 20260505T120033Z | 2026-05-05T12:00:33Z | Hal2600 | 269,776,802 | U_kgDOEBR3og | User | false |
openclaw/openclaw | 4,373,002,812 | I_kwDOQb6kR88AAAABBKa6PA | 76,945 | https://github.com/openclaw/openclaw/issues/76945 | https://api.github.com/repos/openclaw/openclaw/issues/76945 | [Bug]: Cron no-delivery jobs show confusing delivery preview/status output | Bug type: Behavior bug (incorrect output/state without crash)
Beta release blocker: No
Summary:
Cron jobs configured with `delivery.mode: none` still show confusing delivery preview/status output such as `none -> last` and `will fail-closed`, and status also shows old cron sessions alongside the current isolated cron ... | closed | completed | false | 2 | [] | [] | 2026-05-03T21:14:35Z | 2026-05-03T21:50:12Z | 2026-05-03T21:50:12Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | lily-oc | 281,468,826 | U_kgDOEMbfmg | User | false |
openclaw/openclaw | 4,372,981,028 | I_kwDOQb6kR88AAAABBKZlJA | 76,937 | https://github.com/openclaw/openclaw/issues/76937 | https://api.github.com/repos/openclaw/openclaw/issues/76937 | Feature Request: Chat rename + project folders in webchat UI | ## Feature Request
### 1. Chat/Session Rename
Ability to give custom names to conversations in the webchat UI. Currently all chats are unnamed, making it hard to identify them later.
### 2. Project Folders
Group related chats under a project or folder. When working on multiple projects, all chats look the same and it... | closed | not_planned | false | 1 | [] | [] | 2026-05-03T21:04:15Z | 2026-05-04T01:09:09Z | 2026-05-04T01:09:09Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | iahmadov86-spec | 272,632,775 | U_kgDOEEALxw | User | false |
openclaw/openclaw | 4,373,014,145 | I_kwDOQb6kR88AAAABBKbmgQ | 76,947 | https://github.com/openclaw/openclaw/issues/76947 | https://api.github.com/repos/openclaw/openclaw/issues/76947 | [Bug]: file: syntax for env var references in openclaw.json not resolved | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
The `file:` syntax for referencing external files in the `env` section of `openclaw.json` is not resolved by the gateway. Instead, the literal string (e.g. `file:secrets/xai-api-key.txt`) is passed directly to the provider a... | open | null | false | 1 | [
"bug",
"regression"
] | [] | 2026-05-03T21:20:03Z | 2026-05-03T21:50:39Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | EmpireCreator | 12,735,079 | MDQ6VXNlcjEyNzM1MDc5 | User | false |
openclaw/openclaw | 4,373,063,460 | I_kwDOQb6kR88AAAABBKenJA | 76,955 | https://github.com/openclaw/openclaw/issues/76955 | https://api.github.com/repos/openclaw/openclaw/issues/76955 | [Bug]: Message handlers block indefinitely due to fire-and-forget pattern + lane queue deadlock | ## Summary
OpenClaw's Discord integration (and potentially all inbound message handlers) silently deadlocks when receiving messages. The gateway accepts messages from Discord, triggers typing indicators, but never processes them. This blocks the lane queue permanently, preventing any messages from being handled until ... | closed | completed | false | 1 | [] | [] | 2026-05-03T21:44:06Z | 2026-05-03T22:04:20Z | 2026-05-03T22:04:20Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | rogerfsh | 76,795,410 | MDQ6VXNlcjc2Nzk1NDEw | User | false |
openclaw/openclaw | 4,373,101,559 | I_kwDOQb6kR88AAAABBKg79w | 76,960 | https://github.com/openclaw/openclaw/issues/76960 | https://api.github.com/repos/openclaw/openclaw/issues/76960 | [Bug]: 2026.5.2 agent web_search can bypass configured Brave and route to Gemini/stale provider metadata | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
In OpenClaw `2026.5.2`, agent `web_search` can fail to use the configured Brave provider even when:
- `tools.web.search.provider` is set to `brave`
- the Brave plugin/package is installed
- direct runtime/provider resolutio... | closed | completed | false | 1 | [] | [] | 2026-05-03T21:59:51Z | 2026-05-03T22:06:13Z | 2026-05-03T22:06:06Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | Kevin-Li-jgrc | 102,808,002 | U_kgDOBiC5wg | User | false |
openclaw/openclaw | 4,373,072,869 | I_kwDOQb6kR88AAAABBKfL5Q | 76,957 | https://github.com/openclaw/openclaw/issues/76957 | https://api.github.com/repos/openclaw/openclaw/issues/76957 | [Bug]: Control UI typed /new no longer fires command:new hooks after 37aebf612b — session-memory and custom hooks broken | # Issue Draft for openclaw/openclaw
## Title
[Bug]: Control UI typed `/new` no longer fires `command:new` hooks after 37aebf612b — session-memory and custom hooks broken
## Labels
bug, regression
## Body
### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After v2026.5.2,... | open | null | false | 3 | [] | [] | 2026-05-03T21:48:51Z | 2026-05-04T16:16:03Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | origfla | 15,446,573 | MDQ6VXNlcjE1NDQ2NTcz | User | false |
openclaw/openclaw | 4,373,122,827 | I_kwDOQb6kR88AAAABBKiPCw | 76,966 | https://github.com/openclaw/openclaw/issues/76966 | https://api.github.com/repos/openclaw/openclaw/issues/76966 | [Bug]: Control UI webchat does not stay pinned to bottom during streaming output | Bug type: Behavior bug (incorrect output/state without crash)
Beta release blocker: No
Summary:
Control UI/webchat does not consistently stay scrolled to the bottom while assistant output streams, causing the transcript to scroll upward during generation.
Steps to reproduce:
1. Use OpenClaw Control UI/webchat.
2. Sen... | open | null | false | 1 | [] | [] | 2026-05-03T22:10:00Z | 2026-05-03T22:17:29Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | lily-oc | 281,468,826 | U_kgDOEMbfmg | User | false |
openclaw/openclaw | 4,373,126,069 | I_kwDOQb6kR88AAAABBKibtQ | 76,968 | https://github.com/openclaw/openclaw/issues/76968 | https://api.github.com/repos/openclaw/openclaw/issues/76968 | Bug: orphaned user repair leaves persisted transcript out of sync | ## Summary
When a new prompt arrives while the previous turn left a trailing `user` leaf, OpenClaw merges that orphaned user content into the next prompt and moves the in-memory `SessionManager` leaf back to the parent. The file-backed transcript is not rewritten at that point.
That can leave the `.jsonl` transcript ... | open | null | false | 2 | [] | [] | 2026-05-03T22:11:37Z | 2026-05-04T19:49:37Z | null | CONTRIBUTOR | null | 20260504T210031Z | 2026-05-04T21:00:31Z | serkonyc | 6,172,067 | MDQ6VXNlcjYxNzIwNjc= | User | false |
openclaw/openclaw | 4,373,152,865 | I_kwDOQb6kR88AAAABBKkEYQ | 76,972 | https://github.com/openclaw/openclaw/issues/76972 | https://api.github.com/repos/openclaw/openclaw/issues/76972 | plugin-runtime-deps directory uses stale OpenClaw version after gateway upgrade — non-bundled plugins get version-mismatched dependency tree | After upgrading the gateway from OpenClaw 2026.4.29 → 2026.5.2, the `plugin-runtime-deps` directory inside the state-dir still contains only `openclaw-2026.4.29-<hash>/`. No 2026.5.2 directory is created automatically.
```
$ ls /home/node/.openclaw/plugin-runtime-deps/
openclaw-2026.4.29-f53b52ad6d21
```
This causes ... | closed | completed | false | 1 | [] | [] | 2026-05-03T22:25:12Z | 2026-05-03T22:32:21Z | 2026-05-03T22:32:21Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | inxaos | 10,542,202 | MDQ6VXNlcjEwNTQyMjAy | User | false |
openclaw/openclaw | 4,373,096,439 | I_kwDOQb6kR88AAAABBKgn9w | 76,959 | https://github.com/openclaw/openclaw/issues/76959 | https://api.github.com/repos/openclaw/openclaw/issues/76959 | [Bug]: 2026.5.2 external plugin sandbox ping-pong: restoring Brave can remove Feishu, restoring Feishu can break Brave | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
In `2026.5.2`, Brave was externalized and Feishu is also loaded as an external channel plugin. When both `brave` and `feishu` are configured, the external plugin npm sandbox can get into a ping-pong state: restoring one plug... | closed | completed | false | 2 | [] | [] | 2026-05-03T21:57:06Z | 2026-05-03T22:32:24Z | 2026-05-03T22:32:24Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | Kevin-Li-jgrc | 102,808,002 | U_kgDOBiC5wg | User | false |
openclaw/openclaw | 4,373,122,781 | I_kwDOQb6kR88AAAABBKiO3Q | 76,965 | https://github.com/openclaw/openclaw/issues/76965 | https://api.github.com/repos/openclaw/openclaw/issues/76965 | [Bug]: Control UI webchat sometimes displays duplicate assistant output | Bug type: Behavior bug (incorrect output/state without crash)
Beta release blocker: No
Summary:
Control UI/webchat sometimes displays duplicate assistant output for a single response.
Steps to reproduce:
1. Use OpenClaw Control UI/webchat in a direct conversation.
2. Send a message that causes the assistant to respon... | closed | completed | false | 1 | [] | [] | 2026-05-03T22:09:58Z | 2026-05-03T22:32:29Z | 2026-05-03T22:32:29Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | lily-oc | 281,468,826 | U_kgDOEMbfmg | User | false |
openclaw/openclaw | 4,373,174,791 | I_kwDOQb6kR88AAAABBKlaBw | 76,979 | https://github.com/openclaw/openclaw/issues/76979 | https://api.github.com/repos/openclaw/openclaw/issues/76979 | [Bug] CLI backend token counter / context window display is unreliable — shows wildly wrong values (often >100% of cap) on small sessions | ## Description
The displayed token / context window percentage is consistently wrong on CLI-routed sessions (e.g., `claude-cli/*` provider). Numbers stored in `sessions.json.agent:main:main.totalTokens` and shown in the TUI status bar do not match the actual context size — frequently showing >100% of the model's conte... | closed | completed | false | 1 | [] | [] | 2026-05-03T22:36:54Z | 2026-05-03T22:43:33Z | 2026-05-03T22:43:27Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | Adam-Researchh | 261,827,313 | U_kgDOD5sq8Q | User | false |
openclaw/openclaw | 4,373,177,687 | I_kwDOQb6kR88AAAABBKllVw | 76,980 | https://github.com/openclaw/openclaw/issues/76980 | https://api.github.com/repos/openclaw/openclaw/issues/76980 | [Bug]: Nextcloud Talk channel agent silently skips `message` tool, marks reply 'completed' without posting | ### Bug type
Behavior bug (agent fails to deliver reply, no error)
### Summary
In the `@openclaw/nextcloud-talk` channel, the embedded agent harness inconsistently calls the `message` tool to deliver replies back to Nextcloud Talk. When it doesn't, the agent streams generated text via the `text_end` channel — which ... | open | null | false | 1 | [] | [] | 2026-05-03T22:38:34Z | 2026-05-03T22:45:57Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | marfrit | 37,203,914 | MDQ6VXNlcjM3MjAzOTE0 | User | false |
openclaw/openclaw | 4,373,191,401 | I_kwDOQb6kR88AAAABBKma6Q | 76,985 | https://github.com/openclaw/openclaw/issues/76985 | https://api.github.com/repos/openclaw/openclaw/issues/76985 | [cron] announce-on-failure replay should dedup by intent | ## Summary
When a `cron.add` (or `cron.update`) fails validation, the rejection is logged into the gateway's announce-on-failure stream and replayed at the originally-scheduled fire time. If the operator/agent retries with a corrected payload under a new cron job ID, **the stale failure is still announced at fire time... | open | null | false | 1 | [] | [] | 2026-05-03T22:45:10Z | 2026-05-04T00:52:44Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | justinjkline | 8,153,539 | MDQ6VXNlcjgxNTM1Mzk= | User | false |
openclaw/openclaw | 4,373,108,429 | I_kwDOQb6kR88AAAABBKhWzQ | 76,962 | https://github.com/openclaw/openclaw/issues/76962 | https://api.github.com/repos/openclaw/openclaw/issues/76962 | subagent timeout leaves zombie claude -p → late output emitted directly to user transport (bypasses parent agent) | ## Environment
- OpenClaw 2026.5.2
- Gateway: local mode, systemd
- Backend: anthropic/claude-sonnet-4-6 via claude-cli OAuth, fallback openai-codex/gpt-5.4
## Bug chain (3 linked issues)
### Bug A — Subagent timeout does not kill the physical `claude -p` process
When a subagent session expires (`timeoutSeconds`), t... | open | null | false | 1 | [] | [] | 2026-05-03T22:03:15Z | 2026-05-04T00:52:42Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | anibalTNS | 277,111,588 | U_kgDOEIRjJA | User | false |
openclaw/openclaw | 4,373,068,828 | I_kwDOQb6kR88AAAABBKe8HA | 76,956 | https://github.com/openclaw/openclaw/issues/76956 | https://api.github.com/repos/openclaw/openclaw/issues/76956 | CLI cannot approve any new device pairing — device-ownership-mismatch blocks all admin tokens after upgrade | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading from 2026.4.15 to 2026.5.2 (via Watchtower auto-update), the Control UI started rejecting connections with device pairing required (requestId: <UUID>). There is no way to approve the pending pairing request v... | closed | completed | false | 3 | [
"bug",
"regression"
] | [] | 2026-05-03T21:46:50Z | 2026-05-03T22:52:57Z | 2026-05-03T22:52:57Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | neo19482 | 43,018,794 | MDQ6VXNlcjQzMDE4Nzk0 | User | false |
openclaw/openclaw | 4,373,195,590 | I_kwDOQb6kR88AAAABBKmrRg | 76,986 | https://github.com/openclaw/openclaw/issues/76986 | https://api.github.com/repos/openclaw/openclaw/issues/76986 | Resume chain broken: cliSessionId / previousSessionIds / transcriptPath remain null in agent session record | ## Summary
`~/.openclaw/agents/<name>/sessions/sessions.json` keeps `cliSessionId`, `previousSessionIds`, and `transcriptPath` as `null` indefinitely on the main agent record. As a result, when the runtime spawns a fresh `claude -p` for the same agent (e.g. after a cron callback or compaction-induced restart), it can'... | closed | completed | false | 1 | [] | [] | 2026-05-03T22:47:03Z | 2026-05-03T22:53:50Z | 2026-05-03T22:53:50Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | JackDA10 | 276,154,198 | U_kgDOEHXHVg | User | false |
openclaw/openclaw | 4,372,988,970 | I_kwDOQb6kR88AAAABBKaEKg | 76,940 | https://github.com/openclaw/openclaw/issues/76940 | https://api.github.com/repos/openclaw/openclaw/issues/76940 | Add startup-time session-size guard: auto-archive when no context engine is registered | ## Summary
When a context-engine plugin (lossless-claw, etc.) gets disabled — silently or by user error — the gateway loads existing session transcripts with their full message history. For users whose sessions have been growing under an active context engine, the resulting load cascade looks like a generic gateway st... | closed | not_planned | false | 4 | [] | [] | 2026-05-03T21:07:46Z | 2026-05-03T22:55:16Z | 2026-05-03T22:55:16Z | CONTRIBUTOR | null | 20260504T000036Z | 2026-05-04T00:00:36Z | 100yenadmin | 239,388,517 | U_kgDODkTHZQ | User | false |
openclaw/openclaw | 4,373,202,292 | I_kwDOQb6kR88AAAABBKnFdA | 76,988 | https://github.com/openclaw/openclaw/issues/76988 | https://api.github.com/repos/openclaw/openclaw/issues/76988 | [Bug] image_generate MEDIA paths not delivered to Telegram chat | ## Bug Description
image_generate tool successfully generates images and returns valid MEDIA:/path/... paths, but the generated images are NOT delivered to Telegram chats. Text responses work fine. Direct Telegram Bot API calls successfully send photos.
## Environment
- **OpenClaw**: 2026.5.2 (8b2a6e5)
- **Channel**: ... | open | null | false | 1 | [] | [] | 2026-05-03T22:50:59Z | 2026-05-04T00:52:46Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | AS76 | 22,101,267 | MDQ6VXNlcjIyMTAxMjY3 | User | false |
openclaw/openclaw | 4,373,216,784 | I_kwDOQb6kR88AAAABBKn-EA | 76,990 | https://github.com/openclaw/openclaw/issues/76990 | https://api.github.com/repos/openclaw/openclaw/issues/76990 | [Bug]: Successful assistant replies can be missing from the active transcript, causing later turns to re-answer old prompts | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
No
### Summary
A successful embedded agent run can produce visible assistant text in the run trajectory while failing to persist an equivalent visible `assistant` turn in the active session transcript branch. Later turns then... | open | null | false | 5 | [] | [] | 2026-05-03T22:58:17Z | 2026-05-04T12:27:11Z | null | CONTRIBUTOR | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Jackten | 2,895,479 | MDQ6VXNlcjI4OTU0Nzk= | User | false |
openclaw/openclaw | 4,373,183,235 | I_kwDOQb6kR88AAAABBKl7Aw | 76,981 | https://github.com/openclaw/openclaw/issues/76981 | https://api.github.com/repos/openclaw/openclaw/issues/76981 | [Feature]: Support "Core Logic + External References" Architecture to Mitigate Context Bloat in Complex Agents | ### Summary
Support "Core Logic + External References" Architecture to Mitigate Context Bloat in Complex Agents
### Problem to solve
As OpenClaw is increasingly used for complex, multi-phase workflows (e.g., long-running data pipelines, comprehensive code reviews, or automated DevOps tasks), the Agent configurat... | closed | completed | false | 1 | [
"enhancement"
] | [] | 2026-05-03T22:41:34Z | 2026-05-03T23:32:55Z | 2026-05-03T23:32:55Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | Caoxing-max | 266,450,415 | U_kgDOD-G17w | User | false |
openclaw/openclaw | 4,373,278,886 | I_kwDOQb6kR88AAAABBKrwpg | 76,997 | https://github.com/openclaw/openclaw/issues/76997 | https://api.github.com/repos/openclaw/openclaw/issues/76997 | [Bug]: pdf-tool eagerly loads pdfjs-dist on every agent turn in 2026.5.2, causing ~2.5s blocking overhead even when PDF is denied | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
<html>
<body>
<!--StartFragment--><p style="box-sizing: border-box; margin: 0.75em 0px 0px; color: rgb(60, 60, 67); font-family: Inter, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, sans-serif; font-size: ... | closed | completed | false | 1 | [
"bug",
"regression"
] | [] | 2026-05-03T23:31:37Z | 2026-05-04T01:53:33Z | 2026-05-04T01:53:33Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | jamaspoli | 259,723,170 | U_kgDOD3sPog | User | false |
openclaw/openclaw | 4,373,276,675 | I_kwDOQb6kR88AAAABBKroAw | 76,996 | https://github.com/openclaw/openclaw/issues/76996 | https://api.github.com/repos/openclaw/openclaw/issues/76996 | [Bug] Discord channel: agent generates response but never sends it via message tool — reply invisible to users | ## Bug Description
**Regression in Discord channel mode:** The agent generates a complete assistant response internally, but the final text never appears in the Discord channel. The user sees only the typing indicator (`Agent is typing...`) and token usage increasing, but no actual message is delivered.
This appears ... | closed | completed | false | 3 | [] | [] | 2026-05-03T23:30:27Z | 2026-05-04T00:37:51Z | 2026-05-04T00:37:51Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | sierralx | 305,603 | MDQ6VXNlcjMwNTYwMw== | User | false |
openclaw/openclaw | 4,373,279,960 | I_kwDOQb6kR88AAAABBKr02A | 76,998 | https://github.com/openclaw/openclaw/issues/76998 | https://api.github.com/repos/openclaw/openclaw/issues/76998 | Telegram formatter drops spacing before numbered section after bullet block | ## Summary
Telegram replies can lose the blank line between a manually-indented bullet block and the next numbered section. This makes long planning/status messages hard to scan in Telegram.
## Repro
Observed in a live Telegram message from `@chipsclawbot` on OpenClaw `2026.5.3-beta.2`.
Input shape:
```text
2. Mai... | open | null | false | 1 | [] | [] | 2026-05-03T23:32:08Z | 2026-05-03T23:37:39Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | evgyur | 120,411,712 | U_kgDOBy1WQA | User | false |
openclaw/openclaw | 4,373,290,764 | I_kwDOQb6kR88AAAABBKsfDA | 77,000 | https://github.com/openclaw/openclaw/issues/77000 | https://api.github.com/repos/openclaw/openclaw/issues/77000 | [Bug]: Telegram update_id is acked before outbound sendMessage confirms — restart-during-write loses forum-topic replies permanently | ## Summary
Telegram forum-topic messages can be permanently lost in a specific narrow window: openclaw acknowledges the inbound `update_id` to Telegram **before** the assistant's final outbound `sendMessage` has confirmed delivery. If the gateway is interrupted (watchdog restart, OOM kill, intentional `systemctl resta... | open | null | false | 2 | [] | [] | 2026-05-03T23:38:15Z | 2026-05-03T23:48:22Z | null | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | jackiedepp | 31,266,295 | MDQ6VXNlcjMxMjY2Mjk1 | User | false |
openclaw/openclaw | 4,373,169,285 | I_kwDOQb6kR88AAAABBKlEhQ | 76,977 | https://github.com/openclaw/openclaw/issues/76977 | https://api.github.com/repos/openclaw/openclaw/issues/76977 | [Bug]: MediaOffloadError: EPERM operation not permitted, fsync | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
## Bug Description
When attaching a PDF file via the Control UI chat interface on Windows, the gateway fails
with a MediaOffloadError during the fsync operation when trying to stage the attachment to disk.
## Error Message... | closed | not_planned | false | 2 | [
"bug",
"regression"
] | [] | 2026-05-03T22:33:58Z | 2026-05-03T23:50:17Z | 2026-05-03T23:50:17Z | NONE | null | 20260504T000036Z | 2026-05-04T00:00:36Z | hellnano | 7,827,206 | MDQ6VXNlcjc4MjcyMDY= | User | false |
openclaw/openclaw | 4,373,320,296 | I_kwDOQb6kR88AAAABBKuSaA | 77,002 | https://github.com/openclaw/openclaw/issues/77002 | https://api.github.com/repos/openclaw/openclaw/issues/77002 | [Bug]: Timeout | ### Bug type
Crash (process/app exits or hangs)
### Beta release blocker
No
### Summary
No matter what timeout value is set in openclaw.json the requests time out.
fi.:
"timeoutSeconds": 4200
and
"llm": {"idleTimeoutSeconds": 0 } //also tried with 4200
The gateway logs show:
**[agent/embedded] embedded run ag... | closed | completed | false | 1 | [
"bug",
"bug:crash"
] | [] | 2026-05-03T23:52:35Z | 2026-05-04T00:04:33Z | 2026-05-04T00:04:33Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | Dejavu333 | 89,163,562 | MDQ6VXNlcjg5MTYzNTYy | User | false |
openclaw/openclaw | 4,373,418,204 | I_kwDOQb6kR88AAAABBK0Q3A | 77,009 | https://github.com/openclaw/openclaw/issues/77009 | https://api.github.com/repos/openclaw/openclaw/issues/77009 | feat(bluebubbles): per-DM systemPrompt — follow-up to #60665 (#69198 only addressed groups) | ## TL;DR
Follow-up to #60665 (closed by #69198): the merged PR added per-group `systemPrompt` for BlueBubbles, but the original ask explicitly requested "channel root, per-group, **or per-DM**" support. The DM case was silently dropped. This leaves BB asymmetric with Telegram and WhatsApp — both of which support `dire... | open | null | false | 1 | [] | [] | 2026-05-04T00:37:06Z | 2026-05-04T00:43:47Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | jrex-jooni | 266,595,103 | U_kgDOD-PrHw | User | false |
openclaw/openclaw | 4,373,384,947 | I_kwDOQb6kR88AAAABBKyO8w | 77,007 | https://github.com/openclaw/openclaw/issues/77007 | https://api.github.com/repos/openclaw/openclaw/issues/77007 | [Bug]: Claude CLI no-output timeout is collapsed into generic Telegram error; operators cannot distinguish timeout from gateway/provider failure | ## Summary
When OpenClaw runs through the Claude CLI backend (`agentRuntime.id = "claude-cli"`) and a turn produces no output for the configured CLI watchdog window, Telegram receives only the generic message:
```text
Something went wrong while processing your request. Please try again, or use /new to start a fresh s... | open | null | false | 2 | [] | [] | 2026-05-04T00:23:42Z | 2026-05-04T00:58:36Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | nikolaykazakovvs-ux | 233,889,353 | U_kgDODfDeSQ | User | false |
openclaw/openclaw | 4,373,480,983 | I_kwDOQb6kR88AAAABBK4GFw | 77,016 | https://github.com/openclaw/openclaw/issues/77016 | https://api.github.com/repos/openclaw/openclaw/issues/77016 | [Bug] EPERM fsync on Windows — image_generate + Feishu picture receive both broken in 2026.5.2 | ### Bug type
Regression (worked before 2026.5.2, now fails on Windows)
### Beta release blocker
Yes — affects core media functionality
### Summary
Two distinct features fail with the same root cause on Windows after upgrading to 2026.5.2:
1. **image_generate tool** — generated images cannot be saved, throwing `EPER... | closed | not_planned | false | 1 | [] | [] | 2026-05-04T01:04:19Z | 2026-05-04T01:11:07Z | 2026-05-04T01:11:07Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | dxjmailbox-eng | 269,920,111 | U_kgDOEBanbw | User | false |
openclaw/openclaw | 4,373,517,492 | I_kwDOQb6kR88AAAABBK6UtA | 77,024 | https://github.com/openclaw/openclaw/issues/77024 | https://api.github.com/repos/openclaw/openclaw/issues/77024 | [Bug]: messages_read and conversations_list MCP tools return count only — data hidden in structuredContent | ### Bug type
Behavior bug (incorrect output/state without crash)
### Beta release blocker
Yes
### Summary
**Version:** OpenClaw 2026.5.2 (8b2a6e5)
**Description:**
The messages_read and conversations_list MCP tools return actual data inside structuredContent instead of the primary tool response content. MCP clien... | closed | completed | false | 3 | [
"bug",
"bug:behavior"
] | [] | 2026-05-04T01:19:56Z | 2026-05-04T06:40:01Z | 2026-05-04T03:26:54Z | NONE | null | 20260504T090105Z | 2026-05-04T09:01:05Z | GMDEEP | 83,973,271 | MDQ6VXNlcjgzOTczMjcx | User | false |
openclaw/openclaw | 4,373,520,954 | I_kwDOQb6kR88AAAABBK6iOg | 77,025 | https://github.com/openclaw/openclaw/issues/77025 | https://api.github.com/repos/openclaw/openclaw/issues/77025 | [Feature]: session-level chat mode for lower-cost lightweight conversations | ### Summary
Add a session-level `chat mode` as a lighter alternative to full `agent mode`.
This would let users explicitly choose between:
- `chat`: lightweight conversation
- `agent`: full OpenClaw execution behavior
### Problem to solve
For simple chat scenarios, the default full agent path can be heavier than n... | open | null | false | 1 | [
"enhancement"
] | [] | 2026-05-04T01:21:11Z | 2026-05-04T01:27:45Z | null | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | tinawenqiao | 15,771,296 | MDQ6VXNlcjE1NzcxMjk2 | User | false |
openclaw/openclaw | 4,373,534,747 | I_kwDOQb6kR88AAAABBK7YGw | 77,029 | https://github.com/openclaw/openclaw/issues/77029 | https://api.github.com/repos/openclaw/openclaw/issues/77029 | core-plugin-tools per-turn overhead from repeated plugin manifest resolution | ## Summary
`core-plugin-tools` appears to spend significant time per turn repeating plugin manifest/path-loader resolution work that should be cached at gateway boot.
On a production OpenClaw gateway with Discord + Telegram active, per-turn prep is dominated by `core-plugin-tools` even after pruning unused channel/se... | closed | not_planned | false | 1 | [] | [] | 2026-05-04T01:27:19Z | 2026-05-04T01:34:22Z | 2026-05-04T01:34:22Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | unicebondoc | 125,530,614 | U_kgDOB3tx9g | User | false |
openclaw/openclaw | 4,373,512,690 | I_kwDOQb6kR88AAAABBK6B8g | 77,022 | https://github.com/openclaw/openclaw/issues/77022 | https://api.github.com/repos/openclaw/openclaw/issues/77022 | IGNORE | IGNORE | closed | not_planned | false | 1 | [] | [] | 2026-05-04T01:17:54Z | 2026-05-04T02:02:17Z | 2026-05-04T02:02:17Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | spenceryang1996-dot | 251,943,929 | U_kgDODwRb-Q | User | false |
openclaw/openclaw | 4,373,685,458 | I_kwDOQb6kR88AAAABBLEk0g | 77,043 | https://github.com/openclaw/openclaw/issues/77043 | https://api.github.com/repos/openclaw/openclaw/issues/77043 | [Bug]: Parameter-free MCP tool schema regression still reproducible in v2026.5.2 | ### Bug type
Regression (worked before, now fails)
### Beta release blocker
No
### Summary
After upgrading to OpenClaw `v2026.5.2`, the parameter-free MCP tool schema issue from #75362 is still reproducible: `openai/gpt-5.4` rejects `flux-mcp__get_flux_instance` with `object schema missing properties`.
### Steps ... | closed | completed | false | 1 | [
"bug"
] | [] | 2026-05-04T02:26:46Z | 2026-05-04T02:35:32Z | 2026-05-04T02:35:32Z | NONE | null | 20260504T030031Z | 2026-05-04T03:00:31Z | tolkonepiu | 5,084,395 | MDQ6VXNlcjUwODQzOTU= | User | false |
openclaw/openclaw | 4,373,654,825 | I_kwDOQb6kR88AAAABBLCtKQ | 77,039 | https://github.com/openclaw/openclaw/issues/77039 | https://api.github.com/repos/openclaw/openclaw/issues/77039 | registerMemoryCorpusSupplement registrations get wiped on lazy tool-load re-register of non-activating plugins | ## Symptom
When a non-bundled OpenClaw plugin registers a `MemoryCorpusSupplement` from inside its `api.registerService({ start })` callback, the supplement initially registers successfully but is then silently wiped on a subsequent re-register triggered by lazy `memory_search` tool loading. Result: `memory_search cor... | open | null | false | 1 | [] | [] | 2026-05-04T02:15:19Z | 2026-05-04T12:27:13Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | inxaos | 10,542,202 | MDQ6VXNlcjEwNTQyMjAy | User | false |
openclaw/openclaw | 4,373,662,670 | I_kwDOQb6kR88AAAABBLDLzg | 77,040 | https://github.com/openclaw/openclaw/issues/77040 | https://api.github.com/repos/openclaw/openclaw/issues/77040 | channels capabilities --channel discord hangs and pegs openclaw-channels CPU | ## Summary
`openclaw channels capabilities --channel discord` hangs and spawns an `openclaw-channels` process pegged at ~99% CPU.
This happened during a Discord ingress diagnostic on a production gateway. The command did not return promptly, emitted plugin runtime dependency staging output, and left child processes c... | open | null | false | 1 | [] | [] | 2026-05-04T02:18:31Z | 2026-05-04T14:14:26Z | null | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | unicebondoc | 125,530,614 | U_kgDOB3tx9g | User | false |
openclaw/openclaw | 4,373,701,154 | I_kwDOQb6kR88AAAABBLFiIg | 77,044 | https://github.com/openclaw/openclaw/issues/77044 | https://api.github.com/repos/openclaw/openclaw/issues/77044 | Bug: Discord plugin fails to load after update - package.json missing main/exports, extensions point to .ts source files | ## Bug Report
### Summary
After running `openclaw update`, the Discord plugin fails to load and does not appear in the channels status table. The gateway restart does not recover it.
### Environment
- OpenClaw version: 2026.5.2
- OS: Linux 6.17.0-19-generic (x64)
- Node: v22.22.0
### Steps to Reproduce
1. Run `openc... | closed | completed | false | 3 | [] | [] | 2026-05-04T02:32:54Z | 2026-05-04T11:06:03Z | 2026-05-04T06:50:44Z | NONE | null | 20260504T180039Z | 2026-05-04T18:00:39Z | Kuechenfenster | 73,640,595 | MDQ6VXNlcjczNjQwNTk1 | User | false |
openclaw/openclaw | 4,373,427,724 | I_kwDOQb6kR88AAAABBK02DA | 77,011 | https://github.com/openclaw/openclaw/issues/77011 | https://api.github.com/repos/openclaw/openclaw/issues/77011 | Gateway update.run can leave half-installed package, killing live session transcripts | ## Summary
A botched `update.run` (npm upgrade triggered via the gateway) left
the package half-installed. The post-restart maintenance step then
tried to load a module that no longer existed at the path the new
build expected, crashed, and a live Telegram session lost its
`<sessionId>.jsonl` transcript. Every subseque... | closed | completed | false | 3 | [] | [] | 2026-05-04T00:41:22Z | 2026-05-04T05:26:05Z | 2026-05-04T05:26:05Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | ChattanoogaDan | 163,780,712 | U_kgDOCcMYaA | User | false |
openclaw/openclaw | 4,373,716,530 | I_kwDOQb6kR88AAAABBLGeMg | 77,047 | https://github.com/openclaw/openclaw/issues/77047 | https://api.github.com/repos/openclaw/openclaw/issues/77047 | DeepSeek V4: xhigh/max thinking levels not exposed in model profile | ## Issue
DeepSeek V4 Pro supports `reasoning_effort: "max"` natively, but OpenClaw 2026.5.2 only exposes the basic five thinking levels (`off/minimal/low/medium/high`) for the model. The extended levels `xhigh` and `max` are not available.
According to the docs at `docs/tools/thinking.md`:
> DeepSeek V4 models expose... | closed | completed | false | 1 | [] | [] | 2026-05-04T02:39:45Z | 2026-05-04T04:07:53Z | 2026-05-04T04:07:53Z | NONE | null | 20260504T060031Z | 2026-05-04T06:00:31Z | youjie | 499,827 | MDQ6VXNlcjQ5OTgyNw== | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.